Aggregation disaggregation method for optimal-control of multi-zone HVAC systems

The problem of minimum-cost control of a multizone cooling system is considered, for the case where chilled-water storage capacity is substantial and electrical energy is offered at discounted price during off-peak hours of the day. The optimization problem is solved using state-increment dynamic programming, but limited to state spaces of low dimensionality through a sequential aggregation/disaggregation procedure. Near optimal results are obtained, demonstrating that substantial savings of energy cost are feasible by optimal use of storage, that these savings are not highly sensitive to the accuracy of cooling-load forecasts and that the required operating schecules can be determined with modest computational effort even for buildings with large numbers of zones.