4.2 Theoretical Overview of Availability and Maintainability in Engineering Design 373 i) Establishing Maintenance Strategies for Engineering Design From the three fundamental principles of a maintenance strategy, it is evident that all required maintenance work is made up of one or more types of maintenance that accomplish specific technical benefits. As stated previously,it is the combination of these different types of maintenance that constitutes a maintenance strategy. From an engineering design perspective, a maintenance strategy is the establish- ment of the most effective combination of the different types of maintenance to be carried out on specific equipment in order to achieve the most desired technical benefit from that equipment. This is determ ined through designing for reliability, availability, maintainability and safety (i.e. designing for engineering integrity— where in this case, the concept of safety is considered as part of designing for re- liability). On the other hand, the most effective com bination of the different types of maintenance for completed engin eered installations (i.e. a maintenance strategy for operational systems and equipment) is established through a RAMS (reliability, availability, maintainability and safety) program (DoD 5000.2-R. 1997). The deliv- erable results are the establishment of operations and maintenance procedures and work instructions in which the different types of maintenance are effectively com- bined into maintenance strategies for specific equipment. The established mainte- nance strategies for the effective care of the condition of engineering equipment are takenupinaRAMS program. The RAMS program The goal of the RAMS program is to establish policies and strategies for effective care of the condition of engin eering systems and eq uipment through the implementation of various RAMS methods and techniques. The objec- tives of the RAMS program are to: • Ensure effective care of equipment condition. • Optimise the technical benefits derived from equipment reliability, availability, maintainability and safety. • Establish priorities for achieving targeted quality and safety. • Establish maintenance strategies for carrying out the most applicable and effec- tive types of maintenance and use of appropriate maintenance procedures and work instructions. • Ensure a correct balance of costs against desired technical benefits. The immediate benefits of the RAMS program are in the establishment of mainte- nance policies and strategies through an analysis and understanding of the follow- ing: • The systems process, equipment functions, failure modes, failure effects, failure causes and failure consequences, and the criticality of equipment failures result- ing in safety h azards, d owntime, and consequential damage, • Identifying equipment conditions and failure characteristics and establishing ef- fective m aintenance through the correct combination of the different types o f maintenance by prioritising the related technical ben efits to be achieved, 374 4 Availability and Maintainability in Engineering Design • Avoiding consequential damage and establishing the necessary maintenance pro- cedures, work instructions and logistic support for equipment care and product quality, • Comparing design integrity as a benchmark against measures of operational in- tegrity. The benefits achieved through the establishment of maintenance policies and strate- gies can be summarised in three fundamental principles of a RAMS program, each relating targeted results and design requirements (in seque ntial order) of safety, re- liability, availability and maintainability to the desired techn ical benefits, perfor- mance measures, consequential effects on the designed equipment, and the required types of maintenance. Principles of a RAMS program in maintenance strategy The first RAM princi- ple in a maintenance strategy is the following logical sequence: Targeted result: SAFETY | Technical benefit: RELIABILITY | Performance measure: MTBF | Effect on equipment: PHYSICAL CONDITION | Type of maintenance: PREVENTIVE MAINTENANCE The second RAM principle in a maintenance strategy is the following logical se- quence: Targeted result: UTILISATION | Technical benefit: AVAILABILITY | Performance measure: POTENTIAL USAGE | 4.2 Theoretical Overview of Availability and Maintainability in Engineering Design 375 Effect on equipment: OPERATIONAL CONDITION | Type of maintenance: ROUTINE MAINTENANCE The third RAMprinciple in a maintenancestrategy is the followinglogical sequence: Targeted result: QUALITY | Technical benefit: MAINTAINABILITY | Performance measure: MTTR | Effect on equipment: REPAIRABLE CONDITION | Type of maintenance: DEFECT MAINTENANCE j) Maintenance Cost Optim isation Modelling Returning to the definition of the goal of maintenance as “that maintenance ac- tion necessary to achieve the correct balance between the costs of input resources and the benefits derived from the performance of effective maintenance a ction”, an additional principle in the understanding of the goal of maintenance, and of mainte- nance as a whole, is the concept of “the correct balance between the costs of input resources and the benefits ”. In a developed maintenance strategy for engineering design, there are two basic types of maintenance costs that relate to the required input resources for effective maintenance: • Costs arising from corrective maintenance action. • Costs arising from preventive maintenance action. Costs arising from corrective maintenance action are the costs of rectifying defects and fixing or repairing equipment. They increase exponentially according to the extent of usage that the equipment will be subject to, and according to the extent of failures resulting in downtime. 