Though the uncertainties involved make exact predictions impossible, designers can solve many problems based on past experience, trial and error, tests, and the like.
Much experience is accumulated in design rules or recommendations, such as the rules for bolted joint design in the ASME Boiler and Pressure Vessel Code. Many industries and companies have similar guidelines.
Where such rules do not apply, designers must create their own, building on experience with the product as it is manufactured and used. It helps to document decisions and discoveries. If early assemblies fail, careful records of assembly practices can produce better results. Likewise, information about failures and customer problems can evaluate a design more accurately than simple experiments or tests.
The designer should also be intimately involved with the assembly process, at least on critical joints. If part configuration makes it difficult to obtain proper bolt preload, the design is inadequate.
For critical joints, ultrasonic equipment may enable designers to overlook many assembly variables. For example, if bolt tension can be measured directly, variables affecting tension can be ignored. Ultrasonic equipment can also monitor joint response to external loads, load cycles, changes in assembly procedure, and such. This can help designers become familiar with joint characteristics more quickly.
Traditionally, the main weapon against uncertainty is overdesign. More and larger bolts are typically used to keep the joint sufficiently clamped even if the bolts are not all tightened correctly; service loads exceed estimates; stiffness ratios were overestimated, or thermal effects were underestimated. In many cases, overdesign is the only option.
Reliability is expensive. Costs are related to the amount of experiment and analysis performed at the design stage; trade-offs between performance and overdesign; the cost of tools and procedures used at assembly; and the cost of bolts and joint members. In critical situations, costs may be boosted by finite-element analyses, optimized designs, exotic materials, and ultrasonic control of bolt stress.
In most cases, testing requirements are much less extensive. However, although the uncertainties involved mean that the cost of improving reliability is not always recouped, in most cases the money is well spent.