RAM modelling using Optimise” MODULE 1 RAM OVERVIEW LES Move Forward with ConfidenceRAM Modelling using Optimise? Training Course- Module 1 n REVISION HISTORY Written by, Reviewed by Verified by Date Revision G Rocha 'S Daniel A Bennett 10 June 2009 | For Internal Review 7 } G Rocha AHowitt ABennett 24 August 2009 | Issued to training ; i J we 1 J § Bliibh Veidovbihy a witk ~ Sov 3- Sat 7 J 4 3 a (©2009 Bureau Veritas Australia Ply Lid Page 2 4RAM Modeling using Optimise® Training Course- Module 1 FOREWORD This material has been prepared as part of the training course “RAM Modelling using Optimise” which is to be delivered to the Bureau Veritas Network in Late October 2009. This module introduces important concepts in Reliability, Availability and Maintainability modelling. Data contained in this report is provided as examples only. Users must ensure that appropriate data is obtained before applying it. Bureau Veritas IRC accepts no liability for application of the data or methods described in this report. = © 2009 Bureau Veritas Australia Pty Ltd Page 3RAM Modeling using Optimise? Training Course- Module 1 ACRONYMS is TABLE OF CONTENTS FOREWORD. INTRODUCTION Module Roadmap .. RAM Introduction — A Brief History .. RAM TERMS AND DEFINITIONS... Terms and Definitions... MONTE CARLO SIMULATION. RAM BASIC THEORY Overview .. Reliability Basic Theory .. System Configuration... RAM PROCESS OVERVIEW. Overview Planning . Procedure. REFERENCES... (© 2009 Bureau Veritas Australia Ply Lid Poge 4 7 aRAM Modelling using Optimise® Training Course- Module 1 TABLE OF FIGURES Figure 1.1 — The Fieseler Fi 103, better known as V-1 Figure 1.2 The Saturn V Rocket launched Apollo 11 Error! Bookmark not defined. Error! Bookmark not defined. 12 14 15 15 Figure 2.1 — Inherent Availability. Figure 2.2 - Example of A Figure 2.3 - RBD Example 1..... Figure 2.4 — RBD Example 2... Figure 2.5 — Flow Network Example. Figure 2.6 - Theoretical Bathtub curve.. Figure 2.7 - Real “Bathtub” curve (Hot alr system). Figure 2.8 - Basic Relationship for R(t)... Figure 2.9 - Relationship between R(t) and F(t)... Figure 4.1 — RAM overview Figure 4.2 - Six patterns of failure. Figure 4.3 - Exponential f(t) over time. 26 Figure 4.4 — Exponential R(t) over time. Figure 4.5 — Exponential A(t) curve over time. Figure 4.6 - Normal f(t) over time. Figure 4.7 — Normal R(t) over time. Figure 4.8 — Normal A(t) curve over time.... Figure 4.9 — Weibull shape parameter effects on the pdf Figure 4.10 - Weibull shape parameter effects on the reliability. Figure 4.11 — Weibull shape parameter effects on the failure rat Figure 4.12 — Triangular Distribution .. Figure 4.13 ~ Rectangular Distributior Figure 4.14 Series System Figure 4.15 — Parallel System 35 Figure 4.16 — K out of N System Figure 4.17 — Bridge System. Figure 5.1 — Useful life extension.. 6 HEE (© 2009 Bureau Veritas Australia Pty Lid Page 5RAM Modelling using Optimise® Training Course- Module 1 & ACRONYMS A Inherent Availabilty Ao Operational Availabilty BSI British Standard international CDF Cumulative Density Function FEED Front End Engineering Design FMECA _| Failure Mode, Effect and Criticalty Analysis FTA Fault Tree Analysis 'so International Organisation for Standardisation MTBF Mean Time Between Failure MITE Mean Time To Failure MTTR: Mean Time To Repair NASA National Aeronautics and Space Administration Paid Process and Instrumentation Diagram P(x) Exceedance — Probability over value x PDF Probability Density Function PFD Process Flow Diagram QRA Quantitative Risk Analysis RAM Reliability, Availability and Maintainabilty RBD Reliability Block Diagram xT Christmas Tree © 2009 Bureau Veritas Australia Ply Lid Page &RAM Modelling using Optimise? Training Course- Module 1 & 1 INTRODUCTION 14 Module Roadmap INTRODUCTION RAM TERMS AND DEFINITION MONTE CARLO SIMULATION RAM Basic THEORY RAM Process OVERVIEW REFERENCES 1.2 RAM Introduction — A Brief History Reliability, Availability and Maintainabiliy (RAM) analysis is a modelling technique that has its origins in the military [1]. During the Second World War, German rocket scientists were ‘concerned with the lack of success of their V1 missile program (Error! Reference source not found.), which had numerous reliability problems. Figure 1.1 - The Fieseler Fi 103, better known as V-1 PS) © 2009 Bureou Veritos Australia Ply Lid Poge 7RAM Modelling using Optimise® Training Course- Module 1 Lead by a mathematician called Robert Lusser, the team of scientists found that the total reliability of a system depended on the reliability of its subcomponents; and that improvement in the reliability of each individual component and reconfiguring the system to improve reliability resulted in an increased likelihood of success. The introduction of more robust (Teliable) components into the V1 missile turned the troubled program into a success. The allied forces were also interested in improving reliability of complex systems both during and after World War Il. This interest resulted in many advances in the understanding of and improvements in the reliability of complex systems. The Apollo Space Program of the United States of America highlights the application of these advances. The Apollo Space Program was a human spaceflight program undertaken by National ‘Aeronautics and Space Administration (NASA) during the years 1961-1975 with the goal of achieving manned moon landings. Figure 1.2 - The Saturn V Rocket launched Apollo 11 on its journey to the Moon, July 16, 1969 ‘As rocket launch, spaceflight and capsule re-entry operations exposed the human operators to great risk there was a need to ensure a high level of reliability in each rocket component and also for the entire system. The attention to improving component and system reliability by the rocket scientists and engineers was a major factor in the success of the Apollo 11 @ SEE (© 2009 Bureau Veritas Australia Ply Ud Page 8RAM Modelling using Optimise® Training Course- Module 1 mission. This success came on July 20, 1969 when astronauts Neil Armstrong and Buzz Aldrin became the first men to land on the moon [Etror! Reference source not found.]. The lessons learned during the Apollo Space Program spurred advances in many related fields of engineering. This included pioneering work using statistical methods to study the reliability of complex systems comprised of many subcomponents. These new statistical techniques were then reapplied in many fields leading to major advances in avionics, telecommunications and computing as well as other fields. The birth of the nuclear power industry, with the potential for highly adverse and long term impacts on the environment and people in the event of a system failure, spurred further research into the reliability of complex systems. This research led to the development of quantitative risk assessments (QRA), [1]. QRA attempts to mathematically define the probability of failure and the consequence of failure for critical subcomponents of a system then use these values to compute an overall risk matrix for the system. QRA is often used during design to drive reductions in risk (of failure and consequence) between design iterations. The idea of quantitative assessment of complex systems has been further developed over time in a variety of ways. The RAM analysis techniques in use today are the result of one line of development. Other methodologies and tools such as Fault Tree analysis (FTA), Event Trees and Reliability Block Diagrams (RBD) have also evolved from this source. Globally, the Oil and Gas Industry has embraced the idea of reliability analysis and especially RAM analysis for driving design improvements. The high cost of equipment failure (through reduced production) means than a highly reliable system is desired. RAM analysis during the design phases of a project provides invaluable information including which are the most critical equipment items and where design changes may be required to increase reliability and therefore production for the life of the production system. The Oil and Gas industry is by no means alone in embracing RAM analysis. Many industries such as Mining and Power Generation and Distribution are striving to improve efficiency and productivity. This constant drive to improve system efficiency and productivity results in increasingly complex systems. Understanding the reliability and productivity of these complex systems is difficult, RAM analysis allows designers to assess these highly complex systems in a rapid and cost effective manner before costly construction or modification takes place. Society as a whole is coming to rely more and more on complex systems. Consider, for example, the impact of the failure in a power distribution system in a major city, the malfunction of an air traffic control system at an international airport, communication failure between networked systems (Internet), or the breakdown of a nuclear power plant. The importance of reliability at all stages of modern engineering processes, including design, manufacture, distribution and operation, can hardly be overstated. RAM analysis is a 6 os (© 2009 Bureau Veritas Australia Ply Lid Page?RAM Modelling using Optimise® Training Course- Module | powerful and effective tool for Reliability Engineers faced with assessing and improving productivity and reliability in production flow systems. 