Intelligent district heating is the combination of traditional district heating engineering and modern information and communication technology. A district heating systemis a highly complex environment consisting of a large number of distributed entities, and this complexity and geographically dispersed layout suggest that they are suitable for distributed optimization and management. However, this would in practice imply a transition from the classical production-centric perspective normally found within district heating management to a more consumer-centric perspective. This thesis describes a multiagent-based system which combines production, consumption and distribution aspects into a single coherent operational management framework. The flexibility and robustness of the solution in industrial settings is thoroughly examined and its performance is shown to lead to significant operational, financial and environmental benefits compared to current management schemes.

A district heating system consists of one or more production units supplying energy in the form of heated water through a distribution pipe network to a multitude of consumers. District heating systems come in a range of different forms and sizes; from small independent systems within industrial estates or university campuses to large city-wide systems supplying millions of consumers with heating and hot water. The geographically dispersed layout of district heating systems suggest that they are suitable for distributed optimization and management. However, this would imply a transition from the classical production-centric perspective normally found within district heating management to a more consumer-centric perspective. In this work we use multi-agent based systems in order to implement distributed policies for operational planning within district heating systems. We also develop models for simulating the dynamics of district heating systems in order to evaluate those policies and their use in computer-based demand side management approaches for improving operational planning and resource management. These policies are then implemented in real world industrial settings and their performance, as well as implementation issues, are analysed and evaluated. It is shown that distributed policies can lead to significant benefits compared to current schemes with respect to energy usage and heat load management at an operational level.

Detta är en förstudie som har som syfte att jämföra simuleringsverktyget DHEMOS med andra på marknaden förekommande kommersiella verktyg. DHEMOS baseras på öppen källkod och har utvecklats i samband med tidigare forskning.

Utvärdering av fjärrvärmesimulator baserad på öppen källkod. Utförd i samverkan med Swedavia på Landvetter flygplats fjärrvärmenät. Fjärrvärme, simulator, öppen källkod

In this paper we investigate the consequences of using temporary heat load reductions on consumer substations, from the perspective of the individual consumer as well as the district heating company. The reason for using such reductions are normally to save energy at the consumer side, but the ability to control the heat load also lie at the core of more complex control processes such as Demand Side Management (DMS) and Load Control (LC) within district heating systems. The purpose of this paper is to study the way different types of heat load reductions impact on the energy usage as well as on the indoor climate in the individual buildings. We have performed a series of experiments in which we have equipped multi-apartment buildings with wireless indoor temperature sensors and a novel type of load control equipment, which gives us the ability to perform remotely supervised and coordinated heat load reductions among these buildings. The results show that a substantial lowering of the heat load and energy usage during periods of reductions is possible without jeopardizing the indoor climate, although we show that there are differences in the implications when considering different types of heat load reductions.

A district heating consumer substation is a complex entity, consisting of a range of interacting components such as valves, pumps, heat exchangers and control systems. The energy efficiency of a consumer sub- station is dependent on several things, e.g. settings of the control system, dimensions and operational behaviour of hardware and accumulation of sediments in the heat exchanger. Visualizing this operational functionality of consumer substations has been studied in several previous projects. This paper addresses certain shortcomings inherent in those previous works by presenting a novel visualization approach using parallel coordinates and scatter plot matrices. A comparison between these and previous visualization techniques is presented and discussed. Furthermore, the paper presents a scheme for statistical analysis based on n-dimensional relationships found in parallel coordinates and scatter plot matrices, thus providing key performance indicators appropriate for large-scale detection and analysis. It is shown that the presented visualization techniques are at least equal to previous attempts in regards to fault detection and operational analysis, while simultaneously addressing several of their shortcomings. Furthermore, it is shown that the subsequent statistical analysis provides a workable starting point for system-wide fault detection and analysis within any district heating system.

Multi-agent cooperation can in several cases be used in order to mitigate problems relating to task sharing within physical processes. In this paper we apply agent based solutions to a class of problems defined by their property of being predictable from a macroscopic perspective while being highly stochastic when viewed at a microscopic level. These characteristic properties can be found in several industrial processes and applications, e.g. within the energy market where the production and distribution of electricity follow this pattern. We evaluate and compare the performance of the agent system in three different scenarios, and for each such scenario it is shown to what degree the optimization system is dependent on the level of availability of sensor data.

Multi-agent cooperation can in several cases be used in order to mitigate problems relating to task sharing within physical processes. In this paper we apply agent based solutions to a class of problems defined by their property of being predictable from a macroscopic perspective while being highly stochastic when viewed at a microscopic level. These characteristic properties can be found in several industrial processes and applications, e.g. within the energy market where the production and distribution of electricity follow this pattern. Another defining problem characteristic is that the supply is usually limited as well as consisting of several layers of differentiating production costs. We evaluate and compare the performance of the agent system in three different scenarios, and for each such scenario it is shown to what degree the optimization system is dependent on the level of availability of sensor data.

Combined heat and power (CHP) generation is often used when building new district heating production. CHP makes it possible to simultaneously produce electricity and heat, thus maximizing the energy efficiency of the primary fuel. The heat is used in the connected district heating system while the electricity is sold on the local power market. In a CHP plant it is not possible to separate the physical process of producing heat and electricity, which may cause suboptimal behaviour when high spot prices for power do not coincide with high heat load demand. This paper presents the design and implementation of a system which makes it possible to control the heat load demand in a district heating network in order to optimize the CHP production. By using artificial intelligence technology in order to automate the run‐time coordination of the thermal inertia in a large amount of buildings, it is possible to achieve the same operational benefits as using a large storage tank, albeit at a substantially less investment and operational cost. The system continuously considers the climate in each participating building in order to dynamically ensure that only the best suited buildings at any given time are actively participating in load control. Based on the dynamic indoor climate in each individual building the system automatically controls and coordinates the charging and discharging of the buildings thermal buffer without affecting the quality of service. This paper describes the overall function of the system and presents an algorithm for coordinating the thermal buffer of a large amount of buildings in relation to heat load demand and spot price projections. Operational data from a small district heating system in Sweden is used in order to evaluate the financial and environmental impact of using this technology. The results show substantial benefits of performing such load control during times of high spot price volatility.

