Lifetime optimisation of multiple deep closed-loop geothermal wells

作为当前能源和环境危机解决方案的一部分，全世界对地热能的兴趣日益增加。与太阳能和风能相比，地热能具有较高的容量系数，因此能够提供可预测的24/7全天候热能供应，可用于满足直接热对热应用的基本负荷，或与有机朗肯循环工厂（例如）一起产生电力。此外，地热足迹（m2/产生的能源）比风能和太阳能装置所需的足迹要小得多，使其在土地使用和社会可接受性方面都更具吸引力。这里提出的闭环地热系统的优点是可以在任何地方使用/安装，并且消除了与开环系统相关的勘探风险。在这种系统中，流体被包含在同轴热交换器中，因此消除了与水提取和注水（水力压裂）相关的任何环境/社会问题。最近，利用地热能源的选择和经济可行性已经得到了改善，这与利用废弃油气井的需求一致，此外还有新油气井的计划。单口地热井对岩层和井筒之间的热能传递具有径向影响区。这受地质的热扩散率和导电性以及提取的能量的影响。随着时间的推移，热量的提取会导致井的热输出减少。因此，本博士课程第一部分中提出的多井开发可以通过间歇性满负荷时间（可变热流密度和瞬态热传导）或在井之间切换来帮助井进行热补给，以允许原始地层（在井的影响半径范围内）在井筒中回收热量。地热资金的一个反复出现的问题是与新钻孔相关的前期成本。然而，地热能提供较低的能源成本（LCOE）。通过使用多口井来增加给定地点/应用的地热发电厂的整体寿命，将提高LCOE，使地热成为实现净零排放的更有吸引力的选择。机器学习可以在没有人为干预的情况下帮助决策和解决优化和控制过程的问题。在第二部分的工作中，研究计划将包括开发一个深度学习优化框架，以控制地热生产，并确定最佳间歇、工作流量和目标回液温度，以提高使用寿命和产量，同时还研究不同井结构和地质对热降解的影响。

Worldwide interest in geothermal energy is increasing as part of the solution mix in the prevailing energy and environmental crisis. Geothermal energy has a high capacity factor compared to solar and wind, hence being able to provide a predictable 24/7 supply of thermal energy that can be used to meet the base load in direct heat to heat applications or produce electrical power in conjunction with, for example, Organic Rankine Cycle plants. In addition, the geothermal footprint (m2/energy produced) is significantly smaller than that required by wind and solar installations making it a more attractive option both in terms of land use and social acceptability. Closed-loop geothermal systems as proposed here have the benefit of being available/installed anywhere and remove the exploration risks associated with open loop systems. In such systems the fluid is contained in coaxial heat exchangers hence, removing any environmental/societal concerns associated with water extraction and injection (fracking). Recently, the option and financial viability of using geothermal energy has been improved in line with the need to utilise abandoned oil and gas well, in addition to the plans for new ones.A single geothermal well has a radial area of influence for the transfer of thermal energy between the rock formation and the well bore. This is affected by the thermal diffusivity and conductivity of the geology and the amount of energy extracted. Over time, heat extraction leads to a reduced thermal output of a well. Therefore, multi-well developments as the ones proposed to be studied in the first part of this PhD programme can aid in thermally recharging wells using intermittent full load hours (variable heat flux and transient heat conduction) or switching between wells to allow for the original formation (to the extent of the radius of influence of the well) to recuperate heat at the well bore. One recurrent issue with geothermal funding is the upfront cost associated with new bore holes. However, geothermal energy offers a lower cost of energy (LCOE). Increasing the overall geothermal plant lifetime for a given site/application by using multiple wells will improve the LCOE, making geothermal an even more attractive option in the drive to net-zero.Machine learning can aid in decision-making and problem-solving for optimisation and control processes without human interference. In the second part of the work, the research programme will include the development of a deep learning optimisation framework to control geothermal production and to determine optimum intermittency, working flow rates and targeted return fluid temperatures for increased lifetime and production, while also investigating the effects on thermal degradation due to different well architectures and geologies.