2 RAM TERMS AND DEFINITIONS 24 Terms and Definitions Failure Termination of the ability of an item to perform a required function [3] Examples: Mechanical failure, material failure, instrument failure, etc. Failure Mode Effect by which a failure is observed on the failed item [3]. Examples: Leakage, Vibration, alignment failure (all related to a mechanical failure). Reliability Ability of an item to perform a required function under given conditions for a given time interval [3]. Reliability is the capability of parts, components, equipment, products and systems to perform their required functions for desired periods of time without failure, in specified environments and with a desired confidence [9]. Reliability can be thought of as the probability of conformance to a speciation (required function) over a specific time. This is an inherent characteristic of an item and is only 6 ea © 2009 Bureau Veritas Australia Ply Lid Page 10 al 1 al eee saleRAM Modelling using Optimise® Training Course- Module 1 os influenced by the operating context in which it is working. This is most commorily defined in association with the Mean Time Between Failures (MTBF). Maintainability Ability of an item under given conditions of use, to be retained in, or restored to, a state in which it can perform a required function, when maintenance is performed under given conditions and using stated procedures and resources [3]. The term is also used to denote the discipline of studying and improving the maintainability of products, primarily by reducing the amount of time required to diagnose and repair failures (9). Maintainability can be thought of as the probability of returning a failed item to a specific standard level over a specific time after a failure has occurred. This is the typical time spent fixing a problem, allowing it to get back to service. Its most commonly quoted as Mean Time To Repair (MTTR). Availability Availability is an important metric used to assess the performance of repairable systems, accounting for both the reliability and maintainability properties of a component or system [1]. The classification of availability is somewhat flexible and is largely based on the types of downtimes used in the computation and on the relationship with time (i.e. the span of time to which the availabilty refers) [9]. Awide range of availability classifications and definitions exist. The ones presented here are the most common used in RAM modelling, but variations exist and one should be aware of how they are calculated and what they mean so that one can make an appropriate choice for the analysis to be performed. Production Availability Ratio of production to planned production, or any other reference level, over a specified period of time [1] and [3]. It Is defined as the production throughput of a system as a ratio of potential production in a given operation time as shown in equation (2.1). Actual Production Throughput (2.1. Potential Production Throughput (in given operational ime) ©") Production Availability & (© 2009 Bureau Veritas Austraa Pty Lid Page 1!RAM Modelling using Optimiso® Training Course- Module 1 This type of Production Availability is considered by simulating models (e.g. Monte Carlo), that evaluate the effects of failures in terms of production losses. Inherent Availability, A, Inherent availability is the steady state availability when considering only the corrective ‘maintenance downtime of the system [9]. This classification is what is sometimes referred to as the availability as seen by maintenance personnel. This classification excludes preventive maintenance downtime, logistic delays, supply delays and administrative delays. Since these other causes of delay can be minimized or eliminated, an availabilty value that considers only the corrective downtime is the inherent or intrinsic property of the system. Many times, this is the type of availabilty that companies use to report the availabilty of their products (e.g. computer servers) because they see downtime other than actual repair time as out of their control and too unpredictable. The corrective downtime reflects the efficiency and speed of the maintenance personnel, as well as their expertise and training level. It also reflects characteristics that should be of importance to the engineers who design the system, such as the complexity of necessary repairs, ergonomics factors and whether ease of repair (maintainability) was adequately considered in the design. For a single component, the inherent availability is calculated as per equation (2.2). MITF STE MITF +MTIR For a repairable item, reference [6], equation (2) can be expressed as shown in equation (3). Where: MTBF = Mean Time To Failure (MTTF) + MTTR as, Figure 2.1, [7] — Inherent Availability Figure 2- is © 2009 Bureau Veritas Australia Ply Lid Page 12 5 1 aiRAM Modeling using Optimise® Training Course- Module 1 available failure repair failure not MTTR je—MTTR_,| available = Real-Time ! MTBF ! l 1 Usually MTTF (expressed in years) is much higher than MTTR (expressed in hours), thus some authors consider MTBF = MTTF. is (© 2009 Bureau Veritas Australia Ply Lid Page 13,RAM Modelling using Optimise” Training Course- Module 1 Operational Availability, A, Operational availability is a measure of the average availability over a period of time and it includes all experienced sources of downtime, such as administrative downtime, logistic downtime, mobilization downtime ete, [9] Operational availability is the ratio of the system uptime and total time. Mathematically, itis given by equation (1.4). Uptime OperationalCycle Ao (2.4) Where the operating cycle is the overall time period of operation being investigated and uptime is the total time the system was functioning during the operating cycle. The Operational availability example is illustrated by Figure 2.2, [3] Figure 2.2 - Example of A, ase In this example the total operational cycle time is equal to 2000 hours, the total uptime is equal to 1980 hours, and therefore the Operational Availabilty for the system is 99.5%. Production-Performance Analysis, ‘Systematic evaluations and calculations carried out to assess the production performance of asystem, [3]. Deliverabil vy Ratio of deliveries to planned deliveries over a specified period of time, when the effect of compensating elements, such as substitution from other producers and downstream buffer storage, is included, [3]. RAM Analysis The RAM analysis itself is a production-performance technique which can take into account all failure and repair rates of each of the elements that compose a system along with its physical configuration in order to estimate, by the use of a simulation model, the system's behaviour over a given operational time. The main calculated outputs from RAM modelling are, but not limited to, the following 6 a © 2009 Bureau Veritas Australia Ply Lid Poge 14 3 bed ausing Optimise? Training Course- Module 1 predictions: ‘* production availability over plant life; * contributors to production availabilty losses; ‘+ expected number of repairs/replacements/interventions; and ‘+ the average fraction of time a system produces according to the demand. ‘The results from the analysis are used as input to improve the system design. In terms of “value adding’, the RAM analysis assists the design process in two main ways, [1]: + fulfiidemonstrate requested requirements from customers (e.g. a target system availability); and * improve system performance if possible, applying sensitivity cases and supporting engineering decisions and modifications. ‘The RAM process is further discussed in Section 5. Reliability Block Diagram A diagram that represents how the components, represented by blocks’, are arranged and related reliability-wise in a larger system, [9] ‘A RBD is often used to depict the relationship between the functioning of a system and functioning of its components. In a RBD, a book is often used to represent each component. The name of the component may be given in the block. Figure 2.9 illustrates a generic ‘component i block diagram. Figure 2.3 - RBD Example 1 s—G} 3 From Figure 2.3 when there is a connection between point (a) and (b), we say that ‘component iis functioning. A RBD does not necessarily represent how the components are physically connected in the system. It only indicates how the functioning of the components will ensure the functioning of the system. That is why a RBD represents the logic relationship between the functioning of the system and the functioning of its components. A RBD is best interpreted by considering the flows through the components from left to right. A working component allows flow through it while a failed one does not. Figure 2.4 — RBD Example 2 © ae (© 2009 Bureau Veritas Australia Ply Lid Page 15RAM Modelling using Optimise? Tigining Course- Module 1 o-{1 | 0 (3 | Figure 2.4 represents the success-oriented network describing the function of a hypothetical system. For the system to be functioning, component 1 and one of components 2 or 3 have to function. The formulation of the ABD is fundamental to the creation of a successful RAM model. Without an RBD, the interactions between the various asset sub-systems and equipment components cannot be correctly defined without great difficulty. The primary function of the RBD is to simplify the item being modelled and to provide a framework around which the mathematical model can be created. RBDs have been used to represent series structures, parallel structures, series-parallel structures, parallel-series structures, bridge structures, and general network structures. The diagrams of these structures will be given when they are introduced. However, not all systems can be represented by RBD. Flow Network A limitation of an RBD is that partial failure of system is not easily handled. The Flow Network theory covers multi-state systems, which can be taken into account by the simulations, [7].In short, the Flow Network serves the same purpose as the RBD, but we can say that Flow Networks are RBD which are not binary. ‘When modelling in Optimise©, one can use a Flow Network to state the different system relations, though they often are quite the same as an RBD. A subsea structure with 7 wells, 7 subsea Christmas Trees (XT) and 2 subsea manifolds is used as an example illustrated by Figure 2.5, [7]. Figure 2.5 - Flow Network Example a (© 2009 Bureau Veritas Australia Ply Ltd Page 16RAM Modeling using Optimise® Training Course- Module 1 The main difference between RBD and Flow Networks lies at the parallel links, as illustrated by Figure 2.5. In the “Manifold” level, there are two blocks delivering 4/7 and 3/7 of the well production respectively. This comes in conflict with RBD philosophy since only 3/7 or 4/7 can be managed by the system if one block fails. In a RBD parallel link, 100 % of the production could still be manage by other if one block failed (RBD are binary entities). Flow networks are, therefore able to represent partial system failures where the level of redundancy Is represented but not bound to binary states only. The further mathematical definitions involved in RAM modelling are presented hereafter. Fallure rate, A(t) A function that describes the number of failures that can be expected to take place over a given unit of time. The failure rate function has the units of failures per unit time among surviving units, i.e. one failure per million hours, [9]. Limit, if this exists, of the ratio of the conditional probability that the instant of time, T, of a failure of an item falls within a given time interval, (t + At) and the length of this interval, At, when At tends to zero, given that the item is in an up state at the beginning of the time interval The chart of the “failure rate” over time is known as “bathtub curve" Figure 4.2.6, [4]. Figure 2.6 — Theoretical Bathtub curve 6 fi) (© 2009 Bureau Veritas Austrafo Pty Lid Page 17RAM Modeling using Optimise® Training Course- Module 1 ut) Early failure Useful ite Wear-out failure > th te t In the interval (0,t1), which is usually short, a decreasing failure rate function is observed. This is often referred to as the early-failure period. The failures that occur in this interval are called early failures, burn-in failures, or infant mortality failures. They are mainly due to manufacturing defects and can be screened out using burn-in techniques. In the interval (t1,t2), the failure rate function is fairly constant. This section is often referred to as the useful lfe of the device or the constant