Short-term operational stability of 5th generation district heating and cooling substations

Abstract

Energy demand for heating and cooling makes up 73\% of total energy use of buildings [2]. This figure is expected to increase over the coming decades, as demand for cooling energy may rise in the Netherlands due to climate change. The heating and cooling of buildings are at relatively low temperature compared to the industry and thus allowing for a 5th Generation District Heating and Cooling (5GDHC) network with thermal energy storage. Several guidelines are available for the design of 5GDHC substations \cite{ISSO39}, however they do not integrally consider the control design.
Currently, in literature the performance of the substations is generally based on 1-hour measurements yet the control of the short term dynamics have a significant impact on the performance, thus shorter time steps are required to improve the performance.


This study considers the short term dynamics of substations of large utility buildings connected to a 5GDHC grid, allowing for seasonal thermal energy storage and independent bi-directional consumption for all buildings. A hydro- and thermodynamic model including the controllers are compared with a case study of utility buildings at Utrecht University where high frequency operational data is gathered. This provides a view on the causes of the problems and inefficiencies and new insights into improvements on the hydro- and thermodynamics and control of the substation. A new control method for the heat exchanger and for the Heat Pump (HP) group is proposed to maximise the networks capacity.

The use of Model Predictive Control (MPC) controllers is investigated to determine the performance improvements that can be achieved based on a simple physical system model and the energy demand and temperature setpoint predictions. As the short-term dynamics are discussed, the time steps for the MPC optimisation are relatively small while the many on/off switches that are allowed might make it computationally intensive due to large number of binary variables. The flow in some parts might be reversed, which is modelled using binary variables. Finally, as the system is transporting water at different temperatures and different mass flow rates, it is described by bilinear equations. These equations can be linearised by the McCormick relaxations, however they once again increase the computational burden as new binary variables have to be introduced.

The changed proposal for switching conditions could easily be implemented and reduces the temperature limit violations as well as the proposed control method for the warm Heat Exchanger (HEX) as it stabilises the substation control. The proposed method for the temperature setpoint and capacity control showed an improved regulation of the temperature and energy provided by the heat pumps and could even reduce the consumption of conventional heating sources. Finally, the MPC controller shows an increased performance both for the constant and variable flow conditions, but cannot yet be implemented as the computational burden is too large.