Demand response flexibility and flexibility potential of residential smart appliances: Experiences from large pilot test in Belgium

Demand side management (DSM) can be defined as modifications in the demand side energy consumption pattern to foster better efficiency and operations in electrical energy systems. DSM activities, which are classified into “energy efficiency (EE)” and “demand response (DR)” are becoming more popular due to technological advances in smart grids and electricity market deregulation. However, it can be argued that ensuring DSM sustainability requires creating suitable business models. Business models are influenced by different factors such as electricity market regulation, mechanisms, power system characteristics and infrastructure. The proliferation of smart grid infrastructure, distributed generation, intermittent renewable energy resources and energy storage devices has affected DSM business models considerably. Therefore, in this paper, possible business models for EE and DR providers in different electricity market segments are analyzed and reviewed. The analysis covers three types of characteristics: DSM transaction characteristics, renewable energy correlation and DSM load control characteristics. In DSM transaction characteristics, the value proposition of DSM such as added value offered to the DSM purchaser and transaction triggers are discussed. In renewable energy correlation, the effect of increased renewable energy penetration on the business model is evaluated. In DSM load control characteristics, load control and aggregation aspects such as response speed, duration, advance notice, location sensitivity and actual usage frequency are analyzed.

A review of demand side management business models in the electricity market

Driven by recent advances in batch Reinforcement Learning (RL), this paper contributes to the application of batch RL to demand response. In contrast to conventional model-based approaches, batch RL techniques do not require a system identification step, making them more suitable for a large-scale implementation. This paper extends fitted Q-iteration, a standard batch RL technique, to the situation when a forecast of the exogenous data is provided. In general, batch RL techniques do not rely on expert knowledge about the system dynamics or the solution. However, if some expert knowledge is provided, it can be incorporated by using the proposed policy adjustment method. Finally, we tackle the challenge of finding an open-loop schedule required to participate in the day-ahead market. We propose a model-free Monte Carlo method that uses a metric based on the state-action value function or Q-function and we illustrate this method by finding the day-ahead schedule of a heat-pump thermostat. Our experiments show that batch RL techniques provide a valuable alternative to model-based controllers and that they can be used to construct both closed-loop and open-loop policies.

/pdf/residential-demand-response-of-thermostatically-controlled-264zy4550n.pdf

Residential Demand Response of Thermostatically Controlled Loads Using Batch Reinforcement Learning

Residential-scale low carbon technologies (LCTs) can help decarbonizing our economies but can also lead to technical issues, particularly in low voltage (LV) distribution systems. To quantify these problems this work proposes a probabilistic impact assessment methodology. First, realistic 5-min time-series daily profiles are produced for photovoltaic panels, electric heat pumps, electric vehicles, and micro combined heat and power units. Then, to cater for the uncertainties of LCTs (e.g., size, location, and behavior), a Monte Carlo analysis is carried out considering 100 simulations for different penetration levels (percentage of houses with a LCT). This methodology is applied to 128 real U.K. LV feeders showing that about half of them can have voltage and/or congestion issues at some penetration of LCTs. Furthermore, to identify the relationships between the first occurrence of problems and key feeder parameters (e.g., length, number of customers), a correlation analysis is developed per LCT. Crucially, these results can be translated into lookup tables to help distribution network operators in producing preliminary estimates of the LCT hosting capacity of a given feeder.

/pdf/probabilistic-impact-assessment-of-low-carbon-technologies-3p4ouohpzd.pdf

Probabilistic Impact Assessment of Low Carbon Technologies in LV Distribution Systems

The increasing amount of photovoltaic (PV) generation results in a reverse power flow and a violation of the overvoltage limits in distribution networks. PV inverters can curtail active power or consume reactive power to avoid these excessive high voltages. Local controllers of active and reactive power that are based on measurements of the produced PV power have a fast response to the changing production levels of the PV installation. The performance of these local controllers depends on the tuning of the control parameters, which are grid and time dependent. In this paper, local control functions are defined as piecewise linear functions. The parameters of all the local control functions are regularly reoptimized. This results in an optimal use of reactive power and a minimum amount of curtailed active power, while respecting network limitations. The optimization of these parameters is formulated as a convex optimization problem, which can be solved sufficiently fast. The performance of the control is evaluated on an existing three-phase four-wire distribution grid and is compared with different local control methods.

