Another opportunity for the research community to gain feedback on their research projects and gauge the reception of their ideas and opinions among their expert peers came with the Erasmus Energy Forum Science Day. This fourth annual conference on Energy Informatics and Management (EIM 2016) could ultimately make a difference to the ideas and concepts adopted by the international energy industry. The event was held at Rotterdam School of Management, Erasmus University (RSM).
The expert international audience included not only academics but also practitioners from companies. Science day was moderated by Dr Yashar Ghiassi-Farrokhfal of the host research centre, the Erasmus Centre of Future Energy Business (ECFEB), based at RSM.
Interaction between presenters and the audience is one of the hallmarks of the Erasmus Energy Forum – both on Business Day and on Science Day. “Once again, interesting discussions came out of the sessions,” said Dr Ghiassi-Farrokhfal. In-depth scientific work explored the use of energy informatics and management to encourage energy efficiency.
Keynote speaker was Professor Srinivasan Keshav, co-director of the Information Systems and Science for Energy Laboratory at the University of Waterloo. His presentation, Solar + Storage + IoT + LED = $30 Trillion described solar energy and its storage as ‘unstoppable’. “Governments have not found a way to tax sunlight!” he said.
Science Day concluded with a panel discussion Bridging Academic Impact with Practical Relevance: A Crucial Task for Energy Research? with panel members Erik van Heck from RSM; Michael Kaisers from the Centrum Wiskunde & Informatica in Amsterdam; Tom Lawrence from the University of Georgia; and Professor Srinivasan Keshav from the University of Waterloo. The discussion had the participation of the audience asking questions about reconciling industry and research in reaching mutual goals of increasing energy efficiency and flexibility.
- Business Award
- Science Award
Rens Philipsen, German Morales-Espana, Mathijs de Weerdt and Laurens de Vries for their paper and presentation about Imperfect Unit Commitment Decisions with Perfect Information: a Real-time Comparison of Energy versus Power.
A highlight of Science Day was the presentation of the Erasmus Energy Science Award for the paper, which in the opinion of the judges, showed the most scientific rigour, potential for innovation and applicability to the real world. Winner of the award, was Imperfect Unit Commitment Decisions with Perfect Information: a Real-time Comparison of Energy versus Power by Rens Philipsen, Mathijs De Weerdt, Laurens de Vries and German Morales-Espana from Delft University of Technology. Their paper gained the highest evaluations from the reviewers on theoretical and practical implications to the energy sector. The abstract from this paper is listed below.
Research papers were presented in four sessions, structured according to topic. A total of 11 papers were presented and were followed by questions from the audience. Each one is described from its abstract below:
- Risk and Decision Making for Electricity Forward Markets with Volatile Resources
Electricity markets experience increasing uncertainty under the influence of larger shares of volatile resources. The authors propose a multi-stage competitive equilibrium model for future electricity wholesale markets with a high share of renewable energy resources. Large-scale power producers sell electricity on the wholesale markets; and retailers buy on the wholesale markets and sell electricity to end consumers. The authors develop a heterogeneous agent-based approach, considering the fact that one cannot apply the same model to all markets and characteristics of specific markets play a crucial role influencing the risk premium. Results indicate evidence for downward biased forward prices and a component in the forward premium for an indirect storage convenience yield and call for market design rethinking to accommodate the integration of large shares of renewable energy.
- Market-Based Multi-Agent Coordination to Manage Energy Balance in Smart Grids
Widespread adoption of sustainable energy sources is driving electricity grid operators to supplement hierarchical control regimes with market-based control that better motivates stakeholder involvement. However, to prevent market failures, such controls require testing before real-world implementation. The Power Trading Agent Competition is a competitive simulation of distribution grids that mirrors real-world scenarios and tests alternative policy and business scenarios. In Power TAC, broker agents acquire energy through bidding in a forward wholesale market to satisfy their customers’ overall demand on an hourly basis. In addition, a balancing market is intended to resolve real-time energy imbalances caused by broker prediction errors using demand response resources. As part of the annual alignment process, the authors discovered that brokers in the 2015 competition were persistently buying insufficient energy on the wholesale market to satisfy their customer demand. Instead, the balancing market made up the deficit, charging brokers a premium over the wholesale price. Also, demand response resources were heavily underused.
