Professional Documents
Culture Documents
Power Station Maintenance PDF
Power Station Maintenance PDF
Power Station Maintenance PDF
2
8.17 xlO
+04
2.58 xlO
+04
3.96 xl O
403
1. 24x10
05
8.33 x10
+0
5. 36xlO
+
4
4.75 xlO
+01
5. 40xlO
t03
2.44 xl O
1. 51xl O
+0
1.05 xlO
+04
1.38x10-'
2.72 xlO*
04
4.54x10
Current
risk
ranking
11
10
2
5
9
1
13
3
12
8
6
15
7
16
4
14
"Predicted lives are total, from installation.
+
Present service life, t
s
, for original components is 220,000 hours.
*The pipe section welded to T-piece 2, leg k, was replaced after 170,000 hours.
Table 2. Overall cumulative failure probability for each piping circuit
Piping
Circuit
Line 1
Line 2
Line 3
Line 4
Line 5
Line 6
Overall failure probability at t
s
+
+ x years
x = 0
220000 h
1.49 x l O-
4
9.30 x l O"
3
1. 16x lO'
2
6.35 x 10'
8
6.72 x l O"
5
6.69 x l O'
1
4
253288 h
3.30 x l O"
4
i . eox i o'
2
1. 9 6x lO'
2
1.71 x 10'
7
1. 55x lO'
4
7. 57x 10" '
8
286576 h
6.41 x ID"
4
2.50 x l O"
2
3. 03x lO'
2
3. 9 7x lO'
7
3. 14x lO'
4
8.25 x 10
4
12
319864 h
1. 13x 10
3. 64x lO'
2
4.36 x l O'
2
8.23 x 10"
7
5.71 x 10"
4
8.75 x l O'
1
+
Present service life, t
s
, for original components is 220000 hours.
Table 3. Operating conditions for the gas turbine stator vanes
Vane
Row 1
Row 2
Notional entry
temperature, C
702
652
Peak stress,
MPa
12
32
Fig. 1: Isometric of pipeline No. 6
Fig.2: Cumulative failure probability plot for pipeline No. 6
Fig.3: Cumulative failure probability plot for gas turbine stator vanes
This page intentionally left blank
C583/03172000
Cost-effective maintenance for the new millennium
M P SHIPLEY and R J BROWNE
AEA Technology pic, Abingdon, UK
ABSTRACT
As we move into the new Millennium, commercial pressures deriving from deregulation and
privatisation of the electricity supply industry are providing renewed impetus to reducing unit
price to consumers by examining and minimising all controllable costs in both the production
and supply. Reducing power generation plant maintenance expenditure offers an obvious
opportunity to compete more effectively. However, depending on the local generating
capacity/demand situation, improving plant availability may offer substantially higher
rewards. The challenge is to target and minimise maintenance expenditure in order to achieve
(only) the required level of plant availability without compromising personnel safety.
Inspection and maintenance programmes defined on the basis of a systematic risk assessment,
focus expenditure to components where it will be most effective in reducing failure outages,
thereby offering the opportunity to reduce unplanned unavailability. Planned unavailability
can also be reduced either by shortening plant overhauls as a result of risk-focused
maintenance or by using the risk assessment as the basis for extending the interval between
plant inspections. The risk profile for different inspection/maintenance strategies and the
associated costs can be compared to determine the optimum strategy for the particular
generating environment.
This paper illustrates a practical approach that has been adopted to plan plant inspection
programmes on a risk basis. Two applications are described. The first demonstrates the
approach in a situation where increased plant availability was the goal. The second describes
the use of risk-based inspection planning to provide the foundation for a successful
application to increase the interval between statutory inspections of boilers.
109
C583/031/2000 IMechE2000
1 INTRODUCTION
Maintenance programmes constitute a significant proportion of controllable operating costs.
The reasons for inspecting and maintaining components are often not soundly based or have
not been subject to review and revision. Enlightened power generators are seeking cost-
effective maintenance achieved through a non-prescriptive strategy based on risk rather than a
calendar-based approach. Accordingly, risk-based approaches to optimising maintenance
programmes are receiving attention both for new and ageing plant. These have the advantage
that maintenance expenditure is focused to plant areas where it offers the greatest benefit in
avoiding failure of pressure parts and breakdown of other equipment, thereby increasing plant
availability and reducing expected failure costs and/or maintenance costs. Furthermore,
establishing rational, effective inspection and maintenance programmes is fundamental to
gaining approval from safety regulators for extension of the interval between statutory
inspections of plant.
Risk-based planning of routine plant inspection and maintenance programmes must of
necessity address a large number and wide range of equipment items. Accordingly, a
qualitative or semi-quantitative risk ranking approach is invariably adopted for planning
routine inspection and maintenance programmes. Fully quantitative risk analysis is, however,
necessary in order to determine the optimum time to refurbish or replace critical components.
This paper describes how risk-based maintenance (RBM) planning is carried out and
illustrates the approach by way of two applications. The first demonstrates the use of
qualitative risk assessment to optimise the inspection programme for steam turbines from an
economic perspective. The second illustrates the use of RBM planning for boilers where the
primary objective is to develop a compelling safety argument that it is safe to extend the
interval between statutory inspections.
2 RISK-BASED PLANNING OF ROUTINE INSPECTION AND
MAINTENANCE PROGRAMMES
Risk integrates the likelihood and consequences of failure to provide a measure of the cost-
exposure to plant failure:
Risk = Likelihood of Failure x Consequences of Failure
Where time-dependent degradation mechanisms are operative (e.g. creep, fireside corrosion,
fatigue etc.), risk increases as the plant ages due to the rising likelihood of failure. In
qualitative analyses undertaken for optimising routine inspection and maintenance
programmes, likelihood and consequences of failure are determined for each component as
ratings using a systematic 'workbook' approach. In this process, the factors affecting
Likelihood of Failure and Consequences of Failure are assessed and scored using prescriptive
questionnaires. Both safety and financial consequences of failure may be considered
separately. These ratings are determined for each degradation mechanism and its associated
failure mode, and plotted on a Financial or Safety QRR Matrix. General experience is that
typically 80% of the risk in large industrial process plants is found to be associated with
<20% of the equipment items.
110
C583/031/2000 IMechE 2000
The RBM planning process comprises the following steps:
Step 1: Prepare the equipment inventory
The equipment inventory records relevant design and manufacturing details for
each item within the scope of the planning study, summarises the inspection
and maintenance history and records its operating conditions of temperature
pressure and environment.
Step 2: Determine the financial and safety risk ratings
For each equipment item, pertinent damage mechanisms are identified. The
Likelihood of Failure and Consequences of Failure Ratings are then determined
using workbook questionnaires and key quantitative inputs, and the Financial
and Safety Risk Categories determined.
This process is initially carried out in the context of the current
maintenance/inspection regime (practices and periodicity) and repeated for the
increased periodicity, if required.
Step 3: Determine the risk-based maintenance/inspection programme
Inspection and maintenance plans are formulated on the basis of the risk
ratings. Reduction in the extent of intervention is explored for the many low
risk items and the inspection/maintenance programme for the higher risk items
is optimised. The QRR is revisited and revised ratings determined, as
appropriate.
The Financial and Safety QRR matrices are four by four grids with each cell assigned a Risk
Category (Figures 1 and 2). Risk is higher towards the top right hand corner of the matrix and
lower towards the bottom left hand corner.
The following actions are required depending on the Risk Category:
Risk Category
Unacceptable
Undesirable
Acceptable With
Controls
Acceptable
Required Actions
Mitigate risk to an 'acceptable' level by defining and implementing
one or more (as necessary) of the following:
(i) improved inspection or maintenance procedures,
(ii) improved operating practices or controls,
(iii) on-line plant monitoring,
(iv) engineering measures to mitigate consequences.
Mitigate as above within the timescale of the next overhaul.
Define and implement an appropriate revised inspection or
maintenance strategy to support QRR judgements.
No inspection, maintenance or other actions required unless to satisfy
national legislative requirements.
C583/031/2000 IMechE 2000
111
Risk-based inspection and maintenance planning takes into account the likely location and
rate of damage accumulation. These factors determine the sites and features on the
component to be inspected, the sample size, the procedure to be employed and the frequency,
or the maintenance action required. If the potential rate of degradation is such that inspection
or maintenance during plant shutdowns would be too infrequent reliably to detect impending
failure, on-line monitoring is recommended. If the likelihood of failure is high because of the
possibility of mal-operation or beyond design operation faults, operating practices, plant
monitoring and management procedures are reviewed to determine scope for improved
control. Where the risk is driven by the consequences of failure, the scope for reducing this
by improving event termination time, limiting the steam release and/or secondary mechanical
damage to plant, and engineering modifications to provide improved personnel protection, is
considered.
3 RISK BASED PLANNING OF TURBINE INSPECTION PROGRAMME
The following case study illustrates risk-based inspection planning for four ageing steam
turbines required to operate for a further 20 years. This study addressed financial
considerations, not safety. For brevity, the study is illustrated here using only the LP Rotor.
3.1 LP rotor damage mechanisms
The following potential in-service degradation mechanisms were identified:
Position
Centre line or bore.
Blade root fixing.
Disc bore and keyways.
Shaft surface at section
changes and keyways.
Blades
Bearing journals.
Coupling flanges.
Damage
Thermal fatigue crack growth
from manufacturing defects.
Brittle fracture.
Stress corrosion cracking
Stress corrosion cracking
Low cycle fatigue and high
cycle corrosion-fatigue.
High cycle fatigue
Erosion
Fretting fatigue.
Fretting fatigue.
Crack Orientation
Axial-radial.
-
Axial-radial.
Circumferential-radial.
-
-
-
-
3.2 Risk-based inspection planning
Qualitative risk ranking (QRR) was carried out for each location and each identified
degradation mechanism. For brevity, only stress corrosion cracking (SCC) at the LP rotor
disc bores and keyways is considered here.
The Likelihood of Failure was rated at 'Medium' primarily due to the 'poorly effective'
inspection programme that comprised only manual pulse-echo ultrasonic testing during
overhauls. This procedure cannot effectively interrogate the disc bores and keyways at the
centre of the discs. Furthermore, defect sizing accuracy is limited to 6mm. In view of the
potentially small critical size for fast fracture of defects at these locations, this is
unsatisfactory.
112
C583/031/2000 IMechE 2000
The Consequences of Failure were rated at 'Critical' because of the possibility of fast fracture
and break-up of the turbine. The financial risk was thus assessed as 'Undesirable'.
An improved inspection procedure was agreed to reduce the risk. This comprised automated
imaging UT using both pulse-echo and time-of-flight-diffraction (TOFD) techniques. The
automated testing recommended deploys probes mounted on the face/shoulder of the disc
using an indexing system, with the scans effected by rotating the rotor such that the disc bore
and keyway are inspected effectively in both the face and centre regions of the disc. Pulse-
echo testing is deployed for defect detection since this can be more effective than TOFD for
stress corrosion cracks where the morphology is highly branched. The ability to image in
various modes (A-scan, B-scan etc.) reduces the required minimum signal to noise ratio for
reliable detection of defects, thereby allowing defects of 1mm radial height to be identified.
Sizing of defects is undertaken using both imaged pulse-echo testing and TOFD. The
combination of these techniques typically provides an accuracy of 1mm for determining the
radial height of defects.
3.3 QRR matrix for the LP turbine rotor
Figure 3 is the financial risk matrix for the LP turbine rotor; the matrix identifies the Risk
Category for each in-service degradation mechanism, where appropriate amended in the light
of the revised inspection strategy.
3.4 Summary
The risk-based review of the inspection programme for the steam turbines revealed a number
of high-risk integrity issues. These were addressed by identifying revised inspection
recommendations to reduce the risk of failure during the intended future extended service to
acceptable levels. Thus benefit was derived from reduced cost exposure to failure during the
20 year life extension period proposed. These benefits significantly outweighed the
additional costs associated with the more rigorous inspection and plant life management
programmes recommended.
4 RISK BASED PLANNING TO EXTEND OF BOILER INSPECTION
INTERVALS
The authors have carried out RBM planning to extend the statutory inspection interval on
more than 15 boilers. The safety risk assessment for each boiler was in all cases used as the
primary factor in the application for the extended interval, although this was strongly
supported by satisfactory generic precedent.
4.1 Safety Risk Assessment (SRA) procedure
(i) Study Team
The safety risk assessment is undertaken by a Study Team of engineers who, between
them, offer a good understanding of in-service degradation mechanisms for boiler
pressure parts and specific knowledge of the particular boiler design and its history of
integrity problems.
C583/031/2000 IMechE2000
113
(ii) Damage Mechanisms
Each component is assessed in turn. Pertinent in-service degradation mechanisms are
first identified by considering the material(s) of construction and the full range of
conditions that may be experienced during service. Thus, in addition to normal
operating conditions, fault and off-load conditions are considered.
(iii) Procedure Overview
Each component is assessed separately in the context of each in-service degradation
mechanism, according to the scheme set out in Figure 4.
(iv) Assigning the Safety Severity Rating
Each component degradation mechanism is discussed to determine the failure mode.
The Safety Severity Rating is assigned in accordance with the following prescribed
classification system:
Insignificant: No threat to personnel safety
can be envisaged
Minor: Unlikely safety threat to personnel
in vicinity of failure
Local: Safety threat to personnel in
vicinity of failure
Widespread: Widespread safety threat to
plant personnel
In order to assign the Safety Severity Rating, the mode of failure must first be
identified by considering how damage would accumulate as a result of the degradation
mechanism being assessed.
The possibilities are:
1 . leak from a pin hole,
ii. leak from a stable crack,
iii. failure - either by plastic collapse or brittle fracture.
The likely steam release volume and release rate is then considered in the context of
the location of the component, the extent of containment and the likely proximity of
personnel. Based on these considerations, experienced boiler engineers have little
difficulty in categorising safety severity according to the scheme prescribed above.
If the Safety Severity Rating is assessed to be 'Insignificant' for all mechanisms, the
Safety Risk Category is 'Acceptable' regardless of the likelihood of failure (see Figure
2, Safety Risk Matrix). Since the interval between inspections of the boiler has no
1 1 4 C583/031/2000 IMechE 2000
significance for the severity of failure, the Safety Risk Category would not change if
the interval were extended and the component need not be considered further in this
regard.
(v) Assigning the Likelihood of Failure Rating
Where the safety severity for any mechanism is 'Minor', 'Local' or 'Widespread', the
Likelihood of Failure Rating is assessed. This is determined for the time of the next
but one inspection outage using projections of additional operating hours and numbers
of starts based on past experience.
(vi) Current Safety Risk Category
The Current Safety Risk Category for the component is determined using the Safety
Risk Matrix by recording the position of each pertinent damage mechanism based on
the Current Likelihood of Failure Rating and the Safety Severity Rating.
(vii) Revised Safety Risk Category
The Likelihood of Failure Rating is then revisited in the context of the extended
inspection interval required by reconsidering and recalculating (as appropriate) the
inherent likelihood of failure, and re-assessing the 'Inspection Effectiveness' factor in
the Workbook.
If the Revised Safety Risk Category for all the in-service degradation mechanisms
identified for the component is 'Acceptable', there is no need for additional inspection
or other controls in order to justify the extended inspection interval proposed. If the
component is currently scheduled to be inspected, the need for this and the extent of
inspection should be considered (see below). However, any inspection and monitoring
required in order to assess the likelihood of failure must be continued.
Where the Revised Safety Risk Category is 'Acceptable with Controls', it is necessary
to support the judgements made in determining the Likelihood of Failure Rating
associated with the extended inspection interval. This support may derive from
refined life assessment, inspections, on-line monitoring or other controls. The
alternatives are considered by the Study Team to identify the optimum programme to
provide support for the assessment result.
(viii) Risk Mitigation
Where the Revised Safety Risk Category is 'Undesirable' or 'Unacceptable',
inspections and other controls are considered by the Study Team to identify the
optimum risk mitigation measures in order to justify the extended inspection interval.
The starting point for this discussion is any existing inspection programme, life
monitoring system or other controls. Where the Likelihood of Failure has been
determined using an input from a remaining life calculation, either off-line or on-line
using a software system, the opportunity to refine the calculation is considered. This
may involve installing additional thermocouples or other monitoring sensors.
C583/031/2000 IMechE2000
115
The Study Team considers all inspection, re-assessment, on-line monitoring and
engineering modification/refurbishment options and re-characterises the Likelihood of
Failure Rating for each option on the basis of shared judgement. The optimum
solution to reduce the Revised Safety Risk Category to 'Acceptable' or 'Acceptable
with Controls' is agreed and the 'Controlled Safety Risk Category' is recorded on the
Safety Risk Matrix.
4.2 Case study
The SRA procedure is illustrated here in the context of an extension of a boiler statutory
inspection interval from two to three years. For brevity, the illustration considers only creep
failure of a secondary superheater outlet header SSOH in the vicinity of the tube penetrations.
(i) Safety Severity
If creep damage were to lead "to failure of the tube ligament region of the header,
failure could occur by bursting, with the formation of an extended axial split and a
sudden massive and violent release of steam and energy. Although the header is
located in an enclosed vestibule, this could not be relied upon to contain the steam
release into the boiler house. Accordingly, the Safety Severity was rated as
'Widespread' (Figure 4).
(ii) Likelihood of Failure
Inverse-code calculations of creep life consumed were carried out. The results of the
calculations are summarised below:
Assessment Temp. (C)
Value
532
Description
Steam
Operating
Material Rupture
Data
Min. l%Cr-'/
2
%Mo
Region
Ligament
Life Fraction Consumed
(%)
2 year
Interval
49
3 year
Interval
57
Accordingly, the inherent likelihood of failure was assessed to be low for the operation
with the current inspection interval and medium for the extended inspection interval.
Other factors in the workbook questionnaire were addressed qualitatively and scored
to arrive at the following overall Likelihood of Failure rating (Figure 5):
Inspection Interval
Current (2 years)
Revised (3 years)
Likelihood Rating
Negligible
Low
(iii) Safety Risk
Current Inspection Interval (2 years): with reference to the Safety Risk Matrix, the
combination of 'Widespread' Safety Severity and 'Negligible' Likelihood of Failure
resulted in a Current Safety Risk Category of 'Acceptable with Controls'.
116 C583/031/2000 IMechE 2000
Extended Inspection Interval (3 years): the combination of 'Widespread' Safety
Severity and 'Low' Likelihood of Failure resulted in a Current Safety Risk Category
of 'Undesirable' (Figure 6).
(iv) Risk Mitigation
Extension of the inspection interval from two to three years was assessed to increase
the Safety Risk Category from 'Acceptable with Controls' to 'Undesirable' (Figure 6).
It was necessary, therefore, to introduce risk mitigation measures at or before the next
but one inspection outage.
In recognition of this situation, the Study Team recommended that thermocouples
should be attached to the surface of the header during the next outage in order to
record metal temperatures directly (Figure 7). This would allow the creep life
calculation (previously based on steam temperatures with an arbitrary 20 C uplift) to
be refined. If this re-analysis fails to reduce assessed the safety risk to an acceptable
level, more extensive inspections would be carried out at the following outage,
including metallographic replication to investigate the presence of creep cavitation in
the ligament region of the headers. These measures reduced the Safety Risk Rating
for the three-year inspection interval to 'Acceptable with Controls' (Figure 8).
4.3 Summary
The safety risk assessment procedure described above provides a structured, systematic
approach for assessing the risk of failure of each pressure part of a boiler. In particular, it
allows the safety implications of extending the interval between inspections to be assessed.
The combination of experience from precedent elsewhere with the confidence provided by a
formal safety risk assessment, is argued to provide a powerful case for extending inspection
intervals for water tube boilers in power generation service.
5 CONCLUDING REMARKS
This paper describes two applications of risk-based maintenance planning. The first
demonstrates the application of risk assessment to optimise the inspection programme for a
steam turbine. In this example, the assessment revealed that the cost-exposure to failure was
undesirably high and an improved inspection programme was recommended to address this
situation. The second application illustrates the use of RBM planning to provide a sound
basis for extending the statutory inspection interval for steam-raising boilers. In both cases,
the risk assessment methodology was necessarily qualitative or, at best, semi-quantitative.
Rigorous quantitative risk assessment has also been used to determine the optimum time to
replace ageing boiler components but this has not been illustrated in this paper.
Risk-Based Maintenance and Inspection Planning Software (RBMS) includes both safety and
financial risk assessment options. RBMS is a Microsoft Access application that has been
customised for the purpose of carrying out risk-based maintenance and inspection planning
for both safety and financial optimisation. It greatly facilitates the Study Team's
implementation of the safety risk assessment (SRA) procedure described in the foregoing and
provides an auditable record of the study. RBMS is supplied as an executable file so that the
C583/031/2000 IMechE2000
117
SRA can be updated when it is reviewed at each inspection outage, as required in Safety
Cases and Written Schemes of Examination (WSEs) for extended inspection intervals.
Figure 1 Safety and financial risk matrices
118
C583/031/2000 IMechE 2000
Figure 2 Financial risk matrix for LP turbine rotor
C583/031/2000 IMechE2000
119
Figure 3 Safety Risk Assessment procedure
Notes: ' The term 'Inspection action' is used as shorthand here for the range of possible
risk mitigation measures which may be applied. In addition to inspection, these
measures include refined life assessment and installation of an on-line
monitoring system.
The assessment considers safety issues only. Inspection actions are often
specified for financial reasons i.e. avoiding failures to maximise plant
availability and associated generation revenues/availability payments. This is
outside the scope of SRA.
120
C583/031/2000 IMechE 2000
Figure 5 Likelihood of failure rating for creep failure of secondary superheater
outlet header (RBMS screen shot)
C583/031/2000 IMechE 2000
121
Figure 4 Consequences of failure rating for creep failure of secondary
superheater outlet header (RBMS screen shot)
Figure 6 Safety Risk Assessment matrix for creep failure of secondary
superheater outlet header (RBMS screen shot)
Figure 7 Monitoring and inspection recommendations for secondary
superheater outlet header (RBMS screen shot)
C583/031/2000 IMechE 2000
122
Figure 8 Revised Safety Risk Assessment matrix for creep failure of secondary
superheater outlet header (RBMS screen shot) following introduction of
controls
C583/031/2000 IMechE2000
123
This page intentionally left blank
C583/013/2000
Industrial Trent Genset - designing for life cycle
J T BILLINGSLEY
Industrial Trent, Rolls-Royce Power Engineering pic, Coventry, UK
SYNOPSIS
The second generation Trent Generating Set has been designed to improve maintainability
and reduce cost, optimising the relationship between initial unit cost and through life cost.
This paper describes how the author led this design project, with the 'whole engine
modelling' techniques of reliability, life cycle engineering and safety management. It also
covers the approach to project management, the design process and the use of an Integrated
Product Team. Aspects of the design are described to illustrate the achieved design for life
cycle. This overall approach would be applicable to many different types of project.
1. INTRODUCTION
When the Industrial Trent was initially launched. Rolls-Royce supplied only the Trent engine
and a third party supplied the rest of the generating set (genset). Subsequently, Rolls-Royce
developed its own Trent packaging capability. Early in 1999, the 'Phase 2' project was
launched to develop a new package. The scope of this project was to design a new Trent
genset, from the filter intake to the generator terminals, including the acoustic enclosure and
all mechanical and control auxiliaries. This project excluded the engine core itself, although it
involving close working with the engine group, to optimise the design of the genset and
systems integration of the engine into the genset. The main drivers for the new design
included improved maintainability and designing for the life cycle.
C583/013/2000 125
2. INTEGRATED PRODUCT TEAM
This project was one of the pilot Integrated Product Teams (IPTs) within Rolls-Royce Energy
Businesses, a new approach to improve working practices. An IPT is a multi-disciplined
team formed to deliver a defined product or process. As a team, they have shared objectives,
have to work together to achieve these objectives, and believe that if they work together they
will achieve these objectives and are rewarded and recognised for doing so.
The IPT had a clearly defined objective - to develop a standardised Trent genset. The
resources were allocated with the appropriate skills and experience. The team had control
over the budget and resources, and was accountable for delivery of the defined product.
The team was put together from various parts of Rolls-Royce pic, where the various skills and
disciplines were, including Coventry, Liverpool, Derby and Gateshead. Subsequently, when
major suppliers were selected, they too came on board, working together towards a common
objective. The staff included multi-disciplined engineers, procurement, quality and project
management personnel.
The team was mobilised over a period of a week. The purpose of the mobilisation was to get
the team working as effectively as possible as quickly as possible. The whole team were
involved for the first and last day, with a number of sub sessions during the intervening days.
The first day was an introduction to the IPT concept and the challenge of Phase 2 as a
standardised genset design. Team dynamics were also reviewed, using a couple of
instruments for team skills and conflict handling. This helped the team understand the
various team working skills, and also how they and their colleagues would work together.
Several sub groups met over the next few days looking at specific technical and non technical
aspects. The focus of these groups was goal planning, as opposed to activity or task planning.
Goal planning is the intermediate level between a high level objective and an activity based
plan. The output of these sub-sessions was a goal plan or milestone plan, and a clear view of
the role of that sub group and individual team members and their interrelation with the other
teams. These groupings then formed the basis of the work breakdown structure. The final
half day session with the whole team reviewed the feedback of all the subgroups, their
concerns and expectations, and developed an overall goal plan for the project. Following the
mobilisation, a comprehensive activity plan was developed to achieve the goals. Also,
regular meetings were held to maintain the team and its focus, for example brief 'morning
prayer' meetings and weekly review meetings.
3. DESIGN PROCESS
The project was developed with the Derwent process, which is a gated project development
process and part of the Rolls-Royce Good Practice Framework. The gated process is
applicable to all projects whether research and development or specific contracts. The
Derwent process is a specific application of the gated process to a new product development.
Each stage consists of a formal business and technical exit review, where the team gives a
formal presentation to an experienced group of internal auditors for review. Stage 1 is when a
single risk assessed preliminary product concept is agreed. Stage 2 is when enough
C583/013/2000
126
engineering work has been done to confirm the concept, and when costed resourced plans are
in place. Stage 3 is a critical design review, development review and service readiness
review. Stage 4 is an in-service review. As life cycle costs are locked in early on in a project
life cycle, the early reviews Derwent 1 and 2 look at the preliminary design before too much
detail design work has been done. Two key documents used here are the Project Policy
Document (PPD) and the Project Requirements Document (PRD). The former is a high level
business requirements document. The latter is engineering's response to this, outlining how
the targets are to be achieved.
This high level product development process was also the means to ensure that all the various
parts of the business were brought together. In parallel with the new package design that this
paper is addressing, there were engine developments and associated engine tooling changes.
A combined 'Whole Engine and Phase 2 Package' Derwent 2 exit review was held to ensure
that all these aspects were brought together, and interfaces agreed.
Within each system of the work breakdown structure, detailed design reviews were held with
representation from all parts of the IPT. As a minimum there was a design review for each of
the sub systems prior to Hazop (see later). For example for the lubricating oil system, a
design review was held when there was concept design, system description and Process and
Instrumentation Diagram (P & ID). Actions from the design review were addressed prior to
the Hazop. For more complex issues such as mechanical handling, a number of reviews were
held to address detailed aspects, for example engine removal from enclosure, combustor
removal and engine split in situ. All design reviews were led by the relevant designer and had
representation as appropriate from relevant technical experts, applications engineering,
product support, safety, reliability and maintainability, and the customer facing business unit.
4. PROJECT MANAGEMENT
The programme management branch of the work breakdown structure included the
administration of the programme and earned value, processes and quality, product cost
control, requirements and standardisation, and the project risk assessment.
A work breakdown structure was developed from the sub groups from the mobilisation, with
the work broken down into manageable chunks of work. Work packages were defined and
owners were assigned, ensuring roles and responsibilities were clear. Whilst the team was a
multi-site IPT, the core group was co-located, until the concept design was established and
the detailed work plan to completion done, i.e. the Derwent 2 exit. Quality plans were
developed to relate different working practices between the different parts of the organisation.
One work package focused on the overall product cost. A project requirement was to design
to a defined product cost. This product cost was apportioned to the work packages, and the
individual work package owners had to ensure that the costs were achieved, for example at
the design reviews.
The programme was developed from the work breakdown structure. One master programme
was used, broken down into the various work packages. An essential aspect of the breakdown
C583/013/2000 127
was to ensure that each work package was manageable. In this way, work package owners not
in the core location also had f ull visibility and control of their aspect of the plan. This also
allowed easy integration of the dif f erence work packages, with all the links within one
programme. The programme was baselined, then updated at a weekly programme meeting,
when any impacts of changes were identified and addressed. Earned value reports for this
research and development programme were produced every four weeks to monitor and
control costs.
A 'standardised' design in this context means a verified and validated design, which complies
with customer, functional and safety requirements. An approach was developed to provide a
base standard with pre-engineered options. For example, the business requirements require a
genset suitable for a very wide range of ambient temperature. This is addressed in the Phase
2 design by having a base standard, which is suitable for a temperate climate, and 'arctic' and
'desert' options, which are suitable for a shift in the ambient temperature for colder or hotter
climates. This provides a cost effective way of being able to satisfy the various customer
requirements. With the standardised design, there is minimal engineering work required for
f uture similar plant, and also full details of the design are available at the outset of the
contract.
