Risk analysis of a construction

cost estimate
D F Cooper*, D H MacDonald-f and C B Chapman*

was proposed for a region with known hydroelectric

A risk analysis of a construction cost estimate for a potential. Attention was focused on a stretch of river
large hydroelectric project is discussed. The purpose of with a number of possible dam sites. At the prefeasibility
the study was to provide an independent check on the stage, preliminary design studies had isolated the three
reliability of the estimate and the adequacy of the or four most promising alternatives and provided
contingency allowance. The paper describes the base- outline estimates of their construction costs and power
cost estimate, the risk analysis approach, the sources of generation potential. The risk analysis described here
risk considered, the details of the risk analysis method concentrates on the most likely of these alternatives as
and the implementation of the risk analysis. The its base case. It corlsiders only the construction costs for
relationship between this form of risk analysis and the the hydroelectric development; the transmission system
cost estimating process is discussed in some detail. It is is not included.
concluded that a combined approach may be possible The risk analysis was commissioned in part because
and that the benefits could be substantial. the energy environment had become less certain than it
was when the project was conceived. Predictions of
Keywords: estimation, risk analysis, construction future demand for energy in the region had been scaled
down owing to the general economic recession, unstable
world oil and gas prices and an unclear pattern of future
A risk analysis of a construction cost estimate was regional energy production. As a result, there was
prepared early in the feasibility study for a proposed some doubt about the overall commercial feasibility of
hydroelectric development. The purpose of the risk the development, and a need was seen for the capital
analysis was to provide an independent check on the costs to be re-examined. This was reinforced by a
reliability of the cost estimate, using a different general feeling on the part of the utilities involved that
approach from that of the original. In particular, the the original cost estimates may have been optimistic
risk analysis was to examine the expected cost of the and that the contingency allowances might be too
project and to assess whether the contingency allowance small.
was appropriate. Although discussed here in a hydro-
electric context, the risk analysis process is a general
one. It has also been used successfully in a study of the BASE-COST ESTIMATE
construction costs of large structures for offshore The original estimate was derived from a traditional
hydrocarbon production. engineering analysis of the preliminary design plans
and drawings. It decomposed the cost of the project
into line items representing the major activities or
PROBLEM CONTEXT
acquisition costs. Each line item showed labour, mater-
Several years ago, a large hydroelectric development ial, equipment and indirect cost components. Different
levels of indirect cost were attributed to different
classes of line items according to standard practice; for
*Department of Accounting and Management Science, University of example, the indirect cost component of a labour-
Southampton, Southampton SO9 SNH, UK, and Acres International
intensive activity was a higher proportion of the cost
Management Services, UK (D F Cooper is currently with Spicer and
Pcgler Associates. X-60 St Mary Axe, London EC3A 8BJ, UK.) than that of a line item representing the capital cost of a
tAcres International Limited, Niagara Falls, Canada turbine unit.

Vol 3 No 3 August 1985 0263-7863/85/030141~)9 $03.00 @ 1985 Butterworth & Co (Publishers) Ltd 141
 Cost variability and uncertainty was acknowledged
1 Preliminary works
by incorporating a contingency allowance in the esti-
1 .l Site development and associated works
mate. This was calculated as a proportion of the total
1.2 Construction camp
construction cost less engineering, management and 1.3 Construction camp operation
owner’s costs. The contingency proportion reflects past
2 Concrete structures
experience with this kind of project, industry practice
2.1 Common considerations
and the ‘feel’ of the cost estimating team.
2.2 Spillway
The original estimate had been through one review 2.3 Intake
process. This was a broad scan to detect obvious 2.4 Powerhouse
anomalies and discrepancies and to escalate the estimate 2.5 Concrete gravity structures
to current costs. The original estimate was revised only 3 Fill structures
where serious differences or errors were detected. 3.1 Common considerations
Many minor differences were ignored, even if the 3.2 Diversion stage I
original appeared to be lacking, on the basis that the 3.3 Diversion stage II
original estimators’ assumptions and working calcu- 3.4 Main dam
3.5 Other fill structures
lations were not always available to the review team,
and at such an early stage in the feasibility study, one 4 Electrical and mechanical equipment

