Perturbation and sensitivity analysis of an urban model

This work describes a sensitivity analysis of the Forrester urban model. Using small perturbations about the system trajectories, sensitivity matrices were computed at 135 years (growth period) and 240 years (equilibrium). The results show a matrix heavily weighted to give large sensitivity coefficients to the three housing variables. In the housing sectors the feedback loops containing the growth rate parameters were found to contribute most to the sensitivity. Finally, sensitivity coefficients of the level variables with respect to a number of model parameters were found. In general, their contribution to overall model sensitivity was quite small when compared to the loop sensitivity described above.