Part of the processes developed also included a process for configuration and change control
as appropriate for a genset, incorporating the pre-engineered options, and the facility to
address f uture customer specific requirements. A good bill of materials (BOM) and
configuration control is essential for through life management. Similarly, a robust change
control process is of utmost importance. This process for the genset design is carried out at
the Product Change Board (PCB). The PCB manages change to standardised products in
service, and follows similar principles of a gated process as for the product development but
on a smaller scale, for example problem definition, concept design and plans. The PCB will
be used in the future to manage obsolescence for the package design, to ensure that suitable
spare parts are defined and available over the years of service of the plant. Technical Queries
from sites also feed into this process where in service improvements could be made by design
changes.
A formal risk assessment session was held for the project using a trained facilitator. All risks
were categorised as high, medium or low for both likelihood and consequence. Mitigations
were then developed for all the risks, and progressed via monthly risk assessment meeting,
particularly focusing on the top risks and the significant risks to mitigate them as far as
possible, then accepting residual risks if necessary.
The overall project work breakdown structure was divided into seven branches, which were
then subdivided. 'Whole engine modelling' was identif ied as a main work package in its own
right to give it focus. This covered reliability, life cycle engineering, safety, CE marking and
environmental.
Engine enclosure
Auxiliaries
Controls and Instrumentation
AC generator
Whole Engine Modelling
128
C583/013/2000
Programme Management
Contract Deliverables
5. WHOLE ENGINE MODELLING
Some of the initial deliverables from the whole engine modelling work packages were design
guidelines and strategy documents for the rest of the IPT. A Use Study detailed how Rolls-
Royce intends to support its package. The purpose of this document was to identify and
describe the support factors that influence the design of the systems and the equipment
necessary to achieve the required availability at the minimum through life cost. A ' life cycle
engineering design guidelines' covered design guidelines, for example on redundancy and
fault tolerance, maintainability criteria, such as modularisation, accessibility and handling,
and a checklist for the team. Other documents included a policy for access into Trent gensets,
a CE compliance strategy and an environmental strategy.
A life cycle cost model was used to do trade-offs for aspects of the design. Where the initial
first cost may be higher for an additional maintainable feature, the through life cost was also
taken into consideration. Where typically Rolls-Royce are supporting our customers with
maintenance contracts, through life availability and cost can be justified as a business case.
For example, the maintenance envelope at the front of the package was increased moderately
from the initial concept, with an associated cost. The trade-off considered extra manpower,
downtime and support equipment.
A questionnaire was developed for suppliers, aimed at the supplier selection stage to
determine the supplier's approach to safety and reliability data. Reviews with Product
Support and Procurement also addressed Rolls-Royce supplier's strategies with respect to
spares recommendations and / or holdings, and maintenance contract arrangement. Selection
of suppliers took into account their differing capabilities for product support.
The following definitions were used for reliability and availability.
C583/013/2000
129
Reliability, R%
Availability, A%
Where T = Equipment total installed time (hours)
S = Equipment scheduled downtime (hours)
U = Equipment unscheduled downtime (hours)
The reliability target for the genset was split between the engine and the balance of the genset
and further apportioned to the subsystems within the genset, based initially on an
approximated parts count. For each sub system reliability data was either provided by the
supplier, or a fault tree analysis was done using a propriety software package. This was done
when the first draft P & ID was available. Comments were then fed back to the relevant work
package owner with suggestions for design improvements. In this way, the reliability model
provided the information to influence the design at an early stage. The model was
subsequently revised using the final design information to verify that the requirement was
met.
Similarly an availability target was set. The availability was modelled as a genset level and
addressed scheduled and unscheduled downtime. Similarly, early consideration of a typical
maintenance schedule was fed into the design and availability.
6. SAFETY MANAGEMENT
A number of safety issues are closely related to maintainability. For example, Confined
Spaces Regulations require work in confined spaces to be avoided. Where it is not
practicable to do work without entry, then measures need to be taken, and safe systems of
work implemented.
Safety legislation applicable to gensets is covered by product safety legislation and
occupational health and safety legislation. Product legislation has changed over recent years
with the introduction of product related directives and legislation, requiring CE marking. The
Machinery Directive is particularly relevant, as is the associated transposed harmonised
standard BS EN 292 on Safety of Machinery. This standard covers generic machinery
hazards, the required strategy for selecting safety measures and risk assessment requirements.
The emphasis for the safety measures taken by the designer is on risk reduction by design,
then safeguarding, then information for use and additional precautions. In the UK, the Health
and Safety at Work Act is the main piece of legislation on occupational health and safety.
There have been several European Directives which have been implemented by this act, that
specify a risk assessment approach, for example covering manual handling and personal
protective equipment.
Risk assessment is a common theme in most of the Health and Safety Regulations and the
Machinery Directive, and hence became a key part of the new safety methodologies which
were developed for this project. Developing the new safety methodology required
understanding of this legislation, considering the design of the equipment for its end use, and
how the client and end user would operate the equipment.
Hazops were used at the primary tool for hazard identification, and a Risk Assessment
methodology, as appropriate for a genset was developed. Hazops, a hazard and operability
study, are a structured brainstorming approach to address both safety and operability issues,
and are widely used in various industries, such as chemical, water and nuclear. The Hazop
studies were done in two stages, Hazop 1 and Hazop 2. The former looked at the whole
system and generic hazards and safeguards, e.g. design standards. A multi-disciplined team,
with an appropriate knowledge and experience were assembled, representing the design,
safety, maintenance, control and instrumentation, working under a chairman and secretary.
They reviewed the design against a number of keywords. Hazop 2 studies looked at each part
of the system in more detail looking at the design intent and possible deviations. An IPT
design review was done on each of the sub systems and actions resolved prior to the Hazop.
130
C583/013/2000
By doing these Hazops at an early stage in the design development, possible hazard or
operability issues were designed out wherever practicable. Failure Modes and Effects
Analysis (FMEA) were done where appropriate for some systems. Risk assessment then
addressed the remaining risks, to ensure they were reduced to an acceptable level.
7. ENGINE ENCLOSURE
In keeping with the world wide support strategy for the Industrial Trent, there is a
requirement to be able to split the engine into its modular components within the enclosure.
This capability provides operational flexibility to either remove the engine or the modules
separately. This new concept has resulted in fundamental design changes to the package for
these options to be achieved in the most efficient and cost effective manner. The enclosure is
constructed using a modular approach, which is assembled at site. This modular approach to
the acoustic panels also enables different noise emissions standards to be achieved more
easily.
The engine or modules are exchanged by use of a rail system that runs the full length of the
enclosure, with removable rails extending under the filter house. This feature together with
the mechanical handling of the split tooling, makes use of the increase in height and width of
the enclosure, which generally improves the maintainability of the engine.
Inside the enclosure are two monorail systems, one either side, which run from the enclosure
doors up to the inter-compressor duct (ICD) plenum. These are for transporting the three way
split tooling into position. There are two manually operated X-Y hoists in an electrically
driven overhead gantry, providing f ull coverage from the ICD plenum to the rear of the
enclosure. There is an 'over the engine' climbing frame / maintenance platform which can be
positioned using the overhead X-Y hoists to provide easy access to elevated areas rear of the
ICD plenum. At approximately one-third-engine height on both sides of the enclosure, there
are maintenance walkways.
Designated maintenance shipping routes have been planned and designed for combustor and
external gearbox removal. The gas turbine exhaust has redesigned exhaust cones, which
reduce the time to disconnect the LP shaft and duct.
Outside the enclosure under the filter house, there is a covered work area with an electrically
operated X-Y hoist for the mechanical handling of engines, modules and ancillary equipment.
This negates the requirement for any mobile crane during engine changeout (Figure 1).
Throughout the package, ladders and walkways for personnel access and safety for
maintenance have been addressed.
C583/013/2000
131
8. AUXILIARIES
The auxiliaries are accommodated in a separate auxiliary module. The auxiliaries include the
lubricating oil system, hydraulic control, hydraulic start, water wash and fuel skids. A
separate control room contains all the controls and electrical equipment. This includes the
engine management system, package control system, generator control and protection panel,
UPS (uninterruptible power supply), switchboard, control auxiliaries and fire and gas panel.
The auxiliaries module is laid out for ease of access and maintenance. Lifting beams are built
in where required for removal of heavy components, for example the hydraulic start motor
and pump assembly and control oil pumps. The gas fuel skid was removed from under the
engine, where it was difficult to access, into a separate compartment in the auxiliary module.
Each of the sub systems was developed from the existing design systems which were
essentially functional. Where possible unnecessary complexity was eliminated. For example,
the hydraulic start system had a complex arrangement for an alternative flow path if the
engine automatic clutch failed to disengage. This aspect has been modified, to provide the
same effect with less equipment, which is simpler and therefore more reliable. The hydraulic
system also now has an off-line cooling and filtration loop, to keep the tank cool and clean.
This enables maintenance of low particulate contamination and enhances the life of the
hydraulic components. The hydraulic pack was from a specialist hydraulic supplier, and was
based on the successful design for the RB211 hydraulic start system.
9. CONTROL AND INSTRUMENTATION
A number of aspects of the controls and instrumentation provided improvements in
maintainability and availability. A key issue was the introduction of fieldbus technology,
specifically the package control system uses Profibus DP/PA fieldbus architecture. This
gives increased on-line diagnostics, on-line calibration / configuration and improved Mean
Time Between Failure (MTBF) performance. The lower component count improves
reliability. The package control system design also incorporates a condition monitoring
package which monitors, for example, the number of engine starts and stops and time at
various conditions.
The fire and gas protection system uses analogue addressable field wiring. This increases on-
line diagnostics, allows on-line calibration and configuration, and reduces undetected failures.
Selection of all components generally addressed maintainability, for example analogue
transmitters providing increased trending to enable advance warning of routine maintenance
requirements, and the starter motor selected uses intelligent controller which improves
diagnostics. The design reviews described above ensured good practice was employed
throughout. Also, all the controls and instrumentation suppliers were asked to provide
FMEAs to highlight any design weaknesses.
132
C583/013/2000
10. PRODUCTION
The first 'Phase 2' package will be built in the fourth quarter of 2000 at Rolls-Royce Power
Engineering, Liverpool (Figure 2). Validation of the engine tooling and various aspects of the
design will be undertaken there prior to the unit going to a customer site. The project is on
target to achieve its maintainability requirements within the product cost, within the
programme budget and to plan, largely due to the committed approach of the team.
ACKNOWLEDGEMENT
This paper is dedicated to Steve Starnes who passed away earlier this year, and who had the
original inspiration for a 'Phase 2' design.
Copyright Rolls-Royce pic 2000
C583/013/2000
133
Figure 1 - Plan View
Figure 2 - Isometric View
134
C583/013/2000
Turbogenerator Techniques
This page intentionally left blank
C583/027/2000
Problems with modern air-cooled generator stator
windings
S TUCKER and J MILSOM
National Power PLC, Selby, UK
G GRIFFITH
Florida Power and Light, Juno Beach, Florida, USA
G STONE
Iris Power Engineering Inc., Ontario, Canada
ABSTRACT
The widespread use of air-cooled turbine generators has reduced the capital cost of modern
gas-fired and combined cycle plants. However problems have recently been experienced in
some of the stator windings in such generators. These problems include deterioration and
failure of the insulation in the stator slot due to coil vibration. This has occurred in both
conventional and global VPI stators. In addition, some machines have seen problems with
partial discharge in the endwinding and lead area due to insufficient spacing between high
voltage components. This paper presents examples of these problems, and discusses some
repair alternatives.
INTRODUCTION
In the past decade, a very large percentage of the new generation brought into service has
been driven by gas turbines, or based on smaller steam turbines, often in a combined cycle
arrangement. The generators needed for such turbines are usually in the range of 25 MVA to
300 MVA, rather than the 600 MVA and greater generators that were common a decade ago.
The market for supplying these smaller turbine generators has become extremely competitive,
and there is tremendous pressure on the OEMs to reduce production costs.
The OEMs have developed several approaches to make their generators more cost-effective.
One of the most widely adopted approaches is to make the turbine generator air-cooled. That
is, rather than using the traditional hydrogen cooling that in the past predominated in high
speed machines rated more than about 100 MVA, air is used to directly or indirectly cool the
rotor and stator windings [1-3]. The benefit is that all the equipment for the hydrogen
handling is no longer needed, reducing costs. The impact of using air cooling is that the stator
windings are much more likely to suffer from partial discharges (PD), since the breakdown
voltage of air is much lower than the breakdown voltage of high-pressure hydrogen.
Another approach with the new turbo generators is to increase some of the design stresses.
For example, by increasing the design electrical stress of the groundwall insulation, the
groundwall thickness decreases. This enables the stator slots to be smaller, and the whole
machine to be smaller, reducing material costs (copper, steel, insulation) by as much as 20%
C583/027/2000 IMechE 2000
137
[4]. Another benefit is that the thinner groundwall insulation can conduct the heat from the
copper I
2
R losses to the stator core (heat sink) much more effectively, reducing stator
temperature. In addition to increasing the electric stress, some OEMs have also been taking
better advantage of the thermal capabilities of modern epoxy-mica groundwall insulation.
This allows the copper cross-section to be reduced (increasing the copper I
2
R losses), the
amount of steel to be reduced (increasing core losses) and/or the endwinding length to be
reduced (reducing the effectiveness of the endwinding as a heat sink). The consequence is
higher operating temperatures. However one should recall that the common Class F materials
are specified to have an average life of 20,000 hours (about 3 years) at 155 C [5]. Using the
Ahhrenius relationship, where the thermal life increases about 2 times for every 10 C drop in
operating temperature, users can obtain an indication of the life of a winding if it operates
close to the 'rated' temperature of 155 C.
A third approach to reduce the manufacturing cost of modern stator windings has been to
si mpli f y the system used to support the bars or coils in the stator slot. Large turbine
generators used elaborate multi-part wedges, ripple springs and/or conformable packing
materials in an effort to eliminate coil vibration in the slot due to 100/120 Hz magnetic forces.
Such systems are expensive and ti me consuming to install. Since the modern machines have
lower ratings, the bar/coil currents are lower and consequently there may be less need for
elaborate slot support systems.
A variation of this has been the trend to global VPI stators, i.e. stators that are completely
wound with flexible 'green' coils or bars, and then the entire stator is impregnated with an
epoxy in a large tank [2,6]. This greatly reduces the labour costs and production time as well
as usually ensuring the coils are rigidly held within the slot.
All of these innovations have reduced the cost of the generators. However, as occurs when
any new innovations are introduced, some stator winding problems have become apparent in
the operation of recently installed air-cooled machines. The following are some of the
problems the authors have encountered to date, together with some of the options to be
considered for repair.
LOOSE COILS IN CONVENTIONAL STATORS
Coil vibration in the slot has long been a problem in all non-global VPI stators made with
thermoset insulation systems such as epoxy mica. The first instances were reported over 50
years ago [7], The root cause of the problem is that if the coils are not tightly held in the slot,
at f ull load, the twice power frequency magnetic forces will vibrate the coil. Consequently,
the groundwall i nsulati on rubs against the laminated steel core - a very abrasive surface. First
the semi conductive layer of the bar or coil is abraded away, and then the groundwall
insulation. The mechanism is sometimes referred to as slot discharge because once the
semiconductive coating is abraded, partial discharges occur between the coil surface and the
core, further increasing the rate of deterioration. First in hydrogen-cooled turbos, and then in
hydrogenerators, OEMs have developed techniques outlined above to keep the windings tight,
even as the insulation and other slot components shrink with normal aging.
As mentioned above, there has been a tendency in the modern turbos to reduce the cost and
complexity of the slot support system. Unfortunately cases have occurred where the windings
were able to move in the slot when new, or gradually become loose in operation, especially if
138
C583/027/2000 IMechE 2000
oil is present. The result has been some premature failures caused by the slot discharge
process. Figure 1 shows the side of a coil for a 40 MVA generator where about 30% of the
groundwall insulation thickness has been abraded away by the vibration, prior to failure. It is
believed that the coils were not originally installed tightly enough during manufacture. The
ridges where no abrasion occurs at the stator core ventilation ducts are clearly visible. The
larger the machine, in general, the larger the magnetic forces acting on the coils/bars, and thus
the faster the failure process.
Arresting this failure process is fairly straightforward, if somewhat expensive. During a
suitable rotor pull, re-side packing and rewedging are needed, preferably with ripple springs
and/or two part wedges. If the repair can be done before the semiconductive coating is
abraded away, then the winding can often be restored to as-new condition. Unfortunately, if
the semiconductive layer is worn away and the epoxy-mica insulation is exposed, then partial
discharge will continue to occur, gradually (but at a much slower rate) destroying the
groundwall. Injecting into the slots carbon loaded varnishes, silicon rubber or epoxy can
somewhat restore the semiconductive coating, but PD will still occur in areas where the
injection was incomplete.
Figure 1: Photo of a coil removed from a stator bar that failed due to slot discharge.
The original winding was not installed in the winding tightly, and no ripple springs were
used. This was a conventionally made stator, where the insulation was cured outside of
the slot.
LOOSE COILS IN GLOBAL VPI MACHINES
In stators with global VPI windings the coils are manufactured slightly oversized to ensure a
tight fit within the slot. A very simple wedging system is utilised which mainly serves to hold
the bars in place prior to impregnation.
Once assembled the stator is moved into a large tank that is then put under vacuum. A
viscous epoxy resin is then injected into the tank in order to impregnate the whole of the main
wall insulation system of the stator bars. It is also intended to fill all the voids in the slot
between the bar and the core and to lock the bars in place since baking to cure the resin
follows the resin impregnation. Since the stator is manufactured with a nominal interference
C583/027/2000 IMechE 2000
139
fit (subject to tolerances of both bar size and core pack variations) the thin layer of resin is
relied on to hold the bars in place.
However when the machine enters operation and undergoes thermal cycling as well as other
mechanical and electrical stresses, the bars will want to move in the slot. The three main
components of the slot: iron, copper and insulation all have different coefficients of expansion
and often a high number of thermal cycles during the commissioning of a unit will lead to the
resin bond between bar and core shearing. This now leaves the interference fit and the
wedges alone to hold the stator bars in place. From here the failure mechanism is much the
same as for a conventionally wound stator. The wedges may not help to overcome impact of
the changes in magnetic forces as the machine operates and so over time movement of the
bars may become possible. Once this movement is allowed to occur then voids will appear
between bar and core and this allows partial discharge to begin which accelerates the failure
mechanism by increasing the size of the voids.
Fig. 2 shows levels of PD between two different OEM's air-cooled generators. The
difference in levels is of an order in magnitude. Both have similar slot packing systems and
are subjected to the same global vacuum impregnation technique.
Figure 2: Examples of Partial Discharge on 2 Globally VPI Air Cooled Generators from
Different Manufacturers (note difference in y-axis scale)
As movement increases and partial discharges become stronger then the main wall insulation
is gradually worn away, just as described above for conventionally made stator windings. If
there is a weak spot at any point in the insulation, perhaps where the resin did not fully
penetrate during manufacture, then insulation may fail completely with all the attendant
alarms and excursions. Fig. 3 shows an example of such a failure, the outer semi conducting
layer that have been completely eroded along the bar, except for at the ventilation slots.
After such a failure extensive testing was carried out to determine the extent of the damage in
the machine. This included HV testing the winding, checks on the slot wedge tightness, stator
bar to core contact resistance and boroscoping around 50% of the stator vent slots. Until the
boroscoping results were seen the full extent of the damage to the remaining winding, other
than the one failed bar, had not been realised thus altering the extent of the repair necessary.
140
C583/027/2000 IMechE 2000
Figure 3: Example of loose winding failure in large global VPI generator stator.
Due to the nature of the Global VPI process rewedging is difficult, as the wedge will have
been glued to the bar by the resin. A full rewind, despite extensive damage, at this point is
uneconomic. This leaves injection of a semi conducting material as a palliative attempt to
restore the integrity of the insulation system.
ENDWINDING PD
Along with reducing groundwall thickness, manufacturers have also economized further by
requiring the design engineers to place the windings into a smaller volume. Thus in some
cases the coils are in much closer to one another in the endwinding. Unfortunately, one of the
recent problems that has been detected on some large air-cooled machines is partial discharge,
both between coils and in the lead area. The discharges are the result of design or
manufacturing deficiencies where the envelope has been pushed in coil and phase lead
spacing.
The close spacing has resulted in highly stressed air adjacent to phase connections at the
junction where two-phase leads make connection to the connection ring. The resulting partial
discharge will eventually bore a hole though the insulation, triggering phase-to-phase or
phase-to-ground faults. Figure 4 shows PD occurring in the air space at blocking between
two phase leads. Figure 5 shows insulating material use to fill the air spaces between leads,
and indicates that not all repairs can successfully correct the problem. Figure 6 shows similar
problems in the coil endwinding area. The coils are of opposing phases and have the highest
potential along with the highest electrical stress.
C583/027/2000 IMechE 2000
141
Figure 4: Small air space between
blocking at two phase leads has created
PD (white band)
Figure 5: Nomex used to insulate
between phases. PD activity (white
band) has eaten into the insulation.
Figure 6: White powder from partial discharge activity in a small air space in the
endwinding between two coils operating at very different voltages.
Now that these problems have been identified what can be done to either arrest the partial
discharge activity or completely eliminate the PD. For the lead problems, the best repair and
one that would be permanent, is to open the clearance between the two phases, reducing the
li keli hood of PD. Two options that are available to open up the space: 1) redesign the
endwinding utilizing a new connection method or 2) modify the existing design by reducing
the amount of insulation on the phase leads. The second option requires stripping the
insulation from each lead and reinsulating.
Another approach, which would be quicker but not necessarily permanent, is to fill the void
between the two phase leads with a silicon rubber material. The silicon rubber has to be
applied void free or PD activity can reoccur. Similar to the above approach is to install
Nomex (trademark of Dupont) sheets between the phase leads filling all voids with an epoxy
and applying conducting paint to attempt to redistribute the electrical stress at the insulation
contact point between the two phases. Again this approach may not be permanent and could
142 C583/027/2000 IMechE 2000
require further maintenance. These two repairs will require less time and cost initially but
could result in higher maintenance in the long run.
As to the repair of the endwinding PD activity between two adjacent coils of opposing phases
one is left will few choices. As the spacing is set by the current design one may be only able
to reduce the PD activity by inserting a solid insulating material such as silicon rubber in the
affected areas.
CONCLUSIONS
1. Modern stator winding design and manufacturing methods have been successful in
reducing the costs of turbine generators.
2. Problems such as coil abrasion in the slot and partial discharges in the endwinding
have lead to failures in as short as 5 years of operation.
3. To avoid premature failures, users of modern air cooled machines should ensure they
have a good purchase specification, are present at the factory for critical phases of
machine manufacture, operate the machine within specification, keep the windings
clean, and closely inspect the stator winding after about 1 year of operation. The latter
is much more important that it was for hydrogen-cooled machines.
4. Although some problems cannot be corrected, effective means are available to slow
down the deterioration process, if the problems are discovered at an early enough
stage.
REFERENCES
[1] G. Ruelle et al, "Development of Large Air-Cooled Generators for Gas Turbines and
Combined Cycles", CIGRE Paper 11-201, September 1992.
[2] A.P. Kopp et al, "Comparison of Maximum Rated Air-Cooled Turbo generators with
Modern Hydrogen Cooled Generators", CIGRE Paper 11-202, September 1992.
[3] B.E.B. Gott, "Application of Air-Cooled Generators to Modern Power Plants", Proc IEEE
International Electric Drives Conference, Seattle, WA, May 1999, pp 317-319.
[4] B.J. Moore et al, "Utilizing Reduced Build Concepts in the Development of Insulation
Systems for Motors", Proc IEEE Electrical Insulation Conference, Cincinnati, October
1999, p 347.
[5] IEC Standard 60085, "Thermal Evaluation and Classification of Electrical Insulation".
[6] N. Urakawa et al, "New High Voltage Insulation System for Air-Cooled Turbine
Generators", Proc IEEE Electrical Insulation Conference, Cincinnati, October 1999, p
353.
143
C583/027/2000 IMechE 2000
[ 7 ] J.S. Johnson, "Slot Discharge Detection Between Coil Surfaces and Core in HV Stator
Windings", AIEE Trans, Vol. 70, 1951.
144
C583/027/2000 IMechE 2000
C583/009/2000
A new approach to turbine lubrication management
J A PLATT and C J DOLMAN
Castrol Industrial, Swindon, UK
A P SWALLOW
Castrol International Technology Centre, Pangbourne, UK
SYNOPSIS
Effective management of turbine lubricants is widely acknowledged to be a significant factor in
controlling operating costs. Progressive advances in turbine design and output place ever
greater stresses on lubricants, demanding higher specification products and closer control in
service in order to optimise fluid life without compromising effective lubrication or protection
of the turbine.
The paper demonstrates the concept of "lifetime" management of turbine oils, by the lubricant
specialist, as an effective tool in cost control and in facilitating reliable budgeting of lubrication
costs over the duration of the project. It further discusses how this can be achieved by expert
monitoring to accurately evaluate lubricant condition and the use of appropriate physical and
chemical treatments to optimise fluid life.
The Rotating Bomb Oxidation Test (RBOT) has become almost universally accepted as the
industry standard method of evaluating the condition of turbine oils in service, often to the
exclusion of other test methods, despite the fact that RBOT is a lengthy and relatively
expensive test which can be difficult to perform reliably, particularly where high performance
and synthetic fluids with inherently high RBOT values are involved. The paper sets out to
explore a number of available alternative techniques, evaluating their reliability in determining
lubricant condition and in accurately predicting remaining useful life.
Involving liaison with major industry OEM's, work has been driven by close involvement in
lubrication contracts with major partners in the power generation industry and the need to gain
a more precise and detailed understanding of how turbine oils degrade in service, how this is
affected by specific turbine design characteristics and how it can be addressed by correct oil
selection and effective management in service.
C583/009/2000 IMechE 2000
145
1. TURBINE LUBRICATION
The cost of turbine lubricant over the life of a modern power plant is very small in relation to
the initial capital investment and lifetime operating costs of the plant. However, the impact of
poorly managed lubrication can be immense in terms of maintenance and repair costs and the
resultant excess downtime and lost output. The management of turbine oils in service must
clearly focus on minimising overall operating costs by balancing plant reliability with the cost
of monitoring and control of the lubricant to achieve the most cost effective solution. Over a
15 year lifetime the expected turbine lubricant cost for a typical combined cycle gas turbine
power station of around 700 MWe output (2 Gas Turbines; 1 Steam Turbine) would be of the
order of 80 000. Considering this against a downtime cost approaching 20 000 per hour in
lost revenue, clearly plant reliability is of paramount importance. Nevertheless, correct
management of turbine lubrication can significantly impact on overall operating costs by
addressing lubricant costs, disposal costs, maintenance costs and most importantly in
protecting the plant against unplanned outages.
Historically, large steam turbine plant has, for many years, been the mainstay of UK base-load
power generation. Large turbo-alternator sets of some 500 MWe output were common, with
660 MWe being found in some stations, a number of which continue to operate today. In
comparison to many other types of plant these large steam turbines present relatively
undemanding operating conditions to the lubricant. Very large capacity lubrication systems
coupled with high top-up rates, 10% per annum not being uncommon, resulted in lubricant
service life of 15-20 years in many cases.
More recently, the use of gas as a major fuel for power generation and the advent of heavy
industrial gas turbine plant has presented new challenges in the field of lubrication. The
increasing drive for greater efficiency in modern combined cycle gas turbine power plant
presents the lubricant with ever more demanding operating conditions. As new blade
technologies emerge, permitting higher firing temperatures, thermal stresses on lubricants
become ever greater. Turbine design has an important part to play here as, in many cases, the
lubricant feed and return pipes pass close to the hot gas path and maximum temperatures
encountered by the oil can vary considerably from one machine type to another, quite
independently of firing temperature. Hot-end bearing return line temperatures can be in excess
of 500C and even with double skinned pipes and a good standard of lagging the lubricant can
be exposed to elevated temperatures leading to rapid oxidation or the formation of coke. In
addition to this, lower system volumes coupled with very low losses and top-up rates combine
to give the lubricant an ever harder life. Increasingly, double shifting and standby operation
can present other problems associated with variations in lubricant temperature and careful
monitoring and supervision is required to ensure optimum performance is maintained. The
recent trend towards single shaft configuration machines with combined gas/steam turbine
lubrication systems require products combining optimum performance in terms of both thermal
and hydrolytic stability and providing rapid demulsification characteristics.
Under normal running conditions, turbine bearing lubrication is fully hydrodynamic, hydrostatic
lubrication normally being provided by the jacking oil system during the start-up period. Thus,
all metal-to-metal contact of the bearings is effectively prevented and little requirement exists
for the oil to provide any anti-wear performance, except in the case of geared turbine
applications where the turbine lubrication system also supplies oil to the generator drive
146 C583/009/2000 IMechE 2000
gearbox. The lubricant is also required to provide cooling to the bearings and therefore must
resist oxidation and thermal degradation due to the elevated temperatures involved.
Consequently, thermal and oxidation stability are critical performance characteristics of a
turbine lubricant. In addition, the lubricant is often required to perform the secondary function
of control fluid in the turbine hydraulic control systems incorporating spool valves with
extremely fine clearances and requiring very high standards of fluid cleanliness to maintain
reliable operation.