set of assumptions was probably as good as another. 5 Indirect costs

(The possible variability identified here undoubtedly 5.1 Personnel salaries and expenses, and site
contributed to the utilities’ feeling that the contingency expenses
allowances were too small.) 5.2 Bonds and insurance
5.3 Contractors’ financing
The base estimate for the risk analysis was the
5.4 Contractors’ head office expenses
original cost estimate as revised by the subsequent 5.5 Contractors’ profit and contingency
review. Indirect costs were extracted from the line
6 Engineering, management and owner’s costs
items in the base estimate and shown as a separate
7 Reservoir clearing
item. Table 1 summarizes the relative proportions of
major components in the estimate to the whole 8 Reservoir seepage control
estimate. 9 Global risks

10 Escalation risks

RISK ANALYSIS APPROACH

The line items in the revised estimate were used as the Figure I. Activitylcost item structure used in the risk
basis for risk consideration. These line items were analysis
decomposed or consolidated into activity/cost items as
required to provide base costs for which common risks
could be assessed. describe all residual effects not identified specifically.
For some line items having a large impact on the total Some risks common to all activity/cost items were
cost of the project, decomposition was required. distinguished as global risks.
Alternatively, where several line items had a relatively The effect of each risk on an item was assessed as a
small impact on the cost, and where risks were similar, distribution of proportional variations on the base-cost
consolidation was appropriate. The object was to estimate for the item. Proportional variations are
ensure that sufficient detail was employed to assess the usually more convenient to estimate and manipulate
risks accurately, but no more than necessary. Figure 1 than variations expressed in dollar values; they avoid
shows the item structure used in the analysis. the problem of ‘negative dollars’, which can occur when
A limited number of key risks were identified for successive absolute reductions to the base-cost estimate
each activity/cost item. ‘Estimation risk’ was used to are used, and they embody a simple multiplicative form
of dependence that is usually realistic. Statistical
dependence required separate consideration.
Table 1. Summary of base costs, as percentages of the Risks assessed for each individual activity/cost item
total project cost were combined into a total risk distribution for
combinations of items. Each succeeding combination of
risks was added to the aggregation of the preceding
Base cost Total project cost, %
combination to generate the total risk distribution,
incorporating risks common to all activities last.
Preliminary works 11
An outline of the structure of the basic evaluation is
Civil works 28
given in Figure 2.
Electrical equipment
(:87)
Indirect 19 RISKS
Clearing, seepage control 6
The study considered a range of cost variations, due to
Engineering, management and
a variety of causes, referred to as risks. In general,
owner’s costs
these were variations within the normal range for
(9:) projects of this kind. The risks that affect individual
Contingency 10
line-item costs can be grouped according to whether or
Total project cost 100
not they affect the quantity estimate, the unit cost

142 Project Management

 Total
I risk analysis
I

I
2.9
I
Global Direct
risks costs

Figure 2. Risk analysis structure outline

estimate or the schedule. Risks comprising a fourth might affect the abutments and the length of the main
group act uniformly on all activities; these global dam.
risks include labour rates, contractors’ profit margins . Compaction Compaction variations might affect fill
and taxes. quantities for the dam.
l Estimation Drawings might not be accurate or com-
prehensive enough for detailed quantity takeoffs,
Quantity risks
and other considerations like those noted above
Design The engineering design may not have been might not have been identified specifically.
finalized. This refers to design changes that do not
alter the overall concept of the project. For example,
Unit cost risks
changes in concrete slab thickness or pier design were
included here, but changes in the relative placement Engineering approach The engineering approach
of the powerhouse and spillway were excluded. might affect unit costs by varying the mix of labour
Engineering approach The detailed engineering was and equipment. For example, the method for concrete
not complete, and there may be alternative ap- delivery from the batch plant would affect the
proaches. For example, a different approach might processing requirements for aggregates.
vary the number of construction joints in the concrete Weather Weather conditions might affect labour
structures and thus affect the quantity of formwork productivity.
required. Processing The cost of processing quarried or bor-
Definition Sometimes it was not clear in the estimate rowed materials might vary according to the availa-
what was included or excluded from a line-item cost. bility of suitable materials.
Rock quality Poor rock and other geological con- Formwork reuse The ability to reuse or repair
ditions might force excavations to be deeper than formwork would alter the formwork unit cost. (This
planned. is related to the engineering approach.)
Ground contours The profile of the river bed might Royalties Royalties might have to be paid for quarried
not be known in detail. or borrowed materials.
Overbreak Significant overbreak during excavation Placement Some of the fill structures required mat-
would increase the quantity of excavated material to erial to be placed in flowing water. The risk of loss of
be removed, and additional fill or concrete would be fill material was incorporated in the unit costs.
required to compensate. Dewatering Rock and cofferdam porosity might af-
River bank characteristics River bank instability fect the cost of dewatering the cofferdams.

