Clustering in DR A primary step in DR implementation is the detailed - - PowerPoint PPT Presentation

Clustering in DR

A primary step in DR implementation is the detailed knowledge of customer potential through customer aggregation and characterization of demand clusters. Customers whose demand follow the day-ahead or real-time prices are the ones who are more suitable for DR programs selection from the view point of both customer and supplier Clustering provides typical load pattern of each customer class which can be used for choosing DSM program, tariff structure etc. Moreover demand characterization provides an effective tool to estimate the potential demand reduction, loss of service cost and impact of DR programs on demand.

19 / 26

DR in Smart Grid

Technology advancements in smart grid facilitates deployment of DR Early detection : Smart grid enables utilities to detect and respond to load increases early by calling a DR event. Improved Communication : Smart grid promises to increase the efficacy of DR by streamlining the notification process and providing real-time information to customers Accurate and Easy Verification : Smart grid enables utilities to measure and verify customers’ curtailment during a DR event Automation Systems : An automated DR would automatically detect the need to shed load, send signals to participants, and control all devices that use electricity Demand Response Markets : Smart grid creates a market for energy efficiency by enabling large energy users to reduce consumption when pricing rates are higher.

20 / 26

DR in Smart Grids contd...

Demand side management DR Porgrames Energy Conservation Strategic load growth Flexible load shaping Incentive based Load management Market based Direct load control Interruptible load control Demand side bidding Capacity market Ancillary services Price based Time of use Real time pricing Critical peak pricing

21 / 26

Enabling Technologies

It can be broadly captured under following heads- Advanced metering - Advanced Meter Reading (AMR) and Advanced Metering Infrastructure (AMI). Most essential feature of AMI is availability of continuous and automatic bidirectional communication link. Communication - Local Area Network(LAN) for communication within consumer’s premise and between the premise and nearest data aggregators, while Wide Area Network (WAN) for collection of data from aggregators spread over wide area. Meter data management - collecting, organizing and processing data for various applications e.g. billing customer load profiling etc. Control - Control of consumption. Home automation systems, home area network (HAN), In-home Displays (IHDs)

22 / 26

Popular DR programs

Some programs from US and Europe. New York ISO : Reliability based and Economic based DR programs. ISO England, USA : Incentive based DR. Pennsylvania Jersey Maryland (PJM, USA): Economic and Emergency DR programs. Electric Reliability Council of Texas (ERCOT, USA) : Emergency Response service. Nordic Market - apart from similar economic and reliability based DR programs it also has grid tariffs for effective price based also included.

23 / 26

DR Standards

Energy inter-operation version 1.0 -information and communication model developed by OASIS for the coordination purpose between any two parties in power system OpenADR 2.0 - extension of OpenADR 1.0. It had two

• profiles. First one is OpenAdr2.0a, to support basic DR

services in low-end embedded devices while second one is OpenADR2.0b, to support DR for high-end embedded devices. It’s communication is based on the principle of VTN (virtual top node) and VEN (virtual end node). Smart Energy profile 2.0 - specification for different energy transaction in smart grid domain e.g. communication smart meters, energy management systems, electric vehicles.

24 / 26

Practical Challenges

Following are the major challenges faced in the implementation of a DR program.

• perational time-scale at which it should be incorporated into

the system proper tariff and regulation policies cost recovery and profit for economic sustainability automation of different systems reliability of supply and privacy of usage for sensitive customers base line problem mainly in incentive based DR programs security and privacy for both customers and utility

25 / 26

DR in Indian Context

Demand response is in its nascent stages in India. Its benefits will mirror those seen in Western countries, such as reduced electricity blackouts, reduced electricity costs,

• ffsetting the need to build supply resource and the ability to

integrate electric vehicles and renewable energy sources. Regulatory framework also needs to be in place for implementation of DR strategies. It is necessary to identify the appropriate consumers to be roped in for DR to ensure the success of the program due to large diversity. Consumer awareness and maintaining transparency with them must be a priority to win their confidence and ensure acceptance of the DR program. The electrical distribution network must be strengthened to ensure reliability in operations. To cater to the diverse needs of the consumers, a variety of DR modules need to be prepared.

