Abstract
Introduction:
Solar thermal devices are sources of cheap renewable heat with minimal environmental impact. The main disadvantages of this technology are the irregularity of solar irradiation (on which the heat yield depends) and the typically negative correlation between solar heat supply and heat demand. These disadvantages can be lessened, but not completely eliminated, by the use of storage tanks.
As a next step, the heat production can be stabilized by the use of heat pumps, though at the price of significant consumption of electric energy. In practical implementations, solar thermal devices are often embedded in complex heat generation and distribution systems. In general, these range from single homes with an additional biomass boiler to large-scale heat networks which incorporate many types of heat sources. In installation engineering, the challenges of including solar thermal devices in such heat distribution systems are modest and can be regarded as largely solved.
The control of both the solar thermal plants and the heat distribution systems, however, constitutes a major challenge. Standard strategies, typically independent linear controllers for the individual subsystems, are not appropriate for control of transient modes. By disregarding the couplings between the systems, changes in operating conditions in one system can lead to oscillations and suboptimal behavior in others, which again can have repercussions on the first one. A promising path to make full use of solar thermal devices while avoiding such unwanted behavior is the use of subordinate model-based and superordinate model-predictive control.
The Subordinate Model-Based Control Approach:
The main reason why simple linear controllers like standard PID controllers are often unsuited for control even of single solar thermal collectors is the dead time (delay), which is unavoidably present in such systems. In a model-based approach, the dead time can be explicitly taken into account in the control strategy.
Since the velocity-dependent throughput time corresponds to the heating time in the collector, one has some level of control over the outlet temperature by properly adjusting the mass flow. Within certain limits, a model-based control approach can thus provide a fixed outlet temperature in spite of varying ambient conditions. The heat flux, of course, will vary according to these conditions.
Adding an electrically operated heat pump to the system allows to decouple (to some extent) outlet temperature and heat flux. With a model-based control approach it is thus possible to independently prescribe target values both for outlet temperature and heat flux (within technical and physical limits).
The Superordinate Model-Predictive Control Approach:
Superordinate Model-Predictive Control (MPC) takes the model-based approach one step further and combines a (usually very simplified) model of the system with forecasts for possible solar yields and heat demand. By setting the target values for critical system quantities, a superordinate MPC can make best use of all components and avoid unwanted situations (like fully loaded storage tanks at a time when considerable solar yields are possible). The main conditions for reasonable use of MPC in a heat distribution system are:
• Sufficient knowledge of the system, summarized in a mathematical model, is available
• Reliable forecasts for solar irradiation, ambient temperature and heat demand can be obtained.
• The system contains at least one significant device for heat storage.
• The system has non-trivial heat demand characteristics.
Implementation of MPC can be done in different ways. We will present three main approaches:
• A central instance optimizes the operation mode by regarding all actuator settings as degrees of freedom in order to minimize a reasonably defined cost function.
• A central instance defines time-dependent prices for heat in-feed and possibly also heat consumption. Consumers, producers and hybrid prosumers act autonomously based on the price levels.
• Subsystems provide individual prices for heat in-feed and heat consumption. Heat transactions occur based on choice of best offers, taking into account transport losses.
The conditions for reasonable inclusion of MPC and the approaches to its implementation, including advantages and disadvantages, will be presented as a poster at the conference.
Solar thermal devices are sources of cheap renewable heat with minimal environmental impact. The main disadvantages of this technology are the irregularity of solar irradiation (on which the heat yield depends) and the typically negative correlation between solar heat supply and heat demand. These disadvantages can be lessened, but not completely eliminated, by the use of storage tanks.
As a next step, the heat production can be stabilized by the use of heat pumps, though at the price of significant consumption of electric energy. In practical implementations, solar thermal devices are often embedded in complex heat generation and distribution systems. In general, these range from single homes with an additional biomass boiler to large-scale heat networks which incorporate many types of heat sources. In installation engineering, the challenges of including solar thermal devices in such heat distribution systems are modest and can be regarded as largely solved.
The control of both the solar thermal plants and the heat distribution systems, however, constitutes a major challenge. Standard strategies, typically independent linear controllers for the individual subsystems, are not appropriate for control of transient modes. By disregarding the couplings between the systems, changes in operating conditions in one system can lead to oscillations and suboptimal behavior in others, which again can have repercussions on the first one. A promising path to make full use of solar thermal devices while avoiding such unwanted behavior is the use of subordinate model-based and superordinate model-predictive control.
The Subordinate Model-Based Control Approach:
The main reason why simple linear controllers like standard PID controllers are often unsuited for control even of single solar thermal collectors is the dead time (delay), which is unavoidably present in such systems. In a model-based approach, the dead time can be explicitly taken into account in the control strategy.
Since the velocity-dependent throughput time corresponds to the heating time in the collector, one has some level of control over the outlet temperature by properly adjusting the mass flow. Within certain limits, a model-based control approach can thus provide a fixed outlet temperature in spite of varying ambient conditions. The heat flux, of course, will vary according to these conditions.
Adding an electrically operated heat pump to the system allows to decouple (to some extent) outlet temperature and heat flux. With a model-based control approach it is thus possible to independently prescribe target values both for outlet temperature and heat flux (within technical and physical limits).
The Superordinate Model-Predictive Control Approach:
Superordinate Model-Predictive Control (MPC) takes the model-based approach one step further and combines a (usually very simplified) model of the system with forecasts for possible solar yields and heat demand. By setting the target values for critical system quantities, a superordinate MPC can make best use of all components and avoid unwanted situations (like fully loaded storage tanks at a time when considerable solar yields are possible). The main conditions for reasonable use of MPC in a heat distribution system are:
• Sufficient knowledge of the system, summarized in a mathematical model, is available
• Reliable forecasts for solar irradiation, ambient temperature and heat demand can be obtained.
• The system contains at least one significant device for heat storage.
• The system has non-trivial heat demand characteristics.
Implementation of MPC can be done in different ways. We will present three main approaches:
• A central instance optimizes the operation mode by regarding all actuator settings as degrees of freedom in order to minimize a reasonably defined cost function.
• A central instance defines time-dependent prices for heat in-feed and possibly also heat consumption. Consumers, producers and hybrid prosumers act autonomously based on the price levels.
• Subsystems provide individual prices for heat in-feed and heat consumption. Heat transactions occur based on choice of best offers, taking into account transport losses.
The conditions for reasonable inclusion of MPC and the approaches to its implementation, including advantages and disadvantages, will be presented as a poster at the conference.
Originalsprache | deutsch |
---|---|
Publikationsstatus | Veröffentlicht - 9 Juni 2016 |
Veranstaltung | Gleisdorf SOLAR 2016 - Internationale Konferenz für solares Heizen und Kühlen - Gleisdorf, Österreich Dauer: 8 Juni 2016 → 10 Juni 2016 |
Konferenz
Konferenz | Gleisdorf SOLAR 2016 - Internationale Konferenz für solares Heizen und Kühlen |
---|---|
Land/Gebiet | Österreich |
Ort | Gleisdorf |
Zeitraum | 8/06/16 → 10/06/16 |
Schlagwörter
- Solarthermie
- Regelungstechnik
- Modellierung
- Prädiktive Regelung
- Wärmemanagement
Fields of Expertise
- Sustainable Systems