DSM in the Smart Grid-a control perspective - …home.deib.polimi.it/ferrarin/psc/files/DSM in the Smart Grid-a... · 8 Demand Side Management in the Smart Grid: a control perspective

Demand side management in the Smart Grid:a control perspective

Giuseppe Tommaso CostanzoEnergy Systems Operations and Management

Department of Electrical Engineering

Technical University of Denmark

[email protected]

27/05/20132 Demand Side Management in the Smart Grid: a control perspective

Agenda

• Electric energy and power systems

• Toward Smart Grids: the role of control in an intelligent energy system

• Demand Side Management (DSM) for residential applications

• Distributed control for intelligent power dispatch

• Discussion: close the loop, what’s next?


Electric energy and power systems

The key to modern quality of life is the availability of electric energy, foreveryone and in every situation. Imagine what would happen in case offailure of energy supply in such applications like:

• Transportation: car, train, airplane

• Health care

• Information technology

• Industrial production (water handling, fuels, food, etc…)

• Space heating and ambient control

• Security systems

• … and many more…

Almost every human activity nowadays consumes electric power



How Ampere law is used to produce electric energy

Electric current

Magnetic field

Inducted current



How Ampere law is used to produce electric energy

The alternator is used to transform mechanical energy into electric energy







Parts of the power system:

• Production: high voltage

• Transport: medium voltage

• Distribution: low voltage

Why we need high voltages?

To save energy and money!

1. Ohm law:

2. Thermal losses in the cables:

� � � ∗ �

� � � ∗ � � � ∗ ��

Quadratic relationship: by doubling the voltage, we reduce the losses by four times



… and what about frequency?

People cycling

People just sitting



… and what about the voltage?

Bike speedSystem frequency

Steering barSystem voltage


Renewable energy sources (RES)

Factors as:

• rapid industrialization of a number of countries

• interest in reducing the carbon footprint of human activities

led to an increased attention to renewable energy.


Renewable energy sources (RES)

The increasing share of distributed generation from renewable energy sources puts the power system to face problems as:

• Unpredictable production

• Alternated power flows at the LV level

• Grid congestion, peak loads, valley filling


Smart Grid

Integrate information technologies in power systems in order to coordinate all the brunched customers and producers actions to deliver power safely and sustainably for the next century.


Smart Grid

Will accommodate bidirectional fluxes of energy and information, enabling:

• operational efficiency: with distributed generation, remote monitoring;

• energy efficiency: with reduced system and line losses, and accomplishment with governmental policies about energy saving;

• customer satisfaction: Smart Grid will enable customers self-service;

• CO2 emission reduction: by integrating renewable energy sources, PHEVs and by decreasing the usage of support plants.


The role of consumption control inintelligent energy systems

We are witnessing a change from centralized bulk production in large power plants to power systems with an increasing share of distributed production and uncontrollable power feed in from renewables.

In this context, users adaptation is a promising technology allowing to reduce and delay grid reinforcements and enhance reliability power supply.

Some motivations to users adaptation in power systems:

• it improves the distribution grid stability and operation margins;

• it allows users to actively operate in the energy markets;

• it enhances energy saving and conservation;


Direct Load Control (DLC): is a technique based on issuing specific commands to controllable DERs. The decision on DER operation is taken by a controller at higher level of the control hierarchy, has a knowledge of the DER status.

V.S.

Indirect Load Control (ILC): is a technique based on one-way communication of information which might affect DERs operation. Reaction to such signal is determined by decisions taken at DER level (local), and the device is not obliged to react to the external control signal, thus its slight unpredictability.

From:

[1] A.M. Kosek et al., ”An overview of Load Control Policies in Buildings for Smart Grids”, IEEE International Conference on Smart Energy Engineering, 28-30 August 2013, Oshawa (ON), Canada.

Demand side management in the Smart Grid: DLC versus ILC


Tools:

• statistics and probability theory;

• optimization: deterministic, robust, stochastic;

• scheduling: online, offline;

• artificial intelligence and machine learning;

• & more…

Models:

• building consumption;

• devices operation;

• users’ behavior;

• & more…

Demand side management in the Smart Grid


Home Energy ManagerSmart Grid Smart Appliances

A layered control architecture can separate and hide devices operation from grid dynamics. At the same time it can provide the services of responsive demand, among which we find at glance (for active power) [1]:

• Deferred operation

• Delta operation

• Scheduled operation

• Direct power control

Demand side management viaDirect Load Control for residential applications in the Smart Grid


Appliances

In this context, appliances are seen as devices allowing the tenants to execute tasks, the accomplishment of which causes an increase of the comfort level.