26 / 26

Review of Demand Response Under Smart Grid Paradigm

• V. S. K. Murthy Balijepalli, Vedanta Pradhan, S. A. Khaparde and R. M. Shereef

Panel Session on Smart Grid 2011 IEEE PES International Conference on Innovative Smart Grid Technologies 2nd December, 2011, Kerala, India Presented by: Vedanta Pradhan

• I. I. T. Bombay

Outline

1

• Introduction

2

• Motivation and Benefits

3

• Literature Review

4

• Challenges in DR implementation

5

• Conclusions

Outline

1

• Introduction

2

• Motivation and Benefits

3

• Literature Review

4

• Challenges in DR implementation

5

• Conclusions

Introduction

• Definition:

According to Federal Energy Regulatory Commission Demand Response (DR) is defined as: “Changes in electric usage by demand-side resources from their normal consumption patterns in response to changes in the price of electricity over time, or to incentive payments designed to induce lower electricity use at times of high wholesale market prices or when system reliability is jeopardized.”

DR Classification

Trigger Criteria Motivation Method

Load Response Price Response Reliability Economic

• Direct Load Control
• Curtailable Load
• Interruptible Load
• Direct Load Control
• Curtailable Load
• Critical Peak Pricing
• Demand Bidding
• Time-of-Use Pricing
• Critical Peak Pricing
• Real-time Pricing
• Demand Bidding

1

• Introduction

2

• Motivation and Benefits

3

• Literature Review

4

• Challenges in DR implementation

5

• Conclusions

DR Benefits

End users affecting market Savings in electricity bills Incentives Performance payments Deferred need for T&D infrastructure  Managing network constraints Spot price volatility risk reduction Market power mitigation Overall electricity price reduction Customer Network & System Operation Retailer/ Distributor Market More penetration of renewable resources Renewables

Motivation

Advancement

• f ICT and

automation Ample DR potential

1

• Introduction

2

• Motivation and Benefits

3

• Literature Review

4

• Challenges in DR implementation

5

• Conclusions

Literature Review

• DR concepts

and models

Class A

• DR frameworks

directly applicable to wholesale markets

Class B

• DR frameworks

directly applicable to retail markets

Class C

• DR concepts

and models

Class A

• DR frameworks

directly applicable to wholesale markets

Class B

• DR frameworks

directly applicable to retail markets

Class C

The importance of an active demand side in the electricity industry- H. Fraser,

• Nov. 2001
• Market based model; effective DR mechanisms; vary consumption

according to hourly prices Demand side management: Benefits and challenges – G. Strbac, Dec. 2008

• Review of DSM techniques; challenges; dynamic pricing

Behaviour modification – S. Braithwait, May 2010 A comparison of system response for different types of real-time pricing – A.

• K. David, Y. Z. Li, 1991
• Consumer categories SR, LR, RW; system disturbances; spot pricing and

day-ahead pricing; consumer response

Demand response and market performance in power economics – D. Yang, Y. Chen, 2009

• Demand interaction with power flows; adverse effect on price on some

node; congestion on some corridor; market gaming A summary of demand response in electricity markets - M.H. Albadi, E.F. El- Saadany, Nov. 2008

• Definition, classification, potential benefits, cost components; indices

used for DR measurement; utilities experience with DR programs ; effect

• n electricity prices.

Framework for the incorporation of demand side in a competitive electricity market - G. Strbac, E. D. Farmer and B. J. Cory, May 1996

• Redistributable demand; Demand Side Bidding; cost minimization

problem; limits on generation and demand reduction; diversity in load redistribution

Multi-objective demand response allocation in restructured energy market - T.

• T. Nguyen and A. Yousefi, 2011
• Optimal DR location; multiple objectives- max. ATC, min. EENS, min.

losses, min. DR capacity, etc.; NSGA II algorithm Analysing the system effects of optimal demand response utilization for reserve procurement and peak clipping - M. Behrangrad, H. Sugihara and T. Funaki, 2010

• Unit commitment problem; LMIP problem of energy cost minimization;

RSDR and PCDR; effects on total energy cost and market clearing price Demand response management in power systems using a particle swarm

• ptimization approach - P. Faria, Z. Vale, J. Soares, J. Ferreira, Apr. 2011
• Load control based on price elasticity; minimization of cost to the

consumer; constraints are maximum load reduction from a consumer; maximum increase in price & load and generation balance; Particle Swarm Optimization.

