1 > REPLACE THIS LINE WITH YOUR PAPER IDENTIFICATION NUMBER (DOUBLE-CLICK HERE TO EDIT) < 1 Coordination of System needs and provision of Services Hjörtur Jóhannsson, Henrik Hansen, Lars Henrik Hansen, Hans-Henrik Holm-Hansen, Peder Cajar, Henrik W. Bindner and Olof Samuelsson Abstract This paper addresses a challenge associated with large scale deployment of distributed energy resources (DER) to provide system services in future sustainable power systems; namely how to prioritize conflicting interests in a service provided by a DER. For that purpose, different services utilizing the DER in a future system are identified as well as potentially conflicting interests between services. A scheme is suggested for how conflicting interests should be prioritized based on considerations regarding the nature of the service and the system operating state at the time of the service request. Examples are provided for illustrating the functionality of the scheme. Index Terms Smart grids, distributed power generation. I. INTRODUCTION The share of wind power generation in the Danish power system has increased from being negligible in 1980 to represent approximately 20% share of the total domestic electricity supply in 2009 . Danish TSO expects that wind power will cover 50% of Danish electricity consumption in 2025 [2;3] and furthermore, the Danish government has expressed a long term target of achieving a Danish energy supply based on 100% renewable energy by 2050, much of which will be intermittent production. As a consequence the power system is moving from a system with mainly centralized power production to a system which has an increasingly distributed production. This introduces challenges since the central power plants deliver not only electric power, but also a set of ancillary services to support the power system. The ancillary services support system operators in their effort to maintain the power system in a secure and stable state of operation. As the share of conventional central generation is decreasing by means of renewable and distributed power Manuscript received October 05, This work was a part of the ipower SPIR platform (Strategic Platform for Innovation and Research) and is funded jointly by the Danish Council for Strategic Research and the Danish Council for Technology and Innovation. H. Jóhannsson is with the Centre for Electric Technology, Department of Electrical Engineering at the Technical University of Denmark. H. Hansen is with the Danish Energy Association, Copenhagen, Denmark. L. H. Hansen, P. Cajar and H.H. Holm-Hansen are with DONG Energy (a Danish DSO), Copenhagen, Denmark. H. W. Bindner is with Risø-DTU the Danish National Laboratory for Sustainable Energy, Roskilde, Denmark. O. Samuelsson is at Department of Industrial Electrical Engineering and Automation at Lund University, Sweden. generation increasing, it becomes necessary to find alternative means to provide the ancillary services. With the expected development towards a power system based on intermittent renewable energy sources, the need for some of the ancillary services is likely to increase, especially for ancillary services for balancing purposes . In , the Danish TSO published the results of an extensive survey of methods that could make it possible to meet the goal of 50% wind energy in the Danish power system by This report emphasizes on the challenge, which an increased availability of resources for power system balancing is going to be critical. For that purpose, distributed energy resources (DER) that can be controlled in such a way that it supports the system operation are considered to be an important resource in the future power system. Examples of DER within this project that could contribute to system balancing are mainly focused on household sized micro-chp units, flexible loads like electric space heating, electric vehicles and other distributed units, all being too small to act in the current market on individual basis. Demand response is thus considered a part of DER. The main focus is thus on the consumer, and how the consumer can deliver services to grid controllers. These services can either be by actively delivering services, or by changing behavior to match the needs of the grid instead of the grid adapting to the needs of the consumer at all times. During strong wind conditions, where the total wind power production capability becomes highly utilized, the task of maintaining balance between production and consumption will not be the only system stability problem of concern. E.g. if the balance is maintained by utilizing controllable loads, the transmission and distribution system might become overloaded. Hence, the risk of stability problem occurrences that are greatly affected due to stressed network conditions (such as voltage and angle instability) is increased. The increased risk of other forms of stability problems, when a fluctuating power production has to be balanced by DER, has motivated research and development of methods able to carry out a real-time assessment of overall system stability and security [5;6] and research on methods for automatically identifying control actions to avoid instability. In these studies, altering the system consumption patterns using controllable DER have been considered as one of the means for avoiding system instability.