failure rate period. The failures that occur in this interval are called chance failures or random failures. They are usually caused by chance events like accidents, overloading, and a combination of the underlying complex physical failure mechanisms. In the interval (t2, =), the failure rate function is increasing. This interval is often called the increasing failure rate period or the wear out failure period. The failures that occur in this period are due to wear out, aging or serious deterioration of the device. The life of the device is close to its end once entering this period, unless there is preventive maintenance or major overhauls to revitalise the device. Mathematically is given by equation (2.5). _ Number of Failures (2.5) a . Total time For example, the failure rate per 100 hours of the hot air system used by an aircraft motor ignition system is illustrated in Figure 2.7, [10]. Bs © 2009 Bureau Veritas Australia Ply Lid Page 18 a 4RAM Modelling using Optimise® Training Course- Module 1 Figure 2.7 - Real “Bathtub” curve (Hot alr system) 0.20 0.15 AE YL00 ° s 0.05 0.00 o 500 1000 1500-2000 Hours (h) Further discussion regarding the Bathtub curve theory and patterns are presented in Section 40. Failure distribution (pdf), f(t) ‘A mathematical model that describes the probabilty of failures occurring over time. Also known as the probability density function (pdf), this function is integrated to obtain the probability that the failure time takes a value in a given time interval. This function is the basis for other important reliability functions, including the reliability function, the failure rate function, [9]. Cumulative density function (cdf), Fit) ‘A function obtained by integrating the failure distribution pdf. In life data analysis, the odf is equivalent to the unreliability function, [9]. is (© 2009 Bureau Veritas Austraia Ply Lid Page 19RAM Modeling using Optimise® Training Course- Module 1 Reliability Function, R (t) Also known as “survival probability’, it is the probability of an item operating for a given ‘amount of time without failure, as given by the equation (2.6), [9]. R(t") = probability (item life >t*) (2.6) Where t’ is a time equal to or greater than 0. Figure 2.8 - Basic Relationship for R(t) ) t t Error! Reference source not found. illustrates a hypothetical distribution of the times to failure of an item (pdf), [5] then: The probability that item will survive beyond time t* is given by equation (2.7) RP) = [fd en The probability that item will fail in a given time less or equal than t* is given by equation (2.8). : 8) = F)=[fOd gy 0 Also the Relationship between reliability and the probability of failure is given by equation (29). RC) -F(’), (2.9) na (© 2009 Bureau Veritas Australia Ply Lid Page 20 ” al 7 u3 alRAM Modelling using Optimise® Training Course- Module 1 Figure 2.9 - Relationship between R(t) and F(t) ‘The most used statistical distributions (pa) to mode! failures and repair over time will be further discussed in Section 4.0. ie ©2009 Bureau Veritas Australia Ply Lid Page 21RAM Modelling using Optimise Training Course- Module 1 oS 3 MONTE CARLO SIMULATION There are many numerical methods used in the area of prediction and simulation to build simulation models (Markov, Petri Net, etc.), but the Monte-Carlo modelling is the technique most widely used for the simulation of complex and dynamic interactions where a number of restricting elements or time-based activities need to be incorporated, [1]. Monte Carlo is a method of generating values from a known distribution for the purposes of experimentation. This is accomplished by generating uniform random variables and using them in an inverse reliability equation to produce failure times that would conform to the desired input distribution, (9). Curiosity: The name "Monte Carlo" is a reference to the Monte Carlo Casino in Monaco. The use of randomness and the repetitive nature of the process are analogous to the activities conducted at a casino. ‘The Monte Carlo model can be used to model any failure pattern, repair characteristic or conditional logic for which data is available. The outputs are a time-based availabilty curve for the length of the simulation. This can be very useful for predictions of periods of low availability during the system lifecycle, such as major shutdowns/turarounds intervals. In addition, the availability of, and contribution to the overall downtime of each branch, sub- system or individual item can be illustrated. This is particularly useful for identifying the major contributors to production losses for use in the criticality assessment, and ultimately the design process. Optimise® software is a Monte Carlo based simulation tool. * Some possible advantages of simulation that may account for its widespread appeal are: Most complex, real-world systems with empirically stochastic elements cannot be accurately described by a mathematical model that can be evaluated analytically. Thus, a simulation is often the only type of investigation possible. * Simulation allows the performance of an existing system under some projected set of operating conditions to be analysed and predicted. * Alternative proposed system designs (or alternative operating policies for a single system) can be compared via simulation to see which best meets a specified requirement. ena (© 200? Bureau Veritas Australia Ply Ltd Poge 22 us t aed ee eed aSRAM Modeling using Opfimise® Training Course- Module 1 In a simulation we can maintain much better control over experimental conditions than ‘would generally be possible when experimenting with the system itself. ‘Simulation allows us to study with a long time frame ~ e.g., an economic system ~ in compressed time, or alternatively to study the detailed workings of a system in expanded time. * The ability to deal with uncertainties concerning failure / maintenance data. * The ability to dynamically model inputs, i.e. with a varying input of production profile. * The ability to model maintenance logistics and constraints (Maintenance Utilities) As the plant development cycle continues through detailed engineering and into operation, the model can be updated and further utilized to evaluate the following: * reservoir performance (based on actual production profiles achieved); ‘+ equipment reliability and maintainability achieved in operation; and + the effects of preventive maintenance activities, maintenance and spares on the overall plant availabilty. The model can be used as an iterative tool in these respects, allowing the asset management team to make decisions based upon a constantly updated picture of the future production output of the system, Simulation is not without its drawbacks. Some disadvantages are as follows: * The large volume of numbers produced by a simulation study or the persuasive impact of a realistic animation often creates a tendency to place greater confidence in a study's results than is justified. If a model is not a "valid" representation of a system under study, the simulation results, no matter how impressive they appear, will provide little useful information about the actual system. * The main disadvantage is the length of time involved in running a model. In order to achieve a satisfactory level of confidence in the model, itis necessary to run the model simulation many times (lifecycles) over time, with each simulation run-time dependant upon the computer processor being used. MEE (© 2009 Bureau Veritas Austrafa Ply Lid Page 23,RAM Modelling using Optimise® Training Course- Module | (G3) 4 RAM BASIC THEORY 41 Overview ‘Once the required level of detail is achieved in the Flow Network, the associated fallure data, and maintainability and logistic information can be added to each block to prepare for the Monte Carlo simulation of system performance over the required time-period as shown by Figure 4.1, [3]. Figure 4.1 — RAM overview Availability (tem) Production Deliverability Availability (System) Uptime Downtime (System) Reliability Maintainability | [Consequence ‘Compensation Failure Rates | | Repair Rates Configuration Butfors MITE MITR Production Profile Line pack Logistic (Demand) In addition to the item failure information, the maintenance profile and resources have to be established. The equipment involved need to have defined maintenance profiles stating how they are repaired. Maintenance profiles may involve the type of maintenance utilties/vessels required , spare parts, maintenance crew available, lead times for repair vessels and spares, mobilization times for crew and repair vessels, and all issues in the maintenance of critical components, {7]. In Optimise®, it is possible to include this maintenance profile for each unit involved in the system, at the modelled level of detail 42 Reliability Basic Theory The theoretical bathtub curve presented in Section 2.0 represents an important concept in reliability engineering from a didactical point of view. However, in the reality, more than one pattern exists. Figure 4.2 illustrates six different failure patterns from the United Air Lines Boing airoraft study performed in 1968, [11] 6 cd © 2009 Bureau Veritas Austraia Ply Lid Page 24 an | i) "RAM Modeling using Optimise® Training Course- Module 1 Figure 4.2 — Six patterns of failure Figure 4.2 shows the failure rate against operating age for a wide variety of electrical and mechanical items. ‘+ Pattern A is the well-known bathtub curve. it begins with a high incidence of failures (infant mortality), followed by a constant or gradually increases of the failure rate, then a wear-out zone. * Pattern B shows a constant or slowly increase of the failure rate, ending in a wear-out zone. * Pattern C shows a slowly increase of the failure rate, but there is no identifiable wear- ‘out age. ‘+ Pattern D shows a low increase of the failure rate when the item is new or just out of the shop, then a rapid increase to a constant level. ‘+ Patten E shows a constant failure rate at all ages (random failures). ‘+ Patter F starts with high infant mortality, dropping to a constant or slowly decrease of the failure rate. The study done on the civil aircraft showed that 4% of the items conformed to pattern A. 2% to B, 5% to C, 7% to D, 14% to E and no fewer than 68% to pattern F. ‘The number of times these patterns occur in aircraft is not necessarily the same as industry. However, other similar studies has been performed and results show that as assets become more complex, there are a predominance of patterns E and F. @ a (© 2009 Bureau Veritos Australia Pty Lid Poge 25RAM Modeling using Optimise® Training Course- Module | In the majority of the RAM models, the focus of the analysis will be the “useful life” period, where the failure rate is constant and modelled by an Exponential distribution 4.2.1 Probability Distributions ‘Common continuous probability distributions used in Optimise® are: © Exponenti * Normal; Weibull; Triangular; and + Rectangular; 4.2.1.1 Exponential Distribution “The {() is given by equation (4.1) , where (\) is the time expressed as a random variable and (A) is the failure rate. Figure 4.3 illustrates the exponential f(t) over time. FO = Ae (4.1) Figure 4.3 — Exponential f(t) over time 0) t The exponential reliability function is given by equation (4.2). Figure 4.4 illustrates the exponential R(t) over time. RO =e (49) Figure 4.4 - Exponential R(t) over time is (© 2009 Bureau Veritas Australia Py Lid Page 26 4 i) wg 3 tuRAM Modeling using Optimise Training Course- Module 1 RQ) t The F(t) provides the unreliability and is given by equation (4.3) F(T) =1-R(P)=1 = (4.3) The failure rate is given by equation (4.4) it de qe 1@= aa (4.5) Itis easy to see that, for the exponential distribution, the A(t) = A. Figure 4.5, illustrates the A(t) curve over time. Figure 4.5 — Exponential A(t) curve over time 2 at t ‘The MTTF for exponential distribution is given by equation (4.6). 