I see no one has responded so I will give it a try. My first thought is to have you check the behavioral modeler package. I have not used it but I think it is worth a try. If you don't have it then I seem to remember a similar type of issue while designing Instrument Cluster pointers. It's not the same problem but I think some of the Pro/E speak and function may be. I had to create parameters linking the mass and weight to the 0,0,0 point since we could not grab the cg results directly. I don't know if the software allows this now or not. I have not had the task of using Pro this way for quite some time. However, we then, through a series of equations, we were able to have a secondary coordinate system rotate as a function of the location of the weight that was moving around. This all assumes your box is fixed. Ie. the container will not change in dimension. I would assume if you were optimizing the container around a fixed weight then the problem is the same. Trying to optimze both simultaneously is the reason PTC introduced Behavioral modeling, I think. I hope I have helped a little at least.
