21 Manufacturing capability

One of the basic expectations of the customer is conformance to specification, that is, the customer expects output characteristics to be on target with minimum variation (Abraham and Whitney, 1993; Garvin, 1988). Assessing the capability of designs early in product development therefore becomes crucial. The designer must aim to achieve the standards demanded by the specification, but at the same time not exceed the capabilities of the production department. This may not be an easy task because the determinants of quality are frequently difficult to identify at the design stage. The designer must decide how the specifications are affected by features of the production process, as well as by the choice of materials, design scheme, etc. If the designer knows that a certain manufacturing process can only achieve very poor dimensional tolerances, for example, a decision must be made to decide whether such an process can be used for the new product. If not, an alternative solution must be devised (Oakley, 1993).

The available information about manufacturing processes is both extensive and diverse, but it is not easily accessible for designers, as the structure is not suited for design applications (Sigurjonsson, 1992). The designer's job is made easier if standard rules are available for the candidate manufacturing method. A good set of design rules can indicate process capable tolerances for a design. Life is much more problematic when rules are not available or an unfamiliar manufacturing process is to be used. Many designers have practical experience of production methods and fully understand the limitations and capabilities that they must work within. Unfortunately, there are also many that do not have this experience and, quite simply, do not appreciate the systems that they are supposed to be designing for (Oakley, 1993). For example, three aspects of engineering drawing that are often overlooked, but are vital in design practice, are dimensions, tolerances and specification of surface finish (Dieter, 1986). The designer 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 efficient manufacture of a design concept (Vasseur et al., 1992). At the design stage, both qualitative manufacturing knowledge about candidate manufacturing processes for design features, and quantitative manufacturing knowledge on the influences of design parameters on production costs, are needed (Dong, 1993). Only with strong efforts to integrate the product design with the selection of the manufacturing process can the desirable performance characteristics be produced with a minimum of variability and minimum cost (Lewis, 1996).

The goal of this first stage of Conformability Analysis (CA), the Component Manufacturing Variability Risks Analysis, is the provision of support in the early stages of the detailed design process for assessing the tolerance and surface finish capability associated with component manufacture. 'Risk' in this sense is a measure of the chances of not meeting the specification. Recognizing that the relationship between a design and its production capability is complex and not easily amenable to precise scientific formulation, the methodology described has resulted largely from knowledge engineering exercises in manufacturing businesses, including those with expertise in particular manufacturing processes. For example, Poeton in Gloucester, UK, specialize in supplying surface coating technology in the form of design guidance and processing techniques. The Poeton engineers had a major consultation role in the determination of the main issues related to surface engineering process variability. To support other aspects of the analysis, such as the tolerance capability analysis, manufacturing tolerance data was readily available, albeit in a form not useful to designers. During the evolution of CA, many alterations and improvements were made through exhaustive consultation and validation in industry. The development of only the current aspects of the analysis will be discussed in detail here. However, some background topics will be reviewed before proceeding.

2.1.1 Variability factors in manufacturing

Designers have always had to deal with the fact that parts cannot be made perfectly; or if they could, they would not remain perfect for long during use. So defining the 'ideal' component is only one aspect of the designer's job (Hopp, 1993). The designer must also decide by how much a still acceptable component can be from the ideal. A component can vary from the ideal in many ways: in its geometry, its material properties, surface finish - a virtually unlimited list (Alexander, 1964).

In analysing a design at the concept development or early detailing stage, it is only necessary to focus on the main variabilities associated with manufacturing processes. In this way the analysis performs at a level of abstraction which facilitates a rapid assessment without being complex or difficult to comprehend. Engineering judgement and experience are needed to identify potential variability risks or 'noise' associated with manufacturing and assembly routes (Phadke, 1989). Before introducing the main aspects of the component manufacturing variability risks analysis, it is worth exploring variability issues in general. There are two main kinds of variability:

• Common-cause or inherent variability is due to the set of factors that are inherent in a machine/process by virtue of its design, construction and the nature of its operation, for example positional repeatability, machine rigidity, which cannot be removed without undue expense and/or process redesign. When only common-cause variability is present, the process is performing at its best possible level under the current process design.

• Assignable-cause or special-cause variability is due to identifiable sources which can be systematically identified and eliminated (see below).

The last sources of variation highlighted are of paramount concern for allocating and analysing mechanical tolerances (Harry and Stewart, 1988).

At the most detailed level, a variation can belong to the basic design properties: form, dimension, material and surface quality for the components, and structure for the relations between components (Morup, 1993). For example, the levels of dimensional tolerance accuracy and surface roughness associated with industrial manufacturing processes vary widely. In general, tolerances reflect the accuracy specifications of the design requirements. These will inherently reflect the variability in the manufacturing process. That is the design of the tolerance for a certain component characteristic will influence the variability in the measurements of that characteristic. Hence, a tightened tolerance requires higher precision manufacturing devices, higher technical skills, higher operation attention and increased manufacturing steps (Jeang, 1995).

