Token based scheduling of HVAC Services in commercial buildings

This paper proposes a novel distributed architecture for controlling Heating, Ventilation and Air conditioning (HVAC) systems in commercial buildings. Zone Modules use local models and measurements to compute requests for HVAC service over various future time windows. These requests are expressed in terms of the heating/cooling service required which we can conceptually regard as tokens. A Central Scheduler balances token requests and allocates tokens to each zone for the next time slot. This allocation attempts to minimize total energy consumption while respecting operational constraints. Zone modules update their local models based on the measured thermal responses resulting from allocated tokens, and re-compute forward token requests. This proposed token based architecture is inspired by medium access control protocols in communication networks. It offers several advantages in the context of HVAC systems. The architecture is scalable to realistic buildings with 200-500 thermal zones, it is robust relative to non-stationary environmental conditions and unanticipated changes in user needs, and it is modular enabling low-cost deployment without requiring expensive custom thermal modeling of buildings. We develop the zone module algorithms for computing token requests and central scheduler algorithms to allocate tokens. Using simulation studies, we demonstrate that the performance loss of our token based scheduling strategy is modest in comparison to a fully centralized nonlinear optimal control scheme.