Example Run for PosoMAS — Case Study: Autonomous Power Management
We envision a power grid in which small, distributed power plants can be controlled or are able to participate in a scheduling scheme. The aim of such a system is to stabilise the power grid by incorporating distributed energy production, create financial incentives for individual power plant owners, and allow the creation of new and innovative products on the power market.
An autonomous power management system as envisioned in the case study has to fulfill the following requirements:
- System scale: While in the past, system operators only had to deal with a manageable number of power plants which they had a detailed model of and could control directly and without limitations1, the integration of distributed energy resources such as small biogas plants, combined heat and power units, or emergency generating unit will change how schedules are created and how primary, secondary, and tertiary power reserves are provided. Scheduling will have to take many more generators into account, often without knowing their exact control models but only based on threshold values about maximum and minimum output, as well as the generators rate of change.
With the increased number of generators also comes an increased availability of data. In principle, every system connected to the power grid can gather data about the current, local status and collect information such as network frequency and voltage levels. This information can be very valuable, e.g., when trying to avoid local voltage spikes that occur when a lot of power is input into the low-voltage band by photovoltaic installations and no consumption occurs at the same segment of the grid. The sheer volume of this data, however, can hardly be centrally processed and transformed into control actions in a short time.
- Uncertainty about intermittent power production: Scale also affects the input of intermittent power generators such as solar and wind power plants into the system. The growing number of installations and their increased production can no longer be readily offset by power plants that can react quickly to drops or increases in the input. This introduces a level of uncertainty into the system that has, so far, been largely ignored.
In future power management systems, the accuracy of the predictions will have to be factored in, however, since it determines the power that has to be scheduled (the so called residual load is the gap between the power consumption and the production by intermittent resources). If variations of the intermittent input are to be dealt with locally and in a distributed fashion by the power plants, reserves have to be scheduled accordingly as well.
- Self-interested system participants: Each power generator operator is interested in making economical use of its generation capacity. For instance, small cooperatives of farmers that operate a biomass generator have an interest of running the generator at capacity whenever biomass is available. That these times may not coincide with the times in which the input is required is, at the moment, of little concern to them. To achieve stability in a future power management system it would be much better to run the generator when its input is helpful and required. In order to do this, the cooperative of farmers has to divulge the state of its generator as well as its control model. Appropriate financial or legislative incentives may be used to induce such a behaviour, but concerns of privacy and self-interest must be taken into consideration.
- Observation and control: In a system of hundreds of thousands of system components, the flow of information and control has to be clearly defined. Responsibilities and authority has to be clearly regulated. As effective control of the system requires good knowledge of its status, the observation of relevant data about the system on all levels of analysis has to be ensured. The data has to be aggregated and collected at the appropriate locations in the system and used there to make control decisions that influence parts of the system. If these control decisions are made autonomously, the appropriate power to make these decisions has to be delegated by a legal entity such as a utility.
- Longevity of the system: The next generation of power management systems will be gradually deployed and will be in use for the next decades. That means that all solutions that are designed now must be robust with regard to their future evolution and additions to the system, both on the hardware and on the software side.
- Integration of controllable and stochastic consumers: So far, the discussion has mainly focused on power generators. However, power consumers can be a valuable asset in the creation of schedules and the stabilisation of the grid. The developments outlined for power generators also apply to power consumers: they are more often networked, can be potentially controlled from the outside, can provide data about the status of the grid, and can serve as storages (such as refrigerated warehouses or electric vehicles).
We make a number of assumptions about the environment the system will be located in, about available technologies, and the regulatory framework.
- Availability of networked measuring equipment: Sensors that are able to communicate the measured data are widely available, allow to observe the grid status and make decisions based on current conditions.
- Known common interface for power generators: We assume that standardisation efforts have led to a common, widely available interface for power infrastructure, especially distributed generators, that we can use without further technical issues.
- Freedom from regulatory and legal issues: In this case study, we assume that it is legally possible to incorporate distributed energy producers into the scheduling and that regulatory statues do not hinder the deployment of the system.