Clearly, it is a fundamental requirement of any turbine oil management programme to ensure
that these critical performance characteristics are monitored and maintained throughout the
service life of the lubricant and indeed throughout the life of the plant. A number of different
approaches may be considered to the lifetime management of turbine lubricants and these are
discussed later in the paper.
2. OIL DEGRADATION - THE OXIDATION PROCESS
The reaction of oxygen with organic materials is an important and integral part of the
ecological cycle and is essential to life. In contrast, from a lubricants perspective, the reaction
of oxygen with hydrocarbons, under varying conditions of temperature and pressure is
generally undesirable leading to the oxidative degradation of these materials. All such
processes have a common reaction pattern which is attributable to the bi-radical status of
oxygen. The propagation of the oxidation process follows one of two distinct paths depending
on the reaction temperature. (1)
2.1 Oxidation of hydrocarbons at low temperature (30C - 120C)
The self-accelerating oxidation of hydrocarbons is termed auto-oxidation. Its initial stage is
characterised by a slow reaction with oxygen followed by a phase of increased conversion until
this process comes to a standstill. The degradation is driven by an auto-catalytic reaction which
can be described by the well established free radical mechanism. It consists of four distinct
stages:
Initiation of the radical chain reaction
Propagation of the radical chain reaction
Chain branching
Termination of the radical chain reaction
2.2 Oxidation of hydrocarbons at high temperature (>120C)
2.2.1 Primary oxidation phase
Initiation and propagation of the radical chain reaction take place in the same way as for low
temperature oxidation, but selectivity is reduced and the reaction rate is increased.
2.2.2 Secondary oxidation phase
The overall bulk viscosity of the fluid will increase owing to the polycondensation of the
difunctional oxygenated products formed in the primary oxidation phase. Additional
polycondensation and polymerisation reactions of these high molecular weight intermediates
C583/009/2000 IMechE 2000 147
result in products which will no longer be soluble in the bulk hydrocarbon. This precipitate is
called 'sludge'. If the oxidation occurs under thin-film conditions, such as on a metal surface,
varnish-like deposits can be formed.
2.3 Oxidation stability of base oils
The basestocks typically used for lubricants are a mixture of C
20
- C
45
hydrocarbons, which can
be subdivided into three main groups: paraffins, naphthenes, and aromatics. In addition,
intra-molecular combinations of these three groups will also constitute a part of the base stock
composition. Traces of nitrogen-, sulphur- and oxygen containing heterocyles, together with
mercaptans (RSH), thioethers (R-S-R) and disulphides (R-S-S-R) are an integral part of the
complex composition of lubricant base oils. Some of these materials may act as natural
antioxidants, others as pro-oxidants.
2.4 Antioxidants
The majority of base fluids used in modern power plant lubricants are hydrocarbon based and
are prone to oxidative degradation. Additives are used to control such oxidation processes.
Such antioxidants fall into three main categories:
Radical scavengers - sterically hindered phenols and aromatic amines
Hydroperoxide decomposers - organosulphur and organophosphorous
Synergistic mixtures of the above
The most common radical scavengers are aminic and phenolic antioxidants. The phenolics
operate by competing successfully with the rate-determining steps of the propagation reaction,
and the aminics by interaction at low temperatures (<120C). The effectiveness of one class of
antioxidant may be enhanced by combining it with another type of stabiliser. For example,
where an aminic antioxidant is used in combination with a sterically hindered phenol, the
aminic component will react faster than the phenol with the peroxy radical. Although the
phenol is a less efficient antioxidant in its own right, the two components react synergistically
to regenerate the more effective aminic antioxidant. The selection of antioxidants and
determination of the precise relative proportions of each component of the antioxidant system
is a very complex subject, the final formulation being optimised for the particular type of
application and conditions in which the lubricant will be employed.
3. OIL CONDITION MONITORING
Oil sample analysis yields valuable information regarding three aspects of the system:
Condition of base fluid, identifying any degradation which has taken place
Additive content, compared to that of new oil
Wear elements, to identify source and severity of wear of machine components
Much can be learned in these respects from a single sample which gives a "snapshot" of the oil
and machine condition, and individual samples are often useful in identifying a specific problem
or in giving the machine a "clean bill of health". However, very much more meaningful data
148 C583/009/2000 IMechE 2000
can be derived from a planned programme of oil analysis by monitoring changes and trends in
the various parameters measured. In determining an accurate picture of oil and system
conditions, in addition to the absolute value of each parameter, it is important to observe any
changes in these values and indeed any variation in the rate of change. It is also important that
no single parameter is viewed in isolation, but rather that all available data is taken into
consideration. (2)
In assessing the condition and remaining useful life of the lubricant, the principal parameters to
be considered are, physical condition and appearance, viscosity, acidity, antioxidant levels and
Rotating Bomb Oxidation Test (RBOT). It is this latter test which, although originally
designed for testing new steam turbine oils, has been accepted as the industry standard method
of evaluating the condition of gas turbine lubricants in service and which, this part of the work
associated with this paper set out to challenge. The following paragraphs describe the RBOT
test and detail the other methods evaluated in the search for a technique which gives good
correlation with RBOT, or indeed a more reliable indication of the condition and remaining
useful life of the oil.
3.1 Rotating bomb oxidation test (TP229)
The objective of this test is to quantify, under accelerated conditions, the resistance of the oil
to oxidation thereby assessing its current condition and giving an indication of remaining useful
life. In so doing, rather than considering any individual parameter, the oil is evaluated as a
complete entity, i.e. base fluid and additives, together with any contaminants present.
A sample of oil is introduced into a pressure vessel or "bomb" containing a copper catalyst
and a quantity of water. The bomb is then pressurised to 620 kPa with oxygen, immersed in a
constant temperature heating bath at 150C and rotated axially at 100 rev/min at an angle of
30 to the horizontal. The pressure rises quickly to a maximum value and remains relatively
constant at this "plateau value" for a period of time, before sharply decreasing at what is
termed the "induction break". The test is terminated when the pressure has fallen to a value
175 kPa below the maximum pressure achieved and the test duration is recorded and reported
as the RBOT value (in minutes).
Typical test durations for standard turbine oils are of the order of 5-6 hours for new oil
samples. With the introduction of high performance dedicated gas turbine oils much higher
RBOT values are common for new oil resulting in test times of 15-20 hours and with fully
synthetic products test durations of up to 50 hours are not unusual. Where such protracted
testing is involved turn-round time in the laboratory is necessarily extended and, where large
volumes of samples are required to be tested, availability of the test equipment can be a
problem. In addition, much discussion has centred around the reliability of the RBOT test in
respect of its repeatability (10%) and reproducibility (20%), which equate to quite wide
variations where high RBOT values are involved. These considerations have focused attention
on alternative methods of determining the condition and remaining useful life of turbine
lubricants and a number of alternatives are considered in the following paragraphs together
with their respective advantages and disadvantages. It will be seen that as a result of this work
RBOT, despite the outlined shortcomings, still emerges as the most representative and reliable
method currently available for evaluating the condition of turbine oils and quantifying
remaining useful life.
C583/009/2000 IMechE 2000 149
3.2 Differential Scanning Calorimetry
The principle of this test is to determine the oxidation onset temperature of the oil sample
under examination as an indication of its oxidation resistance. The sample is initially
equilibrated at 50C, following which the temperature is ramped up at a rate of 50C/min to
350C in zero grade air at a pressure of 3450 kPa. The result recorded is the temperature at
which the reaction becomes exothermic, indicating the point at which oxidation commences or
significantly accelerates (typically around 240C for a standard grade turbine oil).
This test has certain similarities to RBOT in that it takes account of overall lubricant condition
and both the base fluid and additive levels contribute to the final value. What is unclear from
work carried out to date, and remains to be firmly established, is the precise impact of the
range of potential contaminants on the test result.
Figure 1 shows the results of correlation tests with RBOT. Although a general trend is
observed in that higher RBOT values are associated with correspondingly higher temperatures
of oxidation onset, it can also be seen that the sensitivity of the test is poor as the magnitude of
change from a new oil to one which is unfit for service is relatively small. Thus, this test is not
viewed as a reliable alternative to RBOT.
3.3 Voltammetry - RULER
The oil sample is analysed by a voltammetric technique using a portable, commercially
available voltammograph (RULER). This is equipped with a probe which consists of a
glassy carbon working electrode, a platinum wire reference electrode, and a platinum wire
auxiliary electrode. The test is performed by diluting a small sample (400 ul) of oil with an
alcohol-formulated solvent containing a dissolved electrolyte and a suspended solid substrate
(sand). When the hydrocarbon-based oil/solvent/solid substrate mixture is shaken, the
suspended solids quickly precipitate the oil insolubles, leaving a clear solution for analysis.
The instrument applies a controlled voltage ramp through the electrode inserted into the
diluted oil sample. As the potential increases, the antioxidant additives become chemically
excited. This causes a current to flow which peaks at the oxidation potential of the
antioxidant. The height of the peak is related to the concentration of the additive in the
solution. The current peak height or area is measured and stored in the data acquisition
software as the RULER number. This value may then be compared to that of new oil to
indicate the degree of additive depletion.
This technique relies on there being a simple relationship between the levels of antioxidants
present and the remaining useful life of the lubricant. In practice, this is not found to be the
case as base fluid condition and contaminants are known to have considerable influence on
oxidation resistance and remaining useful life. Figure 2 shows the wide scatter in the results of
tests carried out on used turbine oils to evaluate the correlation of RULER result with RBOT
value.
Antioxidant levels can be measured more accurately by Fligh Performance Liquid
Chromatography (HPLC) in the laboratory and RULER does not appear to have any
advantage over this technique apart from increased portability of test equipment.
150
C583/009/2000 IMechE 2000
3.4. Total Acid Number
Total acid number (TAN), determined by IP139 or IP177, has long been accepted as an
indicator of oil condition. Organic acids formed during the oxidation process cause further oil
degradation with time. Thus, the acidity increase resulting from oxidation is initially slow but
progressively accelerates as the oil nears the end of its useful life. Whilst it is a useful measure
when viewed in conjunction with other parameters, TAN, in isolation, is strictly an indicator of
the condition of the base fluid and the degree of oxidation which has taken place rather than
necessarily an indication of the remaining useful life, though there is often correlation between
the two. As such, TAN is not specifically affected by antioxidant levels (though depleted
antioxidants are usually associated with oxidation having taken place with resulting increased
acidity).
Figure 3 shows the scatter in the results of tests on used turbine oils to evaluate correlation
between TAN and RBOT. Whilst there is a clear general trend, the higher RBOT values being
associated with lower TAN results as would be expected, the degree of scatter observed in the
results is such that TAN is a useful indicator of condition and remaining useful life only when
viewed in conjunction with other parameters.
3.5 Impedance Spectroscopy
This technique exploits the fact that the species formed during the oxidation process will have
a higher conductivity and permittivity than the new base oil and additives. Thus, it is possible
to detect oil degradation by observing changes in these values. Considerable interest is being
shown in this technique currently and a number of instruments are available, some of which are
readily portable.
The "Predict Navigator" is one such portable device which performs measurements of both
conductivity and permittivity at four discrete frequencies. The frequencies are selected so as to
give maximum sensitivity to water, wear metals, and oxidation. A circuit is formed with a
parallel resistor and capacitor, the conductivity of the lubricant and contaminants representing
the resistor and the permittivity of the circuit and the contaminants in the oil forming the
capacitor. The use of impedance spectroscopy combines the use of both high and low
frequencies simultaneously. The low frequency test measures changes in conductivity,
capacitance or permittivity being blocked at these frequencies. The high frequency
measurements show primarily a change in permittivity, where conductivity is blocked.
Figure 4 illustrates the measurements at all 4 frequencies. It should be noted that the water
content and wear metal content was controlled at the same level for all samples. It can be seen
from the graph that the overall trend is an increase in the change of dielectric with decreasing
RBOT value. Whilst reasonable correlation with RBOT is achieved at lower RBOT values, it
is observed that the sensitivity decreases at RBOT values of over 500 minutes. In view of this,
given the current trend towards lubricants with higher RBOT values for gas turbine
lubrication, the method was not considered as a viable alternative to RBOT. Even if this
shortcoming was to be overcome, a considerable amount of further development work would
be required to examine in closer detail the effect of changes in the various potential
contaminant levels.
C583/009/2000 IMechE 2000
151
4. MANAGEMENT OF TURBINE OILS IN SERVICE
4.1 Lubricant selection
Selection of turbine oil type is probably the single most influential factor in controlling the
service life of the lubricant. It is worth noting that currently, the major OEM's have differing
views in this respect. Whilst many modern gas turbines operate on standard turbine oils which
are principally mineral oil based derivatives of traditional steam turbine lubricants, other
OEM's now specify dedicated high performance gas turbine grades which, although still
mineral based, exhibit enhanced thermal and oxidation stability to resist the higher thermal
stresses experienced by the oil in gas turbine service. Increasingly, however, turbine lubricants
incorporating hydro-processed and fully synthetic base fluids are finding application in
industrial gas turbine plant. Such products benefit from improved inherent thermal and
oxidation stability compared with their mineral based counterparts, leading to longer and more
trouble free service life. Polyalphaolefin (PAO) fluids, often referred to as synthetic
hydrocarbons, are increasingly used in the formulation of high performance lubricants for
heavy industrial gas turbines whilst polyol ester base fluids are normally specified for
aero-derivative type machines.
Although the initial fill cost of a synthetic lubricant represents a considerable increase on the
initial cost of a mineral oil product, substantial benefits can be derived in terms of extended
service life and reduced lubricant maintenance and control costs over the life of the oil.
4.2 Partial change
One approach to maintaining the condition of turbine lubricants in service, which has been
shown to be effective, is to carry out planned partial changes of the oil, the effect of which is
twofold; firstly to regularly replace part of the base fluid, and secondly, to replenish the
additive system of the service lubricant. In this way the oil can be maintained in good
condition despite the ongoing degradation experienced through normal service. A secondary
benefit of this method of oil management is that a proportion of the contaminants and products
of degradation dispersed in the oil are carried away with the oil which is removed from the
system.
A great deal of work has been done by others (3) to determine the partial change/top-up rate
necessary to maintain the condition of a turbine oil in service, as quantified by the RBOT
value. Clearly this is dependent on the severity of the operating conditions and varies widely
from one type of turbine to another and according to operating regime (i.e. base-load, standby,
two-shift, etc.). In practice, to ensure reliable data, it has been found that each turbine must be
evaluated individually by empirical testing to establish the necessary partial change/top-up rate
to maintain the RBOT value between defined limits. Identification of appropriate RBOT limits
is clearly crucial to the success of this method of oil management. As previously discussed,
oxidation of a lubricant is a self-catalysing process and accelerates as the lubricant condition
deteriorates. In order to prevent rapid degradation, it is therefore desirable to maintain the
RBOT value at a relatively high level. However, it is evident that, the higher the target RBOT
value is set, the greater will be the partial change volume required to maintain that target and
hence the higher the cost of this means of oil management.
152
C583/009/2000 IMechE 2000
In practice, it has been shown that to maintain the RBOT between 50% and 75% of the new
oil value represents the optimum condition, combining satisfactory lubricant performance and
protection of the plant with economy of operation. Allowing the RBOT value to fall
significantly below 50% of its original value results in accelerated degradation and
consequently the requirement for more frequent partial replacement of the fluid, whilst
attempting to maintain RBOT substantially above 75% of its new value requires
uneconomically high partial change volumes. In the example shown in Figure 5 the lubricant
was maintained by partial oil changes at the rate of 5% of the system volume at approximately
6 monthly intervals.
4.3 Additive replenishment
An alternative approach to lubricant management and achieving effective control of RBOT
value is the periodic addition of an additive concentrate in order to replenish the antioxidant
levels within the service oil. By this means it is possible to restore the additive levels to the
new oil specification and to achieve an increase in RBOT value, often to a level similar to that
of the new fluid. It must be appreciated however that, unlike the partial change treatment
described in the previous section, this action is merely boosting the antioxidant levels and
provides no means of removing any contaminant from the fluid. As previously described, the
oxidation process is self catalysing and, when managing turbine lubricants in this way, it is
therefore of even greater importance to maintain the oil in a healthy condition, thereby
minimising degradation and the resultant contamination with oxidation products.
Extensive laboratory testing and control of lubricants in service, indicates that RBOT
performance following additive replenishment is influenced profoundly by the general
condition of the lubricant before treatment. Figure 6 shows a clear correlation between RBOT
retention following treatment and residual RBOT value prior to additive replenishment. It is
observed that oils giving higher RBOT values before treatment are considerably less
susceptible to RBOT deterioration after additive replenishment than are heavily degraded fluids
with low residual RBOT values. In practice it is found that, in order to optimise lubricant
service life and cost effectiveness of treatment, additive replenishment should ideally be carried
out before the RBOT value falls below 60% of the initial new oil value. Should RBOT be
allowed to fall below 50% of the new oil value, it is likely that considerable degradation of the
base fluid will have taken place resulting in the need for frequent antioxidant additions to
restrain the oxidation process and hold up the RBOT value.
Thus, the target RBOT level for oil management by additive replenishment should generally be
set within the range 60% to 100% of the new oil value. In some instances, depending on the
particular additive technology employed and base fluid condition, it is found that the RBOT
value reaches a "plateau" level somewhat below the new oil value and further additions of
antioxidant result in little change in the RBOT result. In such cases it is usual to restore the
antioxidants to the new oil levels and to manage the RBOT value between the resulting level
and the lower 60% limit.
Most turbine oil additive systems comprise two or more antioxidants, present in carefully
balanced synergistic proportions to optimise performance. It is normal for these antioxidants
to deplete at differing rates and it is therefore important that replenishment should be in the
appropriate proportions to maintain the correct balance and the resulting synergistic
C583/009/2000 IMechE 2000 153
performance. In theory, this needs to be determined individually for each situation by
empirical testing though in practice it is found that, if the policy of maintaining RBOT above
60% of its original value is consistently applied from the outset with new oil, the optimum
additive concentrate composition is similar in each case.
4.4 Control of contaminants
In any turbine lubrication system, contaminants are present in the lubricant from a variety of
sources. Contaminants may be in the form of solids, liquids or gases and may be built in during
the manufacturing process (e.g. machining or welding debris), or they may be introduced into
the system (e.g. during maintenance or oil change, or via a defective breather etc.) or they may
be internally generated (e.g. wear particles, products of oil oxidation etc.). In seeking to
extend the service life of a lubricant and to limit the degradation process, fluid cleanliness and
purity is of the utmost importance. Experience in monitoring and control of gas turbine
lubrication systems has shown that the first charge of lubricant in a system invariably gives
longer service life than subsequent ones, particularly where the initial charge has been allowed
to remain in service until heavily degraded. During operation, products of oxidation from the
oil are attracted to metal surfaces and deposited in the system. Even the most thorough
flushing of the system is insufficient to completely remove all traces of this deposit, the residue
of which is slowly removed and dispersed in the new charge of oil once in service.
Most contaminants and impurities have a detrimental effect on the service life of the oil, acting
as catalysts to oxidation. Fine paniculate material presents a very large surface area to the
fluid and, especially in the case of copper and to a lesser extent iron, are strong catalysts to
oxidation. Water, particularly at higher temperatures and in the presence of copper or iron
particulates, accelerates the oxidation process and promotes the formation of acids. As
discussed already, the oxidation process is auto-catalytic as the products of degradation
resulting from the oxidation process are themselves catalysts to further oxidation. Clearly
therefore, to remove or control contamination from all these sources can have a very
significant beneficial effect on the service life of the oil.
Water, a problem traditionally more associated with steam turbines though, increasingly an
issue in recent times in combined lubrication systems on single-shaft machines, can be readily
dealt with by use of a separator or centrifugal purifier. The on-line filtration system has to deal
with a high volume flow rate to satisfy all the lubrication requirements of the turbine and
therefore is limited in performance in terms of micron rating in order to achieve acceptable
physical dimensions and element life. Well specified main system filters will normally control
particulate contamination down to 15/zm or possibly 10um, thereby removing the larger
abrasive particulates. However, it is the fine particulate material below this size, including the
semi-soluble products of oxidation, which is of greater influence in accelerating the oxidation
process. The industry accepted standards, ISO 4406 (4) and NAS 1638 (5), currently only
quantify particulate down to 4um and 5um respectively and systems displaying very high levels
of fluid cleanliness in terms of these particle count standards may still have high concentrations
of particulate below these size ranges.
Off-line or by-pass filtration systems with high beta ratios at 3/^m or Ijum have been used
effectively to remove fine particulates and can be shown to have a beneficial effect on oil life
resulting from a reduction in the rate of degradation. Electrostatic fluid cleaning systems such
154
C583/009/2000 IMechE 2000
as the "Kleentek" unit are finding increasing popularity with operators. These systems remove
polar material including degradation products from the lubricant by electrostatic precipitation
and have been shown to achieve high standards of fluid cleanliness down to sub-micron levels.
It is important to ensure that any fluid cleaning or purifying device, whether physical or
electrostatic does not damage the oil in any way by removing any of the additives. The results
of extensive laboratory testing, involving the use of fine filtration systems and electrostatic
cleaning equipment in conjunction with both new and used lubricants, indicate that good
quality turbine oils are able to withstand both these treatments without loss of any of the
antioxidant additives. It is noted, however, that in controlled laboratory tests with heavily used
oil, in certain instances after prolonged processing, some reduction in RBOT value is observed
even though no further depletion of antioxidant has taken place. This is considered to be due
to the fact that, whilst in general, the products formed by decomposition of the antioxidants
are detrimental to the lubricant, certain species formed do in fact, themselves, possess
antioxidant properties and serve to retard the oxidation process. Removal of these species,
therefore, along with the general products of degradation has been shown, in some instances,
to result in a small reduction in the RBOT level. In practice, however, this slight negative
effect is normally small in comparison to the overall benefit of improved longevity of the
lubricant as a result of the cleanliness and purity achieved.
4.5 Determining the appropriate method of oil management
As already discussed, the effect of the antioxidant system in the oil is to protect the base fluid
against oxidation during its service life by acting sacrificially, thereby effectively retarding the
oxidation process, the antioxidants themselves being continuously depleted as a result of this
process. At the same time, contaminants present in the fluid, normally increasing in
concentration and effect with time and service, work counter to this in accelerating the
degradation process. The foregoing treatments describe how each of these effects can be dealt
with in order to reduce the rate of degradation and extend the service life of the oil.
It is clear that maximum benefit will be derived from a combination of these approaches.
Periodic partial oil change addresses both requirements by removing contaminated fluid and
replacing with fresh, and, in addition, by the introduction of fresh additive to help maintain
antioxidant levels. However, whilst extremely effective, this approach can prove expensive,
particularly on large systems, due to the addition volumes necessary to maintain RBOT within
acceptable limits, the resultant disposal costs and the labour costs involved. Equally effective
control can be achieved more cost effectively, particularly on large systems, by a composite
approach, continuously removing fine paniculate material by off-line fine filtration or
electrostatic precipitation and additionally implementing a programme of periodic additive
replenishment. This latter regime clearly involves initial capital cost in provision of the off-line
filtration system or electrostatic precipitator, together with running costs and replacement
elements etc. for this equipment, but this will be more than offset by the savings achieved in oil
usage and in disposal and labour costs compared with a partial change oil maintenance regime.
Whilst it is possible to draw a general cost comparison in this way between the different
methods of lubrication management described here, in practice, it is necessary to evaluate each
situation on an individual basis to define the costs more clearly and determine the most cost
effective management regime.
C583/009/2000 IMechE 2000
155
5. CONCLUSION
Effective management of turbine oils in service is of the utmost importance in ensuring that
optimum lubricant performance is maintained throughout the life of the plant and in eliminating
maintenance and plant operational problems resulting from the use of degraded oil.
The work described in this paper set out to challenge the acceptance of RBOT as the "industry
standard" method of evaluating the condition of turbine oils in service and to identify
alternative, more convenient and reliable tests to quantify remaining useful life. Whilst many
other techniques are available and are used to measure different parameters pertinent to oil
condition, some of which are mentioned, none share the unique feature of RBOT in evaluating
the oil as a complete entity in an accelerated test which simulates, within the limitations of the
method, the conditions seen by the oil in service. Nevertheless, RBOT should not be taken in
isolation as an indication of oil condition or remaining useful life but, rather, should be viewed
together with all available data, in particular, acidity and fluid cleanliness.
We have seen that the benefit of appropriate and cost effective management of turbine oil
extends beyond those costs directly attributable to the lubricant and has a major impact on the
far greater costs associated with plant reliability, maintenance and output. By taking a
proactive approach to controlling the lubricant condition and maintaining the RBOT value
between defined limits it has been demonstrated that it is possible to reduce overall oil usage,
disposal costs and labour costs associated with oil changes, whilst ensuring that optimum
performance of the lubricant and protection of the plant is maintained and guarding against the
risk of unplanned downtime and consequential lost output. By taking this planned approach to
oil management, the total cost of lubrication becomes a known quantity, allowing a new
degree of confidence in setting budgets and offering the operator the feasibility of handing over
the lifetime management of the oil to the lubricant specialist who, in turn, is in a position to
reliably estimate the ongoing cost of lubrication management which may then be charged as a
fixed periodic fee or indeed may be linked to plant output.
REFERENCES
(1) March, I; "Advanced Organic Chemistry: Reactions, Mechanisms and Structure".
McGraw, 1977.
(2) Platt, J.A.; "Managing Lubricating Oils in Service to Optimise Life and Minimise
Cost". Institution of Diesel and Gas Turbine Engineers, July 2000.
(3) Ohgake, R.., Sunami, M., Yoshida, T., and Watanabi, H.; "The Reliable Control of Oil
Quality in Japanese Turbine Units". ASTM Special Technical Publication No 1021,
"Turbine Oil Monitoring". ASTM, 1989. pp 32-43.
(4) ISO 4406: 1999 "Hydraulic Fluid Power - Fluids: Method for Coding the Level of
Contamination by Solid Particles". International Standards Organisation, Geneva.
(5) NAS 1638 (Rev 1): 1992 "Cleanliness Requirements of Parts Used in Hydraulic
Systems". Aerospace Industries Association of America, Washington D.C., USA.
156
C583/009/2000 IMechE 2000
Fig.1 RBOT value vs DSC oxidation onset temperature
Fig.2 RBOT value vs antioxidant content (RULERiw)
C583/009/2000 IMechE 2000
157
Fig.3 RBOT value vs Total Acid Number (IP 177)
Fig.4 Dielectric vs RBOT value
158
C583/009/2000 IMechE 2000
Fig.5 RBOT value vs time with partial oil change
Fig.6 RBOT value vs time after additive treatment
C583/009/2000 IMechE 2000 159
This page intentionally left blank
C583/011/2000
Proper maintenance practices involving condenser
cleaning and in-leakage inspection
R E PUTMAN
Conco Systems Inc., Verona, Pennsylvania, USA
R WALKER
KYRE Limited, Bromyard, UK
Abstract
The proper performance of the condenser is critical to minimizing the operating costs of a
turbogenerator unit, in terms of both heat rate and generation capacity. After reviewing the
major causes for deteriorating condenser performance, a distinction is drawn between those
causes that are cyclical or predictable and those that occur at random time intervals. In the
first category, the fouling of condenser tubes and/or tubesheet can be corrected by cleaning,
the intervals between which can be optimized. A number of cleaning techniques are reviewed.
The second category is concerned with water or air in-leakage, the return of the condenser to
normal operation depending on how quickly the source of the leak can be located and then
corrected. Tracer gas techniques for leak detection have become the preferred method.
1.0 CONDENSER ECONOMICS
1.1 Performance of a clean condenser
Condenser design specifications define a maximum effective rate of removal of the latent heat
in the exhaust vapor entering the condenser, as well as its transfer into the circulating water,
given the condenser backpressure, cooling water flow rate and inlet temperature. Variations in
the latter two parameters will change the backpressure and also affect the heat rate for a given
load. In order to minimize condensate subcooling, caused by variations in inlet water
temperature, some control over backpressure (and heat rate) may be achieved by varying the
cooling water flow rate: but the reduced tube velocities can cause silt to become deposited on
the tube surfaces and, thus, negatively affect heat rate. To avoid this, it may be possible to
allow some of the circulating water returned from the condenser to bypass the cooling towers,
so adjusting the inlet temperature to maintain the back pressure, but without reducing the total
water flow rate or tube water velocity. These are the natural responses between these variables
for a clean condenser.
Unfortunately, condensers seldom operate under clean conditions for very long periods of
time and the ills to which they are prone during normal service fall into four major categories:
C583/011/2000
161
Fouling of the tube surfaces
Tube or Tubesheet Fouling due to shell fish or debris
Circulating water in-leakage
Excess ambient air in-leakage
The first two categories are related to fouling and tend to be cyclical in nature. They may,
therefore, be regarded as predictable, although the actual fouling impact will vary from plant
to plant, and even between units in the same plant. The second two categories, concerning
water or air in-leakage, tend to be random in their occurrence. Either kind of leak is almost
certain to develop at some point in the future and a correction strategy can be prepared: when
it will occur can only be anticipated, not predicted.