This paper describes results and experiences from an industrial proof-of-concept installation of a multi-agent based load control system in three major district heating systems in Sweden. A district heating system is a demand-driven system, i.e. the consumption controls the level of energy input which the district heating producer needs to deliver into the system. The basic idea of load control is that the individual consumers can be utilized as heat load buffers which, when coordinated on a system-wide scale, can be used to adjust the total consumption demand instead of having to change the production scheme. Load control leads to several important benefits such as giving the district heating producer the capability to avoid using expensive and environmentally unsound peak load boilers, while at the same time lowering the overall energy consumption at the consumer side. In order for load control to work the system needs to be able to coordinate the behaviour of a large amount of consumer substations in relation to the dynamic status among a range of production units, while continuously maintaining a sufficient level of quality of service among the consumers. The results show that the multi-agent based system was capable of reducing the peak loads with up to 20% of the total load, and to lower the average energy consumption with about 7,5% without any deterioration of the experienced indoor climate. Different theoretical aspects of load control have long been studied, but it is not until recently that technical advances in hardware and communication infrastructure has made it possible to implement these schemes in real-world settings.

Thermal storage is an essential concept within many energy systems. Such storage is generally used in order to smooth out the time lag between the acquisition and the use of energy, for example by using heat water tanks within heating systems. In this work we use a multi-agent system in order to maintain and operate distributed thermal storage among a large group of buildings in a district heating system. There are several financial and environmental benefits of using such a system, such as avoiding peak load production, optimizing combined heat and power strategies and achieving general energy efficiency within the network. Normally a district heating system is purely demand driven, resulting in poor operational characteristics on a system wide scale. However, by using the thermal inertia of buildings it is possible to manage and coordinate the heat load among a large group of buildings in order to implement supply driven operational strategies. This results in increased possibilities to optimize the production mix from financial and environmental aspects. In this paper we present a multi-agent system which combines the thermal storage capacities within buildings in relation to production optimization strategies. The agent system consist of producer agents responsible for valuing the necessary heat load management, consumer agents managing the quality of service in individual buildings while consenting to participate in heat load management and a market agent acting as a mediating layer between the producer and consumer agents. The market agent uses an auction-like process in order to coordinate the heat load management among the consumer agents, while the producer agents use load forecasting in order to evaluate the need for heat load management at any given point in time. A consumer agent uses continues feedback regarding indoor climate in order to uphold quality of service while par- ticipating in heat load management. Real-time data from a district heating system in Sweden is used in order to evaluate the agent system in relation to operational peak load management. The results show clear financial and environmental gains for the producer as well as participating consumers

District heating systems (DHS) is in many countries an important Agent-based industrial applications, Smart Heat Grid, Combined part of the heating infrastructure, especially in and around urban Heat and Power areas. Combined heat and power (CHP) production makes it possible to producer heat while simultaneously producing power. This combination help maximize the energy efficiency in production, often reaching an 80-90% utilization level of the primary fuel, compared to around 30-50% in a traditional power plant. The heat produced in the CHP plant is used to heat the adjacent DHS, while the power is transferred and sold on the power market. The work presented in this paper relates to the Nord Pool Spot power market, which is the leading power market in Europe and one of the largest in the world. On Nord Pool Spot power is bought and sold based on hourly spot prices, facilitated by the primary day-ahead market and the supplementary balancing intraday market. Since it isn’t possible to separate the physical process of producing heat and power in a CHP production facility, the energy company will want to synchronize high heat load production with high spot prices for power whenever possible. This can be done by using large storage tanks where heat is buffered during hours with high spot prices, while then distributedto the DHS as the heat load demand increases. However, such storage tanks are expensive to build and maintain, and they have limited operational dynamics. An alternative is to use the actual buildings connected to the DHS, in order to utilize their thermal inertia by the use of active load control. This paper presents a multi-agent system (MAS) designed to bridge the information gap between energy companies and building owners in order to enable the use of system-wide active load control in order to synchronization heat load and spot prices. The presented scheme provides a self-regulating market analogy in which agents act to allocate load control resources. Each participating building is assigned a consumer agent, while each production unit is represented by a production agent. These agents interact on the market analogy which is in turn supervised by a market agent. The work in this paper is focused on the intraday market although the underpinning synchronization scheme is suitable for the day-ahead market as well as the intraday market. The results show considerable gains for participating entities when applying the presented strategy to the often volatile intraday spot price market.

One common strategy to include more downstream lifecycle dimensions in early design is to enrich modelling and simulation techniques embedded in decision support systems. However, downstream dimensions are difficult to trade against more traditional engineering objectives. This research studied through individual interviews how six disciplines use models to negotiate design trade-offs. References to models were categorised according to whether models supported or hampered the duration of trade-off identification and how they impacted the duration of trade-off resolution. The results point to the difficulty of applying downstream lifecycle issues earlier in the design process because of the characteristics of the models that are used. A list of characteristics promoting and limiting the use of four models as boundary objects (CAD models, simulation results, total cost of ownership and decision matrices) is provided. The cross-analysis of these characteristics provides insights into how models need to be organised in decision support systems.