1 MITF =7- (48) Ha (©2009 Bureau Veritas Australia Pty Lid Poge 27RAM Modeling using Optimise® Training Course- Module | Where the MTTF is defined as the expectation of the time to failure of an item, [3] ‘The same approach is valid for the repair time distribution, where the MTTR, expectation of the time to restoration, [3], is given the inverse of the repair rate as illustrated by equation (4.7). mrrr=+ (4.7) # ‘The repair rate (i) is given by equation (4.8). Number of repairs experienced Total number of hours repaired (48) 4 2 Normal Distribution The normal f(t) is given by equation (4.9). 4 f= 2s 2) (4.9) Where the “yi is the MTTF and “o” the standard deviation. The Normal distribution envelope isillustrated by Figure 4.6. Figure 4.6 ~ Normal f(t) over time f(t) 25 pb W420 5 The normal reliability function is given by equation (4.10). mg =1-0( SE ae © 2009 Bureau Veritas Australia Pty Ud Page 28 a3 GS te bt alRAM Modelling using Optimise® Training Course- Module 1 Where © is the standard normal distribution N(0,1). The reliability function over time is illustrated by Figure 4.7. Figure 4.7 - Normal F(t) over time R(t) 1 u-2o uw ut2o 4 The normal failure rate is given by equation (4.11), 4 == en fe) _V2n0 ay = 19. (6 Rf) Failure rate Overtime is shown by Figure 4.8. Figure 4.8 - Normal A(t) curve over time Mt w2o ou ut2o 5 4.2.1.3, Weibull Distribution eae (© 2009 Bureau Veritas Australia Ply Lid Page 29RAM Modelling using Optimise Training Course- Module 1 This is most used distribution due to its characteristic to adhere any kind of data. The f(t) is given by equation (4.12). pt a t SO= Ae} ex] -¢) (4.12) a\n. 7 Where (8) is the shape parameter and (n) is the characteristic life. Depending on the values of its parameters, the f(t) can assume different shapes as shown by Figure 4.9, [9]. igure 4.9 - Weibull shape parameter effects on the pdf Weibull paf with O4 0100 009 o060 © soso to20 D 20000 0000 enoan e000 100000 Tine If B = 1, the Weibull distribution is equivalent to an Exponential distribution. When B => 3.5, the Weibull distribution is approximately equivalent to a Normal distribution. ‘The Weibull Reliability function is given by equation (4.13). RQ)= ol-(‘) (4.13) Figure 4.10 illustrates the different reliability curves for different B, [9]. © oa ©2009 Bureau Veritos Australia Ply Lid Page 30 I ba ws C3 68 Fa tts eaRAM Modelling using Optimise® Training Course- Module | Figure 4.10 - Weibull shape parameter effects on the reliability ‘Weibull Retaiity Plot w/ O<1, Bet, 4 100 20 00 om oxo 2 os ow om om oo 0 somo zsoco 2000p, ao codo0 soban 7000 The Weibull failure rate is given by equation (4.14). 2-1 al) 42) (4.14) n\n Figure 4.11 illustrates the effects of the shape parameter over the failure rate function, [9] Figure 4.11 - Weibull shape parameter effects on the failure rate ie (© 2009 Bureau Veritas Australia Pty Lid "page aiRAM Modelling using Optimise® Training Course- Module 1 ‘Weibull Fallure Rate wi 04, ft, 4 ono ooo 1 oto ii one oom i § oor 3 cone oco1o owe own ” ' 0 19090 200.00 ome, x00 ep.00 70.00 a " 4.2.1.4 Triangular Distrib al Triangular distribution is typically used as a subjective description of a population for which there is limited sample data. This distribution is particularly useful if limited information is 1 already available with the modal value determined through “educated guess”. Figure 4.12 presents the probability density function of a triangular distribution. i} Figure 4.12 — Triangular Distrib. ” J Ba J 1 J nH J a A € b ra Optimise® requires three inputs, the minimum (a), the peak (c) and the maximum (b) for a + ae (© 2009 Bureau Veritas Australia Ply Ltd Page 32RAM Modeling using Optimise® Training Course- Module 1 oa triangular distribution. The failure of a manual isolation valve is rare, thus limited failure data is available for this equipment. However, through consultation with experienced operation personnel, the following information is known: * the longest known operating duration before failure was 35 years; * onthe average, the valve is expected to failure once every 15 years; and ° ‘the shortest operating duration before failure was ten years. From the information presented above, a failure mode using triangular distribution can be used. 4.2. Rectangular distribution presents equal probability for all intervals between the stated minimum and maximum values. Figure 4.13 presents the probability density distribution of a rectangular distribution. .5 Rectangular Distribution Figure 4.13 — Rectangular Distribution a b Optimise® requires two inputs, the minimum (a) and the maximum valve (b). ‘The mobilisation of a maintenance crew requires a minimum of one hour and must mobilise within two hours of a critical equipment failure. From the information above, the maintenance crew can be modelled using a rectangular distribution with a minimum of one hour and a maximum of two hours. 4.3 System Configuration The equipment configuration will be used as a reference for the development of the Flow 6 An (© 2009 Bureau Veritas Austalio Ply Lid Page 33RAM Modelling using Optimise® Training Course- Module 1 Networks and Block Diagram. Usually the system drawings and technical information are used to identify the system configuration. 4.3.1 Series Blocks Components that directly affect the availability of their parent system are in series, Figure 4.14, If anyone of the components fails, the overall system fails, [4]. Figure 4.14 - Series System = Pa Po Re The reliability of a system with components in series is given by equation (4.15): Rs =JTT Ri (4.5) ea Applying equation (4.15) to the example on Figure 4.14, the system reliability in series is equal to Rs = Ra x Re x Re. If a= Re=Ro= Re= 0.99, then Rs = 0.9703. Assuming a constant failure rate, for n elements, the system reliability and failure rate will be given by equations (4.16) and (4.17), respectively. RS(t)= Ry (8): Ry(t)-ou- Ry) -TT2o=enf x &a} (4.16) Where ‘N’ is the system failure rate. The System MTTF is given by the inverse of the failure rate as shown by equation (4.18). (4.19) 4.3.2 (Active) Parallel Blocks Multiple components that display redundancy toward the availability of its parent system are © SEE (© 2009 Bureau Veritas Australia Ply Lid Page 34 | a a co) te’ 63 63 Ga oa ea & 5&3 eee:RAM Modelling using Optimise? Training Course- Module 1 [anes in parallel, Figure 4.15. If any one or two components fail, the overall system is stil available. If all three components fail, the overall system fails, [4] Figure 4.15 — Parallel System Ry Ro The reliabilty of a system with components in parallel is given by equation (4.20). ~Ta=Ri (4.20) ial Rs Applying equation (4.20) to the example on Figure 4.15, the system reliability in parallel is equal to Rs = 1 —[(1-Ra)(1-Re)(1-Ro)].. If Ras Re=Re= Re 0.99, then Rs = 0.9999, ‘Assuming constant failure rates, system MTTF is given by (4.21). 1 ft 1 1 MITPs=+-++4+-—_1 _ pa aaae (4.21) 43.3 K-out-of-n Parallel Blocks When ‘k’ number of components out of a possible ‘n’ are required for full system availability. For example, for a 2-out-of-3 configuration as shown in Figure 4.16, at least 2 components are required for full system availability, ie. if more than 2 components fail, the overall system fails, [4]. Figure 4.16 ~ K out of N System a ©2009 Bureau Veritas Australia Ply Lid Page 35,RAM Modeling using Opfimise® Training Course- Module 1 outobn Be som b. The reliability and the MTTF of a system with components in this configuration are given by equations (4.22) and (4.23) respectively. RaSa™ SRR 20) sea(n—x)! MITFos"Zy O23) If Ra= Ra=Ro= Ree 0.99, then Rs = 0.9997. 4.3.4 Decomposition Method Decomposition is the method for determining the reliability of complex systems, as the bridge configuration shown in Figure 4.17. The decomposition method is an application of the law of total probability, which involves choosing a "key" component and then calculating the reliability of the system twice: once as if the key component failed and once as if the key component succeeded. These two probabilities are then combined to obtain the reliability of the system, since at any given time the key component will be failed or operating, [5]. Figure 4.17 - Bridge System A B A B Ata ef -{o c D cFlo The reliability is given by equation (4.24). © Sian ©2009 Bureau Veritas Australia Ply Lid Page 36 =a bead td Ca Ss bs tuRAM Modelling using Optimise? Training Course- Module 1 oor rans R= P(E)P(E) + P(E)P(E) (4.24) If Ra= Re=Ro= Re= 0.99, then R= 2R° —SR* +2R° +2R? = 0.9998. Complex arrangements can be very difficult to solve analytically and to model. ae (© 2009 Bureau Veritas Australia Ply Lid Poge 37RAM Modelling using Optimise® Training Course- Module 1 5 RAM PROCESS OVERVIEW 5A Overview RAM analyses should provide a basis for decisi the choice of solutions and measures to achieve an optimum economy within the given constraints. This implies that the analysis should be performed at a point in time when sufficient details are available to provide sustainable results (e.g. Pre-Front End Engineering Design (FEED) and during the FEED phase). However, results should be presented in time for input to the decision process. Figure 5.1 illustrates that good design decisions, standardization and improved technology can extend the assets useful lifetime. RAM analysis is one of the most important tools which can support this process. Figure 5.1 — Useful life extension Pe Ue Er eats Gc Ces sit ra) Peres Gerd fens a © 2009 Bureau Veritas Australia Pty Ltd Page 38 a3 ba ed ta ee " 3 58 aRAM Modelling using Optimise® Training Course- Module 1 The analyses should be consistent with the assumptions and the reliability data should be traceable. Suitable analysis tools, calculation models, data and computer codes that are acceptable to the involved parties should be chosen and validated. Be aware that analysis tools and calculation models are under constant development. RAM analyses should be planned, executed, used and updated in a controlled and organized manner [3]. 