Several different approaches to re-structure power management systems have already been proposed. A key concept is the Virtual Power Plant (VPP), usually defined as groups of power plants that are in most cases controlled by a central entity (see, e.g., Lombardi et al., 2009). Sometimes, membership in a VPP is restricted to certain types of power plants or power plants with predefined properties, such as dispersed generation units and micro-CHPs (combined heat and power units) in (Schulz et al., 2005) or DERs in (Bel et al., 2007). These approaches mainly focus on providing structures to integrate DERs into existing control schemes. Others concentrate on facilitating trading in power markets and distinguish commercial VPPs that participate in power markets and technical VPPs that provide services for the transmission net (Pudjianto et al., 2007). None of these approaches combine all the features that we are looking for in the power management system of the future. However, they give important insights into the organisation and functionality of such a system.
The techniques we propose should not be seen in isolation, but have to be complemented by other technologies and a shift in the legal and regulatory framework (see, e.g., the results of the E-Energy project as well as of several European and international initiatives such as the FP6 project INTEGRAL, the European Electricity Grid Initiative and its associated projects such as DISCERN, or the UK's Autonomic Power Systems project).
A solution that fits many of the requirements and assumptions outlined above has been described in Steghöfer et al., (2013a). A hierarchy of Autonomous Virtual Power Plants (AVPPs) is formed in a self-organisation process (Steghöfer et al., 2013a) to distribute the computational load of power plant scheduling, enabling the integration of distributed resources. Uncertainties in the system are captured by using trust-based scenarios (Anders et al., 2013). While the techniques developed inform the application of PosoMAS for the case study, the simulated development effort is in principle decoupled from these results and aims for the development of a suitable architecture independent of the need to support existing approaches.
Bibliography
- Lombardi, P., M. Powalko, and K. Rudion. ‘Optimal operation of a virtual power plant’. In Power Energy Society General Meeting (PES ’09), pp. 1–6. IEEE (2009).
- Schulz, C., G. Roder, and M. Kurrat. ‘Virtual power plants with combined heat and power micro-units’. In 2005 International Conference on Future Power Systems (November 2005).
- Bel, I., A. Valenti, J. Maire, J. M. Corera, and P. Lang. ‘Innovative Operation with Aggregated Distributed Generation’. In Proc. of the 19th International Conference on Electricity Distribution (2007).
- Pudjianto, D., C. Ramsay, and G. Strbac. ‘Virtual power plant and system integration of distributed energy resources’. In Renewable Power Generation, IET, volume 1(1), pp. 10–16 (March 2007).
- Steghöfer, J.-P., G. Anders, F. Siefert, and W. Reif. ‘A System of Systems Approach to the Evolutionary Transformation of Power Management Systems’. In Proceedings of INFORMATIK 2013 – Workshop on "Smart Grids", Lecture Notes in Informatics. Bonner Köllen Verlag (2013a).
- Steghöfer, J.-P., P. Behrmann, G. Anders, F. Siefert, and W. Reif. ‘HiSPADA: Self-Organising Hierarchies for Large-Scale Multi-Agent Systems’. In ICAS 2013, The Ninth International Conference on Autonomic and Autonomous Systems, pp. 71–76. IARIA, Lisbon, Portugal (March 2013b).
- Anders, G., J.-P. Steghöfer, F. Siefert, and W. Reif. ‘A Trust- and Cooperation-Based Solution of a Dynamic Resource Allocation Problem’. In Seventh IEEE International Conference on Self-Adaptive and Self-Organizing Systems (SASO). IEEE Computer Society, Washington, D.C., Philadelphia, PA (September 2013).
Footnotes
Copyright
This material is made available under the Creative Commons—Attribution-ShareAlike License v3.0.
© Copyright 2013, 2014 by Institute for Software & Systems Engineering, University of Augsburg
Contributors
Contact: Jan-Philipp Steghöfer