For plastic parts, the way the resin molecules are stretched out to go around corners and the speed with which the particular location within the component cools in the mold both have major influences on the actual strength at a given location or feature of the part. Constraint at a sharp corner or very thick section can make it difficult for the part to deform, thereby encouraging sudden and unexpected brittle fracture. When the material is tested, there is nothing “wrong” with it.
For metallic parts that are formed at room temperature, the final strength at a given location will depend on how much deformation has been imparted to the feature and the details of how the final shape was achieved. I’ve occasionally seen high-strength, low-alloy (HSLA) steel used for severely deformed metal stampings. Usually, these types of materials are difficult to stamp without a high percentage of scrap for split parts. I have wondered whether the designer considered that they are naturally increasing the strength of the material when they form it. If they specified the minimum strength to be present in the material as purchased, they may be shooting the poor manufacturing operators in both feet. Maybe they could specify a lower-strength material that will have the required strength in the formed part.
For parts that are plated, the plating can affect the strength of the component. Perhaps cracks in brittle plating extend into the bulk material, reducing the effective strength. For parts that are heat treated, local geometry of a complex part will allow a wide range of strength properties to develop. There is software available to predict the distribution of hardness in heat-treated metal components. To my way of thinking, however, there is no substitute for cutting the part and getting actual hardness profiles or tensile tests at the significant locations. In the end, this is the only way to prove that you have what the designer thinks is required to make the component function well.
This especially applies to induction-hardened steels because the process is very dependent on a lot of parameters, including slight changes in incoming microstructure of the steel.
As the old saying goes, “One test is worth a thousand expert opinions.” Of course there are also times when a single expert opinion helps shed meaningful light on the thousand data points! Perhaps more important is having the single expert opinion help decide how to collect the thousand data points so that they are meaningful.
The other main issue to contemplate when using material property data from published sources is that the data are obtained on certain standardized geometries and at certain standardized temperatures and loading rates. When you have a real part, it can be pretty difficult to get a realistic estimate of the actual loading rates and service conditions. Corrosive substances (sometimes dry air or plain water) can also have detrimental effects that are difficult to predict. The chances of the part behaving in service in a way readily guessed at from published data are pretty slim!