2 > REPLACE THIS LINE WITH YOUR PAPER IDENTIFICATION NUMBER (DOUBLE-CLICK HERE TO EDIT) < 2 The exploitation of flexible demand has been considered for several other purposes as well, ranging from providing local services for avoiding local congestion in distribution grids, to services for supporting overall system stability. Ever since the fundamental idea of letting DER respond to instantaneous, electricity prices was presented in 1980 , it has become a subject in several research and demonstration projects . The concept of control-by-price is well suited for control of DER due to its transparency and simplicity. It could imply that a price signal must be sent to a customer in due time and that a computer can take appropriate control actions that maximize the customer s profit . DER could also be aggregated into VPPs (Virtual Power Plants), which is another way to control and mobilize DER in the market [9-11]. Before a market model can be designed or existing markets be adapted for accommodating services from the DER resources, it is of importance to identify the different services utilizing DER in a future system, identify potentially conflicting interests between services and to determine how such conflicting interests should be prioritized. The identification, prioritization and structuring of services provided by flexibility of DER in a future power system cover the aim of this paper. II. OPERATING STATES AND CONTROL STRATEGIES Figure 1 Power system operating states and the transitions that can take place from one state to another . A properly designed and operated power system should at all time meet a set of fundamental requirements : Meet the continuous changing load demand Supply energy at minimum cost and ecological impact The quality of power supply should meet certain minimum standards for constancy of frequency and voltage, and level of reliability. For meeting these requirements, several levels of controls are introduced at generating units, transmission systems and at control centres. The different levels of control contribute to a satisfactory operation of the power system where voltages, frequency and other system variables are kept within their acceptable limits. The overall control objectives are dependent on the operating state of the power system. Under normal operating conditions, the control objective is to operate as efficiently as possible with voltages and frequency close to nominal values. When an abnormal condition develops, new control objectives must be met to restore the system back to normal operation. For the purpose of analyzing power system security and designing appropriate control systems, it is helpful to conceptually classify the system operating conditions into five states as illustrated in Figure 1. The five states are [12-14]: Normal: All system variables within normal range and no equipment being overloaded. The system operation is secure, meaning that it is able to withstand a contingency without violating any of the constraints. The overall control objective in the normal state is to operate the system as efficiently as possible. Alert: The system enters alert state if the security level falls below a certain limit of adequacy. System variables are within acceptable range and all constrains are satisfied. The system has though been weakened to a level where a contingency may cause the system to enter emergency state or if the disturbance is severe enough, a shut-down of a major part of the system (enters in extremis). In this state, the overall control objective is to restore normal operation. Emergency: If a sufficiently severe disturbance has occurred when operating in alert state, the system enters the emergency state. The system is still intact and may be restored to the alert state if appropriate emergency actions are taken (e.g. load shedding, controlled system separation, etc.). If no counter actions are taken or they are ineffective, it may result in cascading failures and possible shut down of major part of the system. The control objectives in this state aim at saving the system from instability where severe and costly control actions are considered for this purpose. In Extremis & Restorative: a phase when cascading outages have caused a shutdown of a major part of the system and a phase when control action is being taken to reconnect all facilities and restore system load. A characterization of the system conditions into the five states provides a framework in which control strategies can be developed and where operator actions can be taken in order to deal effectively with each state. Furthermore, it provides a good basis for coordinating conflicting interests that request a service from a DER. III. COORDINATION OF CONFLICTING INTEREST A. The Stakeholders The possibility of flexibility provided by DER introduces new resources for providing services that support stable, secure and economically efficient operation of electric power systems. The main stakeholders with interest in these system services are:
