Smart buildings using compact heat pipes with nanofluid in PCM for energy saving via deep clustering of multi-agent

The peak load of power plants occurs in the afternoon due to the cooling load imposed by the building sector, adversely affecting the plant's efficiency and the coefficient of performance (COP) of chillers. Therefore, optimally managing cooling load strategies is crucial, as it significantly reduces energy consumption in buildings and enhances energy efficiency. This study proposed an intelligent building energy-saving strategy that combines dual-stage phase change materials (DPCMs) with methanol-silver nanofluid thermosyphon heat pipes, optimised via deep clustering self-adaptive reinforcement learning (DCSRL) in a MATLAB environment. This approach refined tank dimensions and sequencing to effectively reduce the peak building load, improving cooling-load demand response and building energy efficiency. On the other hand, to avoid the high-dimensional action space and highly nonlinear control response, DCSRL integrates thermosyphon heat pipes of DPCMs (DCSRLITHPDPCMs). Such a strategy is successfully implemented for energy storage at off-peak periods to absorb the cooling load when heating, ventilation, and air conditioning (HVAC) systems are turned off. The DCSRLITHPDPCMs significantly reduced the physical size of the PCM tanks and thus significantly contributed to achieving the desired response of the cooling load demand. Such a compact design provided a remarkable 33.9 % decrease in energy consumption assessed to conventional systems and increased grid reliability.