In this paper we consider a model for the optimal management of a two dam system. Each dam is modelled via a continuous-time controlled Markov chain on a finite control period and linked to the other dam via a state and time dependent water transfer control. The consumption control for the dam system is provided by a time and state dependent price feedback control. This price feedback control takes into account the active seasonal demands of customers. We consider the case where inflow processes and evaporation for each dam are non-stationary as are the customer demands. The general approach to the solution of this problem is to consider this stochastic optimisation problem in the average case and solve it using the dynamic programming method. We show that the use of parallel computing techniques leads to substantial savings in calculation times for the solution of the optimal controls and demonstrate this via a numerical example.