The authors studied the economic impact of this systematic imbalance on brokers and discovered that they were behaving rationally, given the prices they faced in the two markets. They presented the process and results of this analysis, and showed how the balancing market’s pricing mechanism can be adjusted for the 2016 competition to make it rational for brokers to achieve an overall neutral imbalance.
- A Hybrid Method to Forecast Electricity Prices in an Hour-ahead Wholesale Market
As a result of energy transition policies, fossil-based power production has been shifting towards highly volatile renewable energy production, thanks to rapid improvements on renewable technologies. Within that vision, a number of nuclear power plants have already been shut down in Germany. However, the transition will bring a number of challenges as well. One of the challenges that also motivates this paper is price fluctuations which has been challenging all of the actors, involved in the smart grids. This paper proposes a responsive hybrid model for price forecasting, using dynamic programming techniques and Markov Decision Process. The authors use a belief function to adjust predicted values, derived from exponentially smoothed values of Market Clearing Prices. Proposed models were simulated in Power Trading Agent Competition and compared with another forecasting model. The results show that the method outperforms its opponent in terms of prediction accuracy.
- The Impact of Sustainability on Consumers’ Technology Approval – Taking Smart Energy-Saving Systems as an Example of Application
Valerie Graf and Henner Gimpel
The objective of this work is to show, that sustainability – encompassing environmental and social dimensions – impacts consumers’ behavioural intention to use technology. To analyse this relationship, the authors extend the unified theory of acceptance and use of technology of Venkatesh et al. (2012) by sustainability as an additional predictor and assign it to the application context of smart energy-saving systems. They validate the inclusion of a sustainable component in a quantitative study. Their proposed model extension incorporates the relationships of the baseline model. The theoretical and managerial implications of the study are discussed and future work is outlined.
- Imperfect Unit Commitment Decisions with Perfect Information: a Real-time Comparison of Energy versus Power
In order to cope with fluctuations and uncertainty, power systems rely on contracted reserves. The day-ahead Unit Commitment (UC) is the short-term planning process which is commonly used to schedule these resources at minimum cost, while operating the system and units within secure technical limits. This paper shows through the evaluation of deterministic cases that conventional energy-based UC formulations lead to inefficient use of reserves in real-time operation to deal with completely known deterministic events. These inefficient decisions are inherent to the assumptions underlying the energy-based formulation, and the misaligned incentives between markets and real-time operation. Economic efficiency and system security can be improved by adopting a UC formulation which explicitly considers the instantaneous power trajectories of generators.
This paper won the Erasmus Energy Science Award 2016.
- Metaheuristic Approach for Online Optimal Reactive Power Management in Near-Shore Wind Power Plants by Jose L. Rueda, Istvan Erlich and Peter Palensky
Mean-variance mapping optimization (MVMO) is an emerging metaheuristic optimization algorithm, whose evolutionary mechanism performs within a normalized search space. The most remarkable aspect of this mechanism resides in the application of a special mapping function to generate new values of the optimization variables based on their statistical significance throughout the search process. This paper concerns the feasibility of the MVMO to tackle the problem of online optimal reactive power management in near-shore wind power plants. The main challenges reside in the restricted computing budget and mix-integer nature of the problem. To this aim, MVMO is configured to evolve a single solution throughout the search process, and a new mapping function is proposed to improve the global search capability. Numerical tests on a benchmark system proposed by the IEEE Working Group on Modern Heuristic Optimization demonstrate the effectiveness of MVMO.
