Linear FEA can be and is used to design plastic parts, but you need to understand the true nature of what you are analyzing and think about the assumptions you would be making going in. Unfortunately, you dont really describe what you are doing. Some possible problem areas: How are you loading the model? Impact? Polymers generally get stiffer as the strain rate increases. If you use the data sheet value, you would be safe with regards to deflection. But elongation goes down as strain rate increases, so you better be well below yield. Cyclic? Fatigue of polymers is not straight forward. Cyclic loading produces heat which makes applying S/N data difficult. The type of loading (bending vs normal stress) can have a big influence on results. "Support a weight" sounds like a static component is present all the time, so you may need to consider creep. You can generate an isochronous stress strain curve from which you get an apparent modulus, but that can result in an overly conservative solution because the stress and creep will vary throughout the part, while your apparent modulus will be constant. And the apparent modulus will be lower than a data sheet Young's modulus, so you might get into static deflection issues. Then there is anisotropy, temperature effects, ageing effects, weld lines, etc, etc. You might want to read Structural Analysis of Thermoplastic Components by Trantina and Nimmer first. They deal with these and other issues. If you have access to nonlinear FEA, you should consider it.