T c Price Price Price /P Price P (c) Elasticity P

• P

P

• P

P Max Price P Max P : subject to ) Price Price ( ) E E ( Cost Min

T), Energyvar( c) Energyvar( c) Energyvar( load(c) ial(c) Energyinit loadred(c) nc 1 c LoadRed(c) nc 1 c load(c) Reserve Main c) Energyvar( c) Energyvar( loadRed(c) loadRed(c) 1 c) Energyvar( ial(c) EnergyInit loadRed(c) load(c)

          

    

  

nc c

Demand response management in power systems using a particle swarm

• ptimization approach - P. Faria, Z. Vale, J. Soares, J. Ferreira, Apr. 2011

Development of reliability based demand response program in Korea - T. H. Yoo , 2011

• Incomplete market; reliability based DR; energy, reserve and capacity

reward for compensation of DR providers. Demand response and distribution grid operations: Opportunities and challenges - J. Medina, N. Mueller and I. Roytelman, Sept. 2010

• Distribution system topology integration into DR scheduling process;

better monitor and verification of DR; DSO Demand response architecture- Integration into the distribution management system - S. Mohagheghi, J. Stoupis, Z. Wang, Z. Li and H. Kazemzadeh, 2010

• DR implementation from DMS level; additional benefits like congestion

relief

Demand Response for Domestic and Small Business Consumers : A New Challenge - D. T. Nguyen, 2010

• Issues relating reliability and price based DR; individually cannot

maximize the benefits out of DR Novel business models for demand response exchange - M. Negnevitsky, T.

• D. Nguyen and M. de Groot, 2010
• DR as virtual resource; TSO, DSO, MO, aggregators and end consumers

are the key players; bilateral and pool based business models Centralized and decentralized control for demand response - S. Lu, N. Samaan, R. Diao, M. Elizondo, C. Jin, E. Mayhorn, Y. Zhang and H. Kirkham, 2011

• IEEE 34 bus system; detailed household load models; aggregated load

profile; study of performance of control strategies.

Demand response for smart microgrid: Initial results - S. A. Pourmousavi and M. H. Nehrir, 2011

• Islanded microgrid with diesel generator and fixed and flexible loads simulated in

MATLAB; performance for frequency and voltage regulation evaluated; controller based on Adaptive Hill Climbing Improving WFA K-means technique for demand response programs applications - N. Mahmoudi-Kohan and M. P. Moghaddam, 2009

• Fuzzy average K-means approach applied to cluster classification of 316 load

curves of non-residential customers. Customer classification and load profiling method for distribution systems - A. Mutanen, M. Ruska, S. Repo and P. Jarventausta, June 2011

• ISODATA algorithm for customer classification; includes temperature dependency

correction and outliers filtering. Methods for customer and demand response policies selection in new electricity markets - S. Valero, M. Ortiz, C. Senabre, C. Alvarez, F. J. G. Franco and A. Gabaldon

• Self-organizing maps and physically based load modelling for customer

classification and demand characterization

• DR concepts

and models

Class A

• DR frameworks

directly applicable to wholesale markets

Class B

• DR frameworks

directly applicable to retail markets

Class C

Pool based demand response exchange– concept and modeling – D. T. Nguyen, M. Negnevitsky and M. de Groot, 2011

• Demand Response Exchange; supply and demand bids for DR;
• ptimization of net market benefit; constraints include assurance contract

and supply demand balance; non-linear programming problem. Power grid balancing of energy systems with high renewable energy penetration by demand response - I. Stadler, June 2008

• Integration of intermittent energy resources into electricity system;

thermal energy suited for storage; study of potential of such DR activities by modelling of thermal storage devices and lab tests. Quantifying the effect of demand response on electricity markets – Chua- Liang Su and D. Kirschen, Aug. 2009

• Complex bid day ahead market clearing; models for different types of

consumer bids; objective is maximization of social welfare; piecewise linear generation cost curve; MILP; indices for quantifying DR proposed.

Congestion management using demand response programs in power market -

• E. Shayesteh, M. P. Moghaddam, S. Taherynejhad and M. K. Sheikh, 2008
• Redispatch of demand alongwith generation to handle congestion; cost-

effective; objective is minimization of cost of adjustment; constraints are supply demand balance at each bus; line flow limits and generation limits .Demand response scheduling by stochastic SCUC - M. Parvania and M. Fotuhi-Firuzabad, June 2010

• Short-term stochastic SCUC model; scheduling generating units energy,

reserves and demand response reserves also; two-stage mixed integer programming. Role of demand response in ancillary services markets - K. Schisler, T. Sick and K. Brief, 2008

• Demand side participation in hourly reserve market of PJM; reserve

provided within 10 minutes timeline; high level communication; customer site servers; PJM SCADA manager; TCP/IP

A probabilistic reserve market incorporating interruptible load - J. Bai, H. B. Gooi, L. M. Xia, G. Strbac, and B. Venkatesh, Aug. 2006

