Two-stage stochastic programming for the design optimization of district cooling networks under demand and cost uncertainty

Abstract

The major limitations of district cooling systems are the high capital costs, which make design optimization tools necessary to maximize the potential benefits. Decision makers when designing district cooling have to handle cost and demand uncertainties that further increase the investment risks. On the other hand, the possible evolution of cooling demand during the years, shall be taken into account in the first design stages, in order to allow network expansion in the future. In this paper, a novel two-stage stochastic programming model is therefore proposed for the optimal design of district cooling networks under demand and cost uncertainty. The model was also applied to a case study and the results showed that it is more convenient to build smaller district cooling networks (and eventually enhance them in the future if the cooling demand and electricity costs will increase) rather than building larger systems from the beginning. In addition, it was found that the uncertainties in electricity cost and cooling demand are the ones that most influence the optimal solution. The impact of the stochastic model was evaluated with respect to deterministic approaches, resulting up to 5% less expensive in terms of expected cost and with a three years lower payback time. A second model formulation was also implemented, with more rigid constraints, which limit the amount of pipes that can be installed in a single branch. With this formulation, the model tends to connect more buildings and to install larger pipes from the beginning, but the solution in terms of expected cost is only 0.4% more expensive than the more flexible one. Lastly, it was analysed the impact of asset residual value at the end of project life, revealing that neglecting it would lead to connecting more buildings initially, but in most scenarios the network would not be expanded in the future.