5.2 Planning 5.21 Objectives The objectives of the analyses should be clearly stated prior to any analysis. Preferably, objectives can be to: * verify production-assurance objectives or requirements; ‘* identify operational conditions or equipment units critical to production assurance; * predict production availability, availabilty, reliability, etc.; ‘* identify technical and operational measures for performance improvement; ‘* compare alternatives with respect to different production-assurance aspects; * enable selection of facilities, systems, equipment, configuration and capacities based on economic, optimization assessments; and * provide input to other activities, such as risk analyses or maintenance and spare-parts planning. 5.22 Analysis Information The system for analysis should be defined, with necessary boundaries relative to its surroundings. An analysis of a complete production chain can cover both upstream and downstream set of equipment. It would account reservoir delivery, wells, process and utilities, product storage, reinjection, export and tanker off-take. Operating modes for inclusion in the analysis should be defined. Examples of relevant operating modes are start-up, normal operation, operation with partial load and run-down. Depending on the objective of the analysis, it can also be relevant to consider testing, maintenance and emergency situations. The operating phase or period of time for analysis should also be defined. @ a ©2009 Bureau Veritas Australia Ply Lid Poge 39RAM Modelling using Optimise® Training Course- Module 1 The performance measures to be predicted should be defined. In production-availability predictions, a reference level that provides the desired basis for decision-making should be selected. It should also be decided whether to include the effects from planned maintenance shutdowns, as well as those catastrophic events normally identified and assessed with respect to safety in risk analyses. The analysis methodology used should be decided on the basis of study objectives and the predicted performance measures. 5.3 Procedure 5.3.1 Preparation A review of available technical documentation should be performed as the initial activity, as well as establishing liaison with relevant disciplines. Site visits and failure mode, effect and fiticality analysis (FMECA) workshops may be performed and are recommended in cases where the major failure modes are unknown, 5.3.2 Base Case Model ‘The RAM documentation is usually based on two kinds of document: an input report, so called “assumptions” report, and an output report with the results, conclusions and recommendations. The assumptions will describe: + methodology used; ‘+ system description; equipment breakdown; © RBD; and * reliability data register (failure and maintenance). The system description should describe, or refer to documentation of, all technical and operational aspects that are considered to influence the results of the RAM analysis and that are required to identify the system subject to the analysis, e.g. design basis, piping and instrumentation diagrams, process flow diagrams, operation and maintenance strategies, reliability data, maintainability data, equipment criticality information, cause and effect matrices, production profiles, equipment capacities, etc. Areference to the data source should be included. References can be engineering or expert judgement, but historically based data should be used if available. The basis for quantification of reliability input data should be readily available statistics and system/component reliability data, results from studies of similar systems or 6 eee (© 2009 Bureau Veritas Australia Pty Lid Page 40 oo | ia a 3 64 ta be eo os nq al ” aRAM Modeling using Optimise® Training Course- Module 1 ‘expert/engineering judgement. Production, operability and maintainability review sessions ‘can be used to predict plant-specific downtimes (mobilization times, turnarounds, wells management, etc). In the analysis, the approach taken for reliability data selection should be specified, mentioned in the assumptions report and agreed upon by the involved parties. 5.3.3 Model development Develop a base case model that includes the following activities: + functional breakdown of the system; i. evaluation of the impacts of failure and maintenance; * evaluation of events for inclusion in the model, including common-cause failures; ‘* evaluation of the effect of compensating measures, if relevant; and * model development and documentation. The mode! must subject to an independent peer review, as part of Bureau Veritas Quality Assurance practice. 5.34 Results Analysis and Assessment Performance Measures Evaluate the performance measures of the analysed system. Various performance measures may be used; however production availability is the most frequently used. As a predictor for the performance measure, the expected (mean) value should be used. The uncertainty related to this prediction should be discussed and, if possible, quantified. ‘Aspects to be discussed are: * availabilty of systems/subsystems (Mean and standard deviation). . confidence bounds and P(x) values such P(10), P(50) and P(90). Depending on the objectives of the RAM analysis, the project phase and the framework conditions for the project, the following additional performance measures may be reportable outcomes: + main equipment contributors to losses; * expected frequency of repairs and consumption of spare parts; ‘* proportion of time or number of times production is equal to or above demand (demand availability); . proportion of time or number of times production is above zero (on-stream availability); * proportion of time or number of times the production is below demand; and eae © 2009 Bureau Veritas Australia Ply Lid Page 41RAM Modelling using Optimise® Training Couse- Module 1 ‘* proportion of time or number of times the production is below a specified level for a certain period of time. Sensitivity Analyses Sensitivity analyses should be considered to take account of uncertainty in important input Parameters such as alternative assumptions, variations in failure and repair data or alternative system configurations. Sensitivities consider variations to the base case model. It is recommended to perform one change at the time to proper track the impacts on the results. Criticality Analyses In addition to the performance measure, a list of critical elements (e.g. equipment, systems, operational conditions and compensatory means) should be established. This list assists in identitying systems/equipment that should be considered for production-assurance and teliabiity improvement. Complementary analysis, e.g. FTA, FMECA can be used at this stage. When production availabilty or deliverability is to be predicted, relative and absolute losses can be measured to identify contributions to production unavailability from each item/event in order to take account of the effects of compensating measures. 5.3.5 Reporting and Recommendations The various steps in the RAM analysis, as described above and all assumptions should be reported. The appropriate performance measures should be reported for all alternatives and sensitivities. Recommendations identified in the analysis should be reported. Recommendations may concern design issues or further analyses/assessments. Furthermore, recommendations may be categorized as relating to technical, procedural, organizational or personnel issues. Recommendations may also be categorized by whether they affect the frequency or the consequence of failures/events. 5.3.6 Handling of Uncertainty ‘The uncertainty related to the value of the predicted performance measure should be discussed and, if possible, quantified. The quantification may have the form of an uncertainty distribution for the expected value of the performance measure or a measure of the spread of this distribution (e.g. standard deviation, prediction interval). ‘The main factors causing variability (and hence uncertainty in the predictions) in the performance measure should be identified and discussed. Also, factors contributing to the 6 ae (© 2009 Bureau Veritas Australia Ply Lid Page 42 ia) ete ou " a d 1 a 7RAM Modeling using Optimise® Training Course- Module 1 uncertainty as a result of the way the system performance is modelled should be covered. Criticality and sensitivity analyses may be carried out to describe the sensitivity of the input data used and the assumptions made. ae (© 2009 Bureau Veritas Australia Ply Lid Page 43ti (2) 3] (4) 6 7 {8} i) [10] [11] RAM Modelling using Optimise® Training Course- Module 1 6 REFERENCES Strong Gary et al - Bureau Veritas ATL consulting Group — Reliability and Maintenance Strategy Development within Design Projects - A Guide for Management and Engineering Project Teams, Rio 2000. httpv/en.wikipedia.org/ BS EN ISO: 2008 - Petroleum, petrochemical and natural gas industries — Production assurance and reliability management, BSI, 2008. K McFie et al. ~ Optimise training course Manual, Perth 2008. BROOME, Huge; et al. Introduction to Reliability Engineering. American Society for Quality, 1990. O'CONNOR, P. D. T. Practical reliability engineering. 4th ed. London: John Wiley & Sons, 2002. RAUSAND, M. Reliability Theory and Methods, In: Risk and reliability in Subsea Engineering, COPPE/UFRY, Rio de Janeiro, RJ, Brazil, 2007. SMITH, David. Reliability, maintainability and risk. Boston: Butterworth—Heinemann, EUA,1997. VASSILIOU, Pantelis; et al. Life Data Analysis Reference, Weibull ++ Version 6. Reliasoft Corporation, Tucson, 2000. M. Kamins, Rules for Planned Replacement of Aircraft and Missile Parts, RAND Memo, RM-2810-PR, 1962. NOWLAN, F. Stanley; HEAP, Howard F. Reliability-Centered Maintenance. Washington, DC: Department of Defense, 1968. (Report number AD-A066579).. (©2009 Bureau Veritas Australia Ply Lid Poge 44 3 84 SES ba ba ba Ge ba be eS e.3 "1 aRAM Modelling using Optimise® MODULE 2 RAM DATA oa Move Forward with ConfidenceRAM Modelling using Optimise® Training Course- Module 2 REVISION HISTORY J Written by Reviewed by _| Verified by Date Revision n ‘SDaniel/E Yap | @ Rocha ABennett 12 October 2009 | To Internal Revision a Daniel W Fok | G Rocha Bennett 21 October 2009 | Issued To Training 7 ” "7 a " a + ra] J vy ny a vi os a (©2009 Bureau Veritas Australi Pty Ltd Page 1ieee FOREWORD This material has been prepared as part of the training course “RAM Modelling using Optimise@" which is to be delivered to the Bureau Veritas Network in Late October 2009. This module introduces important concepts in Reliability, Availability and Maintainability modelling. Data contained in this report is provided as examples only. Users must ensure that appropriate data is obtained before applying it. Bureau Veritas IRC accepts no liability for application of the data or methods described in this report. aE (© 2009 Bureau Veritas Australia Ply Ltd Page 2RAM Modeling using Optimise® ‘taining Course- Module 2 " ™ i " TABLE OF CONTENTS a FOREWORD. = ” ACRONYMS .. se ee 1 INTRODUCTION... ” 1.1 Module Roadmap di 1.2 Overview 7 2 COLLECTING RAM DATA a 21 Overview 2.2 Equipment Boundary and Hierarchy Detinition.. 