Vol 3 No 3 August 1985 143

. Engineering and management rates Variations in of these risks or their possible impacts on the project
contract conditions and in the number and types of cost and schedule, although it would certainly have
contracts might alter the requirements for site engin- been possible to do ~0’~~.
eering, office engineering and management.
. Estimation Unit-cost estimation variations arise be- Major design changes The base-case layout and
cause of assumptions about productivity, equipment design was examined, and minor design changes
and labour costs, component costs, embedded mater- within this general concept were assessed, but major
ials, etc., not previously considered. changes were not.
Site changes Rearrangement of the site layout and
relocation of the project were not considered.
Schedule risks Water charges No charges for water were considered.
Labour problems Major long-term strikes were ex-
It was assumed that if delays occurred, or if a likelihood eluded.
of schedule overruns was detected, additional resources Land acquisition The purchase costs of land were
would be used to maintain the schedule, if possible. not assessed. The costs of the purchasing activity
Schedule-recovery risk was included to consider the (legal costs, etc.) were assumed to be included in the
consequences for the project cost of such changes to the owner’s costs.
proiect schedule. Maior catastrophes that might cause Major floods Floods that overtop the cofferdam at a
whole season to be <ost were excluded from the study. critical stage were considered to be unlikely, and they
were not included. (If the main dam were to be
Weather Adverse weather conditions might cause washed out, at least one season would be lost.)
delays. Jurisdictional and regulatory processes Jurisdictional
Seasons A late spring or an early autumn might and regulatory delays might be caused by problems
reduce the summer construction period. associated with land ownership, mineral rights, access
River levels High river levels might cause schedule rights (e.g. to fill sites), environmental factors and
delays if they occur at critical stages of the project. local community social factors.
Conversely, lower-than-expected river levels could
be beneficial. The river levels throughout the system
RISK ANALYSIS METHOD
were monitored, so that sufficient warning could be
given of high river levels for adequate preparation The construction cost estimate consists of a set of major
and counter measures to be implemented. line items that specify the estimated cost of particular
Equipment delivery If electrical and mechanical construction and acquisition activities. Many of these
equipment were to be delivered late, there might be a line items are not related directly to one another, and
delay to commissioning. the areas in which they interface or require coordination
can be isolated. For this reason, the line items provide
the basis for the risk analysis approach, and possible
Global risks sources of cost variation are examined for each one of
them. An outline of the approach is shown in Figure 3.
Global risks apply to all construction activities.
The first stage of the analysis is to decompose the
estimate of total cost into a set of base costs for analysis
l Labour related factors These include possible changes
(lines 1 to 4 in Figure 3). The initial decomposition is
to standard work weeks; labour market effects on the
usually into the major line-item costs. In some cases,
availability of skilled tradespeople, and the existence
the line items will be single activities, or they will be
of an all-trades project labour agreement.
composed of relatively similar subactivities to which a
agreement.
set of risks will apply in a uniform way. Here, the line-
. Bidding environment Competitive pressures and the
item cost will be the appropriate base cost. In other
state of the heavy construction market would affect
cases, the line item will refer to a set of cost
contractors’ markups.
components of different kinds, with different risk
characteristics. Here, the line item should be decom-
posed into sub-item costs to determine the base costs.
Risks not assessed
Figure 1 shows the item structure used in this study.
Risks that might be considered as normal variations for Item 4, electrical and mechanical equipment, was
projects of this kind were examined. However, there treated as a single composite element, since the
were also abnormal or catastrophic variations that were acquisitions involved are broadly similar in nature. On
outside the scope of the study. The abnormal variations the other hand, the elements of item 2, concrete
were treated as project conditions, i.e. the results of the structures, are sufficiently different activities to require
risk analysis were conditional on these abnormal separate treatment.
factors not arising. In general, the analysis should be based on as small a
Most of the risks that were not assessed could cause set of base costs as is sufficient to give reasonable
long delays to the project, of at least a season. The analysis. Further detail can be added later where the
impact of a long delay would depend on its cause and analysis indicates that it is important, but too much
when it occurred in the project schedule. For example, detail too early in the process obscures more than it
a six month delay early in the project might be largely reveals.
recoverable, while the same delay late in the project The second stage of the analysis is concerned with
would almost certainly result in a delay to commis- the risks that might affect each base cost (lines 5 and 6
sioning. No attempt was made to assess the likelihood in Figure 3). For each base cost, a set of risks is

