Load and Flexibility Models for Distribution Grid Management

Research output: Book/ReportPh.D. thesis

Abstract

Recent trends in power systems have favored the electrification of residential heating and transportation by means of modern partially flexible loads. Furthermore, environmental concerns have promoted the notion of distributed generation, thus increasing its penetration in low voltage distribution grids. These technological advancements have changed the nature of conventional demand and supply; nonetheless, the grid has remained the same on the grounds that network modifications are considerably more complex and expensive than individual load/generation changes.

It is evident, that by radically changing the energy flow in low voltage distribution grids, several grid complications are bound to emerge like, for instance, capacity adequacy limitations, power quality issues, reverse power flows etc. One of the Distribution System Operator options to manage these complications is to proceed to advanced metering and control infrastructure investments. These investments will transform distribution networks into smarter" grids which will facilitate flexible load/generation control in return for financial and reliability benefits to electricity consumers.

The theoretical idea of modernising the grid so as to cope with load and generation changes is indeed innovative; nevertheless, its implementation is, up to day, impaired by many practical difficulties. In this Ph.D. study, four of these issues are detected, elaborated and consequently solved by dedicated grid and/or load models. These issues involve lack of generic load forecasts, need for flexible load/generation estimation techniques from smart meter measurements, deficiencies in online load distribution observability, and, finally, energy market compatible control algorithms which treat consumer flexibility in a fair manner. These modules will aid the Distribution System Operator in understanding the grid's load distribution, forecast it, identify any upcoming problems, estimate the available flexibility to alleviate them, and, eventually, schedule this flexibility accordingly.

Regarding the first topic, that is to say load forecasting, a simple, generic, and automated load forecasting technique based on machine learning principles is proposed. Unlike other benchmark forecasting models, this non-parametric approach does not introduce any assumptions regarding the system model and is solely based on past observations. Thus, this method is to a great extent flexible, albeit strongly history dependent.

As far as the flexibility estimation is concerned, a new philosophy is introduced. Contrary to contemporary practices, this philosophy is compatible with the low data recording and transmitting rates which smart meters are operated with. Although demonstrated for the Heat Pump case, the methodology is applicable to any flexible device, as long as it has an impactive power pattern.

One other issue which modern distribution grids frequently encounter, is the lack of thorough online monitoring. Though smart meters might be installed, relevant smart meter data are usually collected with substantial delay from the moment of their recording. To compensate for this situation, a power meter allocation methodology is presented. This methodology, which is applicable to low voltage distribution grids, aims at allocating a few extra power meters as a trade-off for a reasonable online load distribution approximation. Since it results in grouping grid areas hierarchically, the applicability of the methodology in formulating aggregated flexibility bids by Aggregators is demonstrated.

As for the scheduling, a two-step control mechanism, which aims at alleviating grid congestions, is proposed. In the first step, a central controller manages flexibility proactively, whereas in the second step a decentralised control scheme deals with flexibility reactively. Both controllers are designed in such a way, that compatibility with contemporary markets is assured, while special focus is given to the fair activation of consumer flexibility. Finally, possible interactions between the entities handling the technical and the economic aspects of flexibility in a market environment are discussed.
Close

Details

Recent trends in power systems have favored the electrification of residential heating and transportation by means of modern partially flexible loads. Furthermore, environmental concerns have promoted the notion of distributed generation, thus increasing its penetration in low voltage distribution grids. These technological advancements have changed the nature of conventional demand and supply; nonetheless, the grid has remained the same on the grounds that network modifications are considerably more complex and expensive than individual load/generation changes.

It is evident, that by radically changing the energy flow in low voltage distribution grids, several grid complications are bound to emerge like, for instance, capacity adequacy limitations, power quality issues, reverse power flows etc. One of the Distribution System Operator options to manage these complications is to proceed to advanced metering and control infrastructure investments. These investments will transform distribution networks into smarter" grids which will facilitate flexible load/generation control in return for financial and reliability benefits to electricity consumers.

The theoretical idea of modernising the grid so as to cope with load and generation changes is indeed innovative; nevertheless, its implementation is, up to day, impaired by many practical difficulties. In this Ph.D. study, four of these issues are detected, elaborated and consequently solved by dedicated grid and/or load models. These issues involve lack of generic load forecasts, need for flexible load/generation estimation techniques from smart meter measurements, deficiencies in online load distribution observability, and, finally, energy market compatible control algorithms which treat consumer flexibility in a fair manner. These modules will aid the Distribution System Operator in understanding the grid's load distribution, forecast it, identify any upcoming problems, estimate the available flexibility to alleviate them, and, eventually, schedule this flexibility accordingly.

Regarding the first topic, that is to say load forecasting, a simple, generic, and automated load forecasting technique based on machine learning principles is proposed. Unlike other benchmark forecasting models, this non-parametric approach does not introduce any assumptions regarding the system model and is solely based on past observations. Thus, this method is to a great extent flexible, albeit strongly history dependent.

As far as the flexibility estimation is concerned, a new philosophy is introduced. Contrary to contemporary practices, this philosophy is compatible with the low data recording and transmitting rates which smart meters are operated with. Although demonstrated for the Heat Pump case, the methodology is applicable to any flexible device, as long as it has an impactive power pattern.

One other issue which modern distribution grids frequently encounter, is the lack of thorough online monitoring. Though smart meters might be installed, relevant smart meter data are usually collected with substantial delay from the moment of their recording. To compensate for this situation, a power meter allocation methodology is presented. This methodology, which is applicable to low voltage distribution grids, aims at allocating a few extra power meters as a trade-off for a reasonable online load distribution approximation. Since it results in grouping grid areas hierarchically, the applicability of the methodology in formulating aggregated flexibility bids by Aggregators is demonstrated.

As for the scheduling, a two-step control mechanism, which aims at alleviating grid congestions, is proposed. In the first step, a central controller manages flexibility proactively, whereas in the second step a decentralised control scheme deals with flexibility reactively. Both controllers are designed in such a way, that compatibility with contemporary markets is assured, while special focus is given to the fair activation of consumer flexibility. Finally, possible interactions between the entities handling the technical and the economic aspects of flexibility in a market environment are discussed.
Original languageEnglish
PublisherDepartment of Energy Technology, Aalborg University
Number of pages108
StatePublished - Dec 2015
Publication categoryResearch

Download statistics

No data available
ID: 224036130