7 23 Data Analysis.. ab 2.4 Qualification and Application of Reliability Data .. 2.5 Production-Performance Data ... J 3 DESIGN DATA. 1 3.1 Overview... 1" q 32 P&lDs.. : "1 3.3 As-Built Diagrams 12 3.4 Block Diagrams 13 | 3.5 Equipment Lists and Manuals 16 = 3.6 Basis of Design. 18 7 4 PERFORMANCE DATA. 19 a 44 Sources of Failure and Maintenance Data. - 42 Failure Data Caloulation.. a 5 PRODUCTION PROFILE DATA . 6 FINANCIAL DATA se 7 DOCUMENTATION OF INPUT TO THE RAM MODEL. 7.1 Assumption Document. a 7.2 Reliability Registers . a 7.3 Reliability Block Diagrams .. 8 REFERENCES........ I Haat © 2009 Bureau Veritos Austria Py Lid Page s a© RAM Modeling using Optimiso® a Troning Couse- Mode 2 TABLE OF FIGURES Figure 3.1 — P&lD Example..... Figure 3.2 — FFBD Example. Figure 3.3 - Amine Treatment Plant PFD... Figure 4.1 -OREDA Figure 4.2 ~ Stress Testing Figure 5.1 — Production Profile Example Figure 7.1 - RBD Example List OF TABLES Table 3.1 — Equipment List Example .. Table 7.1 — Reliability Register Example... a ©2009 Bureau Veritas Australia Ply Lid Page 4eee RAM Modeling using Optimise® Training Course- Module 2 &) ACRONYMS BoD Basis of Design CEA Canadian Electrical Association CMMs: Computerised Maintenance Management System EIReDA _| European Industry Reliability Data Bank. EPRD-97 _| Electronic Part Reliability Data FFBD Functional Flow Block Diagram FMD-97 _| Failure Mode and Mechanism Distributions FMECA Failure Modes Effects and Criticality Analysis HALT Highly Accelerated Life Testing \EEE Electrical and Electronic Engineer Iso International Standard Organisation MCR Maximum Continuous Rating MTTF Mean Time To Failure MITR Mean Time To Repair NPRD-95 _| Nonelectronic Part Reliability Data OREDA _| Oifshore Reliability Database PalD Piping & Instrumentation / Process and Instrumentation Diagrams PERT Program Evaluation and Review Technique PFD Process Flow Diagram PM Preventive maintenance RAM Reliability, Availability and Maintainability RBD Reliability Block Diagrams SPIDR™ __ | System and Part Integrated Data Resource SAC System Reliability Centre ZAP Electrostatic Discharge Susceptibility Data 1995 (© 2009 Bureau Veritas Australia Phy Lic Page 5 n 4s eu os G2 6.3 6.3 Ga bud £3 oe eae: iJ 7 JRAM Modeling using Optimse® Training Course- Module 2 & 1 INTRODUCTION 1 Module Roadmap INTRODUCTION COLLECTING RAM DaTA DESIGN DATA PERFORMANCE DATA PRODUCTION PROFILE DATA FINANCIAL DATA DOCUMENTATION OF INPUT TO THE RAM MODEL REFERENCES 1.2 Overview The challenges of developing and sustaining large complex engineering systems have grown significantly in the last decades. The practices of systems engineering promise to provide better systems in less time and cost with less risk, and this promise is widely accepted in many industries. Reliability, Availability and Maintainability (RAM) analysis is a modelling technique that is totally reliant on its inputs to provide a logical, accurate, consistent and useable result. “Garbage in, garbage out” describes the danger in not paying attention to the accuracy and reliability of data gathering and screening procedures. Technology provides many solutions, 6 2 (© 2009 Bureau Veritas Austria Phy Lid Page 6RAM Modeling using Optimise® Training Course- Module 2 but some aspects require good old fashioned engineering and / or common sense to 1 ‘overcome. For example, an individual input, experience and history can be a vital and useful i ‘source of information. If an onsite technician knows from 20 years of history that a valve fails every X months or years the information can be utilised. In order to provide these results, one needs to gather as much accurate information as possible, from all resources at disposal. There are some of many useful sources of 7 information and are categorised as following: © design data; I + performance data; | © production profile data; and & * financial data. e3 3 6B t.3 ad Se (© 200? Bureau Vertlos Australia Ply Lic Page 7RA RAM Modeling using Optimise® “raining Course- Module 2 © 2 COLLECTING RAM DATA 24 Overview This section presents the guidelines in collecting RAM data as outlined in International Standard Organisation (ISO) 20815:2008 [2]. Systematic collection and treatment of operational experience is considered an investment and a means for improvement of production and safety critical equipment and operations. ‘The purpose of establishing and maintaining databases Is to provide feedback to assist with the following: * product design; © current product improvement; '* establishing and calibrating the maintenance and spare-parts programmes; * condition based maintenance; ‘+ identifying contributing factors to production availabilty, through RAM modelling and performance analysis; and * improving confidence in predictions used for decision support. jlable for RAM modelling. Itis often that the clients do not have good / useable data av 2.2 Equipment Boundary and Hierarchy Definition ‘Aclear boundary description is imperative and a strict hierarchy system should be applied. Boundaries and equipment hierarchy should be defined according to ISO 14224:2006 [3]. Major data categories are defined as follows: * installation data: description of installation from which reliability data are collected; * inventory data: technical description of equipment, plus operating and environmental conditions; ‘+ failure data: failure event information, such as failure mode, severity, failure causes, etc; and © maintenance data: corrective maintenance information associated with failure events, and planned or executed preventive maintenance event information. =z ©2009 Bureau Veritas Australia Ply Lid Page 8RAM Modeling using Optimisoo Training Course- Module 2 2.3 Data Analysis To predict the time to failure (or repair) of an item, a probability model should be determined. The type of model depends on the purpose of the analysis. An exponential lifetime distribution can be appropriate. The model, if it is expected to delineate a trend, should be based on the collected reliability data, using standard statistical methods. For further information regarding the statistical distribution, one can refer to the training Module 1 [1] 24 Qualification and Application of Reliability Data The establishment of correct relevant reliability data (ie failure and associated repair / downtime data) requires a data-qualification process that involves conscious attention to the original source of data, interpretation of any availability statistics and estimation method for analysis usage. Suitable reliability data management and coordination are needed to ensure reliability data collection for selected equipment and consistent use of reliability data in the various analyses. Selection of data should be based on the following principles: ‘+ data should originate from the same type of equipment and, if possible, originate from identical equipment models; * data should originate from equipment using similar technology; * data should originate from periods of stable operation, although early life or start up problems should be given due considerations; * data should, if possible, originate from equipment that has been exposed to comparable operation and maintenance conditions; * the basis for the data used should be sufficiently extensive; + the amount of inventories and failure events used to estimate or predict reliability parameters should be sufficiently large to avoid bias from “outliers”; + the repair and downtime data should reflect site specific conditions; * the equipment boundary for the originating data source and analysis element should match as far as possible (study assumptions should otherwise be given); * population data (eg operating time, observation period) should be indicated to reflect the statistical significance (uncertainty related to estimates and predictions) and the “technology window’; and * data sources should be quoted. Data from event databases (compliant with ISO 14224 [3}) provide a relevant basis for meeting the recommendations above. In case of scarce data, it is necessary to use engineering judgement and sensitivity analysis of input data should be done. 6 ] © 2009 Bureau Veritos Austria Ply td Pages I 1 J af “ " a 7 aiRAM Modeling using Optimise® Training Course- Module 2 25 Production-Performance Data Production-performance data at facility / installation level should be reported in such a way that enables systematic production assurance to be carried out. The type of installation and operation determines the format and structure of performance reporting. Annex G of ISO 20815:2008 (2] outlines the types of events that are important to cover for a production facility. It is necessary to establish the relationship between facility-performance data and critical-equipment reliability data. Assessment of actual performance should be carried out by the installation operation on a periodic basis in order to identify specific trends and issues requiring follow up. The main contributors to performance loss and areas for improvement can be identified. In this context, reliability techniques can be used for decision support and calibration of performance predictions. Comparisons with earlier performance predictions should be done, thereby gaining experience and provide feedback for future and / or other similar performance predictions. a (©2009 Bureau Veritas Austria Pty Lid Page 10RAM Modeling using Optimise® Training Course- Module 2 : 3 DESIGN DATA 341 Overview Information from various sources is often used to understand the process, design and configuration of the system. This information is also used to develop the assumptions document, Reliability Block Diagrams (RBDs) and finally the RAM model. Design data information can be presented in many forms, including but not limited to, the following * Piping and Instrumentation Diagrams (P&lDs); * as-built diagrams; + block diagrams; * equipment lists and equipment manuals; and * Basis of Designs (BoDs). 3.2 P&lDs P&IDs show all equipment and piping including the physical sequence of branches, reducers, valves, equipment, instrumentation and control interlocks. A P&ID should include: * instrumentation and designations; mechanical equipment with names and numbers; + allvalves and their identifications; * process piping, sizes and identification; * miscellaneous - vents, drains, special fittings, sampling lines, reducers, increasers and swagers; * permanent start-up and flush lines; * flow directions; interconnections references; * control inputs and outputs, interlocks; * interfaces for class changes; . 