144 Project Management

 T
I Total cost

2 Major line Items

3 Sub-item costs

4 Base costs

5 For each base cost:

a) Variations(risks /

A
specific to the
base cost)

Vnk

b) total vorlatlon Quantity eshmotion risk

Figure 4. Example quantity estimation risk, shown as a

proportional variation on the base value; the probability
6 Costitemvoriatlw 17 17 --- 17 is that of a particular variation or less occurring, -a-*-
base

7 Cost item compositlon

(allowing for dependence $w
.
between Items)

8 Total ccst distribution

r/, T

Figure 3. Outline of the risk analysis method

identified, and each risk is assessed as a distribution of
proportional variation on the cost estimate (Figure 4).
The total variation in each base cost, from all the
risks that act on it, is the product of the individual
variation distributions. In practice, this product is
formed in a series of two-element multiplications: two
variations are multiplied together, the product is 1.0 I.1
ProportIonal variation
multiplied by the next variation, and so on. The
intermediate variations can be plotted on common axes Figure 5. Example cost variation due to three risks; the
to indicate the relative importance of the individual gaps between the lines indicate that the unit cost risk is
variations in the total (Figure 5). most important while the labour availability risk is
The cost-item variation in dollars is the product of comparatively minor, (a) quantity variation; (6) = (a)
the base cost with the total variation. This is just a plus unit cost variation, (c) = (b) plus labour availability
change of scale for the variation distribution (Figure 6). variation = total variation; -.-.- base cost
The final stage of the analysis involves the compo-
sition of the individual cost-item variations to determine the combination of variations should assume a substan-
the variation in the total cost (lines 7 and 8 in Figure 3). tial level of dependence.
This usually follows a structure that is the reverse of the In practice, the total cost distribution is formed in a
initial decomposition into the base costs. series of two-element additions: two variations are
Cost-item variations are combined by adding pairs of added, the sum is added to the next variation and so on.
distributions. The assumptions that are made about the All intermediate sums can be plotted on common axes
statistical dependence relationships between the distri- to indicate how the total cost is built up (Figure 7). The
butions are important. For example, it may be reason- cost variation curves can also be related to the jointly
able to assume that the cost variations for site defined minimum (shifted to the right in Figure 7 to
preparation and spillway construction are largely inde- give a common zero-probability point), to show clearly
pendent, since they involve different labour forces, in which cost items contribute most to the cost variation,
different locations, and different risks apply. However, as illustrated by Figure 8.
the construction of concrete structures involves many The analysis to this point has been concerned solely
common features: they use the same sites, the labour with the effects that risks might have on construction
forces may be substantially the same, quantity estimates costs. With large-scale construction projects, the effects
have been done by the same people, etc. In this case, of risks on the project schedule are also of major

Vol 3 No 3 August 1985 14.5

 importance. Schedule effects are treated in two ways.
First, the direct cost variations due to schedule delays
are incorporated in the analysis, like any other risk
variation. Second, delay effects can also be extracted
for a more detailed time-based schedule-risk analysis,
which can take into account weather windows and the
cascade effects of losing a season on subsequent
activities. This form of schedule-risk analysis was not
considered in this study, but the direct cost implications
of recoverable schedule delays have been included.

