Thông tin tài liệu
once a product is in production. However, these methods are more useful at the design stage. We have already seen elements of the CA approach when considering the costs due to safety critical failures. A further insight into the way that failure costs can be esti- mated for non-safety critical failures is also used to support the CA methodology. Estimates for the costs of failure in this category are based on the experiences of a sample of industrial businesses and published material as follows. Consider a product whose product cost is Pc. The costs due to failure at the various stages of the product's life cycle have been investigated, and in terms of Pc, they have been found to be (Braunsperger, 1996; DTI, 1992): 0:1 Pc ± internal failure cost due to rework at the end of the production line Pc ± external failure cost for return from customer inspection 10 Pc ± external failure cost for warranty return due to failure with customer in use. The relationship is commonly known as the `10 rule' and is shown in Figure 1.13. The 10 rule demonstrates how a fault, if not discovered, will give rise to ten times the original elimination costs in a later phase of the life-cycle. In other words, products must be designed in such a way that scarcely any defects develop or if they do, they can be identi®ed as early as possible in the product development process and recti®ed (Braunsperger, 1996). Other surveys have found that these costs could be even higher as shown in Figure 1.14. Suppose a particular fault in a product is not detected through internal tests and inevitably results in a failure severity S 5. If around 80% of failures are found by customer testing and 20% are warranty returns, then the expected cost on average for one fault will be 2.8Pc, from Figure 1.13. If the product has been designed such that C pk 1:33, or in other words, approximately 30 parts-per-million (ppm) failures are expected for the characteristic which may be faulty, then for a product costing £100 the probable cost of failure per million products produced would be £8400. Figure 1.12 Global quality±cost model (BS 6143, 1990) The costs of quality 15 Figure 1.13 The 10 rule of fault related costs by percentage (DTI, 1992) 0 1 2 3 4 5 Relative company cost to rectify error Stage where error is discovered Planning Product design Process design Production Test Delivery Figure 1.14 Cost escalation of rectifying errors at down stream stages (Ostrowski, 1992) 16 Introduction to quality and reliability engineering At C pk 1, or approximately 1300 ppm failures expected, the probable cost of failure per million products would be £364 000. At C pk 0:8 (or 8000 ppm) the probable cost of failure would increase by an order of magnitude to over £2.2 million. These failure costs do not take into account the costs associated with damaged company reputation and lost opportunities which are dicult to assess, but do indicate that failure cost estimates associated with product designs are possible. This aspect of the CA methodology is further developed in Chapter 2. 1.3 How and why products fail 1.3.1 Failure mechanisms We have already established that variability, or the lack of control and understanding of variability, is a large determinant of the quality of a product in production and service and, therefore, its success in avoiding failure. In addition, understanding the potential failure mechanisms and how these interact with design decisions is necessary to develop capable and reliable products (Dasgupta and Pecht, 1991). It is helpful next to investigate the link between the causes and modes of failure and variability throughout the life-cycle of a mechanical product. Mechanical failure is any change or any design or manufacturing error that renders a component, assembly or system incapable of performing its intended function (Ullman, 1992). However, it is also possible to suggest several key aspects of failure (Bignell and Fortune, 1992): . A failure is said to occur when disappointment arises as a result of an assessment of an outcome of an activity. . Failure can be a shortfall of performance below a standard, the generation of undesirable eects or the neglect of an opportunity. . Failure can occur in a variety of forms, namely: catastrophic or minor, overwhelm- ing or only partial, sudden or slow. . Failure can arise in the past, present or future. . Failure will be found to be multi-causal, and to have multiple eects. For the product to fail there must be some failure mechanism caused by lack of control of one or more of the engineering variables involved. Most mechanisms of mechanical failure can be categorized by one of the following failure processes (O'Connor, 1995; Rao, 1992; Sadlon, 1993): . Overload ± static failure, distortion, instability, fracture . Strength degradation ± creep failure, fatigue, wear, corrosion. Figure 1.15(a) shows the results of an investigation to ®nd the frequency of failure mechanisms in typical engineering components and aircraft components. By far the most common failure mechanism is fatigue. It has been suggested that around 80% of mechanical failures can, in fact, be attributable to fatigue (Carter, 1986). Failures caused by corrosion and overload are also common. Although the actual stress rupture mode of failure is cited as being uncommon, overload and brittle fracture How and why products fail 17 failures may also be categorized as being rupture mechanisms, distinct from strength degradation. Figure 1.15(b) provides some insight into the reasons for the mechanical failures experienced. Some root causes of failure are found to be improper material selection, fabrication imperfections and design errors. Other causes of failure are ultimately Figure 1.15 The frequency and causes of mechanical failure (Davies, 1985) 18 Introduction to quality and reliability engineering related to variability in production and service conditions. In general, several primary root causes of failure can be suggested (Ireson et al., 1996; Villemeur, 1992): . Design errors . Production errors (manufacturing, assembly) . Handling/transit damage . Misuse or operating errors . Adverse environmental conditions. From the above, there