8 mic category; (© 2009 Bureau Vertas Australia Ply Lic Page 11RAM Modeling using Optimse® Training Course- Module 2 annunciation inputs; ‘computer control system input; * vendor and contractor interfaces; * identification of components and subsystems delivered by others; and * intended physical sequence of the equipment. ‘As a general rule, a P&ID supplied by the client will be the main source where the majority of the configuration information from. From here one can determine process flow, equipments, valves and specifications, piping and specifications, and most other data you will need to begin creating a complete fleshed out model. Figure 3.1 presents an example of a P&ID. Figure 3.1 - P&ID Example us mmet At Sah) ‘ine Teka lees re ‘rit sang Pot {Sa Sign Piping & Instrumentation Diagram (P&ID) vor Engg TolBoncan 3.3 As-Built ‘As Built Diagrams are a variation of P&ID’s that capture any changes that may have been made during construction. They may also be the only source of data if P&ID’s were not used © Ee (©2009 Bureau Veritas Australia Ply Lid Page 12 gramsRAM Modeling using Optimise® Training Course- Module 2 or kept post-construction. As-Buill Diagrams are usually exists for operating plants and one should bear in mind that itis not seldom that they are not updated. 3.4 3.4.1 Types of block ‘There are numerous different types of block diagrams. The common types are: + basic block diagram; - Functional Flow Block Diagram (FFBD); and * Process Flow Diagram (PFD). 3.4.2 Basic Block Diagram A block diagram is a diagram of a system, in which the principal parts or functions are represented by blocks connected by lines, which show the relationships of the blocks. They are heavily used in the engineering world in hardware design, software design, and process flow diagrams. The block diagram is typically used for a higher level, less detailed description aimed more at understanding the overall concepts and less at understanding the details of implementation. Standard high level block diagrams are of limited use, but they can give a good overall view to assist in managing the model development. 3.43 Functional Flow Block Diagram A FFBD is a multi-tier, time-sequenced, step-by-step flow diagram of a system's functional flow. The FFBD notation was developed in the 1950s, and is widely used in classical systems engineering. FFBDs are one of the classic business process modelling methodologies, along with flow charts, data flow diagrams, control flow diagrams, Gantt charts and Program Evaluation and Review Technique (PERT) diagrams. FFBDs usually define the detailed, step-by-step operational and support sequences for systems, but they are also used effectively to define processes in developing and producing systems. In the system context, the functional flow steps may include combinations of hardware, software, personnel, facilities, and / or procedures. In the FFBD method, the functions are organised and depicted by their logical order of execution. Each function is shown with respect to its logical relationship to the execution and completion of other functions. A node labelled with the function name depicts each function. Arrows from left to right show the order of execution of the functions. Logic symbols represent sequential or parallel execution of functions. © nnn © 2009 Bureau Veritas Australia Ply Lid Page 13, = oe ode Acces eae: aRAM Modeling using Optimise® Training Course- Module 2 ‘The purpose of the FFBD is to indicate the sequential relationship of all functions that must be accomplished by a system. FFBDs depict the time sequence of functional events. That each function (represented by a block) occurs following the preceding function. Some functions may be performed in parallel, or alternate paths may be taken. The duration of the function and the time between functions is not shown, and may vary from a fraction of a second to many weeks. The FFBDs are function oriented, not equipment oriented. In other words, they identify “what” must happen and do not assume a particular answer to "how" a function will be performed. ‘A key concept in modelling functional flow is that for a function to begin, the preceding function or functions within the “contro!” flow must have finished. For example, an “eat food” function logically would not begin until a “cook food” function was completed. The logical sequence of functions (je the functional flow) describes the “control” environment of the functional model. In addition to a function being enabled, it may also need to be triggered with an input. So, in the example, the “eat food” function is enabled once the “cook food” function is completed, and once it receives the “prepared food" as input. This second aspect—triggering a function speaks to the “data” environment. Most system functionality can be modelled using standard symbols. If an extended set of symbols is required, then it should be defined in the resulting Functional Analysis Document to ensure that all stakeholders are able to accurately interpret the diagrams. Figure 3.2 presents an example of a FFBD. Figure 3.2 - FFBD Example orestn AltenteFuncton Ea-e-8 eg BREET | eer sormeronee Eeossrane eae —_—— tkstsints ignite —e| teats = (©2009 Bureau Vets Austra Py ik Page 14RAM Modeling using Optimise® Tiaining Course- Module 2 3.4.4 Process Flow Diagram PFDs present similar information but at a higher level and can be used when P&lDs are not available. A PFD is a diagram commonly used in chemical and process engineering to indicate the general flow of plant processes and equipment. The PFD displays the relationship between major equipment of a plant facility and does not show details such as piping details and designations. Another commonly-used term for a PFD is a flowsheet. Typically, process flow diagrams of a single unit process will include the following: + process piping; * major bypass and recirculation lines; ‘+ major equipment symbols, names and identification numbers; * flow directions; * interconnection with other systems; ‘* system ratings and operational values as minimum, normal and maximum flow, temperature and pressure; and * composition of fluids. PFDs generally do not include: ‘+ pipe classes or piping line numbers; ‘+ process control instrumentation (sensors and final elements); minor bypass lines; * Isolation and shutoff valves; * maintenance vents and drains; + relief and safety valves; and * flanges. PFDs of multiple process units within a large industrial plant will usually contain less detail and may be called block flow diagrams or schematic flow diagrams. Figure 3.3 presents the PFD of an amine treatment plant. es (© 2009 Bureau Veritas Australia Ply Lic Page 15 es 3 GS 6&2 ko kw a3RAM Modeling using Optimiso® Training Course- Module 2 Figure 3.3 - Amine Treatment Plant PFD Sweet gas Condenser (Hz + C02) eG Sate Reflux drum Absorber| Sour Gas Steam Reboiler Condensate sbsorber : 35 to 50 °C and 5to 205 atm of absolute pressure Regenerator : 115 to 428 °C and 1.4 to 1.7 atm of absolute pressure at tower bottom 3.5 Equipment Lists and Manuals Equipment lists and manuals provides valuable information regarding equipment type, capacity and configuration. Table 3.1 presents an example of an equipment list. aE (©2009 Bureau Veritas Australia Ply Lid Page 16a a Pa PS S| 5] SS SF] FS oo ee eo Sr fea Gn mete I Aid yousny soWeA ROeING 6002.8 . [sor eeerest ear] or : 2 a = 7 = ‘ajdwexg ys17 juewdinb3 - jg ages. ZeMPON -sunod BULA, @2s1utudo Bulsn BuNISPOW WrvalRAM Modeling using Optimiso® Training Course- Module 2 3.6 Basis of Design A BoD document is prepared by the project team to present high level information to all parties. BoDs typically consist of the following key information: * overview of the development including limited geographical information; + data of the product and other by-products; + design concepts considered to develop the asset; + primary concept considered in this BoD and its overall development strategy; + functional requirements and process description of key systems; ‘+ process engineering diagrams; and hydrate management, planned * descriptions on various management philosophies maintenance). SEE (© 2009 Bureau Veritas Australia Pty Ltd Page 18RAM Modeling using Optimise® Training Course: Module 2 4 PERFORMANCE DATA 44 Sources of Failure and Maintenance Data 4.1.1 Internal Sources Equipment / systems performance data in forms of frequency of failures, maintenance activities, etc can be a credible source for RAM modelling input. Most companies utilise Computerised Maintenance Management System (CMMS) or their own proprietary software to store the performance data of equipment / systems. However, most of this raw data needs statistical processing and treatments before it can be used. Some commonly used statistical treatment methods are parameters estimation and goodness of fit testing. MMS ‘A CMMS software package maintains a computer database of information about an organisation's maintenance operations. This information Is intended to help maintenance and reliability professionals do their jobs more effectively, such as tracking equipment failure modes, repair times, and failure history, and to help management make informed decisions (ie calculating the cost of maintenance for each piece of equipment used by the organisation, possibly leading to better allocation of resources). The information may also be useful when dealing with third parties (when an organisation is involved in a liability case, the data in a CMMS database can serve as evidence that proper safety maintenance has been performed). CMMS packages may be used by any organisation that performs maintenance or reliability services on equipment, assets and properly. Some CMMS products focus on particular industry sectors (ie the maintenance of vehicle fleets or health care facilities). Other products aim to be more general. Different CMMS packages offer a wide range of capabilities and cover a correspondingly wide range of prices. A typical package deals with some or all of the following: * Work orders: Scheduling jobs, assigning personnel, reserving materials, recording costs, and tracking relevant information such as the cause of the problem (if any), downtime involved (if any), and recommendations for future action. . Preventive maintenance (PM): Keeping track of PM inspections and jobs, including step-by-step instructions or check-lists, lists of materials required, and other pertinent details. Typically, the CMMS schedules PM jobs automatically based on schedules and / or meter readings. Different software packages use different techniques for reporting when a job should be performed. oe (© 2009 Bureau Veritas Australia Ply Lid. Page 19 3 26s esRAM Modeling using Optimise® Training Course- Module 2 * Asset management: Recording data about equipment and property including specifications, warranty information, service contracts, spare parts, purchase date, expected Iifetime, and anything else that might be of help to management or maintenance workers. The CMMS may also generate Asset Management metrics such