A Multi-perspective Model for Evaluation of Residential Thermal Demand ResponseMUHAMMAD BASHAR ANWAR

(UNIVERSITY COLLEGE DUBLIN)

04/04/2019

Presentation Overview

04/04/2019

Background

Problem Definition

The Multi-Perspective Model

Results

Key Takeaways

Demand Response (DR) and Residential Thermal LoadsDemand Response (DR) can enhance the flexibility of power systems

Residential Thermal demand – a major driver of Europe’s energy consumption

European Climate Foundation has encouraged the electrification of heat supply

Increasing integration of the electricity and heating sectors

Residential thermal loads are ideal candidates for DR by virtue of their inherent flexibility andsignificant penetration levels

Thermal-Electric Storage (TES) - a promising electricity-to-heat technology with the potential ofenabling DR as shown in several studies in the literature.

04/04/2019

Are centralized optimization results valid/viable in the presence of self-interested decision makers? Residential demand-side resources are distributed in nature – need for aggregation

Aggregators/Retailers - act as intermediary agents between the residential consumers and the SystemOperator (SO)

Residential DR involves a large number of self-interested decision makers and stakeholders e.g the SO,aggregator/retailer, consumers etc.

Centralized models (e.g. Unit Commitment/Economic Dispatch) assume a perfectly competitive marketand, thus, do not take into account the objectives of these stakeholders.

04/04/2019

Individual consumers would aim to minimize their costs

The retailer would aim to maximize its profits

The TSO would aim to maximize total social welfare

Need for a Multi-perspective Model The existing approaches which consider the strategic behavior of DR participants can be classified into:

1. Wholesale Market Models:◦ Typically aim to determine the optimal bidding strategy of a retailer while considering the operation of the System

Operator (SO).◦ Can provide valuable insights on the impacts of a strategic retailer’s on power system operation◦ However, these models do not account for the impacts of consumers’ strategic participation in DR.

2. Retail Market Models:◦ Typically focus on the retailer’s optimal pricing problem while considering the objectives of the consumers.◦ Can provide valuable insights on the operation of retail markets and retail price setting◦ However, these models not account for the feedback impact of the retailer’s actions on electricity prices and power system

operation

The retailer’s optimal bidding and retail pricing problems are inherently linked: they need to be solved using amodel which explicitly couples the wholesale and retail markets.

04/04/2019

The Multi-Perspective Model (MPM)MPM considers the operation of a strategic retailer in the wholesale and retail electricitymarkets in the context of residential demand response.

Integration of the optimization problems of the retailer, SO and the individual consumers withina single framework.

Incorporates detailed representation of thermal demand using state-space models instead ofarbitrarily selected demand elasticities.

04/04/2019

Bilevel Structure The retailer’s problem is constrained by the outcomes of the market clearing problem and consumers’ cost minimization problems.

The retailer connects the individual consumers and the System Operator

The MPM is formulated as a bilevel problem Upper-level: Retailer’s profit maximization Lower-level: market clearing problem and

consumers’ optimization problems

04/04/2019

Market Clearing Problem

04/04/2019

subject to:

Maximize Social Welfare

Power balance constraint

Minimum and maximum limits of

generators and retailer

Storage evolution of large-scale pumped storage plant

Technical constraints of pumped storage

plant

Consumers’ Optimization Problem

04/04/2019

subject to:

Electricity and thermal comfort violation cost minimization

Indoor temperature constraintsThermal comfort violation non-

negativity constraint

Technical constraints of TES devices

State-space model for indoor temperature evolution

Storage level evolution of TES

Retailer’s Optimization Problem

04/04/2019

subject to:

Profit Maximization

Retail demand balance

Retail price limits

Average retail price constraint

Single Level Formulation and Linearization In order to obtain a solution for the bilevel problem, the UL and LL problems need to be solvedsimultaneously.

As the lower level problems (market clearing and consumer cost minimization) are linear, it canbe guaranteed that any solution which satisfies its KKT conditions would also be the optimalsolution of the problem

The lower level problems can be replaced with their KKT conditions as additional constraints inthe UL retailer’s optimization problem

Linearization: Strong-duality theorem and Fortuny-Amat transformations can be used to linearize the bilinear terms

and complementarity slackness conditions, respectively.

04/04/2019

M. B. Anwar, D. J. Burke and M. O’Malley, “A Multi-perspective Model for Evaluation of Residential Thermal Demand Response,” in press, IEEE Transactions on Smart Grid, 2019.

Case Study Irish All Island Power System (AIPS), with a peak demand of 7.2 GW and 4 GW installed wind capacity

12 representative generators

Building archetypes: Old (Energy Inefficient) and New (Energy Efficient) Apartments

Heating technologies: Thermal Electric Storage (TES) and Direct Resistive Heaters (DRH)

04/04/2019

Typical Daily Operation

04/04/2019

*CFD: Centralized Flexible Demand *MPM: Multi-perspective Model

Impact of TES Penetration Level

04/04/2019

*CIFD: Centralized Inflexible Demand *CFD: Centralized Flexible Demand *MPM: Multi-perspective Model

Impact of TES Penetration Level

04/04/2019

*TES: Consumers with Thermal Electric Storage*DRH: Consumers with Direct Resistive Heating

Impact of Wind Penetration Level (MW)

04/04/2019

Key Takeaways It is very important to consider the strategic behavior of DR participants for evaluation of system value of DR.

Centralized optimization models would typically tend to over-estimate the system value of DR in the presence of strategic participants.

Higher levels of consumer flexibility can reduce the retailer’s profits and the cost savings of the flexible consumers (suggesting ‘self-cannibalization’).

Higher levels of wind generation would be beneficial for all the entities involved in DR.

04/04/2019

Thank You!

04/04/2019

Backup Slides

04/04/2019

Impact of Marginal Discomfort Cost (€/°Ch)

04/04/2019

Impact of Retail Price Margin

04/04/2019

Retail Price Margin can be used to an be used to study

the behavior of theretailer under different

levels of retail competition.

Impact of Retail Price Range (€/MWh)

04/04/2019