3 > REPLACE THIS LINE WITH YOUR PAPER IDENTIFICATION NUMBER (DOUBLE-CLICK HERE TO EDIT) < 3 The Transmission System Operator (TSO) The Distribution System Operator (DSO) The Balance Responsible Party (BRP) The TSO is a monopoly responsible for the operation of the transmission system and the overall system stability. The DSO is a monopoly responsible for the operation of the distribution grid and is thereby responsible for ensuring power delivery to customers at all times, without disturbing the transmission system. The BRP is financial responsible for the energy acquired from the power market. In case of deviations from the purchased energy, the BRP has to pay for imbalances to the TSO, since the TSO is forced to activate up or down regulation acquired in the power regulation market in order to balance the imbalances. Other stakeholders affiliated to the BRP are the retailer and the wholesaler. The retailer communicates consumers demands to the power market through the wholesaler, who purchases electricity and resells it to the retailer. A retailer can also be a wholesaler (the same company), and a wholesaler can be a balance responsible Party (BRP). Given the deregulation of the European electricity industry  the way to obtain and make use of these services must be on a commercial basis, i.e. through markets. Regulating consumers in order to provide services to the power grid by regulations and paragraphs is not a possible approach in this light. Thus, the main relation between these stakeholders is: the BRPs provide services for the markets driven by the monopolies. Today, it is mainly the TSO, who has established markets. Managing the power grid is thus moving away from a monopolized system control with exact rules, into a system control based on liberalized markets. This way of control needs to be monitored and governed, to prevent speculation and financial gain from overruling the need for stability . It requires a strong market, with a solid foundation on the technical interworking of the many possible services; to ensure that services cannot be bought in a manner that makes it profitable to cause instability in the power grid. B. Identifying system needs with interest in utilization of DER flexibility The system needs which can exploit the DER flexible demand for achieving its objective are many. The system needs vary in nature, ranging from needs that deploy local DER for solving local problems, to needs that address system wide problems by deploying large amounts of DER spread over a wide area. Furthermore, the system needs differ in respect to time constraints for the service to be carried out. E.g. some services have a fixed continuous demand (MW/Mvar keeping a constant balance), other financial services have to be solved within the hour (MWh/Mvarh as a total amount of energy within a given period of time), and even other services are measured on quality over a period of weeks (e.g. flicker over a week). The system needs utilizing DER flexibility are listed below, where they are grouped according to which stakeholder they belong to: 1) TSO need for services: Frequency control: Covers the various different reserve power markets today, all the way from primary to tertiary (manual) reserve power. o Primary reserves: respond automatically to changes in frequency. These need to activate automatically, without any delays. o Other reserves: The remaining reserves for frequency control have the purpose of releasing the primary reserve when it has been activated. These have less stringent requirements on activation time, and can thus be utilized by triggering them by automatic systems, using them while avoiding e.g. local congestion. Voltage support: In regards to the transmission grid. This covers the various possible services that can assist the TSO in maintaining the overall voltage balance for the entire power system. An example could be the distribution grid reducing the reactive toll on the transmission grid. Other ancillary services: Includes several different things, dominantly the inertia in the system and short circuit capacity. The inertia contributes to increased stiffness of the system during and following disturbances, while the short circuit capacity affects mainly fault clearing. In addition to the TSO needs, the TSO can intervene in the system operation during emergency conditions. The system enters an emergency state, if a sufficiently severe disturbance occurs. The system is in an alert state when the system security has fallen below certain limits  and a sufficiently severe disturbance occurs. In order to avoid a widespread blackout of the system, the TSOs can take appropriate emergency actions e.g. generation tripping, activation of HVDC-emergency reserves and load curtailment. 2) DSO need for services: Peak-shaving: Is initially a local service, aimed at evening out peaks in power consumption where these peaks might otherwise cause a given component (cable, transformer, fuse, etc.) to exceed its capacity. The service is later extensible to any component in the system, which potentially can become overloaded due to high power transfer during the peak interval. Power Quality: Cover the general power quality. While aspects such as harmonic currents and phase imbalance have some effect on the provision of services, aspects such as flicker have little effect on this. Reduction of e.g. harmonic currents would reduce transmission losses in the distribution grid. Local voltage control: The local voltage level in the distribution grid, while an aspect of the voltage quality, is treated separately. This, as it affects power flow greatly in the distribution grid, among things. It can thus have a