376 4 Availability and Maintainability in Engineering Design The manpower costs of corrective maintenance action are partly due to the time taken to restore the equipment to its expected op erational condition within a min- imum period of time or disruption to the overall operational process through the application of defect maintenance. Corrective maintenance costs are thus dependent upon the extent of defect maintenance, the effect of which is determined by MTTR, the performance measure of the equipment’smaintainability.As noted before, main- tainability is primarily a design parameter, and designing for maintainability defines how long equipment is expected to be down after failure, which has a direct impact upon corrective maintenance costs. Costs arising from preventive maintenance action are the costs of detecting po- tential failures and avoiding functional failures. They increase linearly according to the age of the equipment and according to the extent of the maintenance schedules resulting in downtime. The manpower costs of preventive maintenance action, which comprises both scheduled routine maintenance procedures, and scheduled preventive maintenance procedures incur a cost in direct proportion to the amount of routine maintenance being carried out, and to the amount of preventive maintenance being scheduled. Preventive maintenance costs are thus dependent upon the extent of routine and preventive maintenance, the effect of which is d etermined by potential usage and MTBF respectfully, which are the measures of performance of equipment avail- ability and reliability. The inherent availability of equipment is its potential usage with respect to the operable time established from designing for availability, and the inherent reliability of equipment is initially established by its physical design and quality of manufacture estab lished from designing for reliability. By far the largest portion of preventive maintenance costs is associated with scheduled shutdowns and overhauls. Shutdowns and overhauls are scheduled ac- cording to the expected life of the major critical components in process engineering systems and equipment. In certain types of industries, particularly in refineries, sev- eral different types of shutdowns can be scheduled. They are: • Interim shutdowns for vessel inspections. • Open and clean shutdowns. • Annual shutdowns for replacement of worn components. • General overhauls for plant and equipment refurbishment. The scheduled frequency and duration of interim shutdowns for vessel inspections, and of open and clean shutdowns can be determined according to a maintenance strategy in which the most suitable scope of preventive maintenance work is already established during the engineering design stage. The extent and duration of annual shutdowns for replacement of worn components can also be determined during the engineering design stage, and depends not only upon the expected useful life of the critical components of the process engineering design (i.e. failure characteristics) but also on the complexity of integrated systems, the level of equipment and/or component redundancy (i.e. process characteristics), as well as their relevant extent of usage. 4.2 Theoretical Overview of Availability and Maintainability in Engineering Design 377 General overhauls for plant and equipment refurbishment or rebuild are predom- inantly scheduled on the basis of results obtained, firstly, from condition monitoring carried out either periodically or continually and, secondly, from condition mea- surement carried out during interim shutdowns for vessel inspections and open and clean shutdowns. In principle, however, it is obvious that the costs of corrective maintenance action as well as the costs of preventive maintenance action can be ra- tionalised or, in fact, reduced according to the balance o f defect maintenance with routine maintenance and scheduled preventive maintenance, based upon a particular maintenance strategy. Such a strategy has its developed beginnings during the engi- neering design stage, and is progressively modified and improved during the life of the plant. Mathematical model of preventive maintenance replacement costs The opti- mum operational period b etween annual shutdowns for replacement of worn com- ponents can be determined under a maintenance strategy of periodic replacement, irrespective of the age condition of the equipment’s components. According to this strategy, components are replaced at predetermined intervals, CL, typically the length of the preventive maintenance cycle. If a component fails within this preven- tive maintenance cycle, it is minimally r epaired to last for the remaining time of the cycle. Such a minimal repair job, with relatively negligible repair time, implies that the component’s failure rate λ (x), corresponding to its failure probability density function f (x) at the time of failure x (i.e. the instantaneous failure rate), remains the same as it was before the failure (Kececioglu 1995). The cost function for the model is expressed as C pm = C pr +C mr E[ α (T p )] T p (4.106) where: C pm = preventive maintenance cycle costs C pr = the cost of preventive replacement C mr = the cost of minimal repair E[ α (T p )] = the expected number of failures in interval T p and E[ α (T p )] = T p  0 λ (x)dx (4.107) where λ (x)= f(x)/R(x) (4.108) and: λ (x)=the equipment time dependent failure rate f(x)=the equipment failure probability density function R(x)=the equipment reliability function. 