1.2 Predictable maintenance problems
1.2.1 Fouling of the tube surfaces.
Almost every condenser experiences some kind of tube or tubesheet fouling, Most condenser
circulating water sources contain dissolved solids that can precipitate and become deposited
on the inner surfaces of the tubes, so adversely affecting the unit heat rate and/or limiting
generation capacity. These deposits can also contribute to various types of corrosion and, if
not removed periodically, the corrosion may eventually penetrate the tube wall, allowing
circulating water to leak into and contaminate the condensate.
Fouling can affect not only unit heat rate but also the ability of the turbine to generate its
design load capacity. In fossil-fired plants, an increase in heat rate is reflected in higher fuel
costs for a given load and increases of 2% are not uncommon. In both fossil and nuclear
plants, if the fouling becomes severe, it will cause the backpressure to rise to its upper limit,
forcing a reduction in generated power. There are reports of up to 20 Mw having been
recovered by the removal of severe accumulations of deposits.
Rabas et al(l) included a variety of fouling models (i.e. plots of fouling resistance vs. time)
that were very site-dependent and cyclical in nature, sometimes affected also by the season of
the year. Techniques based on ASME single-tube heat transfer calculations(2) can be used to
estimate fouling resistance, although Putman and Karg(3) noted how account must be taken of
the observed variation in the design cleanliness factor with respect to load. Having established
the fouling model, and the mean monthly load and cooling water inlet temperature for a
twelve-month period, Putman(4) described how this historical data could be integrated to
create an optimum tube cleaning-frequency strategy. The strategy is suitable for both linear
and non-linear fouling models and identifies those months in a twelve-month period during
which the condenser should be cleaned, if the total cost of losses over the period due to
fouling plus cleaning is to be minimized.
However, with nuclear plants, there is less flexibility and it is considered good practice to clean
at every refueling outage. Even where a nuclear unit is equipped with an on-line cleaning system,
an annual or biannual off-line mechanical cleaning assures that condenser effectiveness will be
maintained, reduces the risk of pitting from the stagnant water in those tubes which may become
blocked by stuck sponge balls, and ensures that every tube is cleared and cleaned at least once or
twice a year.
162
C583/011/2000
1.2.2 Tube or tubesheetfouling due to shellfish or debris.
When this occurs, the circulating water flow becomes restricted and the thermal conductivity
of the tube side water film is reduced, again affecting heat rate and/or generation capacity.
Putman and Hornick(5) found that a useful estimate of the circulating water flow rate could be
obtained by interpreting condenser operating data in terms of turbine heat balance and low
pressure expansion line relationships. Their reference plant was located on the Gulf of
Mexico, where shellfish attached themselves to both tubes and tubesheet, causing condenser
performance to decrease fairly rapidly. ASME type single-tube heat transfer calculations were
again used to estimate fouling resistance, based on the observed variation in water flow rate.
Plots showed a close and cyclical (saw-toothed) relationship between flow rate, fouling
resistance and backpressure, a study of which allowed intervals between tubesheet cleanings
to be extended, rather than conducted at fixed intervals.
Of course, while the progress of tubesheet fouling is generally predictable, an exception
would occur should a school of small fish pass through the screens and enter the waterbox.
1.3 Random maintenance problems
1.3.1 Circulating water in-leakage.
Circulating water in-leakage can result from penetrations through the tube walls, from joints
between tubes and tubesheet that have developed leaks, or from other penetrations between
the water box and condenser shell that have lost their integrity. The contaminants in the
circulating water change condensate chemistry and/or pH, tending to increase boiler or steam
generator corrosion; or result in an increased consumption of water treatment chemicals in the
attempt to compensate for the change in water chemistry. Poor water chemistry can also cause
stress corrosion cracking of steam turbine components.
Even a small circulating water in-leakage into the condensate can be damaging to the unit as a
whole and is often the cause of an unscheduled outage. The length of that outage will depend
on the means adopted to locate the source of the leak quickly, the on-line and off-line use of
tracer gases (helium or SF6) being the preferred method.
The EPRI Condenser In-Leakage Guideline(6) explores these problems in detail and shows
how the use of the tracer gases referred to above can be used to rapidly locate the source of
either water or air in-leakage, allowing the problem to be corrected quickly.
1.3.2 Excess ambient air in-leakage.
The design of condensers routinely allows for a normally acceptable level of air in-leakage,
often considered to be 1 scfm (2.13 kg/h) per 100 MW, although a new ASME Standard(7)
shows the limit to vary with the number of condenser compartments and exhaust flow rate.
The sources of such leaks can be labyrinth glands on steam turbine shafts, as well as packings
and seals that are less than leak-tight. As with fouling, air in-leakage rates above the
acceptable values can detrimentally affect heat rate as well as limit generation capacity.
Excessive air in-leakage also affects the concentration of dissolved oxygen in the hotwell,
which can cause corrosion damage to other parts of the unit. Of course, high dissolved oxygen
levels can also be caused by a change in the performance of the air removal equipment and
this should be checked before undertaking the search for leaks. In many cases, the increased
reliance on deaeration taking place within the condenser makes minimizing air in-leakage
even more important.
C583/011/2000
163
2. CONDENSER TUBE CLEANING METHODS
Regardless of the tube material, the most effective way to ensure that tubes achieve their full life
expectancy is to keep them clean. Each time the tube deposits, sedimentation, biofouling and
obstructions are removed, the tube surfaces are returned almost to bare metal, providing the tube
itself with a new life cycle, the protective oxide coatings quickly rebuilding themselves to re-
passivate the cleaned tube.
The majority of cleaning procedures are performed off-line, the most frequently chosen and
fastest method being mechanical cleaning.
Among other off-line methods is the use of very high-pressure water but, since the jet can only
be moved along the tube slowly, the time taken to clean a condenser can become extended. Great
care must be taken to avoid damaging any tubesheet or tube coatings which may be present;
otherwise the successful removal of fouling deposits may become associated with new tube leaks
or increased tube sheet corrosion, only observable after the unit has been brought back on-line.
Chemicals are also used for the off-line cleaning of condenser tubes. Several mildly acidic
products are available and will remove more deposit than most other methods; but it is
expensive, takes longer for the operation to be completed, and the subsequent disposal of the
chemicals, an environmental hazard, creates its own set of problems. It has also been found quite
frequently that some residual material still needs to be removed by mechanical cleaning methods.
Very few on-line methods are available to clean condenser tubes but the best known is the
Taprogge system, which uses recirculated sponge rubber balls as the cleaning vehicle. These
systems often operate for only a part of each day and, rather than maintaining absolutely clean
tube surfaces, tend to merely limit the degree of tube fouling. Unfortunately, although the
tubes may become cleaner if abrasive balls are used, tube wear can now become a problem.
Mussalli et al(8) showed some uncertainty concerning sponge ball distribution and therefore,
how many of the tubes actually become cleaned on line. It is also not uncommon to find that
numerous sponge balls have become stuck in condenser tubes and these appear among the
material removed during mechanical cleaning operations. For these reasons, the tubes of
condensers equipped with these on-line systems still have to be cleaned periodically off-line,
especially if loss of generation capacity is of serious concern.
2.1 Mechanical cleaning of condenser tubes
Off-line mechanical cleaning is especially useful where fouling problems exist and are too
severe to be handled by any of the other methods. Obviously, the tool selected has to be the
most appropriate for removing a particular type of deposit. Moulded plastic cleaners (pigs) are
quite popular for some light silt applications. Brushes can also be used to remove these soft
deposits as well as some microbiological deposits. Brushes are also useful for cleaning tubes
with enhanced surfaces (e.g. spirally indented or finned); or those tubes with thin wall metal
inserts or epoxy type coatings.
With harder types of deposit, metal cleaners of various designs have been developed, often with
a particular deposit in mind. Mechanical condenser tube cleaners were first introduced in 1923
and subsequent patents granted over the years to the both the Griffin brothers and to the
Saxon family have improved on the original design. Figures 1.0(a) and 1.0(b) show some of
164
C583/011/2000
the current versions of this cleaner, which consist of several U-shaped tempered steel strips
arranged to form pairs of spring-loaded blades.
These strips are mounted on a spindle and placed at 90 degrees rotation to one another.
Mounted at one end of the spindle is a serrated rubber or plastic disk that allows a jet of water
to propel the cleaners through a tube with greater hydraulic efficiency. The water is directed
to the tube being cleaned by a hand-held triggered device (also known as a gun), the water
being delivered by a pump operating at only 300 psig (2.07 MPa). Since the pump is usually
mounted on a wheeled base plate, the system can be conveniently moved from unit to unit
within a plant or even moved to another plant.
A water pressure of 300 psig (2.07 MPa) is very effective for propelling the cleaning tools
through the tubes, preventing their exit velocity from rising above a safe level. Some other
cleaning systems use air or a mixture of air and water as the propelling fluid; but the expansion
of the air as the cleaner exits the tube can convert the cleaner into a projectile and place the
technicians at risk.
Another advantage of using water as the cleaner propellant is that the material removed can be
collected in a plastic container for later drying, then weighing to establish the deposit density
(g/m
2
) and followed in many cases by X-ray fluorescent analysis of the deposit cake.
Most metal cleaners are designed to have a controlled spring-loaded cutting edge: but, if
effective deposit removal is to be the result, the dimensions of the cutting surfaces have to be
closely matched to the internal diameter of the tube being cleaned, not only to improve the
peripheral surface contact but also to ensure that the appropriate spring tension will be applied as
the cleaner is propelled through the tube. The effective life of cleaners designed in this way can
be as high as 10 tube passes.
However, since such cleaners can behave as stiff springs, loading the cleaners into the tubes
was sometimes rather tedious. To speed up this operation, while also providing the blades
with more circumferential coverage of the tube surface, the cleaner shown in Figure 1.0(c)
was developed. This design not only reduced the cleaning time for 1000 tubes but, due to the
increased contact surface provided by the greater number of blades, it was found to be more
efficient in removing tenacious deposits such as those consisting of various forms of
manganese.
C583/011/2000
165
FIGURE 1
A later development involved a tool for removing hard calcite deposits, which were found to
be difficult to remove even by acid cleaning. This is shown in Figure l.O(d), and consists of a
teflon body on which are mounted a number of rotary cutters, similar to those used for cutting
glass. These are placed at different angles around the body, which is fitted with a plastic disk
similar to those used to propel other cleaners through tubes. Used on condenser tubes that had
accumulated a large quantity of very hard deposits, Stiesma et al(9) described how cleaners of
this type removed 80 tons (72.48 tonnes) of calcite material from this condenser. It has now
become a standard tool whenever hard and brittle deposits are encountered.
The experience gained from using these techniques has allowed the time to clean to be forecasted
with confidence and cleaning to be performed to schedule. For instance, a normal crew can clean
between 5,000 and 7,000 tubes during a 12-hour shift. Clearly, this number can rise with an
increase in crew size, limited only by there being adequate space in the waterbox(es) for the crew
to work effectively
The concern is occasionally expressed that mechanical cleaners can possibly cause damage to
tube surfaces. With cleaners that have been properly designed and carefully manufactured, such
damage is extremely rare. Indeed, Hovland et al(10) conducted controlled tests by passing such
cleaners repeatedly through 30 feet long, 90-10 CuNi tubes. It was found that, after 100 passes of
these cleaners, the wall thickness became reduced by only between 0.0005 and 0.0009 inches
(12.5 and 22.86u). If a 50% reduction in wall thickness is the critical parameter, extrapolating
this series of tests would be equivalent to 2800 passes of a cleaner per tube, or 1000 years of
condenser cleaning!
Clearly, all off-line cleaning methods sometimes need assistance where the deposits have been
allowed to build up and even become hard. In such cases, it may still be necessary to acid clean,
followed by cleaning with mechanical cleaners or high-pressure water to remove any remaining
debris.
2.2 Developing an appropriate cleaning procedure
The selected cleaning procedure should remove the particular deposits that are present as
completely as possible, while also causing the unit to be out of service for the minimum amount
of time. Some other major considerations in the selection process are as follows:
2.2.1 Removal of obstructions
Many tube-cleaning methods are ineffective when there are obstructions within tubes, or various
forms of macrofouling are present and, clearly, those cleaning methods should be avoided.
Attention has already been drawn to the shell-fish, which constitute macrofouling, including
Asiatic clams and zebra mussels. The selected tube cleaner must have the body and strength to
remove such obstructions. The cleaning method must also be able to remove the byssal material
that shell-fish use to attach themselves to the tube walls.
There are certain types of other debris which can become obstructions, among them being
cooling tower fill, waste construction material, sponge rubber balls, rocks, sticks, twigs, seaweed
and fresh water pollutants, any or all of which can become lodged in the tubes and have to be
removed. Meanwhile, experience has shown that, if appropriate procedures are followed,
properly designed cleaners should not become stuck inside tubes, unless the tube has been
deformed.
166 C583/011/2000
2.2.2 Removal of corrosion products
With condensers equipped with copper alloy tubing, copper deposits grow continuously and the
thick oxide coating or corrosion product can grow to the point where it will seriously impede
heat transfer. Not only will the performance of the condenser be degraded but such deposits will
also increase the potential for tube failure. When a thick outer layer of porous copper oxide is
allowed to develop, it disrupts the protective inner cuprous oxide film, exposing the base metal to
attack and causing under-deposit pitting to develop. Such destructive copper oxide
accumulations together with any other deposits must be removed regularly.
2.2.3 Surface roughness
Rough tube surfaces, as are created by the accumulation of fouling deposits, are associated with
increased friction coefficients while the reduced cooling water flow rates allow deposits to
accumulate faster. It has also been found that rough tube surfaces tend to pit more easily than
smooth surfaces. Thus smooth tube surfaces, which result from cleaning, can improve condenser
performance through:
Improved heat transfer capacity and a lower water temperature rise across the condenser,
reducng the heat lost to the environment
Increase in both flow volume and water velocity, often resulting in reduced pumping
power
Increased time required between cleanings, by reducing rate of re-deposition of fouling
material on the tube surfaces.
Reduced pitting from turbulence and gas bubble implosion
3. IN-LEAKAGE DETECTION METHODS
The EPRI Condenser In-Leakage Guideline(6) discusses in great detail the sources of both
water and air in-leakage and their consequences, together with methods for their location and
correction. The techniques have evolved from earlier methods (e.g. use of foam and plastic
wrap), to the current techniques that involve the use of tracer gases, principally helium and
sulfur hexafluoride (SF6,), both of which are non-toxic. Most of the innovations were
stimulated by the need to locate small circulating water in-leaks but, eventually, the same
techniques became used for the location of air in-leaks as well.
3.1 Water in-leaks
The condenser is supposed to form a barrier between the cooling water - which flows between
the waterboxes through the condenser tubes - and the shell side of the condenser, in which the
exhaust vapor is collected as condensate. However, even small circulating water leaks will
quickly find their way into the condensate, contaminating it with undesirable dissolved solids
which tend to cause corrosion in the feedwater heaters, boilers or steam generators. On-line
conductivity or salinity instruments are used to indicate the presence of a leak and steps
should be taken to rectify the problem as soon as possible. Unfortunately, this usually means
taking the unit out of service, the associated loss of revenue depending on the length of the
outage. Thus the time taken to locate and correct the problem can be economically significant.
This time can be reduced significantly if the waterbox associated with the leak can be
identified while the unit is still on-line.
C583/011/2000 167
Among the leak detection methods commonly employed in the past were smoke generators,
foam or plastic wrap applied to the tubesheet, ultrasonics, tube pressure testing and membrane
type rubber stoppers. These earlier techniques also left some uncertainty as to whether the
leak was confined to only one tube; so that adjacent tubes were often plugged as well (often
unnecessarily) as a form of "insurance plugging". All these methods require that the shell
side of the condenser be under vacuum, provided either by the air removal system or, if the
waterbox is divided, by continuing to run the unit at low load, taking each waterbox out of
service in turn and checking it for leaks.
The development of the helium tracer gas technique in 1978 not only reduced the time
required to locate a leak; it also eliminated much of the former uncertainty whether the actual
source of the leak had been found. However, the lowest detectible concentration of helium is
one part per million above the background level, and helium was often unable to detect small
water in-leaks. Thus a tracer gas with greater sensitivity was sought and, in 1982, a tracer gas
leak detection technique using SF6, was developed. It was found that SF
6
in concentrations as
low as one part per 10 billion (0.1 ppb) can be detected, so that small leaks could now be
located and with confidence.
Figure 2 - General setup for tube water leak test
This method is illustrated in Figure 2, in which a tracer gas monitor is connected to the off-
gas stream leaving the air removal system. A technician is stationed at the monitor to observe
the shape of the trace on the strip chart recorder (See Figure 3.0), a typical response time
being 30-45 seconds. Another technician is stationed in the waterbox and dispenses the tracer.
The two technicians communicate through two-way sound-powered radios, chosen to avoid
RF interference with other equipment.
168
C583/011/2000
Once the waterbox is open and the tubesheet exposed, a series of plenums is placed over a
section of the tubesheet, each sized to cover an ever-smaller group of tubes. The technician in
the waterbox injects the tracer gas into the plenum using a portable dispenser. The vacuum
within the condenser allows the tracer gas to pass through any leaks that may be present and
eventually appear in the off-gas stream leaving the air removal system. The technician
watching the tracer gas detector monitor warns the other technician when the presence of the
gas is observed. A smaller plenum is then used, and so on. By using this rigorous process of
elimination, the problem tube can be rapidly identified.
As a guide to tracer gas selection, if the water in-leakage is less than 50 gallons per day (189.2
I/day), SF6, is the preferred tracer gas; otherwise, either gas may be used. Similarly, if the unit
is operating at more than 20% of full load, either gas may be used. If the leak is so bad that
the unit cannot be brought on-line, then the use of helium would be the standard procedure.
Figure 3 - Chart Recording of a typical leak response
Sulfur Hexafluoride can also be used on-line to identify the waterbox, even tube bundle, in
which the leaking tube is located. The SF6 is injected periodically into the circulating water
before each waterbox while the unit is still on-line, and a permanently installed analyzer and
monitor is used to identify the waterbox associated with the leak This reduces the time
required to locate and repair the leaking tube, once the associated waterbox has been opened.
3.2 Air in-leakage
Condensers are designed to perform correctly with the unavoidable and low level of air in-
leakage which is always present(7). However, greater air in-leakage than this low normal
value will increase the concentration of non-condensibles in the shell side of the condenser
and cause the thermal resistance to heat transfer to increase. An increase in backpressure and
unit heat rate will result. The in-lcakage may even rise to the point where the backpressure
approaches its operating limit, forcing a reduction in load. Another effect of high air in-
leakage is often an increase in the concentration of dissolved oxygen in the condensate, a
C583/011/2000 169
concentration that will tend to increase with lower condensate temperatures. The
consequences are increased corrosion of feedheaters, boilers and steam generators and/or an
increase in the consumption of water treatment chemicals. All these consequences have a
negative impact on unit profitability.
Using tracer gas techniques, the source of most air in-leaks can be located with the unit still
on-line. Once again, a tracer gas monitor is installed in the off-gas line from the air removal
system and the technician handling the tracer gas dispenser roams around the unit in a
methodical manner until the technician at the monitor observes a response. The leak detection
survey starts at the turbine deck level and proceeds from top to bottom of the unit, one deck at
a time. Care must be taken when dispensing the tracer gas that only one potential source is
sprayed at a time, otherwise the ability to associate a response with a particular source may
become impaired.
Conclusion
The careful monitoring of condenser performance, the prompt detection of performance
changes, the ability to identify and locate the source of the problem, and the provision of the
appropriate techniques and means for correcting the problem, can minimize unit downtime
and so make an important contribution to unit profitability.
REFERENCES
1. Rabas, T.J., Panchal, C.B., et al, Comparison of Power-Plant Condenser Cooling
Water Fouling Rates for Spirally Indented and Plain Tubes, ASME HTD-Vol.164,
1991,pp.29-27.
2. Code on Steam Condensing Apparatus, ASME PTC12.2 - 1983, publ. American Society
of Mechanical Engineers, NY.
3. Putman, Richard E. and Karg, Dale C., Monitoring Condenser Cleanliness Factor in
Cycling Plants, Proc. IJPGC 1999, PWE-Vol 34, San Francisco, publ. ASME, New
York, pp.133-142.
4. Putman, Richard E., Calculating an Optimum Condenser Cleaning Schedule with either
Linear or Non-linear Fouling Models, IJPGC 1997, Denver, November 3-5, 1997
5. Putman, Richard E. and Hornick, Mark J., Using Turbine Thermal Kit Data to
Benchmark Condenser Performance Calculations, Proc. IJPGC 1998, Baltimore, MD,
August 23-24, 1998, pp. 511-524.
6. Condenser In-Leakage Guideline, EPRI Technical Report TR-112819, published by
The Electric Power Research Institute, January 2000.
7. ASME Performance Test Code on Steam Surface Condensers, PTC. 12.2-1998, 1998,
publ. AMSE, New York, NY.
8. Mussali, Yusuf, Hecker, George E., Cooper, Charles, et al, Improved On-Line
Condenser Cleaning Ball Distribution, Proc. ASME Power Conference, Practical
Aspects and Performance of Heat Exchanger Components and Materials, 1991,
Minneapolis, PWR-Vol. 14, pp.51-56.
9. Stiemsma, R.L., Bhayana, G.K. and Thurston, R.D., 1994, Performance Enhancement by
an Innovative Tube Cleaning Application, Proceedings of IJPGC 1994, Phoenix, AZ,
publ. ASME, New York, PWR-25-1994, pp 7-11.
10. Hovland, Alan W., Rankin, Drew A. and Saxon, Edward G., Heat Exchanger Tube Wear
by Mechanical Cleaners, Proc. Joint Power Generation Conference, 1988, Philadelphia,
PA, September 25-29, 1988.
170 C583/011/2000
C583/014/2000
Troubleshooting bearing problems during the
uprating of an existing steam turbine installation
SMAW
Neale Consulting Engineers Limited, Farnham, UK
J ABBOTT-GARNER
St Regis Paper Co. Limited, Somerset, UK
S KENDRICK
Sulzer Roteq Services, Leeds, UK
SYNOPSIS
Paper manufacturing relies heavily on large quantities of steam and electrical power. Paper
plants historically generate steam for the papermaking process and generate electricity via a
steam turbine and alternator.
This paper details the economic re-rating of a thrust bearing following repeated failures of an
existing steam turbine installation at a UK paper mill. This involved the inspection of failed
components, fundamental analysis of the turbine operating conditions and ultimately a re-
design of the thrust bearing assembly.
1. INTRODUCTION
St Regis Paper Company Ltd recycle paper at their paper mill in Wansborough, Somerset.
The steam turbine at Wansborough Mill is a 6MW machine running at 5500rpm, originally
designed by AEI Ltd (now ALSTOM). The turbine is a "passout" machine driving an
alternator via a reduction gearbox, in which steam between the LP and HP stages is drawn off
for use in the papermaking process.
In 1998, the decision was made to overhaul the turbine and additionally to introduce state-of-
the-art electronic control circuitry to give improved efficiency and process control, such that
one-man operation could be achieved.
The steam turbine also had a history of thrust bearing failures and this had limited the
electrical load to approximately 3MW.
C583/014/2000 IMechE 2000
171
2. HISTORY OF OPERATION
A turbine failure occurred during the 1980's and a strip down revealed that the thrust bearing
had failed resulting in a loss of axial location and rubbing of the turbine internal components.
At the time of the failure it was considered that the failure had been caused by liquid carry-
over from the boilers. Considerable internal damage had occurred, particularly to the rotor
thrust collar and to the HP blading.
A replacement turbine (Figure 1) was installed however this was not the same design as the
failed rotor, in terms of number of stages and in the diameter of the thrust bearing collar. In
order to fit the replacement rotor, the casing and thrust bearing assembly were modified to
suit the replacement rotor (Figure 2).
Figure 1: Replacement rotor
Figure 2: Modified housing
Whilst rated at 6MW, historically the turbine only ran at between 3 and 4MW due to the
limitations of the replacement rotor and the reliability of the thrust bearing.
C583/014/2000 IMechE2000 172
In December 1996, during a routine inspection, it was decided to replace the thrust bearing
pads with existing spares as a precaution (there was some minor surface markings).
In May 1997 the thrust bearing failed catastrophically and this was attributed to axial loading
from liquid carry-over from the boiler.
Figure 3 shows the failed thrust bearing pads and Figure 4 shows the damaged collar. New
bronze backed thrust pads were manufactured, which were noted to have approximately 3mm
thickness of whitemetal.
Figure 3: Failed thrust pads
The overhauled turbine was returned to service, but only ran at 1 - 1.2MW until July 1997
where upon an increase in power to 3.5MW, resulted in a change in axial displacement of the
rotor, the thrust bearing had failed again. The thrust collar was skimmed and new pads were
fitted (steel backing with 3mm whitemetal). The turbine was returned to service, running at
3.5MW.
During the period Aug - Sept 1997, it was observed that the axial displacement probes
indicated an increase in axial displacement from 0.28mm to 0.37mm (0.011" to 0.0145").
The turbine was stopped and thrust pads inspected and found to be in good condition. The
probes were checked and found to be working. There was no obvious explanation for the
change in axial position although it was suggested that there may have been some wear /
bedding-in at the seating of the spherical journal / thrust bearing housing which was inspected
and found to have a poor fit. It was decided to manufacture a new spherical bearing housing
(new outer made to fit existing inner). The turbine was returned to service in an unmodified
condition.
C583/014/2000 IMechE 2000
173
Figure 4: Damage to thrust collar
In December 1997, the new spherical bearing housing was installed and new pads were fitted.
The turbine was returned to service, running at 4.5MW average load. Again, a gradual
increase in axial displacement was observed.
In April 1998 the thrust pads were inspected and found to have deformed at the surface of the
whitemetal (3mm whitemetal on steel backing). The pads were replaced (also 3mm
whitemetal). The new spherical was found to be tight and was dressed in order to improve re-
alignment capability.
The turbine ran at 4 - 4.5MW until June 1998 when increasing axial displacement was
observed. The thrust bearing was inspected but found to be in good condition. The turbine
was restarted running at 4 - 4.5MW.
The turbine was shutdown on 4
th
July for a major overhaul including installation of a new
control system. Upon stripdown, it was observed that the pads were deformed. Figure 5
shows a deformed pad (top pad) compared to an unused pad (bottom pad) for comparison
purposes.
Wansborough management were made aware of reservations regarding the load bearing
capabilities of the existing thrust bearing and it was recommended that the bearing design
should be subjected to an engineering review, however cost considerations limited any design
investigations to the backing materials of the thrust pads.
Figure 5: Deformed surface of a thrust pad
New thrust bearing pads with temperature monitoring thermocouples were installed during
the overhaul, this permitted accurate monitoring of the surface metal temperature of the thrust
pads, coupled with upgraded rotor axial position monitoring allowed a methodical
engineering approach to be applied to the operation of the thrust bearing.
It was observed in post-commissioning operation that the bearing metal temperature was
sensitive to load and that the axial position of the rotor changed with time. A full engineering
analysis was therefore commissioned in order to understand the performance limitations of
the existing thrust bearing design.
174 C583/014/2000 IMechE2000
3. ANALYSIS
As a first stage in understanding the possible cause(s) of failure(s) of the thrust bearing, it was
necessary to perform basic calculations to assess the bearing load capacity and its predicted
operating conditions such as running temperature, oil flow rate and oil film thickness. These
calculated values could then be compared with the actual operating conditions, as measured,
in order to confirm, or otherwise, that the bearing was operating as predicted. Also, since
there was no available data on the design axial load on the bearing, it was necessary to
perform some approximate calculations on the load from the turbine blading together with the
thrust load from the single helical gearbox.
3.1 Thrust bearing load capacity and operating conditions
The approximate bearing dimensions, as measured on site, together with other bearing details
are given below:-
BearinglD = 129mm Bearing OD = 181mm
Number of pads =14 Diameter of thrust collar = 245mm
Pad length at mean diameter =26mm Shaft diameter = 114mm
Speed of revolution = 92.2rev/s Lubricant grade ISO VG 68
Lubricant supply temperature = 35C Lubricant feed conditions Fully flooded
Lubricant supply pressure = 10
5
N/m
2
The unknowns were:-
Applied axial load, W
Lubricant outlet temperature from bearing assembly, T
out
Typically, a design value for the oil outlet temperature is 75C, however in this case, the oil
outlet temperature was measured and typically found to be 50C. However, the design of the
bearing carrier has oil supply holes which allow oil to pass through the radial holes in the
bearing assembly without passing across the thrust bearing surface (Figure 6). Therefore cool
oil will pass straight through and mix with the hot oil from the thrust bearing resulting in an
apparent cooler oil outlet temperature. As an estimate therefore, an oil outlet temperature of
60 C was considered to be more realistic.
Using the ESDU 92035 calculation method for tilting pad thrust bearings, with the above
data, gave the results in Table 1, showing how the bearing surface temperature, oil flow and
oil film thickness vary with applied axial load.
C583/014/2000 IMechE 2000
175
Axial load
(N)
1000
5000
10000
30000
50000
80000
Bearing surface
temperature (C)
61.3
66.3
72.6
96.0
116.0
137.0
Required oil flow
(xl(T
3
m
3
/s)
0.53
0.58
0.62
0.64
0.68
0.70
Oil film thickness,
(urn)
133
56
37
16
11
7
TABLE 1: THRUST BEARING OPERATING CONDITIONS
AS A FUNCTION OF APPLIED AXIAL LOAD
(based on oil inlet temp = 35C and outlet temp = 60C)
Typically, for whitemetal bearings, a maximum bearing temperature of 120C should not be
exceeded. However, it is good practice for the maximum operating temperature (trip
temperature) to be set 10 C lower at 110 C. With the analysis above, a maximum surface
temperature of 110C, corresponds to an axial load of 44 OOON.