4 > REPLACE THIS LINE WITH YOUR PAPER IDENTIFICATION NUMBER (DOUBLE-CLICK HERE TO EDIT) < 4 significant influence on the ability of local assets to provide services. Mvar bands: To prevent large reactive power flow between the distribution and transmission grids, system operators may impose explicit limits on reactive power flow through system transformers. In Denmark the TSO has divided the TSO grid in 15 regions as described in . An operational band for the reactive power flow between the TSO and each region is defined This effectively places upper and lower limits on the reactive exchange at a given point of connection between a distribution grid and the transmission grid. This could also potentially evolve into dynamic bands, in order to allow increased Mvar levels if desired in a given situation. The reason why the peak-shaving issue is only attached to the DSO and equivalently not to the TSO, is mainly due to two reasons: 1) when clearing the day-ahead market (the Nord Pool Spot), the main constraints on the interconnectors between the control zones are taken into account in the final solution, and 2) the Danish TSO has future plans for massive reinforcement of the their grid. The BRP will also experience congestion as it has operations in the distribution grid. Due to the liberalized market setup, the needs of the two monopolies are serviced by bits from the BRPs. However, the BRPs themselves have some needs for services, which also can be provided by DER flexibility. 3) BRP need for services: Imbalance issues: Represents the financial issues of a BRP when not maintaining the forecasted consumption and production within a given hour, within their area. Thereby having an imbalance between the amount of energy contracted for and the actual consumption /production. This need will become more important, as new services emerge, affecting the actual power flows that the BRP has estimated. While less of a technical service, it is a good example of the complex control, where an imbalance of one BRP causes a financial penalty to that BRP, while providing balancing services from another BRP yields financial benefits to that BRP. Work with congestion: The presence of bottlenecks will act as a new type of imbalance. If a bottleneck occurs, the BRP cannot deliver the contracted energy to his customers because his is cut-off by the bottleneck. The BRP would then go into imbalance unless the BRP can temporarily buffer the imbalance at his customers not affected by the bottleneck. As such, historical constrictions will become more important in regards to future planning. The above list offers an initial overview of needs and their stakeholders. The service can be provided by third party (e.g. retailers and aggregators), but the caller of the need are the main stakeholders. Figure 2: Stakeholder interest in DER flexibility. Figure 2 illustrates the interests between the stakeholders and customers with flexible DER. The left side of Figure 2 shows three stakeholders with a need for services. The BRP includes a number of other entities, as the various commercial actors in the electricity market all purchase their power from a BRP. The Fleet Operator would be a commercial entity managing a fleet of flexible consumer appliances, e.g. the charging of electric vehicles, on behalf of the consumer. The right side is the consumer(s) and all flexible appliances. The two sides are connected by something balancing the requests for services from the three stakeholders. C. Commercial stakeholder as flexibility aggregator To further a commercial solution, the co-ordination of services is done through the commercial stakeholders. The commercial stakeholders aggregating all consumption are the BRPs, which should then co-ordinate all efforts on an aggregated level. This model is shown in Figure 3, with the BRP co-ordinating the needs of the three main stakeholders with the aggregated services from flexible consumers. Figure 3: BRP co-ordinating flexible services Figure 3 illustrates how the three main stakeholders access the services, offered by a local asset, through the BRP. Given the current structure of the electricity markets, the BRP is unlikely to directly control the flexible appliances, but instead asserts its control through a hierarchy. The architectural overview is shown in Figure 4. A list is illustrated in Figure 2, detailing the various stakeholders, each accessing the potential services offered by the flexibility of consumer assets.