IMPLEMENTATION OF THE RISK ANALYSIS

Cost
Figure 6. Cost item variation for an example with base
cost $500 000, -a-.- base cost
The major part of the risk analysis was undertaken by
one senior planning and estimating engineer and one
I I
risk analyst, over a period of about ten days. A second
55 60
analyst was involved in the initial design of the risk
Item vanation, $ x IO'
approach, and other professional engineering and
clerical staff provided assistance with computational
aspects of the study.
The initial review of the cost estimate revealed about
15 main activity/cost items, which were later expanded
to the list in Figure 1. For each item, a consideration of
possible sources of variation led to a detailed structure
of the risks to be assessed, verbal descriptions of the
analysts’ assumptions about what was meant by each
risk and how the variations might arise, and a numerical
specification for each variation distribution. A nested
numbering scheme was used to summarize and keep
track of this information; Figure 9 provides an example.

2.3 Intake

2.3.1 Excavation quantity

2.3.1.1 Estimation
2.3.1.2 Design
2.3.1.3 Rock quality
2.3.1.4 Ground contours
2.3.1.5 Overbreak

2.3.2 Excavation unit cost

Figure 7. Total cost as a sum of component cost
Excavation schedule recovery
distributions, (a) item 1; (b) items I and 2; (c) items 1, 2 2.3.3

and 3 = total cost 2.3.4 Foundation preparation and grouting cost

2.3.5 Concrete quantity

2.3.5.1 Estimation
2.3.5.2 Design
2.3.5.3 Rock quality
2.3.5.4 Ground contours
2.3.5.5 Overbreak

2.3.6 Concrete unit cost

2.3.6.1 Cement and flyash

2.3.6.2 Aggregate
2.3.6.3 Batching
2.3.6.4 Delivery

2.3.7 Concrete schedule recovery

2.3.8 Formwork quantity

2.3.8.1 Estimation
2.3.8.2 Design

2.3.9 Formwork unit cost

2.3.10 Formwork schedule recovery

2.3.11 Steel quantity

2.3.12 Steel unit cost

2.3.13 Granular fill cost

Figure 8. Contributions of individual cost items to the
total cost variation, -‘-.-jointly defined minimum cost;
(a) item 1; (b) items 1 and 2; (c) items 1, 2 and 3 = total Figure 9. Detailed structure of the costs and risks for the
cost intake line item

146 Project Management

 Various forms of output were produced by the
models as described in the previous section and
illustrated in Figures 5, 7 and 8. Presentation of
intermediate results in this way as the study progressed
allowed the estimator to review his assessments so far.
Input specification and coding errors could be detected,
specification consistency could be examined across all
cost items in the estimate, and specifications could be
revised as necessary, in an iterative checking process.
The final results from the risk analysis are sum-
marized in Figure 11. Curve (d) shows the distribution
of total project cost, indicating a 54% chance that the
2.3.5 2 project will be completed within the revised estimate
A, and a 90% chance that it will not exceed the revised
Design &
estimate plus contingency B. There is a 10% chance of
cost overrun.
These results indicate that the original estimate (as
2.3.5.3 revised) was a good one, and that the level of
contingency allowance was fair. In this respect, the
Rock quality k utilities’ original feelings about the inadequacy of the
contingency and the optimism of the estimate were
shown to be unjustified.
The risk analysis provided a review of the original
cost estimate, using a different approach. It proved a
cheap audit of the reliability of the estimate, and it
generated greatly increased confidence on the part of
the utilities. It also had a number of other benefits in
the form of feedback to the project design process,
2.3.5.5 which are noted in the next section.