as the Facility Condition Index. ‘+ Inventory control: Management of spare parts, tools, and other materials including the reservation of materials for particular jobs, recording where materials are stored, determining when more materials should be purchased, tracking shipment receipts, and taking inventory. CMMS packages can produce status reports and documents giving details or summaries of maintenance activities. The more sophisticated the package, the more analysis facilities are available, If a CMMS has been utilised correctly, it can provide a complete history of information on individual equipment including failure data, the exact equipment and parts used, a complete Bill of Materials, manufacturer's data, etc. ‘A poor CMMS is nothing more than a reminder to do PM. Examples of common commercial of the shelf CMMS software are; * Enterprise Resource Planning (SAP); © MAXIMO (IBM); and * Oracle database. Many organisations also have proprietary or custom software to suit their individual requirements. Older sites / plants / infrastructure may have manual maintenance management systems. These are still potentially useful, but will require site access, and a considerable time investment to recover data. Parameters Estimation For the performance data of equipment / systems to be useful, it must be quantified in the forms of mathematics probability and statistic. The probability of failure of equipment is expressed within specified confidence limits (certain values above and below the probability). This confidence limit conveys the aspect of uncertainty, and expressed statistically as parameters. Two basic steps used to estimate the variation of parameters are; firstly stating the maximum and minimum values or tolerance and subsequently, describing the nature of variation [4]. The methods of describing the nature of variation of a data sample can be very meticulous, depending how much level of accuracy one wants to achieve. An example of parameters estimation is to use Weibull plot to obtain weibull distribution parameters. 6 ae (©2009 Bureau Veritas Australia Pty Lid Page 20,RAM Modeling using Optimise® Training Course-Module 2 Goodness of Fit Itis important to determine how well the data fit into an assumed distribution. Statistically the goodness of fit can be tested to estimate the level of confidence (s-significance) of the data. Several methods used for goodness of fit testing are s_* goodness of fit test, Kolmogrov — Smimov test, and the least squares test [4]. 4.1.2 Generic Sources Performance data of equipment can be obtained from relevant historic data. This section presents a list of commonly used technical sources. Offshore Reliability Data (OREDA®) Figure 4.1 - OREDA OREDA? is a project organisation sponsored by eight oil & gas companies with worldwide operations. OREDA’s main purpose is to collect and exchange reliability data among the participating companies and act as the forum for co-ordination and management of reliability data collection within the oil and gas industry. OREDA” has established a comprehensive databank with reliability and maintenance data for exploration and production equipment from a wide variety of geographic areas, installations, equipment types and operating conditions. Offshore subsea and topside equipment are primarily covered, but onshore equipment is also included. The OREDA® data are stored in a database, and specialised OREDA® software has been developed to collect, retrieve and analyse the information. 6 ar (©2009 Bureau Veritas Austraia Pty Lic Page 21 no coor) 3 bt fa tet ca tig a3 ]RAM Modelling using Optimnise® Training Course- Module 2 IEEE Standard 493-2007 Gold Book The objective of the Institute of Electrical and Electronic Engineer (IEEE) Standard 493 is to present the fundamentals of reliability analysis applied to the planning and design of industrial and commercial electric power distribution systems. The intended audience for this ‘material is primarily consulting engineers and plant electrical engineers and technicians. The design of reliable industrial and commercial power distribution systems is important because of the high cost associated with power outages. It is necessary to consider the cost of power ‘outages when making design decisions for new power distribution systems as well as to have the ability to make quantitative "cost-versus-reliability’ trade-off studies. The lack of credible data concerning equipment reliability and the cost of power outages has hindered engineers in making such studies. The 2007 edition of the IEEE Standard 493 overcomes these obstacles. SPIDR The System and Part Integrated Data Resource (SPIDR™) is the new Alion System Reliability Centre (SRC) comprehensive database of reliability and test data for systems and components. SPIDR is a revolutionary replacement for these outdated reliability data resources with more than double the amount of data and updated on an annual basis: + Nonelectronic Part Reliability Data [5]; * Electronic Part Reliability Data (6); * Failure Mode and Mechanism Distributions [7]; and + Electrostatic Discharge Susceptibility Data [8]. SRC maintains extensive quantitative and qualitative databases on components and systems from numerous industry and government test and field sources. The extensive amount of data included within SPIDR is testament to SRC's commitment to collecting reliability data over the last 37 years as the Reliability Analysis Centre. PC and server based options of SPIDR are available. The software includes an interactive user interface, graphical reports, extensive on-line help and a user manual. EIReDA European Industry Reliability Data Bank (EIReDA) is a reliability database prepared by the European Commission and Electricité de France. It contains failure rate data for mechanical and electrical equipment based on experience in nuclear power plants operated by Electricité de France. CEA Generation Equipment Data Bank The Canadian Electrical Association (CEA) publishes the Generation Equipment Status ‘Annual Report. This report contains the following information for over 850 generating units in Canada, 6 Sr (©2009 Bureau Veritas Australia Ply Lid Page 22RAM Modeling using Optimise® Training Course- Module 2 Section 1 contains information on data contributors and the scope of the report. Section 2 contains the distributions of the generating units by age and Maximum Continuous Rating (MCR), as well as a summary of the unit types, operating experience and the top five causes of outages. Section 3 lists the top ten generating unit performers for the year. The definition of terms and the tables and graphs containing the performance on the basis of age and MCR, Operating Factor, and totals by unit and fuel types, along with detailed outage and cause statistics are all contained in the appendices. Military Handbook (MIL-HDBK-217F) ‘The purpose of the MIL-HDBK-217F [9] is to establish and maintain consistent and uniform methods for estimating the inherent reliability (ie the reliability of mature design) of military electronic equipment and systems. It provides a common basis for reliability predictions during acquisition programs for military electronic systems and equipment. It also establishes a common basis for comparing and evaluating reliabilty predictions of related or competitive designs. The handbook is intended to be used as a tool to increase the reliability of the equipment being designed. WellMaster Database WellMaster database is a database designed by Exprosott, for the input and analysis of reliability data for completion equipment. The database covers all major completion equipment items with emphasis on surface controlled subsurface safety valves. Detailed data of other vital downhole equipment such as electrical submersible pumps, permanent gauges, tubing and intelligent completion items are also in this database. Subsea Master Database Subsea Master database is a database that covers reliability data on subsea production systems and intervention systems. The database includes the riser terminations, wellhead and surface equipment supporting the subsea equipment. The database is gathered from all deepwater assets with an up-to-date information from deepwater operations at Gulf of Mexico, North Sea and West Africa. 41.3 Reliability Testing A new type of equipment usually needs to be tested in order to ensure the design reliability under the expected operating environments and the expected operating life meet the specification. Some examples of reliability testing are Highly Accelerated Life Testing (HALT), durability testing and vibration testing. Figure 4.2 illustrates the probability density function, f(t) of an equipment in two different usage levels; use stress and high stress. 6 a (©2009 Bureau Veritas Australia Ply Lidl Page 23 tod to t2 8S 62 6 ee eS asRAM Modeling using Optimiso® Training Course- Module 2 Figure 4.2 HALT Example [10] Stress 4.1.4 Workshop Data The reliability of equipment is strongly influenced by decisions made during the design process. One of the most widely used to determine the reliability of a product based on the design analysis is Failure Modes Effects and Criticality Analysis (FMECA). FMECA analysis is usually workshopped to identify reliability, critical failure modes and effects of equipment. 42 Failure Data Calculation The historical information of equipment failure contained in the databases need to be transformed into useable data for RAM modelling. The OREDA handbook [11], presented in section 4.1.2, is one of the most widely used as a failure data source for offshore and subsea equipment. It contains two different approaches to calculate the Mean Time To Failure (MTTF) and Mean Time To Repair (MTTR) based on historical information from OREDA are: * homogeneous sample approach; and + multi sample approach. & (©2009 Bureau Veritas Australia Ply Lid Page 24[PA RAM Modeling using Optimise® Training Couse- Module 2 on Homogeneous Approach This approach assumes that the failure data is from identical items and operated under the same operational and environmental condition. Failure data calculation for homogeneous approach is depicted in equation (1). 10° * Aggregated time in service a Total number of failures 8760 MTTF (years) =| Multi Sample Approach This approach should be used when the data is taken from various different installations, with different operational and environmental condition. Most of the OREDA data is compiled from different installations, and it Is strongly suggested that the failure data calculation is performed using multi sample approach Failure data calculation for multi sample approach is depicted in equation (2). 1 ‘Mean failure rate (2) 10° MTTF (years) = ( @) ) *8760 Where, the failure rate in the OREDA is expressed in failures per 10° hours and the 8760 factor is to convert hours into years. ie (©2009 Bureau Veritas Austrata Pty Lid Page 25 ra B32 88 bs be bk es bt ee ros