5 > REPLACE THIS LINE WITH YOUR PAPER IDENTIFICATION NUMBER (DOUBLE-CLICK HERE TO EDIT) < 5 Figure 4: TSO and DSOs request service at different system levels and communicate with a BPR which activates local assets for supplying the service through a hierarchical structure of different commercial actors. In this hierarchical structure, the consumers flexible appliances are controlled by the lowest level commercial entity (or in some cases, the consumer himself), and each level of the hierarchy aggregates all the flexible components below it passing the total flexibility upwards. The aggregation ends with the BRP as the top-most aggregator with commercial interests. As a stakeholder the BRP has limited interest in the area aggregated under other BRPs while both the TSO and DSO have interests in areas under multiple BRPs. To this end the BRPs make a logical point of contact for the other two stakeholders, as the BRPs combined indirectly have contact to each and every flexible consumer/appliance across the grid. The TSO request a service to be provided at a node at transmission levels, while DSO requests a service to be supplied at a node at distribution levels. This provides flexibility in how the different needs can be met, since the needs of the TSO can be met by exploiting DER over a large part of a distribution grid, while the needs of the DSO must be met locally in the grid. Therefore, the needs of the DSO should in general be prioritized higher than the needs of TSO, since the variety of options for meeting the needs of the TSO is greater than those for the DSOs. The exception from this rule is when the system is operating in an emergency state; during such conditions the TSO must have a priority for carrying out emergency actions without unnecessary delays. D. Potential Conflicting Interests The introduction of multiple technical system needs, which desire to utilize the same asset opens up for potential conflicts between two or more of these stakeholders, as illustrated by Figure 2. Such conflicts arise when two stakeholders have contradicting needs, e.g. such as two active power services where one of them is requesting an increased consumption, while the other one requests a reduction in consumption. Such conflicts are not limited to a fixed set of two specific services. It should be recognized as well that an activation of a given service can have negative influence on system parameters governed by other services. This complex interworking makes the possible conflicts numerous and difficult to predict at any given time. When a commercial solution intended for enabling the flexibility of DER s is to be designed; it is of the utmost importance that the governing market rules are made in such a way that these technical conflicts are taken into consideration. Without treating this problem of conflicting needs, the commercial solution may have a destabilizing effect on the system operation under certain conditions. Situations where delivering a single service have a positive effect on more than one need may also become an issue. This is not examined in technical terms, but should be included in a market model to avoid anyone receiving payment for multiple needs by delivering a single service. E. Prioritizing the needs In order to handle conflicts efficiently between needs that are requesting flexibility of DER for meeting their objectives, it is useful to categorize the needs into the following four categories. Categorized needs: Emergency situations: These are actions taken to avoid impending system instability and eventually a blackout. These emergency actions are actions taken to save the overall grid at times where it is in a state where
6 > REPLACE THIS LINE WITH YOUR PAPER IDENTIFICATION NUMBER (DOUBLE-CLICK HERE TO EDIT) < 6 commercial solutions will not be enough to avoid critical instability. Emergency actions also include alert actions as a minor form. Alert actions are actions taken when the system is in an alert state, but not quite in an emergency yet, and the regular commercial solutions are incapable of providing a means to return the system to a stable operating point. Local constraints/requirements: These cover services for resolving constraints in the distribution grid, where a limited number of local resources are available. Global constraints/requirements: These cover services for re-establishing an adequate level of system security (maintaining margins and reserves), and for resolving non-critical issues in transmission networks. The services can be provided from a number of DER in the multiple distribution systems connected to them to solve the particular system need. Planning and financial aspects: These are not technical constraints, but limitations set by the accuracy of forecasts and financial penalties for not matching contracts. While these do not directly affect the security and stability of the system, they are directly affected by the multitude of possible changes brought on by the services used to maintain secure and stable operation of the system. These four categories enable a priority setup as shown in Figure 5. In the setup any emergency action is put above all other requirements. This is essential since such actions are taken when there might be a very short time before an impending serious