Overbreak , , b ,
COST ESTIMATION PROCESS

0.8 09 I.o I.1 The form of analysis described could have been
undertaken at the same time as the original cost
estimate was obtained. This would have given a direct
measure of the risk contingency, as well as the cost
estimate, with little change to the estimators’ pro-
Figure 10. Variation distributions for intake concrete cedures. A great deal of additional information could
quantity, 2.3.5 have been obtained, without much additional estimating
effort. This section discusses some of the main issues in
this area.
Variation distributions were defined as proportional
variations on the base estimate. Each distribution was
specified initially in terms of a most likely value,
optimistic and pessimistic values, and probabilities of
exceeding these bounds. This form of specification was
used because the engineer providing the assessments
felt comfortable with it; other specification forms have
been used in other circumstance?‘. Each distribution
was then drawn in the form of a histogram which was
used by the engineer to check his initial specification
(Figure 10).
The histogram representations of the variation distri-
butions were input directly to proprietary computer
software. This was an interactive package providing a
range of editing and data storage facilities, as well as
procedures for combining probability distributions in
various ways. The computational procedures were an
implementation of a set of simple controlled interval
and memory (CIM) models, belonging to a more
general family of risk analysis models. Details of the
models are beyond the scope of this paper3s, but it is Figure I I. Risk analysis results - the total cost distri-
worth noting that they are flexible, computationally bution for the hydroelectric development, (a) direct cost;
efficient and precise, avoiding the restrictions of mo- (b) = (a) plus indirect cost; (c) = (6) plus engineering
ment-based approaches and the computation or error management and owner’s cost; (d) total cost; A =
costs of sampling approaches. revised estimate; B = estimate plus contingency