From catalogue data, for a similarly sized bearing, the maximum load capacity was defined at
31 SOON which was slightly lower than the above figure. However since the analysis above is
strongly influenced by both the supply and oil outlet temperature, it was considered that the
values were in reasonably close agreement.
The limits of minimum oil film thickness (ESDU 92035) depends on the degree of
misalignment present and the recommended values are 24um for "high" misalignment and
7um for "good" alignment. From Table 1, assuming misalignment is "good" (thrust collar
run-out measurements showed good alignment and also the bearing is mounted in a spherical
housing) then the calculated oil film thicknesses are within limits, even for the highest axial
load shown in Table 1, albeit at or near the minimum recommended value.
3.2 Axial load from steam turbine
The axial load from the steam turbine arises from the pressure drop across the rotor hubs and
also the reaction forces on the blades. With both the original rotor and the replaced rotor, the
HP stages used impulse blades in which the pressure drop occurred across the stationary
nozzles and there was no pressure drop across the blades themselves. Also there was no net
reaction force on the blades in an axial direction so there was no axial thrust (this is not quite
true since the effects of friction will produce an axial thrust contribution, however this is
small compared to axial forces from other parts of the turbine). There was therefore minimal
axial force from the HP stages on both the original and replacement rotor.
The 1
st
stage Curtis wheel at the LP stage was also of the impulse type, however there was a
difference in the sealing diameters which meant that there was an axial force contribution
176
C583/014/2000 IMechE2000
since the same pressure acted over different areas. The final 3 stages at the LP stage were of
the reaction type which produced axial forces both from the pressure drop across the blades
and the resulting pressure differential across the hubs. It was therefore the axial forces at this
end of the turbine which needed to be calculated.
Approximate calculations were performed to estimate the total axial force to be resisted by the
thrust bearing as shown in Table 2.
Component contributing to axial force
3 LP reaction stage blading
LP Curtis wheel and 3 LP
reaction stage hubs
Gear tooth reaction force
TOTAL
(a) Reaction analysis
(b) Pressure differential
(a) Pressure differential
Axial force
(N)
12 500 to 21 000
47000
56000
-15500
53 000 to 87 500
TABLE 2: ESTIMATION OF AXIAL FORCE CONTRIBUTIONS
The analysis was approximate since the interstage pressures were not known precisely and
without a detailed thermodynamic analysis they could not be accurately determined.
However, it was considered that a reasonable assumption was that the enthalpy drop across
each stage was equal and using this assumption the interstage pressures were estimated.
The analysis of the forces on the hubs was only pressure and area dependent and could
therefore be assessed as being reasonably accurate. Estimating blade forces was more
difficult and again without full blade details and a full fluid mechanics based analysis this
could not be done accurately. However assuming an average blade angle from blade "tip to
toe", a reaction force analysis could be done based on steam mass flow rates etc. This is
shown in Table 2 as (a) Reaction analysis. Also, as an alternative analysis for the blades,
assuming that the pressure drop across the blades is acting over the blade area then an
alternative value for axial force could be estimated which is shown in Table 2 as (b) Pressure
differential. This assumes that 50% pressure drop at an individual stage occurs across the
stationary nozzle and 50% drop is across the rotating blades.
There was also a gear reaction force due to the single helical gear arrangement and this
subtracted from the turbine rotor thrust load. The calculated axial gear reaction force is given
in Table 2 and is shown as a negative value.
The analysis in Table 2 was at full load conditions (6.0MW) since this represents the
maximum axial load condition. This occurs with maximum steam flow through the LP stages
(approx. 5kg/s, or 40 000 Ibs/hr).
The estimated range of axial force depends on the analysis of the blade forces. However, even
at the minimum calculated value of 53 OOON, the specific load on the thrust bearing was 4.2
MPa and typically a maximum recommended catalogue bearing pressure for tilting pad thrust
bearings is 3.5MPa (SOOpsi). The American Petroleum Institute (API) recommend a
C583/014/2000 IMechE 2000 177
maximum load on tilting pad thrust bearings at 50% of the bearing manufacturer's
recommended maximum, i.e. 1.75MPa (250psi) in this case. With an axial load of 87 SOON
the bearing load was 6.9MPa which considerably exceeds the maximum recommended values
above.
From the data in Tables 1 and 2, for a range of axial load from 53kN to 87.5kN, the estimated
operating temperature of the thrust bearing would be 120 - 150C and this exceeds the
maximum recommended bearing temperature of 110 - 120C. From measurements of the
actual thrust bearing temperatures, typically in the range 95 - 105C, this suggested that axial
loads may actually be slightly lower. However since the RTD measured at the back of the
whitemetal and the fact that the bearing temperature was influenced by the oil outlet
temperature from the bearing assembly the discrepancy between calculation and measured
values was not surprising.
4. DISCUSSION
The turbine load analysis and bearing load capacity limits indicated that the bearing had
inadequate load carrying capacity for the required duty, even taking into account any
inaccuracies in the calculations. The evidence from looking at used thrust bearing pads
confirmed this. Figure 5 shows a steel backed thrust pad which had been replaced on the
basis of excessive axial movement and as can be seen, the whitemetal had extruded over the
edges of the bearing pad. The bearing pad had an excessive thickness of whitemetal (approx.
3mm) which had contributed to a lower overall bulk material strength. However, the same
type of extrusion was also observed on bronze backed pads with only 1.5mm whitemetal after
a relatively short period of running (Figure 6). With a gradual loss in surface profile, the oil
film thickness decreased and temperatures rose until ultimately failure occurred by wiping.
Figure 6: Wiping of bronze backed pads with 1.5mm whitemetal thickness
Since the problem was a fundamental one with the thrust bearing design, the fact that thrust
bearing failures had not occurred in the past needed to be explained (apart from those caused
178 C583/014/2000 IMechE2000
by liquid carry-over). It is considered that there are two main reasons why recurring thrust
bearing failures had not occurred historically:-
(1) Before a new boiler was installed in late 1996, the steam mass flow rate to the
condenser had an upper limit of about 3 kg/s (24 000 Ibs/hr) but typically ran at 1.5 to
1.8kg/s (12 000 - 14 000 Ibs/hr). This would result in lower axial force on the thrust
bearing since it is the conditions at the LP stage of the turbine that determine the axial
load. With the uprated steam capacity together with improved condenser capacity, the
mass flow rate was increased to a maximum of 5kg/s (40 000 Ibs/hr), with a typical
operating range being 3.2 - 3.8kg/s (25 000 - 30 000 Ibs/hr). This represented a
significant increase in flow through the LP turbine blades and had consequently
increased the axial load on the bearing.
(2) It was assumed that steel backed thrust pads lined with 1.0 - 1.5mm whitemetal were
used in the past and it was only recently that bronze pads were used as additional
protection to the thrust collar in the event of failure. It is the ratio of thermal expansion
and thermal conductivity of the backing material that affects the load capacity of a
thrust bearing and steel backed pads are superior to bronze pads in this respect. Bronze
has a higher thermal expansion coefficient and this tends to distort the pad surface at the
operating temperature which derates the load capacity, this in spite of the higher thermal
conductivity which would help to reduce pad temperatures. Also, continual thrust
bearing overhaul probably resulted in thicker whitemetal (up to 3mm) being used;
which further reduced the load capacity of the thrust bearing assembly.
An obvious (in retrospect!) piece of evidence which tended to support the hypothesis of
excessive axial load was the fact that the replacement rotor had a larger diameter thrust collar.
The replacement rotor had the same configuration at the LP stages as the original rotor and
therefore exerted the same level of axial thrust on the bearing. The fact that the thrust collar
was larger suggested that it was indeed needed since thrust collars are generally not made
larger than required since they only increase churning losses and result in higher operating
temperatures.
It was considered that there were a few measures that could be taken in the short term to
prolong the life of bearing of the existing size, however for long term reliability, a design
change to a larger bearing was recommended. These are discussed below:
4.1 Short term modifications
The problem with bearing reliability was excessive axial load together with high
temperatures. Therefore measures to increase the load capacity, or reduce temperatures were
suggested.
(1) The combined journal and thrust bearing housing had large diameter radial holes
(approx. 19mm) close to the thrust bearing feed. This meant that there was a "short
circuit" for fresh cool oil to pass without cooling the thrust bearing pads. Restricting
this flow, or even blocking it off totally, could increase the oil flow to the thrust bearing
and consequently reduce pad temperatures. During a subsequent site visit, the radial
holes were reduced from 19mm to 6mm and the bearing temperatures were reduced
slightly to 97C at 4.0MW generating load. The housing outlet temperature was approx.
55 C, so it was considered that the drillings could be restricted further to 3mm and
assuming temperatures were still stable, then they could be blanked off.
C583/014/2000 IMechE 2000
179
(2) To increase load capacity, steel backed pads were recommended. Whilst this increased
the risk of thrust collar damage in the event of failure, with the monitoring equipment in
place (RTD's and axial position indicators), any incipient failure should be detected by
at least one of the monitoring systems.
As an alternative, another backing material which could have reduced running
temperatures is copper chromium. Whilst the material has a coefficient of thermal
expansion similar to bronze, it has a thermal conductivity about 3 times greater. Any
reduction in pad temperature would be significant in reducing the risk of whitemetal
extrusion.
(3) A slight increase in load capacity could be achieved by using thinner whitemetal. A
minimum thickness of 0.75mm was suggested (existing pads had 1mm whitemetal),
however as noted in (2) above, this increased the risk of collar damage but the same
comments about monitoring were applicable.
4.2 Long term modifications
To ensure long term reliability of the thrust bearing, it was recommended that a re-designed
bearing was required using the available area of the existing large diameter thrust collar. It
was recommended to re-machine the thrust collar flat since a small wear step existed at the
inner and outer periphery of the existing thrust pads.
From catalogue data, a thrust bearing of comparable diameter to the existing thrust collar had
a load capacity of 82 O O O N , which is comparable with the highest value calculated in Table 2.
The calculated running temperature was approximately 100C for fully flooded conditions.
Whilst this was still quite high, since the bearing pressure was lower then there should be no
risk of surface extrusion. The calculated oil film thickness (21 um) was about 3 times as thick
as the smaller bearing under the same loading conditions and hence provided a substantial
increased margin of safety.
The mean peripheral speed of the thrust collar with the larger bearing would be 51m/s and it
was recommended to incorporate individual pad lubrication in the redesigned bearing.
Repeating the analysis for individual pad lubrication gives a running temperature of 80 C and
a thicker oil film of 30um, giving a further increase in safety margin compared to a fully
flooded bearing design.
5. POST MODIFICATION EXPERIENCE
The short terms modification of reducing the oil drain hole size, reduced the bearing
temperatures slightly, allowing the turbine to operate at a higher power rating than previously,
whilst the bearing assembly was re-designed and manufactured.
In January 1999 the new bearing was assembled into the machine (Figure 7). Following re-
commissioning, the bearing operating temperature was stable at approximately 80C at full
load (6.0MW). Additionally, and more importantly, there was no gradual increase in
temperature or axial displacement of the shaft, as was the case with the earlier smaller thrust
bearing design.
180
C583/014/2000 IMechE2000
Figure 7: Installation on the new thrust bearing design
(note the larger thrust pad size)
The new bearing has now been in operation for more than 1 year and the increase in turbine
capacity has resulted in substantial annual saving, principally due to the difference in cost
between generated electricity and imported electricity from the Grid.
6. CONCLUSIONS
1. The existing thrust bearing had insufficient load capacity to withstand the maximum
axial loads that were applied by the 6.0MW steam turbine rotor.
2. Thrust bearing failures had not occurred on a regular basis in the past since the steam
flow rate through the LP stage of the turbine was lower than after the uprating. The
increase in boiler and condenser capacity had increased the axial load on the bearing
and this was considered to be the primary cause of the recent failures.
3. In order to ensure long term reliability of the thrust bearing, a larger diameter bearing
was required. Short term modifications were recommended as a means for increasing
turbine throughput whilst the new bearing was being designed and manufactured.
4. A larger thrust bearing was ultimately introduced and this enabled full turbine power to
be achieved when demanded, resulting in substantial financial savings.
C583/014/2000 IMechE 2000
181
This page intentionally left blank
Condition Monitoring
This page intentionally left blank
C583/024/2000
Evolution of health monitoring for CCGT plant
G HEWITT, N M JOHNSON, J MACDONALD, K DORLING, and J WORSFOLD
Powergen UK plc, Nottingham, UK
ABSTRACT
The objective of any health monitoring regime on turbine generating plant is to aid the operation
and maintenance decision making process. This paper discusses the implementation, monitoring
techniques, resourcing, benefits, case studies and ongoing developments for combined cycle gas
turbine (CCGT) plant of a health monitoring system and its use in risk management of station
operation. As such the paper is of interest to all power plant engineers and managers.
PowerGen has at Power Technology Centre a central remote facility for vibration monitoring of
all of its large turbine generators. Experience and new monitoring technologies have shown that
other parameters are also indicative of machine condition and that knowledge and correlation of
such parameters can be used to give a better indication of the health of the machine. To this end
the features of the remote monitoring service have been continuously developed and upgraded in
a pragmatic approach and now cover most of the main gas turbine health indicators as well as the
more traditional shaft line parameters. In this way health, performance and condition for
maintenance can be trended on a daily or continuous basis and feedback provided to operations
and maintenance personnel to aid in management of CCGT unit operation.
1 INTRODUCTION
Historically the health monitoring of rotating machines has been undertaken using vibration
analysis. With the advent of low cost, high-powered computers and signal conditioning systems
vibration analysis is now undertaken exceptionally quickly, although interpretation is still very
much experience based. The current trend in health monitoring is to complement the vibration
analysis with other technologies, which gives a more comprehensive indication of the overall
health of the machine.
2 SYSTEM ARCHITECTURE
Reference 1 described a system that effectively screened large quantities of vibration data for
abnormal behaviour. This project continues and combustion, pyrometry and performance
monitoring have since been added for CCGT plant. The screening techniques currently used for
vibration analysis have been modified and the monitoring schedule adapted to suit the parameter
in question. However, the problem of extracting relevant information from large quantities of
data still remains.
C583/024/2000 IMechE 2000 185
Figure 1 shows a typical monitoring strategy applied to many CCGT stations. The choice of
which systems are fitted is very much based on cost benefit studies associated with known issues
on the units in question. Hence, whilst most installations will have vibration monitoring
supplemented by off-line performance analysis, relatively few will have the whole tranch.
As discussed later, a health monitoring strategy is an on-line and off-line activity with inspections
and component life management forming an integral part of the overall plan.
Figure 1 Elements of a health monitoring strategy
The monitoring systems are in general specialist tools with little or no interface between
themselves and other monitoring systems or the distributed control system (DCS). There is a
need for the information from one system to be viewed with that from another (e.g. vibration and
compressor performance). This can be achieved by sending the data from each system to a
common database, or by using one of the monitoring systems' databases. With the latter, there
may be limitations with database size hence only a subset of the data can be handled for trending.
There is also a lack of what would be regarded as 'high level' status indications from these
systems since they require experience to interpret the results.
Figure 2 shows a schematic representation of how data can be made more accessible and
integrated. Although this sounds easy, it is not always the case due to a lack of open data
186
C583/024/2000 IMechE 2000
architecture. To make most use of the data and minimise visual interrogation, automatic
screening will be required. The screening of the data could take place on the individual
monitoring systems or after the common database. The type of screening (or rule) would need to
be determined by either fault tree analysis or by experience. For example - sudden vibration
change and drop in compressor performance could indicate foreign object damage had occurred.
Figure 2 System database
3 VIBRATION MONITORING
Vibration monitoring based on daily health checks and long term trending remains an integral
part of the regime for CCGT plant. Knowledge built in this way is used to provide improved
thermal management of transients, early indication of problem scenarios developing and
highlight areas deserving attention during maintenance. This has been fully discussed in
sufficient detail in Reference 1.
However, since Reference 1 was written several developments have taken place that enables an
improved service to be offered. These are: -
1. Ability of Vibration monitoring systems to have alarms that can be tailored to check and
protect against machine specific faults. This has been undertaken on several occasions and is
used to raise audible alarms in the Control Room. Two examples are - potential rotor cracking
and governor drive cross-shaft vibration due to excessive wear.
C583/024/2000 IMechE 2000
187
2. Some Vibration monitoring systems now have the capability to raise bleeper or e-mail
messages. This is extremely useful when the first point of contact is preferably with a
specialist or for occasions when a new alarm configuration is being tested. This enables 24
hour off-site cover to be provided. With many people having home PCs the ability for rapid
off-site support has become a practical option.
4 COMBUSTION - EXHAUST GAS TEMPERATURES, DYNAMIC PRESSURES
AND NOx
The exhaust gas temperatures from the gas turbine are a key feature of the control system but are
also used to monitor the health of the combustion system. The monitoring of exhaust gas
temperatures can be used for the early warning of air ingress (possibly due to cracking) into
combustion systems or burner fouling.
One of the problems with monitoring exhaust gas temperatures is that there are in general a large
number of thermocouples and these can have high variability due to operational mode. A recent
project used data from thirty-one thermocouples and a self-learning algorithm to produce one
alarm as illustrated in Figure 3. The ability to use the On Line Monitoring System (TIRESIAS) as
a platform to extract the thermocouple data from the DCS and via Excel perform calculations or
derive alarms is a powerful tool; performance calculations can also be undertaken in this way.
The algorithm was initially tested at a central location and is to be rolled back into the DCS once
false alarms are minimised. The 24-hour cover via a bleeper system was required because
analysis of failure data showed that the first indications of a problem were 8-12 hours prior to the
likelihood of downstream damage occurring. Hence, daily trending on the previous 24-hours data
would not have been sufficient.
The project showed the benefit of having the ability to obtain relevant data and assess and
manipulate it quickly into a manageable form to produce a single alarm for feedback to
operations staff to aid in unit risk management.
Combustion dynamic pressure measurements have been undertaken on several large gas turbines.
For those machines that are susceptible to combustion instabilities, 'humming', a monitoring
strategy that uses frequency banded data encompassing the humming and a level alarm, is used to
enunciate an alarm in the control room. This has proved effective in keeping load below the
damage threshold thereby reducing subsequent maintenance requirements on the hardware.
Dynamic pressure measurements on can annular machines have shown that there is a rich spectral
content, which has the potential for both diagnosis and monitoring. As of yet PowerGen has not
undertaken a permanent installation (others have - Reference 2). However, routine dynamic
pressure measurements are undertaken at 4000-hour intervals as a pragmatic method of assessing
the value of continuously monitoring these parameters.
Emissions can also be used to detect changes in the health of a combustion system. Sudden
changes in the level of NOx emissions for example in a dry low NOx combustion system are
188 C583/024/2000 IMechE 2000
indicative of a burner using more primary gas i.e. flame burning in diffusion. This could indicate
a flashback.
Figure 3 Generation of a single useful alarm from on-line data
5 PYROMETERS
Optical pyrometers allow blade temperatures to be measured by direct visual means rather than
by the more conventional, but less accurate, system of calculation from the measured exhaust gas
temperatures. PowerGen has pyrometers fitted at two of its combined cycle gas turbine (CCGT)
stations with plans for more to be added in the future (Figure 4),
Figure 4 Typical pyrometer installation
C583/024/2000 IMechE 2000
189
The operating temperature of a rotating gas turbine blade is a major factor in determining the
optimum maintenance interval with the associated impact on costs and the overall life span of the
part. Excess temperatures can significantly reduce component life span and threaten plant
integrity whilst operation at temperatures lower than design will reduce turbine efficiency and
increase running costs.
Power Technology has developed the PYRO software package to accept the data gathered by the
pyrometers and perform monitoring routines that offer several advantages:
On-line analysis of blade temperatures allows early detection of significant changes which
can be indicative of problems developing in the blade cooling system
Compilation of a long-term temperature versus time history for each blade enables extension
of blade life based on real operating data rather than formulae for equivalent operation and
helps to monitor and manage the rate of component deterioration, including coating loss -
particularly important when the blade is transferred between machines or is part of a mixed
set
Monitoring of the average temperature of a row of blades enables the detection of changes in
the actual firing temperature with the subsequent possibility of mitigating actions
Figure 5 Typical PYRO output
PYRO analyses the temperature profile obtained by the pyrometer's data acquisition system
during each revolution of the turbine to extract the maximum, minimum and average
temperatures for each individual blade and the average temperature for the row of blades (Figure
5). By recording at 30-minute intervals and storing data to disc, downloading on a daily basis
through the on-line monitoring (TIRESIAS) link to Power Technology and analysing on a regular
190 C583/024/2000 IMechE 2000
basis, unique part histories are built up. Data can be input to the in-house blade life management
software package (BLADE) to predict the impact of actual operating conditions on part life.
As such, pyrometers, coupled with software data screening and specialist interpretation, are a
valuable tool in aiding operation and maintenance decision making/planning on stations.
6 PERFORMANCE MONITORING
Engine performance measures such as inlet pressure drop, compressor massflow and efficiency,
firing temperature and turbine efficiency are valuable in mapping actual performance degradation
as opposed to predicted loss. They can also be used to detect gross faults and source, for example
an oil leak into the inlet resulting in compressor fouling and evaluate the best response
operationally. In addition they can be used to develop the appropriate on-line/off-line washing
strategy/balance.
Figure 6 Typical output sheet from Acceptance test software suite
Power Technology has a two-pronged approach to on-line performance monitoring:
A suite of programmes have been developed for acceptance testing of gas turbine plant in
open cycle, combined cycle or combined heat and power (CHP) applications. This software
has been adapted to accept data from the TIRESIAS on-line link to stations to calculate
compressor massflow, compressor efficiency, ISO2314 turbine inlet temperature (or firing
temperature when appropriate), turbine efficiency, corrected power and heat rate (efficiency).
The method of data collection and calculation is similar to that for exhaust gas temperatures.
C583/024/2000 IMechE 2000 191
Future enhancements planned include the addition of steam turbine efficiency. Figure 6
shows a typical output page from the performance programme suite.
The PROATES performance-monitoring package has been used successfully for a number
of years for off-line performance trending and evaluation of what-if scenarios for entire
stations. The package has now been adapted to monitor performance on-line and is being field
tested at one of PowerGen's CCGT stations with remote on-line monitoring at Power
Technology.
7 INSPECTIONS
Inspections with an engine off-line are very often overlooked as a health assessment regime but
are in fact extremely important as they enable feedback on the condition of the plant and how it
relates to the on-line monitoring strategy employed to be made. A recent example of this is a gas
turbine that has shown significant retardation in vibration related damage following the
elimination of a minor combustion instability.
8 CASE STUDIES
8.1 Combustion Monitoring
Dry low NOx can annular combustion systems can be subject to in service problems associated
with cracking of the combustion liners. There is the potential for metal release and turbine
foreign object damage as a result. Figure 7 shows one such incident that was caught through on-
site monitoring before downstream damage occurred. Careful monitoring of exhaust gas
temperatures looking for small changes in absolute readings that could indicate the onset of
cracking are therefore potentially very valuable. A system offering 24 hour cover based on a self-
learning algorithm that generates a single alarm from small changes in one of many of the
exhaust gas temperatures to counteract this risk was rapidly developed and has been thoroughly
evaluated in a real operating environment.
8.2 Performance Monitoring
During the acceptance test of an LM6000 CCGT operating in an combined heat and power duty it
was noticed that the unit was operating in T48 (LP turbine inlet temperature) control most of the
time as opposed to P3/T3 (compressor discharge pressure/temperature) control. In addition it was
noted that the power output was much higher than expected. From the manufacturer's operating
characteristics it was expected that the unit would operate in P3/T3 until around 5000 operating
hours when degradation would mean it would not reach the compressor set points and it would
operate permanently in T48 control. With the unit operating in T48 control from new it was
considered that the unit was delivering more power than predicted and effectively 'over-firing'.
192
C583/024/2000 IMechE 2000
Figure 7 Example of combustion liner cracking
This was borne out by eliminating other reasons for possible power increases. The measured
compressor efficiency (~ 86%) was considered normal. The machine was delivering a higher
inlet mass flow but this would be responsible for only 20% of the additional power. In addition,
both the exhaust temperatures and NOx values were higher than design indicating over-firing.
The life of turbine components is directly related to the turbine firing temperature. The basic
components affected are the 1
st
stage rotor blade and stator vanes with particular concern for the
coating of these components. Subsequently, the manufacturer accepted that the unit was over-
firing and advised that the operating set points be revised and the exhaust thermocouple harness
inspected.
8.3 Pyrometry
The leading edge temperature of a row of 1
st
stage blades was found to be trending upwards over
a period of weeks with no other obvious changes in operating conditions. The cause was thought
likely to be coating degradation, a fact that was confirmed on off-line boroscope inspection.
Blade life management calculations determined that had this been allowed to progress until the
scheduled outage the parts would have been scrap and might ultimately have failed in service due
to cracking. An early hot gas path inspection was therefore recommended.
Reducing operating costs safely is a key goal on all stations. By extending the life of the limiting
1
st
stage blades maintenance intervals can be increased and costs reduced. Pyrometry is being
used to ensure that this extension is within the bounds of safe degradation in the part.
C583/024/2000 IMechE 2000 193
9 FURTHER DEVELOPMENTS
Further developments are continuously being evaluated with the aim of improving the level of
on-line monitoring services and feedback provided to station teams of the plant health and
condition to aid in management of unit operation.
One such is Exhaust Debris Monitoring System (EDMS). Here a sensor detects the presence of
particles in the exhaust gas due to their electrostatic charge. With time a picture of normal
operation can be built up that allows abnormal conditions to be detected. A trial installation is
currently in place on one of PowerGen's CCGTs.
The ultimate goal of any health monitoring system is to provide the feedback and alarms directly
to the relevant personnel at the right time to minimise plant outages. This may be the unit
operator or specialist support and through experience and continuous development of the overall
on-line and off-line engine health strategy decision making can be facilitated and costs kept
down.
10 SUMMARY
Historically, the health monitoring of rotating machines has been undertaken using vibration
analysis. With the advent of low cost, high-powered computers, signal conditioning systems and
new technologies, the opportunity exists to gather a wider range of data and provide a more
comprehensive indication of the overall health of the machine.
The health monitoring systems that are in place on PowerGen CCGT units and are monitored
remotely from Power Technology have been described. They include vibration analysis,
combustion system monitoring, pyrometry and on-line performance trending and are
supplemented by inspections and life management programmes. To illustrate the practice of the
monitoring regimes a number of case histories have been reviewed.
The results from these data are fed back in a variety of ways to station personnel to aid in the
operational and maintenance decision making process through activities ranging from immediate
risk management on a 24-hour basis to longer term outage planning.
REFERENCES
[1] G HEWITT PowerGen - Case study: Remote monitoring of a fleet of large Turbine
Generators and Feed Pumps - A.I.M. conference October 1997
[2] Dynamic pressure measurement tests by KEMA at EemsCentrale Power Station
Netherlands - A Verhage May 1999 - private visit communication
194
C583/024/2000 IMechE 2000
C583/034/2000
Condition monitoring of cooling tower fan
gearboxes
P J DONNELLAN
National Power PLC, Swindon, UK
ABSTRACT
To reduce visual and environmental impact, modern power stations are built with induced
draft cooling towers replacing the large natural draft cooling towers previously used. The
ability of the cooling towers to provide adequate cooling for the main circulating water of the
power station is affected by the availability of the cooling tower fans.
A key component of the cooling tower fan is the gearbox. Maintenance of gearboxes is
difficult and expensive, given the location and therefore condition monitoring techniques have
been applied to detect common failure mechanisms. Future condition monitoring may be
possible using low cost on-line techniques.
1 INTRODUCTION
1.1 Cooling tower fan operation
Station main circulating water (MCW) provides the primary cooling for the station. Before
being returned to the station by the MCW pumps, the water is cooled. As it flows over the
baffles in the cooling tower, some water evaporates to provide the cooling. The upward
airflow is provided not by natural draft but by horizontally mounted fans, arranged in cells. A
500MW Combined Cycle Gas Turbine (CCGT) station typically has 14 cells.
1.2 Cooling tower fan gearbox location
The cooling tower fan gearbox is located inside the cooling tower cell, the vertical output
shaft below the fan. A typical fan has 7 blades each of 3m radius. A horizontally mounted 3-
phase induction motor, located on the top level of the cooling tower drives the gearbox via an
input coupling, which extends outside the cell. In some examples, the motor and gearbox have
a common mounting frame.
C583/034/2000 IMechE 2000
195
1.3 Cooling tower fan gearbox construction
The typical speed reduction is from 1490rpm - 990rpm (from a 4-pole or 6-pole motor) down
to lOOrpm - 50rprn, The input gearmesh is a 90 bevel and there are either 2 or 3 single helical
reduction gearmeshes on vertically mounted shafts. An integral oil pump, driven from an
intermediate shaft circulates lubricating oil. Rolling element bearings are used throughout the
gearbox.