• Combined day ahead scheduling; sum of generation production cost,

EENS, failure of ILs to supply reserve is minimized; Lagrangian Relaxation module with EENS module used to determine optimal schedules. Interruptible load management using optimal power flow analysis - S. Majumdar, D. Chattopadhyay, and J. Parikh, May 1996

• OPF based model to support IL management; network and generation

constraints; power factor of interruptible load; period of curtailment; result is optimal location of IL of different natures. Scheduling of demand side resources using Binary Particle Swarm Optimization - M. A. A. Pedrasa, T. D. Spooner and I. F. MacGill, Aug. 2009

• System requirement of hourly curtailments; minimizing payment to the

ILs; minimizing frequency of interruption; multiobjective optimization; complex, non-linear and non-continuous.

• DR concepts

and models

Class A

• DR frameworks

directly applicable to wholesale markets

Class B

• DR frameworks

directly applicable to retail markets

Class C

Consumer centric smart grid - W-H Edwin Liu, Kevin Liu and Dan Pearson, 2011

• Aspects of grid modernization; information & infrastructure; instrument &

technology; intelligence & automation A direct load control model for virtual power plant management - N. Ruiz, I. Cobelo and J. Oyarzabal, May 2009

• Thermostatically controlled appliances of customers; optimal control

schedules or load reduction bids; customer classification and study of effect of different control actions on load profile; used in algorithm for DLC Realizing smart grid benefits requires energy optimizatin algorithms at residential level - T. Hubert, S. Grijalva, 2011

• Home electrical system; modelling of HVAC, storage and non-

interruptible appliances; objective is minimization of cost of energy usage

• ver varying price signal; LMIP problem

Autonomous demand-side management based on game-theoretic energy consumption scheduling for the future smart grid - A. H. Mohesenian-Rad, V.Wong, J. Jatskevich, R. Schober and A. Leon-Garcia, Dec. 2010

• Energy Consumption Scheduler; Game theory; communication among

different customers and with the utility; global optimal energy cost

• btained at the Nash equilibrium of the scheduling game.

Demand response model considering EDRP and TOU programs - H. Alami,

• G. R. Yousefi and M. P. Moghadam, 2008
• DR model based on EDR and TOU; load elasticity; single period, multi-

period models; customer benefit maximization; effect of TOU prices, incentives and load elasticity on load curve are analyzed. Real-time demand response model - A. J. Conejo, J. M. Morales and L. Baringo, Dec. 2010

• Consumption scheduling in near real-time; hourly pricing; objective is

consumer utility maximization; price uncertainty of the coming hours modeled; robust optimization

t

• 4

0,......,2 h , d d d t

• 4

0,......,2 h , d d t

• 4

0,......,2 h , d d t

• 4

0,......,2 h , )/2 d (d e e e e e : subject to )] (e u e } [{λ ) (e u

• e

λ Minimize

1(max) h t 1 h t 1(min) h t h t 1 h t 1 h t h t 1 h t h t h t 1

• t

1 h t

• 24

1 h day h t t h h t h t h t t

• 24

1 h h t t t t t

                

                         

  

U D

r r

Real-time demand response model - A. J. Conejo, J. M. Morales and L. Baringo, Dec. 2010

Industrial power demand response analysis for one part real-time pricing - J.

• G. Roos and I. E. Lane, Feb 1998
• Hourly marginal rate; assumptions of adequate storage capacity,

production target, no scheduling losses; minimization of energy cost; linear programming Optimal demand-side response to electricity spot prices for storage-type customers - B. Daryanian, R. E. Bohn and R. D. Tabors, Aug 1989

• Non-simplex algorithm to optimize consumption of storage type industry;

predetermined price schedule; limits on storage capacity; equal storage levels at the beginning and at the end. Smart (in-home) power scheduling for demand response on the smart grid - G. Xiong, C. Chen, S. Kishore and A. Yener, 2011

• Smart appliances; energy management controller; Home area network;

scheduling procedure of real-time and schedulable appliances; communication signals.

Decentralized demand-side contribution to primary frequency control - A. Molina-Garcia, F. Bouffard and D. S. Kirschen, Feb. 2011

• DR as source of primary frequency control; frequency-time

characteristics; aggregation of such loads; fuzziness; algorithm to be built into the load controllers proposed . Stabilization of grid frequency through dynamic demand control - J. A. Short,

• D. G. Infield, and L. L. Freris, Aug. 2007
• Simulation model including model of the grid with aggregate generator

inertia, governor action , load frequency dependence, large no. of refrigerators, and dynamic demand controller. Supporting high penetrations of renewable generation via implementation of real-time electricity pricing and demand response - A. J. Roscoe and G. Cult, July 2010.