Vol 3 No 3 August 1985 147

 words, when the cost distributions in lines 2, 3 and 4 of
Concrete volume II= 1800m3 (I) Figure 12 are added to form the distribution of material
Material cost in line 5, the calculation process must remember
Cement 1 800 x (unit cost) (2)
that the same V has been used to derive each
Gravel and sand 1 800 x (unit cost) (3) component, and the software must remember not only
Additives 1 800 x (unit cost) (4) the component cost distributions but where they came
Total material (2)+(3)+(4) (5) from. Similar considerations apply throughout the
Batch and mix scheme in Figure 12.
Labour 1 800 x (unit cost) (6) The computational models used in the hydroelectric
Equipment 1 800 x (unit cost) (7) risk analysis were simple examples of CIM models, and
Total batch and mix (6)+(7) (8) the structure of the analysis was such that it was not
Placing necessary to remember how particular distributions
Rate (including preparation, were derived. However, the wider family of CIM
excluding cleanup) = 15 m3/h (9) models can accommodate the kind of memory that
Time T= 1 800/15 = 120h (10) would be needed in a combined approach to risk
Labour
analysis and cost estimation. An introduction to CIM
1 X Supervisor (unit cost) (11) models is provided in Reference 3 and a more detailed
2 x Foremen (unit cost) (12)
12 x Labourers (unit cost)
description is forthcoming*.
(13)
Total labour (11)+(12)+(13) x 120 (14) A risk-based cost-estimating procedure does not
Equipment
require major changes in current estimating practice.
Computer databases are frequently used to determine
1 x 25T crane (unit cost) (15)
1 x Bucket (unit cost)
labour, equipment and material rates, and computer-
(16)
4 x Vibrators (unit cost) (17) based estimating systems are becoming more common
(15)+(16)+(17) x 120 (18) for the management and control of the progress of large
Total placing (14)+(18) (19) cost estimation projects. Risk analysis could be built in
Total concrete cost (5)+(8)+(19) (20) to such packages, and with good software it could be
Summary almost transparent to the estimators.
Labour
A combined risk analysis and cost estimating process
(21)
Material 1:;+‘14’ (22) would have a number of major benefits. A direct
Equipment (7)+(18) (23) measure of risk, in the form of a distribution of project
Total (20) (24) cost, could be obtained at the same time as the basic
cost estimate was formed. This would allow any
particular contingency value to be interpreted quantita-
Figure 12. Example cost estimation calculation process tively in relation to the possible spread of project costs
for powerhouse substructure concrete and probability of a cost overrun (e.g. Figure 11).
As well as the traditional cost estimate form of Figure
12, insight into project cost risk can be provided by
Figure 12 shows a typical calculation scheme that enabling sources of risk to be identified and examined
might be used in estimating the cost of concrete for a in the style of Figures 5, 7 and 8. This can also give a
powerhouse substructure. The volume of concrete, V, better feel for the important interdependencies between
has been estimated at 1 800 m3 from the design cost items, which can have large combined effects on
drawings (line 1). V is used in a series of calculations the overall cost distribution, such as might be caused by
(lines 2-8); V is also used to determine placing time, T, common underlying factors like common technology,
in line 10, and this appears subsequently in lines 14, 18, common labour forces or common economic and
19 and 20. market effects.
Now, suppose that there is some risk associated with Feedback of risk analysis results can have important
the estimate, in the form of uncertainty about the exact benefits for the project as a whole. The analysis
concrete volume V (perhaps due to design variations, provides additional information about sources of risk,
estimation errors or overbreak during excavation) and and it highlights sensitive parts of the project plan and
uncertainty about various unit costs. The same calcu- areas where design uncertainty is greatest. Design
lation structure as before can be used to estimate the changes or reconsideration of the base plan may be
total concrete cost, but V, T and most intermediate line indicated, allowing improvements in overall project
costs have become distributions rather than single risk performance.
numbers, and suitable computer software is needed Two examples from the hydroelectric study illustrate
instead of a calculator. In principle, though, the the range of improvements that are possible. At the
estimating process itself has changed very little. detailed engineering level, the risk analysis showed that
The computer software necessary to implement an a different concrete placement method from that
estimating procedure of this kind must contain a specified in the original engineering design might be
number of special features. The operations that combine more suitable in terms of its implications for the
line items to form subtotals and totals must recognize quarrying and processing of aggregates, its equipment
that the individual line costs are not independent. For requirements, its reliability and its consequences for
example, V has been used to calculate the three the costs of all concrete structures. At the level of the
components of total material cost, and consequently broad design concept, the analysis demonstrated that
the costs of concrete, gravel and sand, and additives there were major sources of risk associated with the
will all tend to be low or high together, depending on planned method of access across the river in the early
whether the volume estimate is low or high. In other stages of the project, and that several critical activities