Handling such tasks require to share a common resource: the power consumption.


Power FlexHouse: single room model [2]

( ) ( )

( ) ( ) ( ) ( )

( )

( ) ( )

1 2

1 1 11

1

1

1

1

ii i om i s

a im

imim im s

omom i om om

im am

a i im hi

ii

a

dT

dt

dTC T T T T T T A p

dt R R R

dTC T T

dQ dTQ T T C

Apdt R

dTC T T T

T Cdt R

T

t

dt R R

d

= − + − + − + − Φ + Φ

= − + Φ

= = − = ≈

= − + −

&

[2] A. Thavlov and H. Bindner, "Thermal models for intelligent heating of buildings", International Conference on Applied Energy, July 5-8, 2012, Suzhou, China.

Demand side management viaDirect Load Control in the Smart Grid


Model performances


Models of appliances:

• regular loads: refrigerators, freezers, water boilers, etc…, are modeled via thermal circuits;

• burst loads: dishwashers, tumble dryers, washing machines, etc…, are modeled via Finite State Machines;

• baseline loads (non deferrable loads): lighting, entertainment, small appliances, are modeled as additive stochastic disturbance in the power consumption.

This approach is simple and effective to catch the dominant dynamics of the considered devices


Refrigerator modeling: experimental setup

Setup features:

• two internal temperatures

• freezer temperature

• door status (open/closed)

• P, Q, S, I, V, f, pf, THD


Refrigerator model: vapor compression cycle

( )·

cs cs cs cs

load cs a cs

e c

Q m c T

Q UA T T

Q COPW

== −

≈

&

& &


Refrigerator model:refrigeration chamber and envelope [4]

( ) ( )

( )

1 1

2 2

1 1 1 2 1

1 1

1 1

, ~ (0,1)

i a i e ii ia i ei

e i e c ce ei i

i

dT T T T T dt dwC R C R

dT T T A dt dwC R C

y T e dv w w v N

σ

σ

= − + − +

= − − Φ +

= + ⊥ ⊥

[4] G.T. Costanzo et al., “Grey-box Modeling for System Identification of Household Refrigerators: a Step Toward Smart Appliances”, IEEE International Youth Conference on Energy (IYCE 2013), Siofok, Hungary, June 5-8, 2013.


Refrigerator model: performances (TiTevap)


Generic appliance modeling viaFinite State Machine

Appliance status can be:

• Off: appliance not enabled;

• Ready: enable asserted, appliance ready to start;

• Run: enable asserted and start command received, appliance consuming energy;

• Idle: enable asserted and stop command received, appliance not consuming energy;

• Complete: task completed and transit to “Ready” for being turned off or possible re-invoke;

• Fault: fault detected in the appliance.


Consumption patterns: dishwasher

Total energy flow = 1.103 kVAh

Total energy consumed = 1.079 kWh

Cycle length (hh:mm:ss) = 01:20:00


Consumption patterns: microwave oven

1) 12:36 – Quick (3’) : 91.09Wh

2) 13:03 – Quick (3’) : 89.41Wh

3) 13:09 – 2*Quick (30’’) : 2 * 13.6Wh

4) 13:12 – Quick (1’) : 28.1Wh

5) 13:15 – Quick (3’) : 91.27Wh

1 2

3

4 5


Generic appliance modeling viaFinite State Machine


Appliances models extensions:

Refrigerator:

• physical appliance model;

• food model;

• usage model: load/unload/neutral.

Microwave:

• operation programs;

• food influence;

• usage model.

Dishwasher:

• available programs;

• tableware influence;

• usage model.


The Home Energy Manager (HEM) [5]

• Communicate with the Micro Grid

• Handle cyber security

• Balance load in time

• Provide sufficient quality of service

• Minimize the energy expenses

• Shave the load-peaks

• Control operations in real-time

• Trade energy with upper layers

• Balance user comfort with user utility

Demand side management viaDirect Load Control in the Smart Grid

[5] G.T. Costanzo et al., ”A System Architecture for Autonomous Demand Side Load Management in Smart Buildings”, IEEE Transactions on Smart Grids, Vol. 3, Issue 4, p.p. 2157-2165, 2012.