instability or blackout occurs. The services providing resolutions for local constraints or requirement have the second highest priority in the priority setup. This is due to the limited resources which might be available for resolving the local issue. The services aiming at resolving global constraints have a greater amount of DER to choose from in order to meet its objectives. Therefore, a service resolving local problems should be prioritized before a service resolving global constraints in the case of conflicting interest prompting the global service to be found in another local area. The services categorized under the Planning and financial aspects have the lowest priority in the case of conflicting interests, since those services do not directly support a safer and more secure operation of the system. Figure 5: General priority of services Deploying the suggested prioritization listed in Figure 5, the previously identified services result in the following prioritized list of needs to gain priority access to an individual asset. Prioritised needs: 1. Emergency actions (TSO) 2. Alert actions (TSO/DSO) 3. Local voltage control (DSO) 4. Peak-shaving (DSO) 5. Voltage support (TSO) 6. Mvar bands (DSO) 7. Frequency control (TSO) 8. Other ancillary services (TSO) 9. Imbalance issues (BRP) 10. Power quality (DSO) The pyramid structure is not entirely utilized, as some needs do not affect the others exactly identically. Emergency actions have the highest priority. These are absolute required actions to be taken in order to prevent an imminent black-out, or move the system into a more stable state. Alert actions are placed in second place, although it is a solution that normally only would be applied when the commercial solutions are unable to deliver the needed service. Local voltage control is the highest priority for local services, as poor voltage control can result in less utilization of the local grid, and thus create more widespread congestions. Peak-shaving is important, as it maintains the constraints of a local substation or cable. If this constraint is not obeyed, the local fuse may blow disconnecting the area and creating additional unforeseen disturbances. Voltage support to the transmission grid is needed to maintain the stability of the transmission grid. This is not focussed on actively providing support but mainly on reducing the strain from the distribution grid on a node in the transmission grid. Mvar bands are in place to prevent large reactive exchange between a node in the transmission grid and its subjacent distribution grid. It is not a direct technical constraint, but large deviations can affect the voltage level. Frequency control is important as a service but it has a low priority on the list as it is a global service and it can be obtained at almost any point in the grid. However, obtaining frequency control should not press the voltage or local constraints beyond their limits. Other ancillary services have low priority since many of the services are needed due to a lack of fine control of some of the higher prioritized services, or services that need to be delivered more directly to the transmission grid. These services are unlikely to be possible from influencing consumers.
7 > REPLACE THIS LINE WITH YOUR PAPER IDENTIFICATION NUMBER (DOUBLE-CLICK HERE TO EDIT) < 7 Imbalance issues have low priority, as they are mainly financial. They do not solve technical needs, but attempt to keep a balance should not cause any of the higher prioritized technical needs to move into more unstable states. Power quality is a local problem; however, while a local high flicker can be annoying to a consumer and should be dealt with it does not cause problems with stability of the power system, as a whole. For this reason it has the lowest priority. IV. EXAMPLES Figure 4 illustrates the overall system infrastructure considered, both in respect to communication structure and power lines. The number of commercial entities may vary, but keeping the overall structure with the three main stakeholders. The BRPs are used as coordinating stakeholder, as they are responsible for all consumption within their area other setups could have other coordinating stakeholders, but the higher the number of consumers under the coordinating stakeholder, the easier it will be to balance and provide the services. Each 0.4 kv line has individual consumers by the hundred, some of these may have the same supplier, but this is not for certain. A. Case 1 In the following, a simple example is shown to illustrate how the frequency control is obtained in conflict with peakshaving. 1. The TSO is in need of frequency control. Either production must be curtailed, or consumption must increase. 2. The TSO requests an increase of X MW from the BRPs. 3. Each BRP, knowing their capacity from their planning, offer the TSO their capacity and price. 4. The TSO activates the cheapest bids with the relevant BRPs. 5. As one BRP activates its subjacent partners, an activated 0.4 kv flags a notice of imminent peakshaving needed. 