Paper ID: 402 Simulating Electricity Spot Prices in Brazil Using Neural Network and Design of Experiments 1 Simulating Electricity Spot Prices in Brazil Using Neural Network and Design of Experiments A.R. Queiroz, F.A. Oliveira, J.W. Marangon Lima, Senior Member, IEEE and P.P. Balestrassi Abstract— The electricity price has been one of the most important variables since the introduction of deregulation on the electricity sector. On this way, efficient forecasting methods of spot prices have become crucial to maximize the agent benefits. In Brazil the electricity price is based on the marginal cost provided by an optimization software (NEWAVE). Forecasting the Operational Marginal Cost (OMC) and its volatility has been one big problem in the Brazilian market because of the computational time taken by this software. This work presents a fast and efficient model to simulate the OMC using DOE (Design of Experiments) and ANN (Artificial Neural Networks) techniques. The paper proved that the combined techniques provided a promising result and may be applied to risk management and investment analysis. Index Terms— Electricity Prices, Simulation, Design of Experiments and Artificial Neural Networks I. INTRODUCTION URING the two last decades many transformations occurred on the electrical power systems of many countries, with the objective of introducing market mechanisms. The figure of the electricity market has emerged with the introduction of deregulation on the electricity sector which has turned the electricity price one of the most important variables. Actually, the basic objective of deregulation consists in maximizing the efficiency in electricity generation and transmission in order to reduce electricity prices. In this way, efficient estimating methods of electricity spot prices became crucial for the agents. Generators and consumers need a precise future price information to establish their bidding strategies to maximize their benefits. In Brazil, the power system restructuring process introduced new agents, such as: the National Regulatory Agency (ANEEL), the Independent System Operator (ONS), the Wholesale Electricity Market (CCEE) and the Energy Planning Institution (EPE). More recently it was enacted in 2004 the Law 10848 that established new rules for the Brazilian Wholesale Market. One of the major changes was the introduction of two trade environment: Regulated trade environment (RTE) and Free trade environment (FTE). The regulated trade was designed to the captive consumers that are represented by the distribution companies. The ANEEL and CCEE conduct buying auctions in a centralized way in the behalf of the distribution companies. The prices at FTE named MCP (Market Clearing Prices) are set by the marginal cost of the energy derived from the optimization program (NEWAVE) [1-2]. The MCP is the basis for the bilateral contracts at the FTE and for the auction bids at RTE because the contracted differences at both markets are prized by MCP. As mentioned before, the electricity price is defined by the OMC (Operation Marginal Cost) derived from NEWAVE. This procedure significantly differs from most of other electricity markets around the world. This special market structure was designed because of the great portion of hydro generation in Brazil corresponding to 90% of the total generation. NEWAVE provides the OMC as the Lagrange coefficient of the load balance constraint. The objective is to minimize the operational costs considering the operation of the power plant reservoirs. A stochastic dynamic programming is used because present decisions affect the future costs and the overall optimization. Therefore, the NEWAVE is the core of price computation which is very important at returns and risks of investment portfolios and of selling and buying contracts assessment. However, this computer program is time consuming and one simulation usually takes four hours long (in a Pentium IV, 2 GHz with 1Gbyte of memory). Many tools regarding risk and return analysis use Monte Carlo Simulation (MCS) which implies a great number of NEWAVE simulations in the Brazilian case. Then, the first step to forecast NEWAVE market prices in Brazil is to solve the optimization problem in few seconds. In other words, it is necessary to find a fast substitute for the NEWAVE. This paper will show the design of this substitute built using Design of Experiments (DOE) [3] and Artificial Neural Networks (ANN) techniques [4-5]. The Figure 1 shows the combined methodologies applied to compute the OMC. D