is a strong indication that the design errors are responsible for many mechanical failures found in the ®eld. This is supported by another study, the results of which are shown in Figure 1.16. Through making poor or uneducated design decisions, the designer alone accounted for around 35% of the failures of those investigated. Historically, the failure of products over their life-cycles, sometimes termed the `bath-tub curve', can be classi®ed into three distinct regions as shown in Figure 1.17. A detailed breakdown of the attributable factors in each region are also given below (Kececioglu, 1991). Infant mortality period ± Quality failures dominate and occur early in the life of the product. In detail, these can be described as: . Poor manufacturing techniques including processes, handling and assembly prac- tices . Poor quality control . Poor workmanship . Substandard materials and parts . Parts that failed in storage or transit . Contamination . Human error . Improper installation. Useful life period ± Stress related failures dominate and occur at random over the total system lifetime ± caused by the application of stresses that exceed the design's Figure 1.16 Designer's responsibility for mechanical failures (designer's share is shaded) (Larsson et al., 1971) How and why products fail 19 strength (most signi®cant period as far as reliability prediction activities are con- cerned). In detail, these can be described as: . Interference or overlap of designed-in strength and experienced stress during operation . Occurrence of higher than expected random loads . Occurrence of lower than expected random strengths . Defects . Misapplication or abuse . `Acts of God'. Wearout period ± Failure occurs when the product reaches the end of its eective life and begins to degenerate and wear out. In detail, these can be described as: . Ageing . Wear . Fatigue . Creep . Corrosion . Poor servicing or maintenance. 1.3.2 The link between variability and failure There exists a relationship between the failure characteristics of a product over its life- cycle, as described by the three periods of the bath-tub curve above, and the phenom- enon of variability. It has already been established that the potential for variation in design parameters is a real aspect of product engineering. Subsequently, three major sources of undesirable variations in products can be classi®ed, these being (Clausing, 1994): . Production variations . Variations in conditions of use . Deterioration (variation with time and use). Figure 1.17 `Bath-tub' curve showing typical life characteristic of a product (Priest, 1988) 20 Introduction to quality and reliability engineering The above ®ts in with the overall pattern of failure as described by Figure 1.17. The ®rst two and sometimes even all three parts of the bath-tub curve are closely con- nected to variations. The manufacturing process has a strong impact on component behaviour with respect to failure, and production variabilities arising from lack of precision or de®ciencies in manufacturing processes lead to failures concentrated early in the product's life (Klit et al., 1993; Lewis, 1996). A common occurrence is that correctly designed items may fail as a result of defects introduced during production or simply because the speci®ed dimensions or materials are not complied with (Nicholson et al., 1993). Production defects are second only to those product de®ciencies created by inadequate design as causes of accidents and improper production techniques can actually create hazardous characteristics in products (Hammer, 1980). Modern equipment is frequently composed of thousands of components, all of which interact within various tolerances. Failures often arise from a combination of drift conditions rather than the failure of a speci®c component (Smith, 1993). For example, typically an assembly tolerance exists only to limit the degradation of the assembly performance. Being `o target' may involve later warranty costs because the product is more likely to break down than one which has a performance closer to the target value (Vasseur et al., 1992). This again is related to manufacturing variation problems, and is more dicult to predict, and therefore less likely to be foreseen by the designer (Smith, 1993). Variations in a product's material properties, service loads, environment and use typically lead to random failures over the most protracted period of the product's expected life-cycle. During the conditions of use, environmental and service varia- tions give rise to temporary overloads or transients causing failures, although some failures are also caused by human related events such as installation and operation errors rather than by any intrinsic property of the product's components (Klit et al., 1993). Variability, therefore, is also the source of unreliability in a product (Carter, 1997). However, it is evident that if product reliability is determined during the design process, subsequent manufacturing, assembly and delivery of the system will certainly not improve upon this inherent reliability level (Kapur and Lamberson, 1977). Wearout attracts little attention among designers because it is considered less relevant to product reliability than the other two regions, although degradation phenomena are clearly important for designs involving substantial operating periods (Bury, 1975; Pitts and Lewis, 1993). Many of the kinds of failures described above may be reduced by either decreas- ing variations or by making the product robust against these variations (Bergman, 1992). For example, the smaller the variability associated with the critical design parameters, the greater will be the reliability of the design to deal with unforeseen events later in the product's life-cycle (Suh, 1990). As stated earlier, with reference to an individual manufacturing process, variation is an obvious measure for quality performance and the link between product failure, capability and reliability is to a large degree embedded within the prediction of variation at the design stage. How- ever, two factors in¯uence failure: the robustness of the product to variability, and the severity of the service conditions (Edwards