The general factors influencing variation include the following (Dorf and Kusiak, 1994; Morup, 1993):

• Tool and functional accuracy

• Deformation - due to mechanical and thermal effects

• Measurement errors

• Material impurities

• Specifications

• Method or job instructions

Further variations may arise from the working of the material during the manufacturing process or from deliberate or unavoidable heat treatment (Bolz, 1981). In general, the undesirable and sometimes uncontrollable factors that cause a functional characteristic to deviate from its target value are often called noise factors and are defined below (Kapur, 1993):

• Outer noise - environmental conditions such as temperature, humidity, different customer usage

• Inner noise - changes in the inherent properties of the system such as deterioration, wear, corrosion

• Product noise - piece to piece variation due to manufacturing variation and imperfections.

An example of assignable-cause variability is given in Figure 2.1. Two cases of milling the same component to finished size are shown. In the first case, the component is










Produces this





Produces this

Target Target

Figure 2.1 Example of process capability improvement (adapted from Leaney, 1996a)

relocated a number of times in the tooling to produce the finished product. The tolerance capability of a key dimension is analysed and the results given in the form of a histogram. In the second case, the number of relocations is reduced and simultaneous operations carried out on the component in the form of gang milling. A significant capability improvement is experienced in the second case as shown by the comparatively low process spread and accumulation around the target value. This is mainly due to the set-up and operation procedures for the manufacture of the component. It is evident from this that a good process set-up ensures better product quality (Hallihan, 1997).

2.1.2 Cost-tolerance relationships

Ideally, engineers like tight tolerances to assure fit and function of their design. Designers often specify unnecessarily tight tolerances due to the lack of appreciation of cost and due to the lack of confidence in manufacturing to produce component parts which conform to specification (Phadke, 1989). However, a tightened tolerance requires higher precision and more expensive manufacturing processes, higher technical skills, higher operator attention, increased manufacturing steps and is more time consuming to achieve (Jeang, 1995; Soderberg, 1995). The configuration and material of a part, as well dimensions, tolerances and surface finishes, can also change the amount of work required in part manufacture (Dong, 1993).

All manufacturers, on the other hand, prefer loose tolerances which make parts easier and less expensive to produce (Chase and Parkinson, 1991). The choice of tolerance is therefore not only related to functional requirements, but also to the manufacturing cost. It has been argued that among the effects of design specifications on costs, those of tolerances are perhaps the most significant (Bolz, 1981). Figure 2.2(a) shows the relationship between product tolerance and approximate relative cost for several manufacturing processes. Each tolerance assignment results in distinct manufacturing costs (Korde, 1997).

Cost-tolerance functions are used to describe the manufacturing cost associated with a process in achieving a desired level of tolerance. The shape of the cost-tolerance curve has been suggested by several researchers. Cagan and Kurfess (1992) propose a hyperbolic cost function. Speckhart (1972) 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 capability and tolerances

Product tolerance can significantly influence product variability. Unfortunately, we have difficulty in finding the exact relationship between them. An approximate relationship can be found from the process capability index, a quality metric interrelated to manufacturing cost and tolerance (Lin et al., 1997). The random manner by which the inherent inaccuracies within a manufacturing process are generated produces a pattern of variation for the dimension which resembles the Normal distribution (Chase and Parkinson, 1991; Mansoor, 1963) and therefore process capability indices, which are based on the Normal distribution, are suitable for use. See Appendix IV for a detailed discussion of process capability indices.

There are many variations on the two basic process capability indices, Cp and Cpk, these being the most commonly used. A process capability index for a shifted distribution at Cpk = 1.33 is still regarded to be the absolute minimum (which relates to around 30ppm failing the specification). This increases to Cpk = 1.66 (which relates to approximately 0.3 ppm failing) for more safety critical characteristics (Kotz and Lovelace, 1998). Motorola stipulate Cpk = 1.5 (or approximately 3 ppm) to their suppliers as an absolute minimum (Harry and Stewart, 1988). However, one weakness of the process capability index is that there is no apparent basis for specifying the optimal value (Taguchi et al., 1989). In practice, most designers do not worry about the true behaviour of a manufacturing process and compensate for the lack of knowledge with larger than necessary process capability indices (Gerth, 1997).

The precise capability of a process cannot be determined before statistical control of the actual process has been established. Designers seldom have sufficient data by which to specify the variability of the manufacturing processes and therefore

Tolerance (mm)

Tolerance (mm)

Cost Tolerance Models

Figure 2.2 (a) Tolerance versus production costs of various processes; (b) comparison of cost-tolerance models (Dong, 1993)

Tolerance (mm)

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 to mention less costly, to create robust designs based on known process capabilities than it is to track down and subsequently reduce sources of variation during the manufacturing phase (Harry and Stewart, 1988).

0 0

Post a comment