1. I had a case where deflection of the steel structure members was causing big problems in nozzle loading

2. I have a current case where old pipes have excessive stresses in FP (Steam pipe). If I would model steel stifness I most probably would get much lower stresses. I do not do this to make sure that pipe is monitored in the future (NDT, replica). Modelling would never be accurate and would only hide potentially risky area.

3. I know that some stress engineers use sometimes steel modelling to get stresses down in the pipe.

4. I have current case where too flexible long support steel members have excessive deflection and other problems. Design has to be changed.

Conclusion: Have to use some common sense but where possible I would follow Craig's method.