- Towards a descriptive Framework of Demand Side Flexibility
The objective of this paper is to contribute to the understanding of Demand Side Management (DSM) in future energy markets. DSM is considered one of the most promising concepts to meet the challenge of balancing demand and supply in the future power grid. In contrast to traditional electricity systems where manageable supply is used to ensure grid stability, DSM strives to influence demand in accordance with current electricity production. From a market perspective, this generates a new, turned-around supply chain: While electricity is traded from generation to consumption, demand side flexibility (DSF) is offered by consumers to grid operators. The authors address the requirement of a theoretical foundation for DSF by conducting a first step towards a descriptive theory to classify different dimensions of DSF. Further, they suggest an overarching, interdisciplinary unification for further research in the topic and enable various research disciplines to join forces on this challenge. The targeted descriptive theory shall be fundamental research on DSM (e.g. electric vehicle charging, flexible industry processes).
- Decoupling Temporal Resource Constraints in Multi-agent Sequential Decision Making for Demand Response
In order to incorporate more uncontrollable sources of energy like wind and solar, demand is required to become more flexible. Unconventional energy storage such as thermal mass present in houses can provide such flexibility. The authors consider demand response of Thermostatically Controlled Loads as a planning problem where multiple independent agents need to co-ordinate their consumption, subject to hard temporal resource constraints. They present resource assignment strategies that decouple agents so that they can plan individually. They compare four approaches to decouple the agents: an optimal pre-allocation strategy, a greedy pre-allocation strategy, a best-response decoupling and a marginal utility-cost decoupling. The optimal and best-response approaches are the state-of-the-art from literature. The greedy algorithm and the utility-cost decoupling both use the marginal utility of resource consumption to assign resources. The authors show that a fictitious-play inspired history of plans allows agents to converge to an expected cost curve which lets them compute effective plans individually. Compared to the optimal pre-allocation, their approaches are shown to be significantly more scalable, with the utility-cost decoupling providing the best trade-off between performance and quality.
- Information Systems Needs for Adaptation of Existing Campus Facilities for Automated Demand Response
The University of Georgia (UGA), which pays real-time price tariff for electricity, has tested concepts for a demand response system (DRS) to handle hot summer days. The DRS takes advantage of the network of pipes for a district cooling system to store ‘coldness’ and modifications to the HVAC control set points. Initial tests show a saving of a minimum of 11%. The study is also identifying barriers to retrofitting existing campus facilities with the capability to participate in a DRS.
- Electric Vehicle Fleets as Virtual Power Plants
This study designs and evaluates a decision support system that turns a car-sharing fleet of electric vehicles (EV) into a Virtual Power Plant (VPP). VPPs are a means to balance volatile renewable energy sources. These energy sources pose a challenge to the electrical grid’s stability and increase the chances of blackouts. The system makes real-time decisions based on expected profits for car sharing operators. It decides whether a particular EV at a certain location should be made available for rental, used to provide energy to the grid, or charged when the grid has excess capacity. The novelty of this approach lies in the car sharing context which explicitly assigns monetary values to the cost of immobility. A dataset of 497 EV provided by Daimler is utilized to assess the profitability of the system. Under current circumstances providing energy from the car to the grid is unprofitable, whereas charging the EVs when excess capacity is available is profitable.
- Modelling Electric Vehicle Owners’ Willingness to Pay for a Charging Service
Although today Tesla S owners can enjoy the benefit of charging their electric vehicles (EVs) for free and fast at the supercharger stations, owners of other EVs have to charge their cars at normal Level-1 and Level-2 chargers. It’s not only that it often takes several hours to refill their vehicles, but they have to pay for the electricity as well. This paper uses Bayesian network approach for modelling EV owners’ willingness to pay for a charging service. The authors characterize EV owners’ charging behaviour through the following variables:
i) EV battery capacity;
ii) EV battery status;
iii) battery capacity of an average EV owner;
iv) charging speed; and
v) electricity price.
Furthermore, the proposed model is instantiated with the real-world data describing the EV market in Australia. Finally, limitations of the proposed model as well as possible improvements have been commented.