• Modelling and simulation of load shifting implemented on elastic

domestic loads in response to high prices when only thermal generation is present and forecasted demand is close to the supply capacity.

An event-driven demand response scheme for power system security enhancement - Y. Wang, I. R. Pordanjani and W. Xu, Mar. 2011

• Action table consisting of credible contingencies, load reduction amount

and location; objective is to minimize cost of load reduction; non-linear sensitivity of reserve to load reduction, non-linear demand cost function of demand reserves; multistage optimization; SCADA/EMS Commercial building load modeling for demand response applications - M. McGranaghan, A. Didierjean and R. Russ, 2009

• Modelling and simulation of commercial building loads especially HVAC

loads; estimation of the control margin obtained with certain load control event; results can be used in optimization algorithm by the central controller Business case for a consumer portal - M. McGranaghan, A. Didierjean and R. Russ, 2005

• Gateway between the utility and the consumer; flexible pricing , remote

energy use monitoring and control, etc. ; standardized technology with common information model; use of IP and existing planned communication infrastructure

Coordinating regulation and demand response in electric power grids using multirate model predictive control - H. Hindi, D. Greene and C. Laventall, 2011

• Integration of intermittent energy resources into electricity system;

thermal energy suited for storage; study of potential of such DR activities by modelling of thermal storage devices and lab tests. Demand Response from household customers: Experience from a pilot study in Norway - H. Saele and O. S. Grande, Mar. 2011

• Experience from DR pilot project at Malvik Everk; household customers;

ToD tariff; smart metering; token indicating peak hours; estimate of DR potential 1000 MWh/h from 50% household customers; DR could be bid into the day ahead market

• The ToD formulated as:

 

   

    

365 1 d 24 1 t 365 1 d 24 1 t t d, tp d, t d, spot t, d, en

W α W p γ β Cen

• Objective was to motivate the residential

customers to change their load curve

• Reduction of total consumption in periods of

expected peak load

• By giving the customers a predictable price that is

also dynamic and market based

• Traditionally the network tariff to household

customers:

part losses Network part Constant Thousehold  

Norwegian Power Market System

Histogram of the occurrence of maximum spot prices for Trondheim 2005–2009 (data source: NordPool).

 

   

    

365 1 d 24 1 t 365 1 d 24 1 t t d, tp d, t d, spot t, d, en

W α W p γ β Cen

ToD Network Tariff used in the pilot study:

Constant Part Network Losses Part Variable Energy part

• 40 household customers under Malvik Everk
• Hourly metering
• Hot water space heating system with electrical

boiler; standard electrical water heaters

• Information meetings; website with information

about network tariff

• DSO had to get exemption from regulation
• Duration of one year

Dynamic price signal to household customers (week 6–2007).

• Remote Load Control via the AMR

was offered

• Each household in the pilot study was

equipped with three small tokens, the “El-buttons”

Load profile for a household customer with hot water space heating system and RLC

• The registered average DR in the pilot study

described in this paper was 1 kWh/h

• By aggregating the response, the total potential

for DR from households in Norway can be estimated at up to 1000 MWh/h.

• This is 4.2% of registered peak load demand in

Norway (23 994 MW, 6 January 2010).

An integrated architecture for demand response communications and control

• Meter gateway architecture; framework integrating building automation

system and advanced meter infrastructure to implementing DR; Unified hub for control and communications An architecture for local energy generation, distribution, and sharing

• LoCal grid- peer-to-peer technology based on the concept of “packetized

energy” introducing routing of energy; Components- Energy storage technologies, intelligent power switches A new architecture for reduction of energy consumption of home appliances

• Smartgrid hierarchical framework involving residential users, network
• perators, and energy utilities; reference architecture includes- AIM

gateway, interfaces to outdoor network ; energy management devices

1

• Introduction

2

• Motivation and Benefits

3

• Literature Review

4

• Challenges in DR implementation

5

• Conclusions

Challenges in DR implementation

Demand Response

Proof of concept Governme nt fund Policies, infrastruct ure ICT Consumer education Standards Rate design