Some designs have an axial cooling fan on the input shaft to force air over the casing of the
gearbox. Another design feature on some gearboxes is a device to stop reverse rotation of the
fan when not being driven, which may occur when the upward airflow from operational fans
then returns down through the "idle" cell. Figure 1 shows a typical gearbox layout.
Figure 1: Typical cooling tower fan gearbox layout
2 OBJECTIVES OF CONDITION MONITORING
The objectives of condition monitoring are as follows:
To collect the minimum information needed to plan predictive maintenance and to avoid
high cost breakdown maintenance.
To assist in identifying appropriate actions when alarm levels are exceeded or significant
changes are detected.
To maximise proactive maintenance by the detection of faults in plant installation or
reinstallation which, if left uncorrected, would lead to plant damage.
In many cases, a combination of techniques is required to fulfil these objectives.
3 CONDITION MONITORING TECHNIQUES IN USE
3.1 Hand held vibration monitoring
The majority of auxiliary plant on a power station is periodically monitored using hand held
data collectors, collecting vibration data using a portable accelerometer. The data is then
loaded into a software package on a PC in order to analyse the data.
196
C583/034/2000 IMechE 2000
Access to the cooling tower fan gearbox is prohibited during operation, so the only data
available to the operator is from the driving motor. Experience has shown that the timely
detection of gearbox faults from analysis of motor vibration readings alone is difficult, but
readings at the drive end of the motor can be analysed to provide corroborating evidence.
3.2 Location of permanently installed accelerometers on gearboxes
The attenuation of gearbox vibration signals (e.g. from a damaged rolling element bearing)
through the gearbox, input shaft and motor housing to an accelerometer placed on the motor
means that an accelerometer must be placed on the gearbox to obtain adequate information.
In most cases, accelerometers have been mounted on the gearboxes radially in line with the
top output shaft (i.e. lowest speed) bearing. This location was initially selected due to
concerns over failures of the top output shaft bearing, but experience has shown that this
location is also suitable for monitoring of the higher speed input and intermediate shafts. The
vibration signals from these shafts are higher in magnitude that from the low speed output
shaft and, even with attenuation through the gearbox, are detectable at the output shaft
accelerometer. Conversely, accelerometers mounted at the bearing locations of the input shaft
are not able to effectively monitor output shaft vibration signals.
3.3 Velocity vibration measurements
Vibration of auxiliary plant has traditionally been measured in units of velocity (mm/s RMS),
which gives an indication of the condition of the plant relatively insensitive to speed (1).
Initially, velocity readings on gearboxes were set up to detect 2 failure mechanisms:
Fan and installation defects (e.g. unbalance, misalignment).
Defects appearing in the vibration spectrum at gearmesh frequencies.
2.3.1 Fan and installation defects
Fan and installation defects (e.g. unbalance, misalignment) appear at low frequencies, usually
less at less than 10 orders (multiples) of running speed. For the case of the output shaft, the 1
st
order frequency can be less then IHz. Detection of the low frequency defects is difficult,
firstly due to the poor response at low frequencies of standard accelerometers and secondly
because there is significant low frequency vibration of the cooling tower structure, saturating
the accelerometer input to the data collector.
2.3.2 Defects appearing at gearmesh frequencies
The maximum frequency of data collected is above the 3
rd
harmonic of the gearmesh
frequency (the product of the shaft speed and the number of teeth on the gear). In all cases of
cooling tower fan gearboxes, the highest frequency of interest is less than 2000Hz, so high
frequency accelerometers are not needed and data is collected in units of velocity mm/s RMS.
3.4 Enveloped Signal measurements
Enveloped Signal Processing (2) is a well proven technique for the detection of defects in
rolling element bearings. Despite the location of a single accelerometer at the output shaft top
bearing location, it was hoped that rolling element bearing defects on other shafts within the
gearbox could be detected. The envelope most commonly used is 5kHz - 10kHz and the
maximum frequency of interest is 20 orders of input shaft speed. The data is collected in units
of acceleration g's Peak.
C583/034/2000 IMechE 2000
197
3.5 Motor current analysis
Motor stator current analysis (3) has traditionally been used to monitor the condition of
induction motor rotors and the sensitivity of the technique to load fluctuations is known. A
variation on this technique has been used, analysing the time waveform of the stator current
over a short period (e.g. 10s). A normal cooling tower fan gearbox motor current has little
variation of current over time but a gearbox or fan with a serious defect (e.g. sheared gearbox
holding down bolts, unbalanced fan blades) shows modulation of motor current at output shaft
frequency.
3.6 Oil Analysis
Oil analysis has been a prime condition monitoring technique for gearboxes, often able to
detect gearbox wear before vibration analysis (4). The difficulty in applying this technique to
cooling tower fan gearboxes has been the collection of a representative, repetitive sample.
One station installed dedicated sample collection and return piping from the gearbox to a
location external to the cooling tower cell. When a sample is needed, a sample pump is
connected to the collection and return piping and, after allowing time to flush the sample
piping, a representative sample can be taken. The disadvantage of this method is that the
small bore piping connected to the gearbox is vulnerable to damage, which presents an
environmental risk from 801 of oil leaking from the gearbox.
The other method in use is to use the oil filler pipe as the connection to the gearbox. The pipe
is flushed into a container until warm oil is detected, so that the sample is reasonably
representative of the oil inside the gearbox. The flushed oil is then put back.
Standard industrial tests are performed on the oil (kinematic viscosity, water content, total
acid number, elemental analysis and particle quantifier).
The results of oil analysis have generally been good but an equally important use of oil
analysis has been the ability to prolong oil life in gearboxes, based on the oil condition. In one
application, the increased cost of using synthetic oil was justified because the frequency of oil
changes was reduced (saving the associated maintenance costs of the oil change as well as the
cost of the oil). Quadrupling of oil life was achieved.
4 COMMON FAILURE MECHANISMS
It is not the intention of this paper to provide a list of the all potential failure mechanisms of
cooling tower fan gearboxes. Some examples of the common failure mechanisms are
included, together with the suggested condition monitoring techniques.
4.1 Gear teeth failures
Some gearboxes experienced single teeth failures on the intermediate shaft bevel gear.
Metallurgical analysis of the damaged teeth suggested a manufacturing defect or a transient
overload on the gear, possibly when the motor was started.
The defect was detected using Enveloped Signal Processing of the signal from the
accelerometer mounted at the output shaft. Figure 2 shows the envelope spectrum, with
harmonics of intermediate shaft speed indicating the repetitive event of a single tooth defect.
198
C583/034/2000 IMechE 2000
Figure 2: Envelope spectrum of gearbox output shaft accelerometer
Figure 3: Velocity spectrum of gearbox output shaft accelerometer
C583/034/2000 IMechE 2000 199
The output shaft accelerometer signal is integrated to give a velocity reading. Figure 3 shows
the velocity spectrum, with the 2" harmonic of the input gearmesh frequency modulated by
intermediate shaft speed, giving the characteristic "sidebands". The precise nature of the
defect cannot be determined from this spectrum, but the change from a "healthy" reading
would lead to a maintenance intervention.
Figure 4 shows a typical gear with the damaged tooth clearly visible.
Figure 4: Damaged tooth on intermediate shaft bevel gear
4.2 Input and intermediate shaft bearing failure
Possible causes of bearing failure are solid or water ingress via a shaft seal, poor lubrication
or contamination of oil.
A gearbox was returned to service after repairs to gear teeth but failed soon after being
installed. Using Enveloped Signal Processing an input shaft bearing defect was detected using
the accelerometer mounted at the output shaft. Figure 5 shows the envelope spectrum, with
harmonics of motor / input shaft speed and sidebands indicating impacting, typical of a rolling
element bearing defect. In this case a more detailed analysis of the spectrum was needed,
shown in Figure 6, to isolate the individual bearing component, but a defect is noticeable,
even without such analysis.
Figure 5: Envelope spectrum with harmonics of motor / input shaft speed and sidebands
200
C583/034/2000 IMechE 2000
Figure 6: Detailed envelope spectrum showing gearbox bearing defect frequency
4.3 Output shaft bearing failure
A number of different failure mechanisms are responsible for output shaft bearing failures.
Some designs of gearbox are prone to oil starvation of the top output shaft bearing when the
fan rotates in reverse. This has been addressed by the fitting of anti-reverse rotation devices.
If the fans are left stationary false brinnelling damage can take place. Over long periods (>3
weeks) the thickness of the oil film reduces to the point where fretting corrosion can occur if
water ingress occurs via the shaft seal. Such damage can occur during commissioning, when
some fans run but others are left stationary for long periods.
Figure 7 shows an output shaft top bearing removed from a gearbox, with markings on the
bearing outer race showing where the bearing had remained stationary. Enveloped Signal
Processing is recommended for detection of these bearing failures.
Figure 7: Output shaft top bearing with markings on the bearing outer race
C583/034/2000 IMechE 2000 201
4.4 Gearbox overheating
At one station, over half the gearboxes experienced failures of the output shaft bottom
bearing. One of the root causes of failure was oxidation of the oil at the elevated temperatures
at which the gearbox was running. The gearbox did not have a cooling fan mounted on the
input shaft and the result was a "dead space" in the airflow around the gearbox. Oil analysis
detected the oxidation and the grade of oil was changed from ISO VG220 to ISO VG320.
4.5 Fan unbalance
Waterlogging of fan blades can cause fan unbalance. Porous GRP material delaminates and
small pieces break off inside the hollow fan blade, lodging at the end of the blades and
blocking the water drain holes.
With no accelerometer installed on the gearbox, vibration on the driving motor was monitored
using a hand held accelerometer. A harmonic of the fan speed triggered an alarm on the motor
drive end bearing velocity spectrum reading. Figure 8 shows the motor drive end velocity
spectrum, with the harmonic of fan speed labelled. The fundamental fan frequency is below
the high pass filter used for the measurement, and therefore does not show on the spectrum.
Figure 8: Motor drive end velocity spectrum showing harmonic of fan speed
4.6 Unusual failure mechanisms in temperature extremes
At one station in the US, oil analysis detected the production of wear metals from gearbox
teeth. Subsequently, high levels of vibration were detected on the drive end bearing of the
driving motor. The gearbox was stripped down and severe adhesive wear was found on the
back of the output shaft gear teeth.
During normal operation, the gearbox temperatures were over 90C and so ISO VG680 grade
oil was used. A detailed investigation revealed that, while the cooling tower fans run forwards
202 C583/034/2000 IMechE 2000
in normal operation, in extremely cold weather (-10C), the fans are run backwards at half
speed in order to de-ice the cooling tower after a short period off-load. At this temperature,
the kinematic viscosity of the oil is over 20000 centiStokes (cSt) i.e. near its pour point (the
temperature at which it stops flowing). The splash lubrication regime for the gear teeth did not
work resulting in severe adhesive wear.
5 FUTURE CONDITION MONITORING
5.1 Online vibration analysis
The advantage of using hand held data collectors for vibration monitoring is that it provides
cost effective condition monitoring for the majority of auxiliary plant. On power stations, the
periodic (e.g. weekly or monthly) vibration survey is backed up by the station operators
during routine or ad hoc plant visits. A "noisy" bearing or change in vibration is logged and
follow up instigated. With cooling tower fan gearboxes, as with other inaccessible plant,
operator plant visits are less likely to be able to detect impending failures.
With the reduction in cost of Vibration Isolated Measurement Pod (VIMP) type technologies,
on-line monitoring of cooling tower fan gearboxes is becoming cost effective. A 16-channel
VIMP type unit wired up to accelerometers on gearboxes can monitor all the gearboxes on a
500MW CCGT station. The VIMP type unit is connected to an on-line version of the normal
off-line vibration analysis software and the data displayed in the plant control room.
The correct setting of alarm levels is important. While an experienced vibration analyst
looking at monthly collected data can disregard readings causing "false" alarms, the control
room operator may be overloaded with alarms in the on-line system. The development of
multivariable alarms in software packages may help solve this problem.
Laying communications cabling on an existing installation represents the biggest problem
(and cost) of an on-line system. A number of condition monitoring companies are developing
radio link technology to solve this problem.
5.2 Online oil condition monitoring
As previously discussed, oil analysis is a potentially powerful diagnostic tool in gearbox
condition monitoring, but can be limited by the ability to collect a sample. A recently
developed product is on trial (6), which provides a continuous indication of oil condition.
The principle of operation is that the loss tangent (Tan6) of the dielectric constant (e) of oil is
affected by oxidation degradation, liquid or solid contamination and wear metal production.
The loss tangent is related to the dielectric constant (or permittivity) as follows:
Where e' is the real component of the dielectric constant.
A low voltage dc powered sensor is located in the oil flow and gives a voltage output and an
alarm output when a threshold is exceeded. If successful, the units could be wired into the
station distributed control system to prompt action once an alarm is triggered.
C583/034/2000 IMechE 2000
203
6 CONCLUSIONS
With a combination of techniques, correct setting of alarm levels and interpretation of data,
gearbox failures can be detected, even if it is not possible to distinguish the individual
components that are failing.
A combination of enveloped signal processing and high resolution velocity spectra are
suitable techniques for detecting gear tooth failures, using vibration monitoring.
7 REFERENCES
(1) A.R.Crawford and S.Crawford "The Simplified Handbook of Vibration Analysis
Volume 2", 1992, Computational Systems Inc.
(2) B.K.N.Rao "Handbook of Condition Monitoring", 1996, Elsevier Advanced
Technology
(3) C.Hargis et al "The Detection of Rotor Defects in Induction Motors", Proc. IEE 213
Electrical Machines Design and Applications Conference, London, July 1992
(4) A.Davies (ed.) "Handbook of Condition Monitoring", 1998, Chapman & Hall
(5) CJ.Collister and P.Weismann "Onboard Oil Condition Sensing" Proc. 12th Int.
Colloquium, Tribo2000, Esslingen, January 2000
204
C583/034/2000 IMechE 2000
C583/035/2000
Permanent partial discharge assessment of power
station high-voltage switchgear
P M BROWN
EA Technology, Chester, UK
MC JONES
Edison Mission Energy (First Hydro), Dinorwig Power Station, Llanberis, UK
SYNOPSIS
Non-intrusive partial discharge monitoring has been installed on two 16kV switchboards at
the Ffestiniog hydro-electric pumped storage power station. The circuit breakers control the
generators/ motors each operating about 3 times per day and therefore require a very high
reliability and availability. The permanent monitoring is intended to provide warning of
impending insulation failures and to assist with the management of the assets.
The monitor has to be capable of detecting a partial discharge signal of sufficient level to
signify possible failure occurring at any time at any of the monitored coupling devices. At the
same time the monitor has to be able to discriminate against partial discharge activity from
the generator/motors and the 275kV overhead lines.
1 INTRODUCTION
Non-intrusive partial discharge detection methods have been applied to high voltage
switchgear since 1982 (1 and 2). The methods have resulted in a range of instruments from
various manufacturers. The instruments may be classed into two categories, electromagnetic
or ultrasonic detectors. The latter mainly involves detection of airborne noise at 40kHz and is
applicable to surface partial discharge activity in air insulated chambers with a clear passage
from the source to the ultrasonic microphone.
The detection of the electromagnetic emission from partial discharge activity within high
voltage switchgear is normally via probes or aerials placed on or around the switchgear. This
technique will detect partial discharge activity in any media provided there is a discontinuity
in the metal cladding of the switchgear. The discontinuity may consist of gaskets, through
bushings or openings due to the irregularities in metal covers, ventilators or direct passage
from circuit breaker bushings or the fixed portion isolating spouts.
C583/035/2000 IMechE 2000 205
At the Ffestiniog hydro-electric pumped storage power station the two main switchboards
contain GEC type KDH 35 oil filled switchgear installed in 1960/61. The switchgear is rated
at 24kV but operates at 16kV. The generator switchgear has a breaking capacity of
1500MVA and a current rating of 3500A. As the circuit breakers, busbars, VTs and cable
boxes are all oil filled the electromagnetic detection technique is the most appropriate non-
intrusive technique available.
2 REQUIREMENT FOR CONTINUAL PARTIAL DISCHARGE MONITORING
First Hydro had for a number of years considered the requirement to replace the 16kV OCB's
at Ffestiniog Power Station to maintain high availability and reliability, virtually 100% at
current levels. From a cost benefit point of view the decision to replace the equipment
appeared marginal. Therefore the decision was made to procure equipment which would
provide data to either support or defer replacement of the switchgear. The equipment was also
to assist with early detection of impending failures, through integration into the existing
Power Station alarm system. This would give the opportunity to repair/replace equipment
suspected of nearing failure, during planned outages rather than unplanned forced outages.
Due to the continual, 365 day utilisation of the plant, it was of up-most importance for the
installation to be a real time, on line, partial discharge condition monitoring facility for both
of the Power Station 16kV OCB switchboards.
3 GENERAL OUTLINE OF FIRST HYDRO SPECIFICATION FOR
PERMANENT PD MONITOR
All equipment, for a complete installation including all necessary enclosures, mounting
accessories, connectors, flexible conduit, computers and peripherals, software, modems, and
workstation were to be provided by the supplier (EA Technology).
The system had to monitor two installations of metalclad switchgear with coupling devices
monitoring continuously and to operate unattended.
The system had to be capable of detecting a partial discharge signal of sufficient level to
signify possible failure occurring at any time at any of the monitored coupling devices.
The installed system had to generate separate alarms to indicate the following conditions:-
Detection of signals indicating high partial discharge
A failure of any part of the system which prevents the capture or
transmission of data.
Alarm thresholds had to be presettable by the operator (First Hydro) and protected by
password.
All the above facilities were to be accessible by modem link from computers at remote
locations.
206
C583/035/2000 IMechE 2000
4 DESCRIPTION OF PD SYSTEM
The main difficulty in carrying out non-intrusive partial discharge measurements on the
switchgear at Ffestiniog is the number of other partial discharge sources that surround the
switchgear.
The switchboards are on the outside of the generator hall, which contains four 90MW
generators/ 75MW motors. The generator/motors are connected to the switchgear via 16kV
cables. Three single core, paper lead insulated, l.Ssq.inch conductors per phase. Adjacent to
each switchroom is a Super Grid Transformer compound containing a 16/290kV 190MVA
transformer with a 275kV tower and overhead lines.
In addition there are common paper insulated, oil impregnated, busbars from the switchgear
towards the transformers which terminate in a through wall bushing out into the outside
switchyard on solid copper open busbars to the secondary terminals of the Super Grid
Transformer. Finally there are overhead lines and insulators within a few metres of the
switchroom roof.
Therefore, on one side of each switchroom there are generators which when they are
operating have a partial discharge level of the order of 60,000pC. On the other side there are
275kV bushings and overhead lines which in wet conditions also discharge at a lower level
but closer to the switchgear.
4.1 Partial Discharge Monitor
The Partial Discharge Monitor detects the electromagnetic signals with non-active capacitive
probes that are magnetically clamped onto different items on the switchboard. The probes
detect the transient earth voltage as the electromagnetic wave from the internal partial
discharge escapes through openings in the switchgear and impinge on to the surface of the
switchgear. However, the probes will also detect electromagnetic signals from external
sources as they impinge onto the surface of the switchgear. Therefore, in addition to eight
probes the monitor has four aerials that are positioned around the switchgear.
The monitor can discriminate to within 7ns the arrival time of pulses detected on any of the
twelve channels. Pulses that arrive within 200ns are treated as the same event and a pulse is
only allocated to the channel or channels at which the pulse arrived first within the resolution
of the instrument. Therefore, if electromagnetic signals from an external source are detected
by an aerial first then only that aerial channel has a pulse allocated to it and the pulses
detected by all the other aerials and probes are ignored. In this manner all the pulses detected
by the monitor are allocated to the channels at which they arrived first providing a location
function.
The three panel switchboards are very large being 3.8m high, 7m long and 3.6m deep.
C583/035/2000 IMechE 2000
207
The probe layout is as follows:
Fig. 1 Front view of switchroom
Fig. 2 Plan view of switchroom
Key to Figs. 1 and 2
208
C583/035/2000 IMechE 2000
There is one Partial Discharge Monitor per switchroom as shown below:
Fig. 3 Layout of monitoring equipment
The monitors, data storage and communication equipment are installed in cubicles, (see Figs.
3 and 4)
Fig. 4 Cubicle for the South Switchroom
C583/035/2000 IMechE 2000
209
The alarm operates on the short term severity values of channels 3 to 10 which are connected
to the probes. The value may be altered using the set up menu of each monitor and initially
the alarm level was set to a short term severity of 100.
Short Term Severity = Maximum pulse amplitude x Number of pulses per cycle during a 5
minute monitoring period.
The short term severity relates to the potential damage caused by the discharge activity.
5 EARLY RESULTS
The monitors were installed on the 29 February 2000 at Ffestiniog, but with the switchgear
operational it was not possible to install probes on the circuit breakers or busbars. Normally
probes on the circuit breakers would be clamped onto the circuit breaker tank, however, due
to the large number of operations that these circuit breakers perform they are serviced every 2
months. Therefore, it was considered to be inappropriate to install probes where they would
interfere with frequent servicing. The proposed probe positions above the circuit breakers can
only be reached during an outage and have yet to be fitted.
The early results did not have probes 3, 4, 7, 8 and 9 in position on either switchboard. The
only probes on the switchgear were 5, 6 and 10 as shown in Fig. 2. Probes 5 and 10 being on
the cable boxes from the generators and probe 6 being on the busbar VT. The four aerials
were in position as shown in Fig. 2.
The North switchroom results for March are as follows:
Date 29/02/2000
Time 17:05:38
Location FFESTINIOG
Substation NORTH 16KV OCB RM
Plant GEC KDH 35
Company FIRST HYDRO
Installer PDM37 3 4 7 8 9 >
Integration Period 5 mins
Operation Mode Location
Alarm Level 100
PDM Version 6
Start time 01/03/2000 00:04:00
Finish time 31/03/2000 23:59:00
C583/035/2000 IMechE 2000
210
Total number of pulses = 91852586
Total number of sets of data = 8928
12 channels connected, 14400 cycles per 5 minutes
In the summary table above for the North switchboard only probes 6 and 10, highlighted,
have a short term severity >0 and are therefore considered as possibly detecting partial
discharge activity. Fig. 5 shows the short term severity of pulses allocated as arriving first at
these channels only exceeds the alarm level of 100 for two five minute periods for each
channel.
Detailed analysis of the data shows that over the month there were external electromagnetic
sources allocated to the aerials and two internal sources allocated to probes 6 and 10.
However, the pulses allocated to the probes 6 and 10 may have come from other parts of the
switchgear where probes have not yet been installed. Therefore, further analysis awaits the
installation of the other probes.
The South switchroom results for March are as follows:
Date 29/02/2000
Time 15:40:25
Location FFESTINIOG
Substation SOUTH 16KV OCB RM
Plant GEC KDH35
Company FIRST HYDRO
Installer PDM36 3 4 5 1 8 9>
Integration Period 5 mins
Operation Mode Location
Alarm Level 100
PDM Version 6
Start time 01/03/2000 00:03:00
Finish time 31/03/2000 23:58:00
C583/035/2000 IMechE 2000 211
Ch Max
Level
Nos of
Pulses
Av
Level
per cycle
1
2
3
4
5
6
7
8
9
10
11
12
34
52
0
0
40
52
0
0
0
40
46
28
0.
0.
0.
0.
0.
0
0.
0.
0.
9.
0.
0.
.002
.300
.000
.000
002
.604
.000
.000
.000
642
107
009
7
19
0
0
1
17
0
0
0
15
16
4
Short
Term
Nos of %
Pulses Pulses
Severity
0
119
0
0
0
241
0
0
0
964
21
0
1743831
34154878
2
1
47669
39688689
0
1
1
10701138
8173080
636186
2
37
0
0
0
43
0
0
0
12
9
1
Max
Pulses
Assoc
Level
%
Time
per cycle
0.
4.
0.
0.
0.
1.
0.
0.
0.
9.
4.
0.
642
049
000
000
032
.778
000
000
000
642
505
421
25
25
0
0
13
16
0
0
0
40
25
16
96
99
0
0
45
100
0
0
0
98
97
92
Severity
Long Max
Term Short
0
2
0
0
0
2
0
0
0
0
0
0
11
273
0
0
0
241
0
0
0
964
196
3
Ch Max
Level
Nos of
Pulses
Av
Level
per cycle
1
2
3
4
5
6
7
8
9
10
11
12
25
52
0
19
52
52
0
19
16
52
40
40
0
0
0
0
0
0
0
0
0
0
0
0
.314
.306
.000
.000
.259
.010
.000
.001
.003
.209
.268
.001
13
20
0
0
5
15
0
0
0
15
16
4
Short
Term
Nos of %
Pulses Pulses
Severity
6
122
0
0
103
4
0
0
0
83
27
0
404085
36083802
91
5
388479
5492021
41
30
90
3306769
44947821
234139
0
44
0
0
0
7
0
0
0
4
54
0
Max
Pulses
Assoc
Level
%
Time
per cycle
0.
1
0.
0.
0.
1
0.
0.
0.
0
1
0.
314
.600
.003
000
259
.139
,002
.001
.003
.672
.726
.115
25
19
0
19
52
25
0
0
16
25
19
19
99
100
0
0
60
100
0
0
0
100
100
71
Severity
Long Max
Term
0
3
0
0
0
0
0
0
0
0
2
0
Short
6
122
0
0
103
23
0
0
0
83
27
1
Total number of pulses = 82637227
Total number of sets of data = 8928
12 channels connected, 14400 cycles per 5 minutes
In the summary table above for the South switchboard only probes 5, 6 and 10, highlighted,
have a short term severity >0 and are therefore considered as possibly detecting partial
discharge activity. Fig. 6 shows the short term severity of pulses allocated as arriving first at
these channels. The short term severity only exceeds the alarm level of 100 for channel 5
only and then only for one five minute period.
Comparing the summary tables and Figs. 5 and 6 shows that many fewer pulses were
allocated to the probes on the South switchboard compared to the probes in the same positions
on the North switchboard.
Detailed analysis of the data collected in March from the South switchroom shows that the
pulses allocated to the aerials dominated the small number of pulses allocated to the probes
and that all the significant sources were external to the switchgear.
6 CONCLUSIONS
We are still at the learning stage when considering the data from the permanent non-intrusive
Partial Discharge Monitors installed in the two switchrooms at the Ffestiniog Hydro Electric
power station. When all the probes are finally installed a clearer picture may emerge.
However, the results to date do indicate that the technique is capable of detecting internal
partial discharge activity in the very harsh environment found with the switchgear surrounded
by other sources of high partial discharge activity.
Although the monitor does not record load or other operational data the time stamping of the
data enables comparison between partial discharge activity and operational conditions.
212
C583/035/2000 IMechE 2000
7 ACKNOWLEDGEMENT
This work is published with the permission of Mr D Frankland, Plant Director, Edison
Mission Energy, First Hydro Company.
8 REFERENCES
1. Brown, P.M. "Non-intrusive partial discharge measurements on high voltage switchgear"
IEE Colloquium on "Monitors and condition assessment equipment", 5th December 1996,
Leatherhead, Surrey
2. Brown, P.M. "Operational experience of non-intrusive partial discharge measurements on
high voltage switchgear" INSUCON/ISOTEC 1998 The 8
th
BE AM A International
Electrical Insulation Conference
C583/035/2000 IMechE 2000
213
Fig. 5 Short Term Severity of pulses allocated to channels 6 and 10 in the North
Switch room
Fig. 6 Short Term Severity of pulses allocated to channels 5, 6 and 10 in the South
Switch room
214
C583/035/2000 IMechE 2000
Sweating the Asset - II
This page intentionally left blank
C583/032/2000
An enterprise wide implementation of a
maintenance/materials management systems
O VALSA
CEZ, as, Prague, Czech Republic
S P JOHNSON
Indus International, Chertsey, UK
SYNOPSIS
Enterprise-wide maintenance/materials management systems represent the core information
systems for utilities companies. With information about capital assets, maintenance work
orders, materials inventory and procurement, health and safety, and cost reporting, they touch
nearly every key business processes in the utility. The ability to effectively realize the potential
of a maintenance/materials management system throughout the enterprise is clearly dependent
upon effective system implementation.
In 1995 CEZ, a. s., the Czech Republic Power Generation Company embarked on an Enter-
prise Wide Maintenance and Materials management system implementation - the ISE Project.
This project is based on the Indus International Passport Maintenance and Materials Manage-
ment System and its implementation required the use of a well prepared methodology based on
the extensive experience gained within number of previous projects of the similar kind.
This paper details the implementation methodology adopted by CEZ and Indus during this
4 year project. Details are given of the experience of both end user and vendor gained during
the implementation of a fully translated generic Maintenance and Materials Management
System into a large multi site Power Generation Company.
C583/032/2000 IMechE 2000
217
1. INTRODUCTION
Enterprise-wide maintenance/materials management systems represent the core information
systems for utilities companies. With information about capital assets, maintenance work
orders, materials inventory and procurement, health and safety, and cost reporting, they touch
nearly every key business process in the utility. The ability to effectively realize the potential of
a maintenance/materials management system throughout the enterprise is clearly dependent
upon effective system implementation.
In 1995 CEZ, a. s., the Czech Republic Power Generation Company embarked on an Enter-
prise Wide Maintenance and Materials management system implementation - the ISE Project.