378 4 Availability and Maintainability in Engineering Design Substituting Eq. (4.107) into Eq. (4.106) gives the following result C pm = C pr +C mr  T p 0 λ (x)dx T p (4.109) In the case where λ (x), the equipment time-dependent failure rate, has an exponen- tial failure probability density function, i.e. λ (x)= f(x) R(x) = λ e − λ x e − λ x = λ differentiating with respect to T p and setting the resultant equal to zero gives the following: dC pm dT p = −C pr +C mr λ T p T 2 p = 0 . The optimum operational period between annual shutdowns for preventive replace- ment is then T p = C pr /C mr 1/ λ . (4.110) Optimal preventive replacement age of components subject to functional fail- ure In many cases, systems and equipment are subject to functional failure, where- by the equipment or a component of the equipment has to be replaced. Where such functional failure is unexpected, it is not unreasonable to assume that a failure re- placement is more costly than a preventive replacement. For example, a preventive replacement is planned, and arrangements are made for it to be conducted with- out unnecessary delays, o r the unexpected failure may have caused consequential damage to other components. In order to reduce the number of failures, preventive replacements are made. However, a balan ce is required b etween the amount spent on preventive replacements, and the resulting benefits, i.e. reduced failure replace- ments. Such a preventive replacement policy, or preventive maintenance strategy, is one where preventive r eplacements are mad e according to the ‘right’ age of the compo- nent, and failure replacements are done only when n ecessary, to minimise the total expected cost of replacing the component over a period of time. In this optimisa- tion approach, when functional failures occur in equipment, failure replacements are made. The time at which preventive replacements are made d epends upon the age of the component.The problem is to balance the cost of preventivereplacements against their benefits of reduced failure replacements, which is done by determining the optimal preventive replacement age for the component so that the total expected costs are minimised over a period of time. This is achieved with preventivereplacement m odelling with the following prop- erties (Vajda 1974): • C p is the cost of preventive replacement. • C f is the cost of failure replacement. • C c is the total expected replacement cost per cycle. 4.2 Theoretical Overview of Availability and Maintainability in Engineering Design 379 • T c is the expected cycle length. • f(t) is the probability density function o f failures of the component. The replacement policy is to perform preventive replacement once the component has reached a specified age, plus failure replacements when necessary, where the specified age is represented by t p . The objective is to determine the optimal replace- ment age of the component to minimise the total expected replacement cost over a period of time. In this problem, there are two possible cycles of operation: one cycle is deter- mined by the component reaching its planned replacement age, t p , and the other cycle is determined b y the component ceasing to operate due to functional failure occurring before the planned replacement time. The total expectedreplacementcost, C(t p ), over a period of time t p is given by C(t p )= Total expected replacement cost per cycle Expected cycle length (4.111) C(t p )=C c /T c where the total expected replacement cost per cycleC c is given as (the cost of a pre- ventive replacement cycle multiplied by the p robability of a preventive replace- ment) + (the cost of a failure replacement cycle multiplied by the probability of a failure replacement) C c = C p R(t p )+C f [1−R(t p )] (4.112) where R(t p ) is the reliability of the component succeeding to last over the period of the preventive replacement cycle t p . R (t p ) is the probability of no failure occur ring in the time period t p , and the expression [1 −R(t p )] is the probability of failure occurring in the time perio d t p , which is the failure density function . Thus: C c =[C p ×Reliability]+[C f ×Failure density] . The expected cycle length T c is given as (the length of the preventive replacement cycle multiplied by the probability of a preventive replacement) + (the expected length of a failure replacement multiplied by the probability of a failure replace- ment) T c = t p R(t p )+t f [1−R(t p )] . (4.113) In this case, t f is the mean time to