6. The DSO informs the BRP of this, to which the BRP shifts the activation of flexible consumers to another area. If the BRP had activated the consumers under the station requiring peak-shaving they would have had to procure the consumption elsewhere to counter the peak-shaving or fail to maintain their balance of power. In extreme cases, the BRP will ignore the peak-shaving; the local area will be overloaded and disconnected. This creates an even larger problem and even more flexible consumers will be needed to control the frequency. It is thus entirely an unwanted scenario for all stakeholders. The cost of peak-shaving should be higher than the benefit of increasing consumption on market terms, encouraging the BRP to locate the service elsewhere, in order to prevent this scenario. B. Case 2 The second example illustrates the activation of a peakshaving service, conflicting with the BRP who has to maintain his power balance. 1. The DSO notices a local area that needs peak-shaving and requests the BRPs to do this. 2. Each BRP knowing their capacity from their planning offer the DSO their capacity and price. 3. The DSO activates the cheapest bids with the relevant BRPs. 4. As the BRP activates his subjacent partners, the BRP will go into an imbalance situation. The BRP will then have to locate increased consumption elsewhere and activate it, as the peak-shaving takes priority over the imbalance. Solving the peak-shaving should be more profitable than getting an imbalance, as leaving the peak-shaving unsolved would cause a disconnection and an even larger imbalance situation. The imbalance should still be solved by the BRP himself, to avoid a penalty for being in imbalance and possibly trigger frequency control. V. USING BEHAVIOUR DESCRIPTIONS FOR PRIORITIZATION The prioritization list, which was developed in section III, however, has the problem that the prioritization does not help in determining which services a particular asset can actually provide, and under which circumstances. Introducing a flexible scheme for activating services will help to solve these problems. This flexibility comes from the possibility to negotiate the list of services a device will offer, and to possibly renegotiate services when circumstances for the system change. If the system, for example, moves into a state of emergency the priorities shift away from commercial operation, to operation with focus on stabilizing the system irrespective of commercial interests. Such an approach is based on so-called behaviour descriptions. A behaviour description  is basically a description of expected future behaviour that is sent from a supervisory controller (such as a BRP) to the controller of the local device. A behaviour description contains a set of items. Each item refers to a service and describes the conditions for activating that service. Communication based on behaviour descriptions implies that the communication link between the supervisory controller and the controlled asset needs only to be available during the negotiation of the behaviour description, but not necessarily during the execution of the behaviour description. This is possible because the execution of the behaviour description, which can be triggered by locally measured or broadcasted signals, happens autonomously.
8 > REPLACE THIS LINE WITH YOUR PAPER IDENTIFICATION NUMBER (DOUBLE-CLICK HERE TO EDIT) < 8 A behaviour description can be understood as a kind of contract: Both parties, the supervisory controller and the controlled asset, agree on a contract, and then both sides have to adhere to the contract: The controlled device has to provide the service that has been agreed upon, and the supervisory controller must provide the compensation for provision of this service, e.g. a lower power price. Behaviour descriptions are also similar to contracts in the sense that a behaviour description is a binding document which forces the asset to behave in a certain way. If the component fails to do so, it breaches the contract, which will result in a penalty. The types of behaviour that can be put into a behaviour description depend on the capabilities of the asset: Which data it can obtain from its environment and what services it can provide based on the data available? Behaviour descriptions offer several kinds of flexibility: Firstly, they do not depend on a certain type of equipment; they only depend on the ability to communicate using a negotiation protocol that is used to negotiate a suitable behaviour description. Secondly, behaviour descriptions can activate any service that is supported by the asset. Thirdly, behaviour descriptions can be tailored to the capabilities of the specific asset. This includes the ability to respect the constraints which an asset has to comply to. Finally, many different measurements or events can be used to guide the behaviour of the asset. The flexibility of behaviour descriptions comes from the fact that the communication is not based on an exchange of measurements and set points, but rather on an exchange of capabilities, constraints and services. A controlled asset sends to the supervisory controller: 1) A set of services it can support, and 2) A set of constraints it has to adhere to. Then the supervisory controller can create a behaviour description and send it to the asset. The behaviour description will refer to the services the component can support, and give the necessary parameters to the service so it can be executed on the asset. An example of parameters for a service is: the droop value