Publication | Conference Papers | Sonstige

Modellprädiktive Regelung eines solar-und biomassebasierten Fernwärmenetzes

Moser A, Muschick D, Lichtenegger K, Gölles M, Hofer A

Published November 2017

Citation: Moser A, Muschick D, Lichtenegger K, Gölles M, Hofer A. Modellprädiktive Regelung eines solar- und biomassebasierten Fernwärmenetzes. Zukunft der Gebäude: digital - dezentral - ökologisch. 23 Nov 2017; Leykam;16:151-159.

Abstract

The use of renewable-energy-based heat producers within district heating grids is getting more and more popular. In order to benefit from the advantages and compensate for the different disadvantages of the various types of heat producers powered by renewable energy sources like biomass, solar energy or waste heat, a combination of these systems could be favoured over using, for example, only one main biomass-based boiler. Furthermore , in many cases, the additional use of buffer storages is necessary to fully benefit from the use of these kinds of heat producers. A major challenge with such multi-producer heating grids is the cost optimal management of all heat producers and buffer storages. Therefore , a high-level control strategy is necessary, which is able to plan ahead the use of slowly reacting and/or weather dependent heat producers while minimizing operational costs and pollutant emissions. This article shows the development of a linear model predictive controller (MPC) for a district heating grid with several (renewable) decentralized heat producers and heat storages. In order to provide the MPC with the required forecast of the future heat demand, an adaptive load forecasting method has been designed. Additionally, in order to be able to incorporate solar panels, the MPC needs to have a forecast of their possible future heat output. Therefore, a physically motivated solar yield forecasting method has been designed. The required prediction models for the MPC were represented by so-called mixed logical dynamical (MLD) system models. MLD system models combine the modelling power of discrete state system models (finite state machines) and discrete time system models by the extension of the regular linear state-space system model approach with integer and continuous auxiliary variables and linear inequality constraints. The occurrence of both integer and continuous variables within the resulting optimization problem of the MPC leads to a mixed-integer linear program (MILP), which can be solved efficiently using modern MILP solvers. The resulting control strategy is tested in a thermo-hydraulic simulation environment of an actual small-scale multi-producer district heating grid consisting of a medium-scale wood chip boiler with buffer storage, a solar collector with buffer storage and a high temperature heat pump, an oil boiler and 25 heat consumers. Additionally, a state observer was designed and connected with the MPC in order to detect control errors and to incorporate feedback from the heat producers and the buffer storages. The simulations have indicated that the designed MPC and the state observer work properly. Therefore, these elements have been implemented on-site on the actual heating grid, with the first test run scheduled for October 2017.
Modellprädiktive Regelung eines solar-und biomassebasierten Fernwärmenetzes | Request PDF. Available from: https://www.researchgate.net/publication/321314304_Modellpradiktive_Regelung_eines_solar-und_biomassebasierten_Fernwarmenetzes [accessed Feb 21 2018].

External Link