Further understanding ground source heat pump system design using finite element methods and machine learning techniques

Abstract

Ground-source heat pump (GSHP) systems can efficiently provide renewable energy for space heating and cooling. Even though these systems have shown great potential, contributing towards the continuously increasing energy demand and reducing greenhouse gas (GHG) emissions, our understanding of how they can be best utilised and designed can still be improved. This research adopts detailed numerical modelling and statistical approaches to provide further insights on these systems and contribute towards their worldwide adoption, focusing on three main areas. Firstly, due to the nature of their installation, there can exist disparities between the designed and installed systems. One such design-installation disparity, variable geothermal pipe separation, is addressed, aiming to reduce the gap between theory and practice. Secondly, due to the relatively recent emergence of energy geo-structures, such as energy piles or retaining walls, there currently exists little information on their utilisation/design. Therefore, an in-depth numerical analysis on energy geo-structure thermal performance is provided, focusing on the less well-researched energy retaining walls and providing suggestions on important factors such as the thermal demand, structure geometry and pipe configuration. Finally, two statistical approaches are presented that complement numerical modelling (often adopted for energy geo-structure analysis) and significantly reduce the computational time/resources associated, making numerical analysis and design of GSHP systems more accessible to engineering practice.