1

• Introduction

2

• Motivation and Benefits

3

• Literature Review

4

• Challenges in DR implementation

5

• Conclusions

• Classification of literature based on applicability

to wholesale or retail markets

• End user participation in energy supply chain can

be promoted

• Design of DR program depends on prevailing

market conditions of a particular region

• DR models reported are mostly academic in

nature

• Practical viability requires ample amount of

research

• Projection on future research growth in DR

Residential Demand Response Modeling in a Dynamic Pricing Environment

• Towards Smart Homes

9/17/2018 1

MTP Stage-1

Vedanta Pradhan Roll no: 10307006 Thesis Advisor- Prof. S A Khaparde

IIT Bombay

Outline

9/17/2018 2

MTP Stage-1

1

• Background and Motivation

2

• Literature review

3

• Proposed framework

4

• Model formulations and Simulation Results

5

• Conclusions and Future Work

IIT Bombay

Outline

9/17/2018 3

MTP Stage-1

1

• Background and Motivation

2

• Literature review

3

• Residential DR framework

4

• Model formulations and Simulation Results

5

• Conclusions and Future Work

IIT Bombay

Background

• Definition:

According to Federal Energy Regulatory Commission Demand Response (DR) is defined as: “Changes in electric usage by demand-side resources from their normal consumption patterns in response to changes in the price of electricity over time, or to incentive payments designed to induce lower electricity use at times of high wholesale market prices or when system reliability is jeopardized.”

9/17/2018 4

MTP Stage-1

IIT Bombay

DR Classification

9/17/2018

MTP Stage-1

5

Trigger Criteria Motivation Method

Load Response Price Response Reliability Economic

• Direct Load Control
• Curtailable Load
• Interruptible Load
• Direct Load Control
• Curtailable Load
• Critical Peak Pricing
• Demand Bidding
• Time-of-Use Pricing
• Critical Peak Pricing
• Real-time Pricing
• Demand Bidding

IIT Bombay

DR Benefits

9/17/2018

MTP Stage-1

6

End users affecting market Savings in electricity bills Incentives Performance payments Deferred need for T&D infrastructure  Managing network constraints Spot price volatility risk reduction Market power mitigation Overall electricity price reduction Customer Network & System Operation Retailer/ Distributor Market More penetration of renewable resources Renewables

IIT Bombay

Motivation

9/17/2018 7

MTP Stage-1

Advancement

• f ICT and

automation Ample residential DR potential

IIT Bombay

Outline

9/17/2018 8

MTP Stage-1

1

• Background and Motivation

2

• Literature review

3

• Residential DR framework

4

• Model formulations and Simulation Results

5

• Conclusions and Future Work

IIT Bombay

Literature Review

9/17/2018 9

MTP Stage-1

• How to

schedule the loads? Scheduling Models

• How DR should

be implemented? DR Architecture

• How to

mathematically model the end- use loads? Load Modeling

IIT Bombay

Scheduling models

Real-time demand response model- A.J. Conejo et. Dec. 2010

• Maximize consumer utility; hourly loads; linear optimization; robust optimization; bi-directional communication

Realizing smart grid benefits requires energy optimization algorithms at residential level- Hubert et. Aug. 2011

• Home electrical system; grid connected; ,minimization of cost of energy; MILP;

Autonomous DSM based on game theoretic energy consumption scheduling for the future smart grid- A. Hamed et. Dec. 2010

• Decentralized; Energy Consumption Scheduler; appliance level; nonlinear model for energy cost minimization and

PAR minimization; Game theoretic basis for solving .

Residential demand response using reinforcement learning- Neill et. Oct. 2010

• Energy prices and device usage modelled as Markov chains; Q-learning is used to make optimal energy usage

decisions; reduce long term energy costs.

Optimal residential load control with price prediction in real-time electricity pricing environments- A. Hamed et., Feb . 2011

• Optimal appliance scheduling ; price prediction in real-time; minimizing payment and waiting time.

9/17/2018 10

MTP Stage-1

IIT Bombay

DR architecture

An integrated architecture for demand response communications and control-Lemay et. Jan. 2008

• Meter gateway architecture; Building automation system; Advance meter infrastructure technologies; Zigbee

Home area networks for electricity demand management- Wacks et. June 2008

• Residential DSM systems; EMA; centralized controller separated from the smart meter

Get Smart- Lui et. May 2010

• Integration of smart appliances; DR architecture including both HAN and smart grid domains.

Simulated demand response of a residential energy management system- Roe et. May 2011

• REMS; automation of DR activities; battery energy storage integration

Residential load control through real-time price signals- Zhang et al. Aug. 2008

• RTP system design; Appliance interface unit; Data concentrator; Display unit integration

9/17/2018 11

MTP Stage-1

IIT Bombay

End-use load modeling

Generalizing direct load control program analysis: Implementation of duty cycle approach-

• N. E. Ryan et, Feb. 1989
• Duty cycle model of end use load data; evaluating load reductions.