for a droop controller or an actual consumption schedule when the service is scheduled consumption. A. A behaviour description example In this example, the heaters of a household are controlled by a supervisory controller. The owner of the household has two goals: 1. to participate in frequency control 2. to reduce his cost of heating The incentive to participate in frequency control is due to the owner being paid to deliver this service, and the reduction of the cost of heating is to heat the house, at the lowest expense. The constraint under which the heaters have to operate is that room temperatures should have an acceptable level a setting that is controllable by the household residents. The services that heaters provide are frequency control and e.g. price-based, thermostat-controlled heating, the local events that they respond to are system frequency, measured at the point of common coupling and the power price which is broadcasted using various means. The two services can be offered at the same time because the frequency control algorithm has a deadband. When the measured frequency is within the limits of the deadband, the frequency controller does not have to operate; this gives the thermostat controller time to control the room temperature. To additionally reduce the time a heater is subjected to frequency control, the heater only responds to frequency control part of the time. The behaviour descriptions for this case are a list of rules for a rule-based system. Rules have the form "if condition - then action". The action activates a certain service, while the condition encodes the constraints the service has to operate under. In this case, the constraints encode the deadband and the time period for the frequency controller, whereas the pricebased controller is used whenever the frequency controller is inactive. VI. CONCLUSION This paper identifies the three main stakeholders with regards to their need for services and individual needs. Instead of simply forcing compliance and with focus on commercial solutions a scheme for co-ordinating the services are set up at increasingly aggregated levels. The scheme covers the communication and co-ordination of the needs for services. Each need is then considered under normal operations, and a priority list is created for the needs. This priority list is constructed in such a way that in case of a conflict between two services, a service should not be able to be activated, if it would create the need of a higher priority service in the process. With the set-up of the priority list, two examples are given on conflicting services and their activation. Lastly, the concept of behaviour descriptions is introduced as a means of implementing the control. Bibliography  Danish Energy Agency, "Energy Statistics 2009," Danish Energy Agency,2010.  K. Nørregaard, J. Østergaard, P.-F. Bach, M. Lind, P. Sørensen, B. Tennbakk, M. Togeby, and T. Ackermann, "Ecogrid.dk Phase 1, Summary Report - Steps towards a Danish Power System with 50 % Wind Energy," Energinet.dk,2008.  P. E. Sørensen, M. Togeby, T. Ackermann, D. K. Chandrashekhara, J. P. F. Horstmann, H. Jóhannsson, A. H. Nielsen, P. Nyeng, T. W. Rasmussen, and Z. Xu, "Ecogrid. dk Phase 1 WP4 report: New measures for integration of large scale renewable energy," Forskningscenter Risø,2008.
9 > REPLACE THIS LINE WITH YOUR PAPER IDENTIFICATION NUMBER (DOUBLE-CLICK HERE TO EDIT) < 9  Preben Nyeng, "System Integration of Distributed Energy Resources - ICT, Ancillary Services, and Markets." PhD The Technical University of Denmark,  H. Jóhannsson, "Development of Early Warning Methods for Electric Power Systems." PhD The Technical University of Denmark,  H. Jóhannsson, R. Garcia-Valle, J. T. G. Weckesser, A. H. Nielsen, and J. Østergaard, "Real-Time Stability Assessment based on Synchrophasors,"  F. C. Schweppe and et al, "Homeostatic Utiity Control," IEEE Trans. Power App. Syst., vol. PAS-99, no. 3, pp ,  P. Nyeng and J. Østergaard, "Information and Communications Systems for Control-by-Price of Distributed Energy Resources and Flexible Demand," Smart Grid, IEEE Transactions on, vol. 2, no. 2, pp , June2011.  P. B. Andersen, B. Poulsen, C. Træholt, and J. Østergaard, "Using Service Oriented Architecture in a Generic Virtual Power Plant,"  S. You, C. Træholt, and B. Poulsen, "Generic Virtual Power Plants: Management of Distributed Energy Resources under Liberalized Electricity Market,"  S. You, C. Træholt, and B. Poulsen, "A Market-Based Virtual Power Plant," 2009, pp  P. S. Kundur, Power System Stability and Control McGraw-Hill, Inc.,  "Dynamics of Interconnected Power Systems: A tutorial for System Dispatchers and Plant Operators," EPRI Report EL L,May1989.  L. H. Fink and K. Carlsen, "Operating under Stress and Strain," IEEE Spectrum, pp , Mar  The European Parliament and the Council of the European Union, "Directive 2003/54/EC of the European Parliament and of the Council of the European Union,", 2003/54/EC ed  C. L. DeMarco, "Strategic behavior in electric generation markets via dynamic governor control design," Decision support systems, vol. 24, no. 3-4, pp ,  Energinet.DK, "Teknisk Forskrift: TF Dansk Mvarordning,"2010.  D. Kullmann, O. Gehrke, and H. W. Bindner, "Implementation and Test of Demand Response using Behavior Descriptions," 2011.