Home Energy Manager:Admission Controller


Load peak shaving via A.C. (Exp.) [6]

[6] G.T. Costanzo et al., ”A Experimental study on Load-Peak Shaving in Smart Homes by means of Online Admission Control”, IEEE Innovative Smart Grid Technologies (ISGT Europe), Berlin, Germany, October 14-17, 2012.


Home Energy Manager:Load Balancer


Home Energy Manager:Load Balancer


Home Energy Manager:Demand Response management (Exp.) [7]

( ) ( ) ( )( ) ( ) ( )( )( ) ( )( )

( ) ( ) ( )

1

log

,

i i i i i i

i i i

k

ii C

x k x k u x k x k p k

u k w x k

xp k f x k f x a

C

α

∈

+ = + −

= −

= =

∑

[7] G.T. Costanzo , ” Assessment and experimental validation of a real- time energy pricing algorithm for the Smart Grid”, DTU-CEE internal report, available upon request, May 31 , 2012.


DERri – Distributed Energy Resource research infrastructure

Project: SEMBAD – Smart Energy Management in Buildings and Districts

Partners: Politecnico di Milano, Risø DTU

Dates: 15/8-19/8, 2011 and 28/11-2/12, 2011



Project: SEMBAD – Smart Energy Management in Buildings and Districts

Partners: Politecnico di Milano, Risø DTU

Dates: 15/8-19/8, 2011 and 28/11-2/12, 2011



Project: SREI-MG – Smart Buildings and Renewable Energy Integration in Micro Grids

Partners: Politecnico di Milano, Technical University of Sofia

Dates: 09/7-13/8, 2012 and 17/12-21/12, 2012



Project: SREI-MG – Smart Buildings and Renewable Energy Integration in Micro Grids

Partners: Politecnico di Milano, Technical University of Sofia

Dates: 09/7-13/8, 2012 and 17/12-21/12, 2012


iPower – a Strategic Platform for Innovation and Research within Intelligent Electricity

The iPower Platform will develop intelligent control of decentralized power consumption and production tools to manage millions of flexible consumption units, and methods of operation of a distribution with flexible power generation.

Methods for identification of user needs and acceptance of flexible consumer units will be tested in practice


iPower is a joint project in collaboration with 32 partners, universities and research institutions as well as industrial companies from various countries. Department of Electrical Engineering at the Technical University of Denmark is the coordinator and has the overall responsibility that the iPower platform achieves the desired results.

iPower platform has been granted a support of 60 million DKK over the next 5 years from SPIR - Strategic Platform for Innovation and Research. SPIR is an initiative of the Danish Council for Strategic Research and the Danish Council for Technology and Innovation.

http://www.ipower-net.dk/


Distributed control for intelligentpower dispatch: an MPC approach

The increase of small PV installations in buildings can threaten the voltage stability of LV distribution feeders.

At the same time, buildings equipped with modern HVACs, smart energy control and EV charging posts open new possibilities for balancing the local production so as to offer voltage and regulating power services to the grid.

This approach strives at defining a control structure that allows to integrate DERs of heterogeneous type in the same Smart Building with minimum integration effort (no central intelligence).


Case study

• 5kW PV array

• 3 batteries (15kWh, 3kW)

• 1 office building (10 kW)

• 16A fuse

Objective: do not exceed fuse max current


Control architecture

Each DER is equipped with an MPC, which reads information from a database and sends back the production/consumption plan for a generic timeframe. The information exchanged is simple time series and the MPCs can have a different utility functions.

In this setup they are all soft-economic MPCs, so that they use information on price and feeder power capacity (fuse limitation).

Each MPC computes the control action so as to minimize the costs and limit the current at the fuse.


Simulation study - Building

The economic MPC runs the heating system at minimal costs. Offset-free implementation and feed forward compensation of external temperature guarantee the comfort bands are respected at any time.

Ts= 1h

Pred. hor. = 24h


Simulation study – Batteries

The economic MPC operates the battery in order to balance the other devices power and respect the constraint of 3kW at the PCC.

At the same time, being economic MPC, it minimizes the energy costs.

Ts= 1h

Pred. hor. = 24h


Simulation study – Power at the PCC


Simulation study – energy bill

Without coordination With coordination

Demand side management in the Smart Grid:a control perspective

DISCUSSION…

DSM in the Smart Grid-a control perspective - …home.deib.polimi.it/ferrarin/psc/files/DSM in the Smart Grid-a... · 8 Demand Side Management in the Smart Grid: a control perspective

Documents