and McKee, 1991). To this end, it has been cited that the quality control of the environment is much more important than How and why products fail 21 quality control of the manufacturing processes in achieving high reliability (Carter, 1986). In order to quantify the sometimes intangible elements of variability associated with the product design and the safety aspects in service requires an understanding of `risk'. The assessment of risk in terms of general engineering practice will be discussed next. This will lead to a better understanding of designing for quality and reliability, which is the main focus of the book. 1.4 Risk as a basis for design The science of risk, and the assessment and management of risk, is a very complex subject and one that covers a wide diversity of disciplines. Society is becoming more aware of the risks related to increased technological innovation and industrial- ization. Recent reports in the media about environment (global warming), health (BSE) and technological (nuclear waste processing) risks have played their part in focusing attention on the problem of how to assess risk and what makes an acceptable risk level. As a result, risk and risk related matters are becoming as important as economic issues on the political agenda. There is, therefore, an increasing need for a better understanding of the topic and tools and techniques that can be used to help assess product safety and support the development of products and processes that are of essentially low risk (EPSRC, 1999). To meet these needs, a new British Standard, BS 6079 (1999), has been published to give guidance to businesses on the management of risks throughout the life of a project. The term `risk' is often used to embrace two assessments: . The frequency (or probability) of an event occurring . The severity or consequences of the event on the user/environment. The product of these two conditions equals the risk: Risk Occurrence  Severity We can demonstrate the notions of risk and risk assessment using Figure 1.18. For a given probability of failure occurrence and severity of consequence, it is possible to map the general relationship of risk and what this means in terms of the action required to eliminate the risk. For example, if both occurrence and severity are low, the risk is low, and little or no action in eliminating or accommodating the risk is recommended. However, for the same level of occurrence but a high severity, a medium level of risk can be associated with concern in some situations. The level of occurrence, for some unknown reason, changes from low to medium and suddenly we are in a situation where the risk requires priority action to be eliminated or accommodated in the product. The aim of a risk assessment is to develop a product which is `safe' for the proposed market. A safe product is any product which, under normal or reasonably foreseeable conditions of use, including duration, presents no risk or only the minimum risk com- patible with the product's use and which is consistent with a high level of protection for consumers (DTI, 1994). In attempting to protect products against failure in service and, therefore, the user or environment, diculty exists in ascertaining the 22 Introduction to quality and reliability engineering degree of protection most suitable for a given application. First, the choice should be expressed in terms of risk and probability of failure as shown above, but the determination of an acceptable level for a product depends on many factors, such as (Bracha, 1964; Karmiol, 1965; Welling and Lynch, 1985): . Safety . The costs of failure . Criticality of function that it supports . Complexity ± number of component parts, subsystems . Operational pro®le ± duty cycle or time it operates . Environmental conditions ± exposure to various environmental conditions . Number of units to be produced . Ease and cost of replacement . `State of the art' or present state of engineering progress . Market sector/consumer category. Suggestions have also been made as to the number of people that are aected by the risk at any one time, and, in principle, it should be possible to link the acceptable level of individual risk to the number of people exposed to that risk (Niehaus, 1987). For example Versteeg (1987) provides risk levels associated with three areas: acceptable risk, reduction desired and unacceptable risk, and the number of people exposed to Figure 1.18 Risk and the identi®cation of priority action Risk as a basis for design 23 the risk. This further compounds the problem of assigning acceptable risk targets, but implies that safety is of paramount importance. Businesses make decisions that aect safety issues, but which are only considered implicitly. In an increasingly complex world, the resulting decisions are not always appropriate because the limits of the human mind do not allow for an implicit consideration of a large number of factors. Formal analyses are needed to aid the decision-making process in these complex situations. However, the application of a formal analysis to safety issues raises new questions. The risks perceived by society and by individuals cannot be captured by a simple technical analysis. There are many reasons for this; however, it is clear that decision making needs to account for both technical and public values (Bohnenblust and Slovic, 1998). The decision to accept risk is not based on the absolute notion of one acceptable risk level, but has some ¯exibility as the judgement depends on the cost/bene®t and the degree of voluntariness (Vrijling et al., 1998). The notion of safety is often used in a subjective way, but it is essential to develop quantitative approaches before it can be used as a functional tool for decision making (Villemeur, 1992). A technique which `quanti®es' safety is FMEA. 1.4.1 The role of FMEA in designing capable and reliable products In light of the above arguments, it has been found that there are two key techniques for delivering quality and reliability in new products: process capability analysis and FMEA (Cullen, 1994). FMEA is now considered to be a natural tool to be used in quality and reliability improvement and it has been