A stochastic computer model for heating and cooling loads- Mortensen et. , Aug. 1988

• Stochastic discrete time dynamic model of house temperature; to study discomfort to the customer

Probabilistic calculations of aggregate storage heating loads- Hatzyargyriou et. ,July 1990

• Probabilistic model for aggregate storage heating loads; uncertainty in the external temperature forecasts.

A state queuing model for thermostatically controlled appliances- Lu et., Aug. 2004

• State queuing model for TCAs; response to electricity prices; loss of load diversity.

Developement and vaildation of physically based computer model for predicting winter heating loads- Nehrir et.,Feb. 1995

• Physically based model for electric heating loads; predicting loads on distribution feeders; load management

strategy

9/17/2018 12

MTP Stage-1

IIT Bombay

Gist of the Literature Review

9/17/2018 13

MTP Stage-1

• Literature covers immediate as well as remotely

related aspects

• Several practical and theoretical models for

residential DR

• Several DR architecture models including

information and communication technology aspects

• A comparison of practical DR models not covered

VSK Murthy, V Pradhan, and S A Khaparde,“ Review of Demand Response under smart grid paradigm” IEEE PES international conference on ISGT , Dec.2011 (under review)

IIT Bombay

Outline

9/17/2018 14

MTP Stage-1

1

• Background and Motivation

2

• Literature review

3

• Residential DR framework

4

• Model formulations and Simulation Results

5

• Conclusions and Future Work

IIT Bombay

Residential Demand Response Framework

9/17/2018 15

MTP Stage-1

Simulated in MATLAB

IIT Bombay

Outline

9/17/2018 16

MTP Stage-1

1

• Background and Motivation

2

• Literature review

3

• Residential DR framework

4

• Model formulations and Simulation Results

5

• Conclusions and Future Work

IIT Bombay

Model Formulations

• Load Characteristics:

– Appliances can be categorized as shiftable and non- shiftable loads – Each shiftable load is characterized:

• With an operational time band within which only it
• perates and for the rest of the day it is “OFF”
• A minimum and maximum demand at any period of time

within the operation time band

• An average daily consumption

9/17/2018 17

MTP Stage-1

IIT Bombay

Model 1

A a for ρ ρ ρ A a for β α t for ρ A a for E ρ ρ ρ ρ λ Min

a at a at a t T t a a A a t where, t T t t

a) a, ( S.T. t

             

  

 

max min 1 1

9/17/2018 18

MTP Stage-1

t = index for time block T= total no. of time blocks in a day λ = price of electricity

ρ = total consumption from LSE ρa = consumption of appliance ‘a’

αa = first hour of operational time band for appliance ‘a’ βa = last hour of operational time band for appliance ‘a’ ρa min = minimum consumption level of appliance ‘a’ ρa max = minimum consumption level of appliance ‘a’ Ea = average daily consumption of appliance ‘a’

Linear Programming solved in MATLAB

IIT Bombay

Model 2

.

waiting

• f

cost

1  





t T t t

ρ λ Min

ameter stable par is an adju σ ) ,β (α t A, a for E β t ) (σ θ where ρ θ

, a a a a a a at at t T t A a a

1 ,

1

      



 

Cost of waiting:

9/17/2018 19

MTP Stage-1

Linear Programming solved in MATLAB

IIT Bombay

9/17/2018

MTP Stage-1

20

Cost of waiting Cost of waiting

, σ For

a

1  

a

σ





a

σ



, σ If

a

1 

a a a a at

/E E β t ) (σ θ 1   

for all time blocks

 user does not lay any importance on the comfort

level as far as the particular appliance is concerned.

IIT Bombay

Model 3

demand

• verall
• f

elasticity price self average the is E ρ E ρ ρ ρ ρ λ B ) B(ρ where, ρ θ ρ λ ) B(ρ Max

t t) t

• t)/

t (

• t)(

t (

• t
• t

t at t T t A a a t T t t T t t

2 ) (

1 1 1 1          

   

  

9/17/2018 21

MTP Stage-1

λo = nominal price of electricity

ρo = nominal overall demand

Bo = nominal benefit B(ρt) = Customer benefit from the use of ‘ρt’ kWh of electrical energy