CEZ senior management recognized that the unified information system is necessary for an
effective assets management. This information system should serve as the real repository for
knowledge. The better the knowledge, the better the system; and the smarter as a whole your
organization can be.
All the technology in the world does not help an organization achieve its objectives unless the
tools rapidly reach the desktops and work sites of those who need them. The advantages of an
enterprise-wide strategy are clear: reducing the time and money spent trying to get different
systems to work together (and fix them when they break or get upgraded), greater efficiency,
informed decision making, and distribution of best business practices. However, an organiza-
tion must make an enterprise-wide commitment in order to successfully implement systems
across the organization.
2. THE CHANGING LANDSCAPE
In the past, large in-house IT organizations were responsible for integrating and operating a
variety of software systems supplied by several software vendors along with home-grown
legacy systems. These systems were poorly documented, awkwardly integrated, and expensive
to maintain. Many software providers delivered replacement systems into this environment,
including extensive customization services so that the replacement systems provided essentially
the same business function as the retired legacy systems, thus having little positive impact on
improvement of business processes. Implementation of these systems was largely handled by
the in-house IT organization with little outside assistance.
Increased global competition, deregulation in the utility industry, and a drive to reengineer
business has changed everything, virtually overnight. Reengineering studies have centered
around cost cutting, downsizing, and reorganization, with all strategies and justifications
focused on business improvement.
The impacts on the traditional IT organization have been dramatic - most have been down-
sized such that they can no longer handle implementation of enterprise systems. Furthermore,
the re-engineers have emphasized that companies must focus on core business competencies
and get out of all non-core competency areas such as IT and systems integration. The now
leaner IT organization must rely on software suppliers to provide not only state of the art
software solutions, but also pre-integrated suites of baseline software which require no
customization. Today's IT organizations must rely more on outside software and technology
218
C583/032/2000 IMechE 2000
suppliers to provide both initial technical support during implementation of enterprise-wide
systems and ongoing technical support throughout the software system life cycle.
The related impact of reengineering on the business side is that selected baseline software
solutions must represent industry 'best practice' and business processes must be modified to
adopt those built-in practices. Success in implementation of these systems requires that a
comprehensive implementation methodology is used to identify and implement the required
business changes.
3. CEZISE PROJECT PRINCIPLES
3.1 Scope of the Proj ect
The main goal of the ISE Project was to implement a unified system to manage maintenance
work and other supporting activities at all thirteen power plants of CEZ including the
necessary IT support. This ISE System should be based on minimal customization of a
commercially available software package. As part of the implementation analysis a process
reengineering of all processes covered by the ISE System should be provided and all the ISE
System functionality should be analyzed in detail. All necessary modifications both of the
application and of the business processes should be identified well in advance. Modifications of
the software package should be accepted only when necessary to meet the existing legislation.
ISE System should be implemented in the unified manner at all thirteen CEZ's plants including
code tables unification. The migration of data from the existing CEZ's legacy systems into the
newly implemented ones, the end user training and assistance are also within the scope of ISE
Project.
Such an implementation inevitably results in change of the work culture and affects almost all
personnel working in the CEZ's plants. By maximizing the users involvement during each step
of the ISE Project the developed solution is consequently accepted by users as their own.
3.2 Covered Functionality
The ISE System covers the following functional areas:
Maintenance Management (Work Mgmt, Tag Out)
Materials Management (Inventory, Purchasing, Contract Mgmt)
Document Administration (Document Mgmt, Engineering Change Control)
Operational Support (Ops Log, Dozimetry, Action Tracking, Personnel Qualif)
The Indus International Passport Maintenance and Materials Management System has been
selected as the core of ISE System.
C583/032/2000 IMechE 2000
219
4. AN IMPLEMENTATION METHODOLOGY
4.1 General Principles
The CEZ ISE Project was designed to implement business process improvement and new
technologies simultaneously across a large/complex business enterprise, involving several
business units and thousands of employees. Success in such a complex project depends upon a
clear, proven methodology that addresses the entire implementation project lifecycle of tasks
and responsibilities. An exhaustive Implementation Plan must include all activities, regardless
of who will perform those activities (customer, vendor or third parties). Since the
implementation methodology must cover all aspects of implementation, it is the first priority
and is used as the 'checklist for success.'
Indus within the last few years have been converting their experiences of Enterprise Asset
Management System Implementation into a formal methodology. This methodology has been
called ABACUS and is Indus International's set of tools and implementation methodology
developed for enterprise-wide maintenance/materials management systems. It consists of
software-driven analytical tools, implementation plans, and educational tools that encapsulate
Indus's extensive experience in implementing enterprise management software solutions.
The CEZ ISE Project followed many of the principles laid out in this methodology. Further in
this paper the methodology is presented and the hands on experience of CEZ is discussed.
4.2 Structure of the Methodology
The methodology provides planning tools and templates which are pre-loaded with all planning
details and relationships already established. All the planning team has to do is determine who
will perform each activity, and perform resource leveling, and planning is essentially complete.
This method of leveraging existing tools and templates saves significant time as compared to
starting the planning of a complex enterprise system implementation and business review from
scratch.
Fast-Track execution of major activities simultaneously provides for management, business and
technical activities to progress in parallel. Integrated planning and control offered through
ABACUS, all activities are coordinated to meet mutual milestones with little or no dead or
wait time. Several teams of specialists can work on their focus tasks while ensuring that all
activities are done efficiently and for the common good. Traditional implementations have the
business review completed before any technical tasks are initiated, which has the effect forcing
these tasks to take place in series and adding significant time to the overall implementation
schedule - delaying Return on Investment.
220
C583/032/2000 IMechE 2000
4.2.1 Organization & Terminology
The ABACUS has been built to provide an enterprise-wide methodology which addresses
implementation activities from both a technical and a business process improvement perspec-
tive. This objective necessitates a 'bird's-eye' view of the complete activity matrix with the
ability to drill down on specific tasks to gain depth where it is needed, yet provide a global
context for all team members. The methodology matrix is organized into Tracks, which iden-
tify critical roles across the organization, and Phases, which facilitate the proper timing of
activities throughout the enterprise as the implementation progresses.
4.2.2 Tracks
The methodology tracks assist area managers in focusing on their tasks and in understanding
the relationship of their team's activities to the overall implementation. This approach provides
for several specialized teams working in parallel with common goals and end-points resulting in
a fast-track implementation.
The ABACUS methodology is divided into the following tracks:
a) Management - The Management track defines the scope of the solution, helps ensure that
strategic goals are addressed and ensures continuous executive sponsorship. A team of
implementation managers is responsible for the overall implementation success, including
strict accountability for the budget and schedule and all communications with company
senior management, the implementation team and members of the end-user community.
b) Technical - The Technical track validates the overall information strategy, defines the
technology upgrade path and provides for operational support. A designated technical
team is responsible for technical support of the implementation, including all traditional
Information Technology (IT) and client/server activities. This track is critical as many
companies are implementing client/server technology to support enterprise systems for the
first time.
c) Business Process Improvements - This track employs change management techniques to
reengineer processes within the company and embody best practices in a cost-efficient
manner, improving overall company efficiency. A team of expert business analysts is
responsible for this critical set of activities including the identification and implementation
of business process changes, including how software systems and tools will facilitate those
changes. The goal of this track is to quickly implement positive business change,
supported by the application software, returning cost savings and other benefits to the
organization while minimizing the trauma normally associated with such change. At the
conclusion of this analysis, several follow-on activities are initiated including a change
management program responsible for communications, training and ultimate culture
change throughout the company. Other activities initiated at this time include procedures
review/modification, identification of metrics for measuring implementation benefits and
systems integration and configuration activities.
d) Integration - The Integration track incorporates the use of software tools to rapidly
establish the enterprise-wide resource which provides interoperability between the Indus
Solution Series and other systems. This includes all aspects of systems integration needed
to support the business process improvement program, such as system interfaces, data
C583/032/2000 IMechE 2000
221
collection and application software implementation. This track also provides for quick
identification of the 'current state' of data and initiates a variety of activities for the data
collection, consolidation and clean-up so that the implementation is not delayed and is
ultimately supported by quality data.
e) Education - The Education track ensures user acceptance of both the new procedures and
the application software through a comprehensive education program, computer-based
training tools and courseware tailored to the organization's training environment and
specific educational needs. This track is managed by the training team who are responsible
for the definition of both technical and end-user training requirements. The team also
develops and deploys training tools and programs to support both the system
implementation and the business process improvement program. In this age of installing
packaged software solutions and changing business processes to match best practice rather
than changing software to mirror current work practices, the education track plays a
critical role in the required culture change. Rather than having a training program that
focuses on software training, company employees are trained in the use of software tools
to improve both their individual skills and overall company performance.
f) Implementation Support - This track covers potential outsourcing of either selected or all
IT functions, plus the ongoing Help Desk and on-site assistance services. These functions
are on-going and must be in place during the entire life-cycle of the implementation.
Support personnel are trained to assist users on a day-to-day basis. It is very important
that this support be provided so that users not only utilize the new software tools
efficiently, but also continually learn new methods to use this powerful set of tools to
improve business processes and job performance.
4.2.3 Implementation Phases
The implementation methodology represents the implementation steps in phases so that activi-
ties are scheduled and emphasized in the optimal time frames. It is strongly recommended that
the Implementation Schedule identify all activities in these phases and that pre-production and
production activities be included. Ongoing support activities should be included to identify the
level of commitment required to support the fully integrated system for the entire life time, not
just during the initial implementation and start-up phases.
Implementation phases in the ABACUS matrix are defined as follows:
a) Mobilization & Planning - Covers all activities initiated early in the implementation cycle.
The deliverable for this phase is a comprehensive Implementation Plan supported by a
series of topical or specialty plans. A rapid, cost efficient implementation can only be
achieved through careful planning. On the other hand, the plan should be flexible enough
to absorb some delays which, despite of all careful planning, will happen.
b) Initial Implementation - This is the phase where significant activity commences on all
implementation tracks. The emphasis of this phase is to begin as many coordinated
activities as are appropriate and reasonable to support a fast-track implementation in
accordance with the Implementation Plan.
222
C583/032/2000 IMechE 2000
c) Pre-Production - Covers the activities that are intentionally scheduled to be performed just
prior to system start-up. These activities should occur just-in-time, so that key elements,
such as data and education, are still fresh and ready to support the system in production.
d) Production - Production is the initial period in which the software tools are used in a
production environment. This phase must be carefully controlled and managed so that any
identified problems are addressed early and adjustments are made prior to rolling out the
system to the entire company.
e) Roll-Out - This final phase represents the complete roll-out and operation of enterprise
software tools. In addition to the standing teams providing technical and hands-on user
support, this track also covers on-going business and training reviews. The purpose of
these reviews is to both monitor the effectiveness of the implementation and to identify
additional opportunities for increased efficiency and business process improvement. It is
not enough to get software into production, declare victory and go home. Substantial
improvements in overall operating efficiency can be gained through adherence to the
continuous review and improvement program as described in the methodology.
5. CEZ IMPLEMENTATION EXPERIENCE
CEZ has accepted for the ISE Project a methodology based on the same principles as the
ABACUS methodology. The main areas covered by the implementation methodology and the
lessons learned during this large project are discussed in more detail below.
5.1 Basic Information about CEZ, a. s.
The leading utility company at Czech Republic
Total installed capacity 10,999 MW
Production of 75 % of electricity for Czech Republic
Two nuclear plants (4xWER440, 2xVVER1000-under construction) (1,760 MW)
Ten fossil plants (7,367 MW)
1 hydro complex of 35 units (1,872 MW)
approx. 10 thousand employees
C583/032/2000 IMechE 2000
223
5.2 ISE Project - llands-On Experience
Further the hands-on experience of CEZ with the ISE Project is discussed. As it has been
mentioned above, the methodology of the CEZ's ISE Project followed the same principles as
the ABACUS methodology. Therefore the main parts theoretically covered at the previous
chapter are discussed from the practical point of view here.
5.2.1 Mobilization & Planning
When the ISE Project was commenced, CEZ has operated at its plants five different legacy
systems for the maintenance and materials management. The first main task of the ISE Project
was to decide which overall approach will be accepted. The following options were analyzed in
detail:
To select one of the existing legacy systems used at some of CEZ's plants as a standard
and to implement this system without any modification in the unified manner at all
plants of CEZ. This approach would provide easy-to-achieve low cost solution, which
would not meet all functional expectations and it would just fix the current business
procedures. The expected benefits of ISE Project would not be met.
To enhance functionality of one of the existing legacy systems used at some of CEZ's
plants to meet all functional expectations and to implement such an enhanced system in
the unified manner at all plants of CEZ. This approach would provide probably the full
desired functionality, but does not include any business process reengineering. The
process of enhancing any of the existing systems would be complicated, expensive and
high risk.
To develop a new system from scratch based on functional specifications and users
expectations. This completely new system then to implement in the unified manner at all
plants of CEZ. This approach would probably provide the best result and it was heavily
supported by end-users. All dreams and desired functionality would have chance to be
included. But such a development includes a lot of serious risks - the main is how to
make end-users to agree on what they do really want. Once all possibilities are open,
there is no limit on users expectations and requirements. Moreover, CEZ is an utility
company and not a software house to develop such an application on its own.
To analyze the market and then to purchase an off-the-shelf system with appropriate
references. The system meeting all these requirements to implement without any
modifications at all plants. This approach includes a lot of business process
reengineering which is necessary to modify an organization to meet requirements of the
selected application. But, provided the selected solution is really well proven, it is the
low risk, relatively low cost (compared to the cost of in-house development of the full
application) and fast to implement solution. This option has not been very popular
within the end-users community, but it was supported by top management of the
company. The top management committed to make all the necessary modifications of
the company and business processes affected.
Not to do anything - to provide a full picture of the initial analysis, even this option has
to be mentioned. This option would mean that ISE Project had not been successful. It
features no additional cost, no risk related to migration to any new environment and/or
224
C583/032/2000 IMechE 2000
application. But it does not provide any expected benefits. The existing business
processes would be frozen, no reengineering would be run and it would be just
question of time when a project with the same goals as ISE Project would be initiated
again.
As it has been already mentioned above, the approach of selection of a commercially available
system with the appropriate references has been adopted. Based on this approach the Indus
International Passport Maintenance and Materials Management System has been selected as
the core of ISE System. All fundamental principles of the ISE Project (e.g. time schedule of
main milestones, budget, IT standards) have been fixed in a Project Definition document
defining all roles, rights and responsibilities. Based on this document the ISE Project progress
is reported and evaluated. A contract was formed around this document for and a Systems
Integrator Selected to deliver the defined solution. The Selected Systems Integrator was
Westinghouse Electric CR.
5.2.2 Initial Implementation
Business improvement was one of the main goals of the ISE Project. Therefore the initial
implementation stage included the most important part of the project - the ,,should-be"
analysis. During this analysis, the functionality of the Indus/PASSPORT application, best
practice used by plants already using Indus/PASSPORT and the knowledge of CEZ's specialists
have been thoroughly analyzed. The ,,should-be" analysis was run with participation of about
100 specialists for affected functional areas from all power plants of CEZ. This analysis
represented a confrontation of the current practice, the best practice used in the industry and
the Indus/Passport functionality. The desired solution was discussed, agreed and documented
in the should-be analysis output - document ,,Project of Implementation". This Project of
Implementation document represents a ,,bible" of ISE Project describing in detail how the
system will be used at all CEZ power plants, its benefits and impacts.
The Project of Implementation document consists from two main parts:
a) List of modifications to be implemented within CEZ - these are business process
modifications which should be accepted to enable CEZ to utilize the selected
Indus/PASSPORT application in the most efficient manner. Within the ,,should-be" analysis
about 200 modifications of the business procedures within CEZ have been identified and
proposed for implementation. CEZ finally committed to implement all of these.
b) List of software modifications to be included into the Indus/PASSPORT application. As the
result of the ,,should-be" analysis, fifty two software modifications have been proposed.
These modifications mainly improved an end-user comfort, but some of them were
necessary ISE System to conform with the existing Czech Republic legislation. After in-
depth analysis, twelve software modifications were approved for implementation, the rest
of forty have been rejected.
In parallel with the ,,should-be" analysis, the technical track showed significant progress. The
IT infrastructure specifications were developed and tested, the IT infrastructure has been
installed and the IT environment for all further steps has been finalized.
C583/032/2000 IMechE 2000
225
The education track trained the first group of ,,functional specialists"- key technical personnel
covering all ISE System functional areas. These functional specialists were trained for the full
ISE System functionality. All other end users were trained just for the segment of the
functionality and/or processes they need for their everyday work. The original intent was to
use these functional specialists as tutors for the end users training, but because of their low
availability the end users training has been taught by professional trainers form the CEZ's
training center.
5.2.3 Pre-Production
The methodology says that all activities should be finalized just before the real exploitation of
the systems starts. It is clear and understandable for everybody, that for example end users
should be trained just before the system commission not to forget the knowledge they just
gained. The main problem of the project of this scope is to make all, or at least the majority of
the activities to meet this requirement.
The ISE Project of CEZ encountered a number of serious problems within the pre-production
stage. Not all of the business procedures modifications have been finished in time. The code
tables were not finally agreed and each plant desired to use its own ones. The application has
not provided the stability, reliability and response time expected. The pre-production phase
was the most painful phase of all the project.
5.2.4 Production
CEZ decided to implement the ISE System at two ,,pilot" plants first. One fossil and one
nuclear have been selected as pilots. All the business procedures have been finalized at these
two plants. The system integration track focused to the data conversion process. The useful
data stored within the legacy systems has been extracted and consequently loaded into the ISE
System databases. Not all data has been transferred. Some of the historical data has been kept
at the legacy system for read-only access. Sometimes it was more simple to reenter the data
into the ISE System then to try to run the data extraction - data loading procedure.
The end users were trained at the existing training rooms equipped by PCs, data projector and
with access to the server with the Indus/PASSPORT application. The mobile classroom has been
used, too.
5.2.5 Roll-Out
Once the ISE System is fully tested at the pilot plants and it is approved by the CEZ
management for the routine implementation, the roll-out stage starts. The ISE System is
implemented in the same manner as it has been implemented at the pilot plants. In this stage it
is necessary to ensure the unified manner of the implementation, not to enable plants to modify
the solution to fit better to their ,,special conditions".
Within the roll-out stage the long term support must be established. This support provides the
enhanced help desk, ensures the necessary approved modifications of the system (for example
modifications of the printed reports) and implements all these modifications in the unified
manner at all plants. In practice, provided some end-user from some plant requests some
modification, this modification must be approved by the ,,steering committee" and once
226 C583/032/2000 IMechE 2000
approved it is developed by the centralized support team. The finalized, tested and approved
modification is implemented at all plants in the unified manner.
5.2.6 ISE Project Experience in Detail
a) ISE Project Team
The ISE Project team consisted from specialists for different areas:
functional specialists for each of the functional areas
IT specialists
training organization specialists
trainers
data conversion team
communication, PR
helpdesk support
project management support (finance, legal, etc.)
The total number of persons who were full time involved in the ISE Project averaged 15.
b) Data Conversion - CEZ used four different legacy systems with large amounts of data held
in each system. The quality of data varies from plant to plant and it is often easier to
recreate the data instead of using conversion of data from the legacy system to the new
one. A dedicated team has been established to create and test all tools necessary to
prepare data for conversion. Data was retrieved from the legacy systems to ,,flat" files
using tailor made extraction programs. The flat files were consequently processed by the
standard data loading utilities provided as part of the software package. This approach
enabled the basic data to be migrated.
c) Training - Two different approaches have been accepted for functional end users training.
At the initial stage of the project the functional specialists were trained. These specialists
were trained for the full functionality of appropriate ISE System modules and they gained
the comprehensive in depth knowledge of ISE System functionality. These functional
specialists were involved at the plants in the work procedures preparation and ISE System
implementation at CEZ plants. The full training of functional specialists required
approximately two weeks of training per module.
The second approach was used for standard end-users training. The ISE System end users
were trained for Indus/Passport system basics and then for the work procedures they will
need for their day-to-day work. End users were trained by CEZ professional trainers at
their sites and the duration of training varied from 5 to 12 days based on number of
processes the end user needed. The dedicated regions and ,,flying classroom" were used
for the end-user training.
The quality end-users training represents one of the most important factors for overall
success of the project.
d) Implementation Support - The implementation support within the course of
implementation is provided both by the central ISE Team and the local functional and IT
C583/032/2000 IMechE 2000
227
specialists at the plants. It is crucial the local specialists to be able to contact the central
help desk and support calling them for advise and/or help. The fact that the
implementation is supported directly by local people helps them to understand the system
and then to provide perfect support for regular operation. The local implementation teams
are not allowed to run any modifications of the system, they setup the system using the
standard functionality. Should any more complicated modification be required, it must be
submitted for approval to the central level and when approved it is implemented in the
unified manner at all plants.
e) Communication - One of the very important activities is communication between
management, project team and end-users at the plants. It is necessary to ,,sell" them the
system, to make them to accept the solution, to abandon their ,,lovely" legacy systems.
Despite the initial reluctance, all end-users, once they had learned the system functionality
in depth, became the real supporters. The communication program included short video
clip, series of articles and the special issue of CEZ newspaper and even a plush mascot of
ISE Project.
6. CONCLUSION
Pressure to reduce operating costs, provide improved quality of service, and meet health and
safety requirements continues to increase for utility companies. Congressional and state
legislative initiatives associated with deregulation of the electric industry have had a cumulative
effect on both the public and private segments of the industry. Cost reduction must be achieved
by reducing labor and materials costs and/or by improving operational efficiencies.
Enterprise-wide maintenance/materials management software systems that integrate the various
business processes that comprise a utility's workflow represent a key opportunity to improve
processes while reducing costs and meeting customer service objectives. In order to realize the
potential benefits offered in such systems, however, critical implementation issues must be
successfully addressed.
An effective implementation methodology is required to manage the broad range of business
process, technology, infrastructure, education, and cultural change tasks that must be
completed.
228
C583/032/2000 IMechE 2000
C583/033/2000
Power station maintenance for safety and reliability
J J LEWIS and L SHARP
Eulech Engineering, Billingham, UK
SYNOPSIS
The changes in the regulatory framework, which historically governed the historic pattern of
power plant maintenance, have allowed advantage of actual life performance to be taken into
account when considering maintenance schedules.
This paper examines how, by developing a holistic approach to the maintenance requirements
of a power plant, significant reductions in maintenance costs and reliability improvements can
be achieved.
1 INTRODUCTION
Increased freedom for operators to determine shutdown intervals has highlighted the need for
a structured approach to maintenance. Risk based inspection regimes for boiler systems have
been determined based on a focussed scheme of examination supported by in house expertise
in material engineering, fracture mechanic analysis, creep assessments and finite element
modelling. Taking account of the overall impact on the facility these studies have included
ancillary equipment, such as valves and relief streams, to exploit the flexibility permitted on
examination intervals.
The overall performance of the power plant is also dependent on the performance of the
rotating equipment. The processes developed by Eutech for the assessment of safety and
reliability of Safety Critical machinery in chemical process plant, have been extended to cover
application to boiler plant. These assessments have shown where requirements for hardware
improvements and development of maintenance inspections were required to enable
maintenance activities to be targeted to ensure safety, reduce the risk of failure and extend the
interval between overhauls.
C583/033/2000
229
2 BACKGROUND
Historically, power plant overhauls and maintenance has been dominated by the need to
conform to statutory regulation covering examinations of the pressure parts. The legislation
in the UK in the form of the Factories Act required that boiler plant inspection intervals
should be no greater than 26 months, or reduced intervals for older plants. With such frequent
overhauls, other aspects of equipment maintenance could be covered in the planned outage
and attention focussed on planned routine changes of items such as relief valves at these
intervals.
This situation was in strong contrast to the chemical industry, where legislation was less
prescriptive regarding inspection of equipment containing hazardous chemicals. Instead,
chemical operators were responsible for ensuring that adequate maintenance was carried out
to ensure safety. This meant that large chemical producers established systems to determine
and quantify the risks associated with operation of different types of equipment and from that
starting point developed appropriate maintenance policies.
When the Pressure Systems and Transportable Gas Regulations 1989 came into force, the
onus for setting an appropriate inspection interval fell on the inspection authority, with
historical legislation to be taken as a guide. These regulations applied to a range of industries
with all relevant fluids being incorporated, as well as the traditional steam.
Work then began in several areas to look technically at the case for pressure part inspections
to determine the most appropriate actions to carry out and appropriate frequencies. In many
cases, this resulted in the interval between full internal inspections being extended. This
process offered significant scope to the power industry, where with the barrier to longer
overhaul intervals removed, the focus moved onto other items of ancillary equipment such as
rotating equipment and relief valves.
It became apparent that in many cases, it was the maintenance requirements of these items
rather than pressure parts that would limit shutdown interval. Detailed studies are therefore
required to determine the most economic interval for shutdowns as part of an integrated
maintenance strategy to improve reliability. This paper will consider several case studies that
demonstrate how appropriate maintenance strategies have been reached.
3 POWER PLANT ISSUES
Power plants have several issues common to other process plant, and some which are industry
specific.
In common with the heavy chemical industries, there are
* High temperatures
High pressures
* High power rotating equipment
* Need for continuous production
* Demand for high availability
* Possible corrosive media
* A lot of equipment dating from 60s-70s
230 C583/033/2000
Those items which are specific to the power industry include
* Fuel problems
* Need for flexibility
* High number of cycles
* Steam specific issues.
4 THE IMPROVEMENT JOURNEY
Application to the Power Industry of the methodology developed and used for many years in
the chemical industry by Eutech Engineering gives a structured approach to determining
maintenance needs. The approach can be applied at several levels to look strategically at a
whole plant, or in greater detail to optimise the work carried out around individual items.
4.1 Consider criticality
This should be based on assessing the risks to both safety and production. The consequence
of failure is assessed.
Reviews of the selected family of equipment are carried out to identify Criticality for any of
business, safety, and reliability reasons for an installation. (Process reported by author in
IMechE Guide Process Machinery - Safety and Reliability edited by William Wong) The
reviews give a background for future decisions and a sound base for further study. Safety
assessments carried out in this way can provide initial assurance for suitability of purpose,
maintenance, training, and inspection particularly when considering the Provision and Use of
Work Equipment Regulations (PUWER 1998) Regulations.
The flow chart shown below outlines the classification process to determine safety criticality
based on potential hazard and requirement for additional detailed reviews to assess risk from
the safety critical equipment
C583/033/2000
231
MACHINES SAFETY
CLASSIFICATION PROCEDURE
232
C583/033/2000
4.2 Determine deterioration mechanisms
This requires full knowledge of the operating envelope and should include input from process
operators, maintenance technicians, etc as required. The necessary detail of such studies are
time and resource hungry, however, applied in conjunction with the definition of criticality
for the family of equipment the effort can be directed to the particular concerns on the
important items.
What could go wrong can go wrong, but by applying a HAZOP style approach consideration
can be given to all eventualities. This is especially important at start up and shutdown
conditions, where the effect of temperature transients etc can be determined.
Wherever possible, quantification of the degradation by time, cycle etc, should be carried out.
For plant that has been operation for a period of time, actual data for instance on erosion rates
can be gathered.
4.3 Develop preventative measures
Prevention is preferred to correction and preventative measures should be aimed at picking up
any deterioration mechanisms as soon as possible. For example, confirming feed water
quality by sampling to prevent corrosion rather than monitoring thickness of material lost.
The wide range of condition monitoring techniques are then applied selectively to identify
change, with preventative measures chosen to supply actual measurements to support
predictive models of life time performance.
4.4 Develop appropriate maintenance routines and set intervals
These are aimed at maintaining the equipment condition by carrying out proactive tasks.
Understanding of real life operational problems by the application of RCA (Root Cause
Analysis) techniques has led to the development of a tailored FMEA (Failure Mode and
Effect Analysis) approach to rotating equipment issues. The application of these generic
assessment structures developed for equipment in the chemical processing industry to similar
equipment in the Power Industry produces benefit through the implementation of best
appropriate practice. Enhancement of the tailored FMEA using probabilistic assessment
software such as APT(Woodhouse Partnership) then can be used to determine the most
economic maintenance strategy.
5 CASE STUDIES
* De-aerator inspection interval
* Super-heater header life assessment
* Turbo Alternator Reliability
* Super-heater tube reliability
5.1 De-aerator Inspection
Problem
De-aerators are known to be susceptible to an environmental cracking mechanism on un-
coated carbon steel surface. The discovery of this mechanism in the 1970s lead to an
enormous increase in the amount of NDT carried out. In the case of low pressure de-aerators
which are not stress relieved, these detailed inspections often highlight areas of defects
associated with the heat affected zones of welds below the liquid level. Historically, the
C583/033/2000
233
defect areas have been removed and weld repairs carried out, to restore the full shell
thickness. This approach has several drawbacks;
* A large amount of effort in grinding and re-welding small areas of metal removal is
required at each full inspection.
* All areas of weld repair introduce additional heat affected zones, increasing the area
susceptible to further damage.
Achievement
De-aerators are now inspected every 4 years rather than 2, with targeted inspection reducing
the initial NDT requirements and before full weld repair, areas of grinding are assessed on a
fitness for purpose basis. For a typical inspection, this reduced the locations of weld repairs
from 18 to 3.
