MASTER THESIS Nr. 315 A SIMULATION PLATFORM FOR POWER TRADING

Department of Telecommunications

MASTER THESIS Nr. 315

A SIMULATION PLATFORM FOR POWER TRADING

Jurica Babić

Zagreb, July 2012

Department of Telecommunications

Outline

¨ Background and motivation¨ Power Trading Agent Competition (Power TAC)¨ CrocodileAgent 2012

2 of 14

Department of Telecommunications

Background and MotivationTransition from traditional to smart grids

¨ Some limitations of traditional power grids: centralized production energy losses not able to cope with intermittent and decentralized renewables

(e.g., wind turbines and solar panels) ¨ Smart grids:

ICT layer - advanced grid management new market opportunities

3 of 14

Department of Telecommunications

1. market design needed

Motivation and background (2)Smart grid as an enabler of market applications

4 of 14

Traditional grids

Smart grids

2. risk-free environment for testing needed

Department of Telecommunications

Power Trading Agent Competition (Power TAC)Introduction

¨ Power market simulator¨ International project (6 universities)¨ Purpose – test market designs for smart grids¨ Combines market agents and the element of competition¨ Software agent in a role of self-interested broker:

buys and sells energy for customers intention of earning a profit minimizes energy imbalance

5 of 14

Department of Telecommunications

Distribution utility – owns/operates local grid

(regulated monopoly)

Large Energy Suppliers

Retail CustomersProducers, Consumers, Prosumers

Power TAC

Brokersbuild portfolios,buy and sell power

Tariff Market

Sim

ulati

on e

nviro

men

tPa

rtici

pant

s

Balancing

MarketWholesale Market

Power TACGame scenario

6 of 14

CrocodileAgent 2012

Visualizer

Department of Telecommunications

CrocodileAgent 2012Overview

¨ A software agent for power trading in Power TAC¨ Emphasis was placed on:

customer market (i.e., tariff market) activities wholesale market activities forecasting using Holt-Winters method

¨ Modular design¨ Main advantage is its active tariff offering adjustments

7 of 14

Department of Telecommunications

CrocodileAgent 2012Tracks customers energy usage patterns

8 of 14

No periodicity – WindmillCoOpWeekly periodicity – OfficeComplex

Daily periodicity – CentervilleHomes

Department of Telecommunications

CrocodileAgent 2012Holt-Winters: customers energy usage forecasting

¨ = Holt-Winters (trained by minimizing NRMSE)¨ = energy usage values¨ = periodicity¨ = number of forecasted values

9 of 14

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 240

5

10

15

20

25

30

35

Usage

Predicted usage

TAC hours

Ener

gy u

sage

(MW

h)OfficeComplex: actual vs predicted values

Department of Telecommunications

CrocodileAgent 2012Activities in the customer market

¨ Chooses and applies an appropriate tariff model: fixed price multi-rate flate rate (with periodic payment*) special tariffs

¨ Evaluates existing tariffs: deriving tariff utility replacing low-rated tariffs with new ones

* amount of money each of the subscribed customers pay in each TAC hour to the broker

10 of 14

Department of Telecommunications

CrocodileAgent 2012Wholesale market in Power TAC: day-ahead market

11 of 14

1 2 3 4 5 6 7 8 9 1011121314151617181920212223242526272829303132333435363738394041424344454647480

10

20

30

40

50

TAC hours

Pric

e [€

]

Hour-ahead display of market clearings

Department of Telecommunications

CrocodileAgent 2012Activities in the wholesale market (2)

¨ = past wholesale clearing prices¨ bidding strategy: higher prices for closing markets

Energy usage forecast + wholesale price forecast + bidding strategy = wholesale order

12 of 14

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 2420

30

40

50

60

Clearing price

Predicted clearing price

Pric

e [€

]

TAC hours

Wholesale prices: actual vs predicted values

Department of Telecommunications

CrocodileAgent 2012Improving competition design

¨ CrocodileAgent 2012 has identified two main drawbacks in the current game design: infinite periodic payment

customers are ignorant toperiodic payment

PAC-MAN syndrome frequent spawn of similar

tariffs yields to a highermarket share

¨ more rounds in a yearly cycle of Power TAC needed13 of 14

Department of Telecommunications

Conclusion and Future Work

¨ Power TAC is a promising simulator for the evaluation of market design ideas

¨ CrocodileAgent design focus: customer and wholesale market activities

¨ Holt-Winters method for forecasting¨ Proposal: more rounds of Power TAC ¨ Future Work: more sophisticated portfolio and wholesale m.

management

14 of 14