fail (MTTF) of the component. Here, it is im- portant to take note of the description of MTTF, compared to MTBF, the mean time between failures. The differencebetween MTTF and MTBF is in their usage. MTTF is applied to items that are not repaired but replaced, such as components, whereas MTBF is applied to items that are repaired. Therefore: T c =[Replacement age·Reliability]+[MTTF·Failure density] 380 4 Availability and Maintainability in Engineering Design The replacement model relates replacement age t p to the total expected replacement cost over a period of time, where C(t p )= C p R(t p )+C f [1−R(t p )] t p R(t p )+t f [1−R(t p )] (4.114) C(t p )= [C p ·Reliability]+[C f ·Failure density] [Age·Reliability]+[MTTF·Failure density] . Thus, the essential integrity measures for determining the total expected replace- ment cost, C(t p ), over a period of time t p , in addition to the cost of preventive re- placement C p and the cost of failure replacement C f are the component (or equip- ment) reliability and failure density. Values for the specific costs of C p and C f as well as component reliability and the failure density (or 1–reliability), and MTTF must be evaluated in order to determine the minimum total expected replacement cost C(t p ) over the period of time t p . Preventive replacement age is where C(t p ) is minimum. Cost of input resource of spares A significant portion of preventive maintenance costs, during ramp-up and the specified warranty period, as well as the remaining life-cycle stages of an engineered installation, is the input resource of spares. Spares for engineered installations can be grouped according to two categories: • Contract spares • Maintenance spares. Contract spares are normally part of the initial procurem ent of systems and equip- ment, and are determined by available reliability data from the manufacturer or ven- dor. The main concern with contract spares is not so much the quantity,or individual cost, but rather their identification. Determination of maintenance spares is achieved through the method of maintenance spares requirements planning (SRP). SRP can be defined as “a strategy involvingthe purchasing, supply, identification, storage and issue of spare parts which improves system maintenance and results in an increase of plant availability”. SRP is different from inventory control. SRP is better suited to maintenance spares that have a high-risk component failure and estimated equipment failure rate. Inventory control is better suited to maintenance spares with low-risk component failure and estimated stock levels. With SRP, the required spares are calculated ac- cording to the estimated failure rate of the relevant equipment, and according to the criticality of the equipment with regard to downtime costs. Inventory control is a resource management system that makes use of calculated order-points, reorder quantities, and forecasts of the stock level at which stock must be replenished as well as the quantity to be ordered. It is evident that SRP considers single items of spare parts for equipment when they are needed, whereas inventory control considers many items to be placed into stock until they are needed. SRP deter mines the efficiency level of the availability of spar es for maintenance, and thus minimises downtime as well as avoids holding unnecessary spare parts in stock. Inventory control determines the service level of the stores in not being out of 4.2 Theoretical Overview of Availability and Maintainability in Engineering Design 381 stock with spare parts, and thus also optimises on spares stock levels (Orlicky et al. 1970). Both SRP as well as inventory control are important to managing spare parts for maintenance, but it is essential to understand that each of these methods are ap- plied to specific types of spares. The types of maintenance spares that are m anaged through SRP and inventory control are determined from the demand for these spares by the type of maintenance action. There are two types of d emand for maintenance spares: • Dependent demand • Independent demand. Dependent demand for maintenance spares relates to the need for the replacement of other components of which the maintenance spare is a part. Dependent demand is based on the systems h ierarchy structure of the process or equipment that forms the b asis of a bill of spares for a spares requirements planning system. Independent demand for maintenance spares relates to the demand of the maintenance spare on its own, and is not subject to the need for other components or parts. Independent demand is based on forecast usage of the spares that forms the basis of order-points and reorder quantities for an inventory con trol system. It is evident fro m these de- scriptions that different categories of spares can be grouped under the two types of demand. There are several general categories of maintenance spares: • Consumable materials (materials that are used up through the maintenance ac- tion, such as oils, greases, waste cloth, etc.). • Consumable spares (spares that are used up in the operation of the equipment or process, such as filters, pump impellors, turbine blades, tube bundles in coolers, etc.). • Replacement spares (parts that become worn through excessive usage or