Results show that subsequent inspections show a reduction in the number of defects
developed.
How?
An alternative approach has now been developed, based on assessing the risks and taking a
longer term view.
Taking information from a wide study of de-aerators in ICI, an overall growth rate for crack
growth was determined. Experience showed that cracking was generally very shallow but
long on the surface. Severely affected vessels tend to have a very large number of defects, but
these can generally be removed with fairly simple grinding.
The removal of cracks generally reduced the thickness to below that acceptable to the original
design code. In these cases, the localised areas were assessed using appropriate codes or by
using finite element analysis to demonstrate that the stresses in the areas remained acceptable.
A fracture mechanics approach was also used to assess the significance of postulate cracking.
It was demonstrated that cracking likely to develop within the inspection period would not
give rise to an unacceptable situation.
234
C583/033/2000
Figure 5.1.1 Deaerator cracking found in- situ
Below: Figure 5.1.2, de-aerator cracking around a nozzle
C583/033/2000 235
Below: Figure 5.1.3 Typical cross section through crack
236 C583/033/2000
5.2 Superheater life assessment
Problem
The high temperatures associated with boiler operation mean that many super-heater headers
operate in the creep range for their materials of construction. Consequently, they have a
limited design life. At the design stage, however, pessimistic temperatures are often assumed
and hence low allowable stresses are applied to achieve an acceptable creep life. After a
number of years operation, the headers may be in quite acceptable condition, but the original
design calculations suggest replacement will be required. A systematic approach is required
to avoid unnecessary replacements.
How?
During real operation, the actual super-heater metal temperatures are usually lower than
design, and so creep damage accumulates at a slower rate. The effects of even a small
difference in temperature can be significant as a 10 degree increase in temperature can double
the rate of creep life usage when items are operating well within the creep range.
a) All headers in a site power station were assessed to compare the normal steam
operating temperatures against the temperatures where creep may be significant and
against the assumed maximum temperatures used during original design. This analysis
allowed those headers where largest benefit might be obtained from a more detailed
analysis to be identified, and determine positions for thermocouples to be fitted.
An estimate of the creep life usage already used was determined by examining records
of historical steam temperatures. These were available for over 15 years, and hence
provided a good basis for backward extrapolation of the usage from start up. For the
years where no actual data was available, it was pessimistically assumed that the boiler
ran for a whole year, ignoring any overhauls.
b) At the plant overhaul, thermocouples were fitted and material checks were carried out
to confirm materials of construction and weld metal (there have been historical
problems with lean metal welds being identified). The plant control system was then
used to store the maximum temperatures in the header on an hourly basis. For
calculation purposes, these temperatures were then sorted into 5 degree temperature
bands and differences between different systems were highlighted and the Robinson
rule was applied to calculate the creep life usage for the time spent in each temperature
band.
C583/033/2000 237
The graph above shows the distribution obtained for two similar boilers, with total
hours shown totalling 45541 and 63714 hrs respectively. It is clear that although in
terms of design the two units are almost identical, the operating policy has resulted in
very different distributions of temperature.
Where real data was not available (early years of operation) the creep life usage was
calculated on the assumption of maximum operating temperature for the whole period
on line.
c) Annual creep life assessments are now carried out to confirm the rate of creep life
usage. As well as calculation of the percentage creep life used, anticipated dates when
60% used header life and 80% used header life will be reached are calculated. This
allows any changes in the operating regime to be identified. These requirements for
these checks form part of the written scheme of examination.
d) The schemes of examination for the boilers now include detailed checks for early
signs of creep damage.
e) The same calculation approach was then used to assess the consequences of proposed
changes to the boiler operation, which would increase the operating temperature.
Achievement
Although operating hours for the boiler under consideration are well in excess of the original
design life of 200,000 hrs, it has been demonstrated that remaining lives for all headers
considered are in excess of 5 years.
The calculation based assessment, using real plant data, allows the affect of different
operating cases to be evaluated without shutdown of the unit.
5.3 Turbo alternator reliability
Problem
Assessment of a Turbo Alternator train on a medium sized Power Station indicated that the
system was critical to the process for a number of safety and business reasons.
238
C583/033/2000
In particular:
the potential release of kinetic energy from a high power machine system.
the possible deterioration and failure of the alternator end bells due to stress corrosion
cracking
higher output demands required the identification and protection of the machine
against deterioration mechanisms to enhance the equipment reliability .
How?
A series of structured reviews were carried out by a rotating equipment expert. Firstly bench
marking the installation against good practice for such installations based on equivalent
machines working in both Power and Process Industries. These showed that the potential
issues for release of energy through over speed, coupling failure, barring gear operation
(where failure to disengage would result in overspeed of the barring gear drive), bearing
failure, bolting failure, and shaft failure were addressed either by the original design standard
of the machines, or by installed protective equipment, or by operating and maintenance
procedures.
The process to review the equipment is based on a series of structured reviews. These relate
elements of the equipment considered, through a series of guide words determined by the
hazard potential identified in the classification process, to the risk and ways in which the risk
is managed. A typical flow model for this process is:
C583/033/2000 239
-
-
-
-
The issue of the alternator end bells with a potential for stress corrosion cracking had also
been addressed by modification of the material and requirements for ongoing routine
inspections.
The system had been assessed critical for:
the potential release of kinetic energy from a high power machine system.
the possible deterioration and failure of the alternator end bells due to stress corrosion
cracking
higher output demands required the identification and protection of the machine
against deterioration mechanisms to enhance the equipment reliability.
The initial reviews carried out allowed consideration of the machine for the potential release
of energy and danger due to the cracking of the alternator end bells. Further examination of
the system was required to identify potential failure modes which could result in significant
financial impact through machine breakdown.
A series of structured FMEA studies covering both design and operations of the unit were
then carried out, again by a rotating equipment expert. The risk from the potential failure
modes were evaluated, either by comparison to industry experience - supported by in house
databases, and other published data (OREDA), or by analysis using direct statistical methods,
or application of software such as APT, which allows economic evaluation and optimisation
of issues from general reliability descriptions of the equipment. From these reviews further
developments of the installed system were identified, either for changes in hardware or
maintenance practice.
240 C583/033/2000
RELIABILITY BY DESIGN
FAILURE MODE REVIEW
Causes / Consequences
Mechanisms for failure
Preventative Measures
Effectiveness of Prevention
Outstanding Concerns
Vulnerable components
Guide Words
Item
Achievement
Introduction of additional vibration and temperature monitoring to provide indication of
deterioration of rotor dynamic performance were justified against long term probability of a
high cost failure.
However, equally significant were the number of minor changes to operating practices, and
local instrumentation. Changes to operating and maintenance practices with introduction of
some local indication of oil flows and pressures significantly reduce the risk of common mode
failure where there is the potential to result in major machine damage.
In the case of the oil system, loss of oil inventory through relief valve venting to drain, and
the potential for oil starvation from blinding of intake filters were possible on the installed
system. Minor changes to the system will in the first case eliminate the hazard by routing the
relief valve vent back to the oil feed tank, and in the second case provision of operating
procedures to monitor and trend screen pressure drop will minimise the risk.
Another particular example following the review of maintenance practices was on the routine
trip testing procedure. These were defined to confirm operability of the trip system (
avoiding hidden failure modes), but, required that the linkage was lubricated prior to testing to
ensure it was free to move, effectively eliminating the faults that would be identified by the
test. The test should have confirmed the operability of the trip linkage system after a period of
operations. However, the testing method gave no indication that the trip would have worked
on demand, and only provided false assurance that the machine was protected against
overspeed.
The application of the learning from the process industry through a developed methodology
ensured that the learning from equivalent machines could be focussed to support the
assessment of this machine system. This allowed the study to be carried out effectively within
two days, and the use of a rotating equipment expert required little input of effort from the
operating team apart from the provision of standard equipment information and instructions.
5.4 Superheater tube failure
Problem
A plant boiler suffered from several outages over the course of a year, due to failure of
superheater tubes. The number of outages seemed to be increasing, and the plant realised the
need to stand back and consider the reasons why.
C583/033/2000
241
Figure 5.4.1 : As received failed bends showing the splitting around the extrados of
the bend
Figure 5.4.2: Cross sections through failed bends showing the extent of oxide
formation on the bores.
242 C583/033/2000
How?
In addition to traditional metallurgical analysis of the tube failures, a full system review was
carried out to identify the root causes of the problem. The assessment involved water
technology specialists, mechanical engineers and metallurgists alongside plant operations and
maintenance representatives.
Following the method outlined above, the criticality of parts of the boiler system was
examined, and mechanisms of failure for each part of the system were considered. The review
looked at start up and shutdown procedures as well as the normal operating regime, and
highlighted those regimes that made the tubes more at risk. A programme of immediate,
medium term and long term improvements were highlighted to remove the root causes of
problems. The highlights were;
* Redesign of the weld attachment of a selection of the end tubes to minimise the stress
concentration.
* Changes to the water dosing system, to minimise the risk of blockages (many of the
failures were due to localised creep damage following tube blockage)
* Fitting of automated blowdown system to ensure consistency in purging and give
improved water quality control.
The need for consistency in water quality was the key finding, and providing documentary
evidence of satisfactory operation between defined limits became a responsibility of the
operations team detailed on the Scheme of Examination for the boiler.
Achievement
Following short term actions at the next planned overhaul, the number of outages due to tube
failures was significantly reduced in the next year on line.
6 CONCLUSION
The examples given in this paper have highlighted that a structured approach for determining
maintenance strategy, developed by the chemical industry, can be applied to give benefit in a
wide range of issues in the power plant environment.
The overall effect of these reviews has been to take advantage of actual life performance
when considering maintenance schedules. This holistic approach to the maintenance
requirements of a power plant, enables both the achievement of significant reductions in
maintenance costs and reliability improvements.
C583/033/2000
243
References
1 IMechE Guide for the Process Industry :- Process Machinery - Safety and Reliability
Edited by William Wong ISBN 1 86058 046 7
2 Provision and Use of Work Equipment Regulations 1998 Statutory Instrument Health
and Safety 1998 -2306 HMSO
3 The Pressure Systems and Transportable Gas Regulations 1989, Health and Safety
executive, No.2169, HMSO, London, UK, 1990.
4 OREDA Offshore Reliability Data Handbook (SINTEF Industrial Management)
ISBN 82-14-00438-1
5 Selva and Amphlett, High integrity pressure systems in service - a balanced approach
to cost effective risk management through legislative compliance, Int conference on
Pressure systems and operation, IMechE, ISBN 0 85298 951 2
6 LH Larsson, High Temperature structural design, ISBN 0 85298 771 4
7 Examination of de-aerators in steam raising plant - EUT.222C, Eutech Engineering
Solutions Ltd.
8 An Overview of NACE Task Group T-7H-7'S Report on De-aerator Cracking
(Corrosion 88 - Paper Number 350) - J A Kelly, C E Guzi, T M Laronge.
9 Recommended Practice for Prevention, Detection, and Correction of De-aerator
Cracking (NACE Standard RP0590-96 Item No 21046) ISBN 1-57590-0111-4.(1996).
10 Environment Assisted Cracking of Steels in High Temperature Water. University of
Newcastle Upon Tyne Materials Division. Sere/Case Award Ref No 9056155X. J J H
Fegan-May 1994.
11 Stress Corrosion and Corrosion Fatigue of Metals in High Temperature Aqueous
Environments by J Congleton ('Strain', February 1990)
Eutech 2000
244
C583/033/2000
C583/016/2000
How to contract for a gas turbine major inspection
R W BONSALL
Siemens Power Generation Limited, Knottingley, UK
1. Introduction
As one of the worlds largest suppliers of equipment and services to the power generation
industry, Siemens Power Generation has a great deal of experience in working with power
generation companies around the world designing, constructing and servicing power stations.
Owners and operators of power stations demand exacting standards of service from supply
companies. Since 1990, Siemens has supplied over 6,000 MW of large power stations in the
UK and all of this capacity has been CCGT plant. At the heart of the CCGT power station lies
the gas turbine. The gas turbine requires more maintenance than the conventional steam
turbine, more regularly and the spare parts (particularly turbine blading) are more expensive
than steam turbine equivalents.
Modern CCGT power plants generally operate at base load and it is therefore imperative that
the CCGT plant runs reliably with a high operating availability. Gas turbine maintenance
periods must therefore ensure that all the necessary work is undertaken to achieve the required
reliability and availability over the next operating period. Traditionally, power station operators
have performed whatever maintenance is necessary to ensure high availability and operating
reliabilities. In today's world of complex electricity markets and profit driven companies,
power station operators need to match the plant performance to the commercial issues
affecting their company.
Service providers have traditionally come from a risk averse, conservative background as their
traditional customers have also had a similar mindset. Also a conservative standpoint allowed
more spare parts to be sold and more service work to be undertaken. In the modern world of
the power generation industry, those companies that can manage their risks the best will be
well placed to succeed. Operators and service providers need therefore to challenge their
traditional way of performing business and look at ways of sharing the risks and the rewards.
Major scheduled maintenance takes place on modern gas turbines every 3-4 years, therefore on
a power station comprising four gas turbines, a gas turbine major inspection will be performed
C583/016/2000 IMechE2000
245
every year.
This paper
* Reviews the approach taken by power station operators and service providers when
approaching a gas turbine major inspection
* Looks at the many options that are available to a power station operator
* Explains how Siemens Power Generation is working with our customers in the UK to
develop a partnership approach to performing major turbine inspections.
2 Preparation for the gas turbine major inspection
As the power station maintenance manager begins to plan the gas turbine major inspection, a
number of questions will be asked
* What work should be included in the work package?
* Who will perform the work?
* What services do I need to procure?
* How do I ensure that I have the right amount of spare parts available?
The maintenance manager should be asking these questions early enough to allow full planning
to be performed. Some spare parts can take over 12 months to supply and therefore the
planning process should ideally start about 18 months prior to the start of the planned
inspection period.
2.1 Agreeing the workscope
The workscope is usually developed from many different sources and may include
* Standard recommendations from the Original Equipment Manufacturer (OEM)
* Operational feedback from the fleet (from the OEM or customer users groups)
* Upgrades/modification work previously performed on other units or new packages offered
by the OEM
* Defect rectification work
The OEM clearly has an important role to play in recommending what work should be
performed. Upgrade and modification work can often improve the plant performance and
feedback from other plant operators provides a useful source of information. However at the
end of the day the power station operator has to decide what goes into the work package. Only
they know the ' full picture' of constraints that they are facing including the budgetary, time,
and resource constraints. At this point in the planning process OEM's should avoid the ' We' re
the experts - we know what' s best for you approach' because service providers often do not
know what restraints and limitations have been placed on a maintenance manager when
arranging a gas turbine overhaul. The full range of options should be discussed and then the
power station operator can decide what should be included in the work package. The power
station operators generally welcome a partnership approach from their service provider.
246 C583/016/2000 IMechE 2000
2.2 With whom and how shall I contract?
Once the workscope has been agreed (at least in a draft form) the power station operator is
faced with the question 'With whom and how do I contract?' There are different types of
service organisations active in the power station maintenance field and they generally fall into
three categories.
* OEM
* Non-OEM
* Power Station owners service organisations
Traditionally, power station operators have performed a large part of the plant maintenance
using existing 'in house' resources. However over the last 10 years in the UK, power stations
have reduced their 'in house' resources and are therefore more reliant on contracting
companies. These companies may be the Original Equipment Manufacturer (OEM) or may be
another company offering services on that equipment. The power station operator may also
have the option of having the work performed by an in-house service company or group.
Generally only the larger power generation companies have this option. It is very unlikely that
Independent Power Producers (IPP's) with a total station staff of about 50 will have this in
house option. A decision to contract with any of the three types of service organisations is a
complex one and will only be taken after the budgetary, risk management, technical
competence and commercial guarantees issues have been fully considered.
Another issue the power station operator is probably facing is whether to place all the planned
work into the responsibility of one contractor. Again a multitude of questions will be raised in
the mind of the power station operator and will involve the usual variables of risk, money,
technical competence and resources.
The single contractor turnkey approach may offer the power station operator the lowest
demand on their resources in managing the numerous interface issues however may not be the
cheapest option as the co-ordination effort is then passed to the contractor.
A multi contractor approach with the interface management being performed by the power
station operator may offer a cheaper but more resource intensive approach and will require the
power station operator to shoulder the risk of successfully managing the interfaces.
An intermediate option may be a model where the main contractor manages the overall
programme and workscope risks but utilises sub-contractors that are contracted to the power
station operator but are 'free issued' to the main contractor.
Another approach may be for the power station operator to contract on an individual basis
with a number of contractors and employ a managing contractor to co-ordinate and manage
the other contractors. These four approaches are summarised in Table 1.
C583/016/2000 IMechE 2000
247
Table 1: Advantages and disadvantages of types of contract
Contracting Model
One Contractor accepting
the Ml risk of work and
programme
One Main Contractor with
other contractors 'free
issued' to main contractor
Multi Contractor
Managing Contractor
Advantages (to customer)
Less co-ordination required -
only deal with one contractor
Favoured sub-contractors are
used
Gets all the contractors he
wants
Least cost option
Reduces reliance on 'in
house' resources.
Disadvantages (to
customer)
Additional 'Co-ordination'
fee to pay for
Main contractor may be
unhappy with choice of
nominated contractors.
Risk Margin included in
contract price
Power Station operator has to
manage interfaces
Higher risk of contractual
problems if management
effort is not present
Additional co-ordination fee
to pay.
Risk remains with power
station operator
Assuming the power station operator has decided what work he wishes to contract out, he has
to decide how he intends to procure these services. If the decision is taken to contract the
service with an outside Service Company then the operator must follow certain procurement
obligations. For all countries in the European Union, legislation was introduced in 1992 to
ensure an open and transparent procurement process for ALL contracts greater than 200,000
Euro (120,000). The process requires customers to advertise their procurement intention in
the European Journal (EJ), which is issued several times per week. Suppliers of equipment and
services must therefore monitor the European Journal regularly to ensure that they do not miss
an opportunity to tender for work. Power station operators may use the EJ procurement
process for every gas turbine major inspection (providing the contract value is greater than
120,000) or they may 'bundle' a number of major inspections together or even stipulate a
contract duration which may last many years.
As the effort and cost of following the EU procurement process is not insignificant, power
station operators must therefore decide on their contracting strategy. Figure 1 shows the stages
in the development of contracting and relationship arrangements available to customers and
shows the One-off approach to contracting at one end of the continuum to the Long Term
Service Agreement (LTSA) at the other end of the continuum. In deciding what contracting
approach to take, power station operators have to consider many different issues including, in-
house resources and capabilities, risk allocation, the long-term business prospects, the
changing marketplace and of course cost. Again there are constraints on the power station
operators when deciding the length or number of overhauls to be included in any contract.
Generally it is beneficial to both parties to place a contract for a longer period of time (typically
3-6 years) as the not insignificant costs of specification writing and tendering can be reduced
248 C583/016/2000 IMechE 2000
for both parties. The downside of a longer time period is if the contract does not meet the
expectations of both parties then there is every chance of contract difficulties. The only way for
a longer-term contract to succeed is if both parties are in a Win-Win position. Even if this is
not achieved during the early stages of the contract, then a re-negotiation of the contract may
be an option to ensure the 'Win-Win' situation is achieved.
Having decided the contract route and duration and issued the invitation to tender (or more
usually pre-qualify) the power station operator will normally receive expressions of interest
from a number of companies. The power station maintenance manager should be aware from
knowledge of the marketplace which companies are competent to perform the proposed work
and therefore will not be surprised which companies respond to the European Journal
notification. Following the usual contract tendering, clarification and negotiation activities the
power station operator will decide which company they want to place an order with.
Irrespective of what end of the continuum is selected as the appropriate contracting method
the contract should be written to ensure that the appropriate drivers and adjustment
mechanisms are in place to ensure that both the customer and the contractor achieve positive
outcomes. In establishing what is important to both customers and contractors, it is useful to
review the needs and wants of both the power station operator and the service provider. It is
fundamental to the success of any maintenance contract that the aims are aligned (wherever
possible) and Key Success Factors are identified. Table 2 shows a typical review.
Table 2: Needs and success factors
Customer Needs
High quality of workmanship
High quality of technical
advice
Work completion on time
No accidents
No surprises
Defect free plant
Value for money
Supplier Needs
Repeat business
Satisfied customer
A profitable business
High reliability of plant
Defect free plant
Reputation for technical
Competence
Reputation for integrity
Key Success Factors
Joint planning
Open and trusting
relationship
Apportioning risk correctly
Avoid running to traditional
contracting positions
Appropriate incentivisation
Be tolerant of human
fallibility
Wherever possible some of these key success factors should be incorporated into the contract
and adjustment mechanisms introduced to ensure both contracting partners share in the success
of the project or that that both parties are penalised for not achieving success. The popular
expression often used today is 'gain share, pain share'. The actual amount of the contract that
is available to be shared or to be lost will be the subject of discussion.
Table 3 shows a number of different options which may be considered in ensuring that the
appropriate drivers are in place.
C583/016/2000 IMechE 2000
249
Table 3: Options for Contract Types
Standard Contracting
Performance Based Contracts
1.
2.
3.
4.
5.
6.
Fixed price quoted against specified work
Additional work quoted on a fixed price, schedule of
rates or time and material basis.
Bonuses / penalties may apply against certain
performance criteria.
Target cost
Ceiling costs
Cost plus fee
2.3 The target cost approach to performance based contracts
The target cost approach to contracting has been used successfully in the power generation
industry over the last few years. The largest single cost in performing a gas turbine major
inspection (apart from the spare parts) is the cost of labour. Therefore, in agreeing the price for
the work the amount of man-hours projected for the work will be closely scrutinised by both
the power station operator and the contractor. The target cost example shown below uses a
man-hours approach however can be applied to any other measurable quantity.
A labour only job to be contracted is agreed at taking 6000 hours. At an hourly rate of
25/hour the job will cost 150k. If the job actually takes only 5200 hours, then a saving has
been achieved of 800 hours or 20k. According to the scheme shown graphically in Figure 2,
the power station operator will save 10k off the contract price and the contractor will get
10 k providing the agreed performance indicators of safety and quality are achieved. If not then
this 10k will be reduced in accordance with the agreed schedule. If the actual hours varied
less than 300 hours (using a 5% deadband) from the agreed 6000 hours, then the contract price
would not be adjusted.
The ceiling cost approach is a variant on the target cost approach and caps the customer price.
Cost saving mechanisms may apply in the same way as for target cost.
2.4 What is in this for the contractor ?
The performance style of contracting has generally been introduced by power station operators
in an attempt to reduce the cost of his maintenance by ensuring costing transparency. If the
price secured by service companies is reducing, then isn't this affecting the service companies
margins. The answer is 'not necessarily so'. Service companies servicing the power station
sector have generally had to cope with great bursts of activity (usually during the summer
outage season) followed by periods of relative inactivity. Running any business with such
varying levels of business requires a certain amount of built in inefficiency during the relatively
quiet periods.
Traditionally the approach to uneven work loads has been the shedding of staff during
relatively quiet periods and the recruitment of staff during the busy periods. This approach has
traditionally worked against a background of large quantities of well-trained and skilled manual
250 C583/016/2000 IMechE 2000
engineering workers. With the reduction of traditional British engineering companies
throughout the 1980's and 1990's the large pool of skilled workers which traditionally existed
is no longer present and the age profile of those currently working in the industry is continuing
to increase. Service companies therefore need to recruit, train, incentivise and retain the core
skills required to service the modern power stations. It can no longer afford the 'hire and fire'
approach.
In these days of low unemployment, people do not always see six months per year of intensive
site work away from home as an attractive career prospect. There are often many more
attractive employment opportunities closer to home. Companies therefore need to offer staff
long-term work prospects with appropriate incentives and benefits. To achieve this companies
need to have long term relationships with their customers offering continuity of work to allow
the necessary investment in our people. If performance based contracts offer the opportunity to
have longer-term contracts then this is usually a 'Win-Win' for both power station operator
and service provider.
2.4 Ensuring I have the right spare parts
A related outage issue that the power station operator is faced with is the acquisition of spare
and replacement parts. The OEM is the traditional source of spares for the power station
however alternative spare parts suppliers may be available in the marketplace. The availability
of spare parts is a very important issue if maintenance activities are to be performed without
delay. The location of the spare parts holding is also an issue. Again, a number of options are
available to the power station operator.
* The power station operator may procure and store all the required spares in their stores at
the power station. Engineers generally prefer this approach as the spares are then available
for immediate use and maintenance work can proceed without delay. Unfortunately the
accountants don't always like this approach as a lot of capital can be tied up in site spares.
Over the past ten years most operators have reduced their spare parts holding to an
'optimum' level which generally ensures that sufficient spares are held on site for normal
maintenance activities.
* Some customers try and 'pool' certain high value spares. The benefits of this approach are
that the pool members can share the spares purchase costs. However the pool has to be
administered and there can be complex pooling agreements. Another disadvantage may be
that another member of the pool may secure the spare part you require before you do.
Nevertheless 'pooling' may offer what certain customers require.
* Another option is to agree with the spares supplier to reserve spares at the suppliers
factory or store. These spares may or may not physically exist however the supplier will
guarantee that the spares can be supplied to the customer within a stated time period. The
problem to supply the spares within the agreed time period will be with the supplier. The
supplier will normally require a fee to provide this service.
* The simplest and usually cheapest solution is to purchase the spares parts when required.
The supplier would then provide the spare to his usual delivery dates. However, this may
C583/016/2000 IMechE2000
251
cause delays to the power station operator if he requires the spare part for his ongoing
maintenance activities.
Table 4: Spares holding advantages and disadvantages
Type of spares holding
Owned by customer in store
on site
'Pooled' with other
customers
'Reserved' in Suppliers Store
Rely on Supplier
Advantages (to customer)
Engineers can see and touch
spare - gives confidence
Immediately available for use
Owned by pool members
Shared purchase costs
Available when required
Exclusively available for
customer
Least Cost Option
Disadvantages (to
customer)
Accountants hate the tied up
capital
Pool management costs
Other pool members may use
before you
Complex pooling agreements
and administration
Time to get to site
Reservation fee incurred
Delay to supply may be
unacceptable
Most power station operators maintain a good level of spares at site for immediate use. They
usually also have arrangements in place with suppliers to ensure that spares can be supplied to
site at short notice. As in the previous discussion, operators have to weigh up the issues and
decide what options are right for him and his particular circumstances.
3 Summary
Gas turbines require more maintenance than steam turbines and therefore modern CCGT
power stations perform major outage work more often than on conventional and nuclear
power stations. The increased frequency of these major inspections offer the opportunities to
both power station operators and service providers to develop long terms contracts which can
be closely aligned to the complex operational issues facing power stations operators in today's
electricity markets.
Ensuring that the inspection period achieves the technical, commercial, environmental and
safety targets is a complex process which requires a lot of planning and management expertise.
There are many different contracting options available in deciding which company and in what
commercial framework to contract with. In deciding the best option for a particular power
station, most decisions taken by the power station will be decided by considering the financial,
risk, technical competence and resource issues. Siemens Power Generation Ltd believes that
developing long term relationships with clients allow both parties to invest in the people that
ensure that turbine maintenance is performed on time to the right quality and safety standards.
We favour a partnership approach whereby we discuss the various options available to clients
and after discussing the various options agree what is the best all round solution.
252
C583/016/2000 IMechE 2000
It takes a lot of time, effort and money to procure and manage a turbine overhaul contract
which meets the customers expectations in terms of programme, quality and cost. Success can
be best achieved by
* Selecting a service partner in which you are prepared to develop a long-term relationship.
* Work together in a non-adversarial and trusting way
* Invest in individuals, tooling and processes to ensure continuous improvement.
* Structure a contract which allows for flexibility and can accommodate changing market
conditions.
* Continuously work at the relationship.
* Learn from experiences gained and don't reinvent the wheel every time you carry out an
outage.
C583/016/2000 IMechE2000
253
Figure 1 Stages in developing contractual relationships
Fig 2 Target cost approach to performance based contracts
C583/016/2000 IMechE2000
255
This page intentionally left blank
Authors' Index
A
Abbott-Garner, J 171-182
B
Billingsley, J T 125-134
Bons al l . RW 245-256
Brear, J M 97-108
Brown, P M 205-214
Browne, R J 109-124
c
Clarke, S F 13-22
Copeland, C 53-62
Crane, O 63-74
Grassland, A W 3-12
D
Dixon, MR 75-84
Dolman, C J 145-160
Donnellan, PJ 195-204
Dorling, K 185-194
G
Grif f ith, G 137-144
H
Hewitt, G 185-194
J
James, P J 23-32
Johnson, N M 185-194
Johnson, S P 217-228
Jones, MC 205-214
K
Kendrick, S 171-182
L
Lant, R P D 97-108
Lewis, J J 229-244
Lindsell, G 13-22
M
Macdonald, J 185-194
Maw, S 171-182
Milsom, J 137-144
Myers, A 85-94
P
Payne, B 13-22
Platt, J A 145-160
Putman, R E 161-170
S
Sharp, L 229-244
Shipley, MP 109-124
Stone, G 137-144
Swallow, A P 145-160
T
Tucker, S 137-144
V
Valsa, O 217-228
W
Walker, R 161-170
Wood, M1 97-108
Worsfold, J 185-194
Wright, H 33-50
257