Combined Central and Local Active and Reactive Power Control of PV Inverters

The LINEAR project (Local Intelligent Networks and Energy Active Regions) focuses on the introduction and implementation of innovative smart-grid technologies in the Flanders region, and aims at a breakthrough in the further development and deployment of these solutions. It consists of a research component and a large-scale residential pilot, both focusing on active demand-side management of domestic loads. This paper describes the unique approach, the main objectives and the current status of this project. Selected business cases and target applications are discussed in detail.

/pdf/linear-breakthrough-project-large-scale-implementation-of-2168byj81h.pdf

LINEAR breakthrough project: Large-scale implementation of smart grid technologies in distribution grids

In this paper, the impact of residential distributed energy resources (DER) on the power quality is investigated in four feeder types of the electrical LV distribution network in Flanders, Belgium. The investigated power quality issues are over-voltage, under-voltage and unbalance. The results of the simulations are discussed in detail. The paper leads to an estimation of the compliance to the power quality standard EN 50160, and a summary of issues in the distribution grids when increasing the amount of DER.

LV distribution network feeders in Belgium and power quality issues due to increasing PV penetration levels

Demand Response is seen as important to support the integration of renewable energies into the grid. In Flanders, a residential Demand Response setup is realized in the Linear pilot. The aim is to assess the potential benefits and ways of technical realization of residential Demand Response. In this paper, household devices, like washing machines, are used to offer a flexible load. These are also devices which are used in the pilot. The effect of using flexible loads on the lifetime of a low-voltage transformer is assessed. An IEEE transformer model is used to calculate the lifetime. To calculate the effect of Demand Response, aging is first calculated based on the load of a group of customers and then based on their load being optimized by Demand Response. In this paper, devices are scheduled based on the transformer temperature. The temperature is optimized by using a simulation model based on a mixed integer quadratic programming (MIQP) scheduler. To assess the effect of Demand Response on the transformer lifetime, aging for the improved load curve is compared with aging for the initial load curve. To demonstrate the impact, realistic data for household load curves and the usage of household devices are employed. Results for this input data show reductions in aging of up to 75 % for transformers operating at rated load. The setup will be used to calculate a benchmark for the setup in the Linear pilot, which will use an on-line scheduler. It will be also used to determine potential outcome of a business case.

Effect of Demand Response on transformer lifetime expectation

This paper proposes an algorithm to detect the stealing of electricity by illegal connections in smart grids with unknown or uncertain cable lengths. Many new applications in the rising smart grid context will require information of the grid topology. We show that with measurement data of smart meters, the grid can be identified, as well as the phase of connection. One of the applications requiring grid information is the detection of electricity theft by double feeding. Electricity theft is a problem faced by all power utilities. Financial impacts are a reduced income for the system operator and the necessity to charge more to other customers. Stealing of electricity by double feeding can not be detected by the smart meter or by analysing the load profiles. Therefore it is suggested in this paper to use measurements of smart meters to identify the grid parameters and detect irregularities of specific customers.

/pdf/parameter-identification-of-unknown-radial-grids-for-theft-3ryznz58bv.pdf

T. De Rybel

Papers

LINEAR breakthrough project: Large-scale implementation of smart grid technologies in distribution grids

LV distribution network feeders in Belgium and power quality issues due to increasing PV penetration levels

Effect of Demand Response on transformer lifetime expectation

Parameter identification of unknown radial grids for theft detection

Technical assessment of centralized and localized voltage control strategies in low voltage networks