Optimal Control of Central Air-conditioning Systems: From Time-driven to Event-driven Paradigm

Description

Climate change creates international pressure to improve building energy efficiency. Because central air-conditioning systems always present the largest primary energy consumer in buildings, it is important to improve the energy efficiency of such systems. Optimal control has thus been developed. It minimize the energy use of central air-conditioning systems by searching the optimal values for the controllable variables that significantly affect the system energy efficiency. Because the optimal values changes with the operating conditions, real-time optimization is needed. Currently, real-time optimal control always adopts a periodic mechanism with a fixed frequency to trigger optimization actions.Using a ‘clock’ to trigger optimization actions is easy to implement. However, it has two major limitations for applications in central air-conditioning systems. Firstly, most changes in the operating conditions of central air-conditioning systems are aperiodic or stochastic, such as changes in weather, load and occupancy conditions. Thus, a “clock” is not an ideal trigger and, consequently, time-driven optimal control may lead to delayed optimization actions that impair energy efficiency. Secondly, real-time optimal control always presents a large-scale mathematic programming challenge and is implemented as standard in a building energy management system (BEMS). Although one can attempt to capture the random changes with minimal delay by choosing an extremely high optimization frequency, a higher frequency will inevitably increase the computational burden of the BEMS, which is considered as a barrier to the implementation of real-time optimal control. Additionally, a higher frequency may waste computational resources when the operating conditions are stable and optimization actions are not necessary.To address the limitations mentioned above, we propose to develop a new real-time optimal control paradigm known as event-driven optimal control. This optimal control paradigm will trigger optimization actions in response to the occurrence of events other than fixed time instants. Instead of searching for a suitable optimization frequency (which is difficult in practice due to aperiodic changes in operating conditions), this new paradigm will define appropriate events to capture the aperiodic changes in the operating conditions of central air-conditioning systems. This will ensure that optimization actions are carried out when and only when required. The outputs of this study will be a systematic method of defining events for the optimal control of central air-conditioning systems and a toolbox including the optimal control algorithms corresponding to each event, which can be used to improve the energy efficiency of central air-conditioning systems and thus enhance building energy performance.