Performance Assessment of Black Box Capacity Forecasting for Multi-Market Trade Application

Pamela MacDougall, Bob Ran, George B. Huitema and Geert Deconinck
Additional contact information
Pamela MacDougall: Monitoring and Control Services, TNO, 9701 BK Groningen, The Netherlands
Bob Ran: Monitoring and Control Services, TNO, 9701 BK Groningen, The Netherlands
George B. Huitema: Monitoring and Control Services, TNO, 9701 BK Groningen, The Netherlands
Geert Deconinck: Department of Electrical Engineering, University of Leuven, 3001 Leuven, Belgium

Energies, 2017, vol. 10, issue 10, 1-19

Abstract: With the growth of renewable generated electricity in the energy mix, large energy storage and flexible demand, particularly aggregated demand response is becoming a front runner as a new participant in the wholesale energy markets. One of the biggest barriers for the integration of aggregator services into market participation is knowledge of the current and future flexible capacity. To calculate the available flexibility, the current aggregator pilot and simulation implementations use lower level measurements and device specifications. This type of implementation is not scalable due to computational constraints, as well as it could conflict with end user privacy rights. Black box machine learning approaches have been proven to accurately estimate the available capacity of a cluster of heating devices using only aggregated data. This study will investigate the accuracy of this approach when applied to a heterogeneous virtual power plant (VPP). Firstly, a sensitivity analysis of the machine learning model is performed when varying the underlying device makeup of the VPP. Further, the forecasted flexible capacity of a heterogeneous residential VPP was applied to a trade strategy, which maintains a day ahead schedule, as well as offers flexibility to the imbalance market. This performance is then compared when using the same strategy with no capacity forecasting, as well as perfect knowledge. It was shown that at most, the highest average error, regardless of the VPP makeup, was still less than 9%. Further, when applying the forecasted capacity to a trading strategy, 89% of the optimal performance can be met. This resulted in a reduction of monthly costs by approximately 20%.

Keywords: machine learning; predictive trade; virtual power plants; flexibility; demand response; electricity markets (search for similar items in EconPapers)
JEL-codes: Q Q0 Q4 Q40 Q41 Q42 Q43 Q47 Q48 Q49 (search for similar items in EconPapers)
Date: 2017
References: View complete reference list from CitEc
Citations: View citations in EconPapers (2)

Downloads: (external link)
https://www.mdpi.com/1996-1073/10/10/1673/pdf (application/pdf)
https://www.mdpi.com/1996-1073/10/10/1673/ (text/html)

Related works:
This item may be available elsewhere in EconPapers: Search for items with the same title.

Export reference: BibTeX RIS (EndNote, ProCite, RefMan) HTML/Text

Persistent link: https://EconPapers.repec.org/RePEc:gam:jeners:v:10:y:2017:i:10:p:1673-:d:116021

Access Statistics for this article

Energies is currently edited by Ms. Agatha Cao

More articles in Energies from MDPI
Bibliographic data for series maintained by MDPI Indexing Manager ().