SECTION 1.6
Selection of Tolerances

The selection of tolerances has a dual effect on both economics and on quality. The tolerance effects:

1. Fitness for use (the salability of the product)
2. Costs of manufacture (facilities, tooling, productivity)
3. Costs of quality (equipment, inspection, scrap, rework, material review).

Figure 1-2.
An approach to functional tolerancing.

SECTION 1.6.1
Methods of selecting tolerances

In theory, the designer should, by scientific study, establish the proper balance between the value of precision and the cost of precision. In practice, the designer is unable to do this for each tolerance--there are too many quality characteristics. As a result, only a minority of tolerances are set scientifically. Some scientific tools for tolerancing include:

1. Regression studies. For example, a tool may be required to produce a feature at a certain dimension with a specified micro finish. Several tools are built and used. The test data recorded are (1) the dimension produced, (2) the micro finish produced, and (3) the physical characteristics of the tool which affect the test criteria. These data permit scatter diagrams to be prepared and regression equations to be computed to aid in establishing tolerances on a basis which is scientific, within the confidence limits for the numbers involved.
   
   
2. Tolerances for interacting dimensions. Numerous designs involve "interacting dimensions". A cutting tool may consist of a chain of 10 dimensions, located about a single centerline. A pot assembly may consist of a buildup of 15 elements. What they have in common is the existence of interaction among these elements or dimensions. Each element or dimension has its own tolerance. However, the variation of the composite (the tool or the assembly) will be related to the variations of the elements according to the laws of probability; i.e., it is very unlikely that all the extremes will come together simultaneously. This unlikelihood makes it possible to establish wider tolerances on elements of such designs without significantly increasing the extent of non-conformance.

Most tolerances are established by methods which, in varying degrees, are less than scientific. The principal methods include: precedent, bargaining, and standard tolerance levels which have been defined at the company, industry, national, and international levels.

SECTION 1.6.2
Unrealistic tolerances

In some companies, the accumulated specifications contain an extensive array of unduly tight tolerances, i.e., tolerances not really needed to achieve fitness for use. Whenever this condition creates problems in manufacturing, the toolmakers will often respond by exceeding the tolerances to meet delivery dates. This becomes a situation of "unrealistic tolerances loosely enforced". Such a situation has evolved from many historic forces and unfortunately has resulted in distrust between design and manufacturing departments. More recently, companies have tried to reduce the distrust in order to convert to a situation of "realistic tolerances rigidly enforced".

A conservative approach in the short term is to make provisions for handling specifications having unrealistic tolerances. In the long term, the evolutionary solution to unrealistic tolerances lies in providing designers with the information needed to set realistic tolerances. This includes/ process capability data and data on the cost of achieving various leve