Cost-optimal Management of a Residential Heating System With a Geothermal Energy Storage Under Uncertainty

TL;DR

This paper develops a stochastic optimal control framework for managing a residential heating system with geothermal storage, accounting for uncertainties in renewable energy and environmental conditions to achieve cost efficiency.

Contribution

It introduces a novel modeling approach combining PDEs, SDEs, and model reduction for cost-optimal control of geothermal storage in residential heating systems under uncertainty.

Findings

Effective discretization methods for complex PDE-SDE systems.

Approximate solutions for cost-optimal control policies.

Demonstrated potential for reducing operational costs.

Abstract

In this paper, we consider a residential heating system with renewable and non-renewable heat generation and different consumption units and investigate a stochastic optimal control problem for its cost-optimal management. As a special feature, the heating system is equipped with a geothermal storage that enables the intertemporal transfer of thermal energy by storing surplus heat for later use. In addition to the numerous technical challenges, economic issues such as cost-optimal control also play a central role in the design and operation of such systems. The latter leads to challenging mathematical optimization problems, as the response of the storage to charging and discharging decisions depends on the spatial temperature distribution in the storage. We take into account uncertainties regarding randomly fluctuating heat generation from renewable energies and the environmental conditions that determine heat demand and supply. The dynamics of the multidimensional controlled state processes is governed by a partial, a random ordinary and two stochastic differential equations. We first apply a spatial discretization to the partial differential equation and use model reduction techniques to reduce the dimension of the associated system of ordinary differential equations. Finally, a time-discretization leads to a Markov decision process for which we apply a state discretization to determine approximations of the cost-optimal control and the associated value function.