Non- Linear Programming solved in MATLAB

IIT Bombay

Simulation Data

• Price data

9/17/2018 22

MTP Stage-1

IIT Bombay

Appliance data

Appliance Number ρamax (kW) ρamin (kW) α β Ea Lamps 5 0.10 0.00 1 24 25.00 Tubelights 5 0.04 0.00 1 24 25.00 Electric Iron 1 1.00 0.00 1 24 1.00 Toaster 1 0.75 0.00 1 24 0.75 Television 1 0.20 0.00 1 24 7.00 Fans 5 0.10 0.00 1 24 10.00

Non – shiftable appliance data

9/17/2018 23

MTP Stage-1

IIT Bombay

Shiftable/Schedulable appliance data

Appliance Number ρamax (kW) ρamin (kW) α β σ Ea Dishwasher 1 1.20 0.100 7 10 10.0 1.50 A/C 1 1.00 0.100 9 12 10.0 2.50 Cloth dryer 1 4.00 0.800 11 17 1.0 1.50 W/M 1 0.40 0.000 15 22 10.0 3.50 PHEV 1 5.00 0.500 17 24 10.0 2.50 PHEV 1 5.00 0.500 1 6 10.0 2.50 W/H 1 1.50 0.250 7 10 30.0 1.50 R/H 1 1.00 0.300 22 24 1.0 1.00 R/H 1 1.00 0.300 1 5 10.0 3.00 Refrigerator 1 0.15 0.075 2 23 1.5 1.50 A/C 1 1.00 0.300 12 17 1.7 2.75 Pump Motor 1 0.75 0.200 5 9 1.5 1.50

9/17/2018 24

MTP Stage-1

IIT Bombay

Steps in obtaining the results

Process the raw appliance data Condition the formulated problem as per the

• ptimization solver and Run the solver

Calculate

• Cost of unschedulable consumption
• Cost of schedulable but unscheduled consumption
• Cost of typical consumption
• Cost of scheduled consumption

Total optimal consumption cost

9/17/2018

MTP Stage-1

25

IIT Bombay

Simulation Results

9/17/2018 26

MTP Stage-1

Optimal schedules and normal consumption data of appliances 1-6 with model 1

IIT Bombay

9/17/2018 27

MTP Stage-1

Optimal schedules and normal consumption data of appliances 7-12 with model 1

IIT Bombay

9/17/2018 28

MTP Stage-1

Optimal schedule and normal consumption data of appliance 1 with model 2 and σ=1, σ=10

IIT Bombay

9/17/2018 29

MTP Stage-1

Optimal schedules and normal consumption data of appliances 1-6 with model 2 and all σ’s = 1

IIT Bombay

9/17/2018 30

MTP Stage-1

Optimal schedules and normal consumption data of appliances 7-12 with model 2 and all σ’s = 1

IIT Bombay

9/17/2018 31

MTP Stage-1

Optimal schedules and normal consumption data of appliances 1-6 with model 2

IIT Bombay

9/17/2018 32

MTP Stage-1

Optimal schedules and normal consumption data of appliances 7-12 with model 2

IIT Bombay

9/17/2018 33

MTP Stage-1

Optimal schedules and normal consumption data of appliances 1-6 with model 3

IIT Bombay

9/17/2018 34

MTP Stage-1

Optimal schedules and normal consumption data of appliances 7-12 with model 3

IIT Bombay

Results consolidation

Cost (in Rs. ) Model 1 Model 2 Model 3 Objective function value 58.8904 86.6324 56.4775 Cost of unschedulable consumption 33.3100 33.3100 33.3100 Cost of unscheduled consumption 65.3147 65.3147 65.3147 Cost of scheduled consumption 58.8904 59.5197 62.9466 Cost of typical consumption 98.6247 98.6247 98.6247 Optimal consumption cost 92.2004 92.8279 96.2566 Saving in payment 6.4243 5.7968 2.3681

9/17/2018

MTP Stage-1

35

IIT Bombay

Outline

9/17/2018 36

MTP Stage-1

1

• Background and Motivation

2

• Literature review

3

• Residential DR framework

4

• Model formulations and Simulation Results

5

• Conclusions and Future Work

IIT Bombay

Conclusion

• Exhaustive literature review on residential

demand response

• A background about the topic is built to introduce

the recent advances in the field

• Few pragmatic models are simulated and tested
• n the empirical residential data and results

compared

9/17/2018

MTP Stage-1

37

IIT Bombay

Future work

• Extending the concept of an active residential

controller to an interactive controller

• Proposing a moving window scheduling model

taking into account the near real-time prices

• Incorporating the price classification models for

the scheduling algorithm selection

9/17/2018

MTP Stage-1

38