suggested that between 70 and 80% of potential failures could be identi®ed at the design stage by its eective use (Carter, 1986). For example, performing a comprehensive FMEA well will alleviate late design changes (Chrysler Corporation et al., 1995). FMEA was ®rst mentioned at the start of this chapter. It is recommended that the reader unfamiliar with FMEA refer to Appendix III and several other references provided to gain a ®rm understanding of its application in product design. In general, an FMEA does the following (Leitch, 1995): . Provides the designer with an understanding of the structure of the system, and the factors which in¯uence quality and reliability . Helps to identify items that are of high risk through the calculation of the Risk Priority Number (RPN), and so gives a means of deciding priorities for corrective action . Identi®es where special eort is needed during manufacture, assembly or main- tenance . Establishes if there are any operational constraints resulting from the design . It gives assurance to management and/or customers that quality and reliability are being or have been properly addressed early in the project. Of the many characteristics of a product de®ned by the dimensions and speci®cations on a drawing, only a few are critical to ful®lling the product's intended function. 24 Introduction to quality and reliability engineering [...]... failure costs associated with non-safety critical and safety critical applications in production and service through linkage with FMEA Comprehensive guidance on the application of the technique is given and a number of case studies and example analyses are used to illustrate the bene®ts of the approach 2 Designing capable components and assemblies 2. 1 Manufacturing capability One of the basic expectations... which aid product developers in formulating quality objectives and speci®cations 3 Q Synthesis techniques which aid the designer in generating ideas and in detailing solutions 3 Qq 27 28 Introduction to quality and reliability engineering Figure 1 .21 Hole in plate analogy to quality and the Q/q concepts Veri®cation techniques which verify and evaluate the quality of solutions in relation to the speci®cation... an FMEA in terms of designing capable and reliable products are the potential failure modes, severity rating and critical characteristics for the design By identifying the capability of the critical characteristics, and the potential failure mode, a statistical analysis can then be performed to determine its reliability The FMEA Severity Rating (S) is crucial for setting capability and reliability targets... needs to understand when required tolerances are pushing the process to the limit and to specify where capability should be measured and validated Tolerances alone simply do not contain enough information for the ecient manufacture of a design concept (Vasseur et al., 19 92) At the design stage, both 38 Designing capable components and assemblies qualitative manufacturing knowledge about candidate manufacturing... variability of the manufacturing processes and therefore 41 42 Designing capable components and assemblies Figure 2. 2 (a) Tolerance versus production costs of various processes; (b) comparison of cost-tolerance models (Dong, 1993) during the design phase, the designer must use the best available process capability data for similar processes (Battin, 1988; Chase and Parkinson, 1991) It is far easier, not... should play an important role in the design and manufacture of reliable products (Amster and 33 34 Introduction to quality and reliability engineering Table 1.1 Competing issues in deterministic and probabilistic design approaches Deterministic design Probabilistic design Dominated design for 150 years Design parameters treated as unique values Underlying empirical and subjective nature Factor of safety... consequences and solutions for an actual or potential design problem This means making predictions, where appropriate, based on evidence from testing, experience or other hard facts using statistical probabilities, not vague guesses The use of FMEA to evaluate all the potential risks of failure and their consequences, both from normal use and foreseeable misuse, is a key element in designing capable and reliable. . .Designing for quality Figure 1.19 The FMEA input into designing capable and reliable products Hence, a critical characteristic is de®ned as one in which high variation could signi®cantly aect product safety, function or performance (Liggett, 1993) In order to assess the level of importance of the characteristics in a design, a process of identifying the critical characteristics and then using... researchers Cagan and Kurfess (19 92) propose a hyperbolic cost function Speckhart (19 72) and Dong (1997) suggest an exponential cost function The inverse quadratic form was advocated by Spotts (1973) A comparison of various researchers' cost models is provided in Figure 2. 2(b) Determining the parameters of the cost±tolerance models described, however, is by no means a trivial task (Dong, 1993) 2. 1.3 Process... parameters 31 32 Introduction to quality and reliability engineering Figure 1 .22 Hole in a plate analogy of probabilistic design More plausible representations of stress and strength distributions for a given situation will enable meaningful failure predictions to be produced, and will be particularly useful where test to failure is not a practical proposition, where weight minimization and/ or material . the potential risks of failure and their consequences, both from normal use and fore- seeable misuse, is a key element in designing capable and reliable products (Wright, 1989). 1.5 Designing for quality The. FMEA input into designing capable and reliable products Designing for quality 25 published in the ®eld of Design for Quality (DFQ) compared to Design for Assembly (DFA), for example, and little methodology. 1.19, the important results from an FMEA in terms of designing capable and reliable products are the potential failure modes, severity rating and critical characteristics for the design. By identifying
Ngày đăng: 13/08/2014, 08:21
Xem thêm: Designing Capable and Reliable Products Episode 2 pptx
