I have a long-term (100 year) dynamic model with a large number of stochastic inputs. I need to carry out Monte Carlo analysis with each realization optimized independently (optimal decision variable values are different from one realization to the next). It seems that I can achieve this using a submodel containing the deterministic dynamic model set as an optimization, within a parent static model containing the stochastic elements. However, I need to see the full probability time histories of several outputs, not just final values. Is there a way to do this?
Also, for the number of realizations I need to carry out to adequately sample the stochastic element input distributions, this approach would be very computationally expensive. Is there any way I can carry out a staged optimization - for example, break the 100 year duration into 5 periods of 20 years each, optimize the first 20, then the second, and so on. This would mean fewer timesteps calculated and has the additional benefit of allowing the decision variables to change over the full realization duration. Would this be possible with an additional submodel or looping container?