insuf- ficient routine maintenance, or that need to be replaced due to defects, damage or failure. These spares are mostly the p arts of components such as bearings, sleeves, liners, etc.). • Repairable spares (assembled units that are repaired or overhauled through the replacement of parts and then returned to stores (RTS) for later re-issue, such as electric motors, valves, pumps, etc.). • Critical spares (spares that are kept in stores for insurance against hazardous fail- ures of critical equipment, such as special high-pressure or acid resistant valves, high-voltage electrical parts, etc.). • Strategic spares (spares that are kept in stores for insurance against high down- time costs due to long ordering lead times, such as special alloy parts, specialised engineered parts, etc.). There is a further category that is called capital spare s, which are not really main- tenance spares and consist of assembled units that are very expensive and are usu- ally categorised by very high capital equipment industries such as power generation plants. Most stores in industry make use of an ABC classification system to cate- gorise the types of stock being held but, in many cases, this ABC classification has proved to be inadequate to support effective maintenance strategies. 382 4 Availability and Maintainability in Engineering Design Dependent demand maintenance spares usually consist of some replacement spares, repairable spares, critical spares and strategic spares that are stocked be- cause of the risk or frequency of failure of the relevant equipment. These spares are controlled through a spares requirements planning (SRP) system. Preventive maintenance makes use of dependent demand maintenance spares, and is therefore associated with SRP. Independent demand maintenance spares usually consist of consumable mate- rials, consumable spares and some replacement spares that need to be stocked ir- respective of the frequency of component replacement. These spares are controlled through an order-pointand reorder quantityinventory control system. Routine main- tenance makes use of independent demand maintenance spares, and is thus associ- ated with inventory control. Because the sort of maintenance spares that are controlled through an SRP sys- tem are typically the logistic support spares required for shutdowns and general overhauls (i.e. some replacement spares, repairable spares, critical spares and strate- gic spares that are stocked because of the risk or frequency of failure of the relevant equipment), SRP is extremely important for the effective application of preventive maintenance, and also for the effective use of contracted maintenance crews during shutdowns and overhauls (Hillestad 1982). Mathematical modelling of spares requirement Most spares requirements opti- misation models assume the constant failure rate to be a good approximation for a constant demand rate, even if components have non-constant failure rate distribu- tions. Such a failure rate is fundamentally a measure of the intrinsic failure charac- teristics of a componentbrought about by usage stress and load over time. However, it is not quite correct to express the demand rate for a spare simply by the intrinsic failure characteristic of a component. In most cases, the demand for a given spare is the result of a number of fac- tors. Firstly, there may be several different items of equipment that require the same spare. Secondly, there could be several similar parts in each component. Thirdly, there are usually a large number of similar components within each system. Clearly, it is cumbersome to derive the exact spares demand based o n the component fail- ure rate. Furthermore, it is somewhat unrealistic to assume a specific failure rate of a component within a complex integration of systems with complex failure pro- cesses. At best, the intrinsic failure characteristics of components are determined from quantitative probability distributions of failure data obtained in a somewhat clinical environment under certain operating conditions. As indicated before, the true failure process depends upon many other factors, including, for example, rou- tine and preventive maintenance. It is generally accepted that preventive mainte- nance affects the failure properties of components, although it is debatable whether the end result is positive or negative from the point of view of equipment residual life. When modelling spares requirements, the foremost criterion to take cognisance of is that the need for spares is determined by a spares demand. This demand is formed by and dependent upon several factors, such as (Alfredsson et al. 1999): . Theoretical Overview of Availability and Maintainability in Engineering Design 373 i) Establishing Maintenance Strategies for Engineering Design From the three fundamental principles of a maintenance strategy,. designing for engineering integrity— where in this case, the concept of safety is considered as part of designing for re- liability). On the other hand, the most effective com bination of the different. maintenance a ction”, an additional principle in the understanding of the goal of maintenance, and of mainte- nance as a whole, is the concept of “the correct balance between the costs of input resources