148 Project Management

might be affected. The cost and schedule implications REFERENCES
were such that it was recommended that this aspect of
1 Chapman, C B ‘Large engineering project risk
the access plan be re-examined and that an alternative
analysis’ IEEE Trans. Eng. Manage. EM-26 (1979)
site layout that reversed the relative positions of the
powerhouse and spillway be considered. pp 78-86.
2 Chapman, C B ‘Risk analysis of offshore ventures’
AS well as identifying major cost and economic
Proc. Channel Offshore 80 Southampton, UK
uncertainties, analysis of cost risks frequently provides
(February, 1980)
valuable indications of where risks might profitably be
3 Chapman, C B and Cooper, D F ‘Risk engineering:
spread or shed. Depending on who is doing the
basic controlled interval and memory models’ J.
analysis, this might influence how the owner calls for
Opl Res. Sot. Vol 34 No 1 (1983) pp 51-60.
tenders, how the contractor responds to tender invita-
4 Chapman, C B, Cooper, D F and Cammaert, A B
tions and how detailed contract negotiations are
‘Model and situation specific OR methods: risk
conducted. These aspects were considered explicitly in
engineering reliability analysis of an LNG facility’ 1.
the hydroelectric study under items 5 and 6 in Figure 1:
Opl Rex Sot. Vol 35 No 1 (1984) pp 27-35
indirect costs and engineering, management and
5 Cooper, D F, Chapman, C B and Cammaert, A B
owner’s costs.
‘Reliability analysis of an LNG facility’ Proc. ASOR
6th National Conf. Brisbane, Australia (29-31
August 1983) pp 43-53
CONCLUSIONS 6 Chapman, C B, Cooper, D F, Debelius, C A and
A general method for the risk analysis of a construction Pecora, A G ‘Problem solving methodology design
cost estimate has been described. Although it has been on the run’ J. Opl Res. Sot. Vol 36 No 9 (1985) in
discussed here in the context of a hydroelectric press
development project, the approach has wider applica- 7 Cooper, D F, Chapman, C B, Debelius, C A and
bility. It has been used in substantially the same form to Pecora, A G ‘An overview of the risk analysis for a
examine offshore hydrocarbon production structures’. major hydroelectric development’ Private communi-
The study implementation used examples from a cation
general family of risk analysis models, CIM models, in 8 Chapman, C B and Cooper, D F ‘Risk analysis
a specific context. The models themselves have been models’ Private communication
used in a wide range of risk analysis studies; their 9 Cooper, D F and Chapman, C B ‘Construction cost
method of application varies according to the situation, risk analysis for offshore structures’ Private com-
but the same basic models are appropriate ‘~+~,i’ munication
Risk analysis of a project cost estimate has a number 10 Chapman, C B and Cooper, D F ‘Risk analysis:
of benefits. It provides an indication of the reliability of testing some prejudices’ Europ. J. Opl Res. Vol 14
the cost estimate and the adequacy of the contingency No 3 (1983) pp 238-247
allowance, in an independent audit process”. Sensi- 11 Pomeranz, F ‘Control costs with preemptive auditing’
tivity analysis can be undertaken with respect to Power Vol 127 No 4 (1983) pp 65-67
important economic parameters; in the hydroelectric
context, for example, the analysis was conducted
Dale Cooper joined Spicer and Pegler Associates, UK in 1984.
assuming a very competitive heavy construction market, Prior to that, he was Senior Lecturer in Management Science at
and again under more normal market conditions. the University of Southampton, UK. He has research and
Distributional results can be used directly in more consulting interests in many aspects of management science,
general economic risk analyses and related to wider particularly in risk analysis, command and control systems, and
the design of business games. Most of his recent work has been
regional power generation planning. A range of benefits concerned with risk analyses of large engineering construction
at the project level is possible in terms of feedback to projects in Canada and the USA.
the design, tendering and contract setting processes.
Other less specific organizational benefits appear in the David MacDonald is Head of the Planning and Estimating
Department at Acres International in Canada. He has more than
form of study documentation, which provides a struc-
20 years experience in the project management of heavy, civil
tured database of corporate knowledge that usually projects and nuclear power plants, design, project engineering,
resides in the minds of various individuals and might planning and estimating for major hydroelectric projects; design
otherwise not be revealed explicitly. and project engineering of steel rolling mill systems and equip-
Risk analysis of this form could be combined with ment and construction supervision of high-voltage transmission
lines. While the majority of his experience has been in the
regular cost estimation procedures, to produce both a consulting engineering role, he has also been employed by a
risk analysis and a cost estimate at the same time, with major American steel producer and a major Canadian provincial
little change to the estimators’ methods of working. electrical utility.
The general technology exists in the form of CIM
models, although specific software would have to be Chris Chapman is Reader in Management Science and Head of
the Department of Accounting and Management Science at the
written. The benefits of a combined approach could be University of Southampton, and a Director of Acres Inter-
substantial. national Management Services, UK. He holds a BASc in
engineering from the University of Toronto, Canada, an MSc in
operational research from the University of Birmingham, UK,
and a PhD in mathematical economics from the University of
ACKNOWLEDGEMENTS Southampton, UK. He has taken a particular interest in risk
analysis in a wide range of contexts for the past decade, with
This paper was prepared while the first author was responsibility for the methodology of a large number of studies in
Visiting Fellow at the Australian Graduate School of the UK, Canada and the USA.
Management.

Vol 3 No 3 August 1985 149