Juan Guillermo Lazo

Decision Support Methods

This section presents a summary of the main concepts on which evolutionary algorithms are based. First, the operating principle of Genetic Algorithms (GAs) is explained and their main parts and their evolution parameters described. Next, a description of Cultural Algorithms (CAs) is presented and its main components are pointed out.

Portfolio Selection and Management Using a Hybrid Intelligent and Statistical System

This paper presents the development of a hybrid system based on Genetic Algorithms, Neural Networks and the GARCH model for the selection of stocks and the management of investment portfolios. The hybrid system comprises four modules: a genetic algorithm for selecting the assets that will form the investment portfolio, the GARCH model for forecasting stock volatility, a neural networks for predicting asset returns for the portfolio, and another genetic algorithm for determining the optimal weights for each asset. Portfolio management has consisted of weekly updates over a period of 49 weeks.

Real Options Theory

Economic investment decisions, such as purchasing new equipment, increasing the work force or developing new products, as well as project economic valuation, are affected by economic uncertainty, technical uncertainty and by the managerial flexibilities embedded in the project. Economic uncertainty is caused by factors external to the project and is generally represented by stochastic oscillations in product prices and by costs. Technical uncertainty is caused by internal factors, such as uncertainty regarding the production size and the project’s performance as a result of the technologies employed. The managerial flexibilities that are built into projects give managers the freedom to make decisions such as to invest, to expand, temporarily shut down or abandon a given project. Such flexibilities are called real options. If any one of these possibilities is ignored in the economic analysis, the project may perhaps be under-assessed and this may lead to irreversible decision-making errors. Therefore, managerial flexibility has a value which is not taken into account by conventional techniques such as the net present value (NPV) and the internal return rate (IRR) techniques. In addition to uncertainty, real options also consider managerial flexibility and their objective is to maximize the investment opportunity value.

Evolutionary Computation for Valves Control Optimization in Intelligent Wells Under Uncertainties

This work presents a new decision support system for intelligent wells control considering technical uncertainties. The intelligent control of valves operation tends to become a competitive advantage for reservoirs development. Such control refers to the opening and shutting of the valves that distinguish the intelligent wells. The strategy consists in identifying a valve configuration that maximizes the net present value. The developed system uses Genetic Algorithms, reservoir simulation, Monte Carlo simulation, techniques of sampling variance reduction and uncertainties representation by probability distribution and geologic sceneries. The theoretical concepts applied and the implementation of a system capable of supporting, managing and developing the intelligent fields, constitute an advance to petroleum exploration area. The obtained results demonstrate that the approach given to the problem and the used methodologies deal with the control valves in an efficient and practical way.

Real Options and Genetic Algorithms to Approach of the Optimal Decision Rule for Oil Field Development Under Uncertainties

A decision to invest in the development of an oil reserve requires an in-depth analysis of several uncertainty factors. Such uncertainties may involve either technical uncertainties related to the size and economic quality of the reserve, or market uncertainties. When a great number of alternatives or options of investment are involved, the task of selecting the best alternative or a decision rule is very important and complex due to the considerable number of possibilities and parameters that must be taken into account. This paper proposes a new model, based on Real Option Theory, Genetic Algorithms and on Monte Carlo simulation to find an optimal decision rule for alternatives of investment regarding the development of an oil field under market uncertainty that may help decision-making in the following situation: immediate development of a field or wait until market conditions are more favorable. This optimal decision rule is formed by three mutually exclusive alternatives, which describe three exercise regions through time, up to the expiration of the concession of the field. The Monte Carlo simulation is employed within the genetic algorithm to simulate the possible paths of oil prices up to the expiration date. The Geometric Brownian Motion is assumed as stochastic process for represents the oil price. A technique of variance reduction was also used to improve the computational efficiency of the Monte Carlo simulation.

Estimation of the displacements among distant events based on parallel tracking of events in seismic traces under uncertainty

The method we propose in this paper seeks to estimate interface displacements among strata related with reflection seismic events, in comparison to the interfaces at other reference points. To do so, we search for reflection events in the reference point of a second seismic trace taken from the same 3D survey and close to a well. However, the nature of the seismic data introduces uncertainty in the results. Therefore, we perform an uncertainty analysis using the standard deviation results from several experiments with cross-correlation of signals. To estimate the displacements of events in depth between two seismic traces, we create a synthetic seismic trace with an empirical wavelet and the sonic log of the well, close to the second seismic trace. Then, we relate the events of the seismic traces to the depth of the sonic log. Finally, we test the method with data from the Namorado Field in Brazil. The results show that the accuracy of the event estimated depth depends on the results of parallel cross-correlation, primarily those from the procedures used in the integration of seismic data with data from the well. The proposed approach can correctly identify several similar events in two seismic traces without requiring all seismic traces between two distant points of interest to correlate strata in the subsurface.

Processo Estocástico Neural Aplicado em Séries de Afluências Mensais

The generic model of stochastic process based on neural networks, called Neural Stochastic Process (NSP), was applied to the treatment of series of monthly inflows. These series correspond to Affluent Natural Energy (ANE), which is the aggregation of the inflows to the plants, comprising a reservoir equivalent of a subsystem of National Interconnected System (NIS). The series of ANE presents temporal correlation and spatial correlation. The NSP model in its original version can capture the temporal correlation, however, does not incorporate the spatial correlation of the series. This paper presents a variant of the NSP model aimed at the incorporation of spatial correlation of the series of ANE. The results indicated that the model is able to capture the behavior of the time series of all NIS subsystems, providing different scenarios for the next 5 years that embody the same temporal and spatial correlation of the historical data.

Análise de um modelo inteligente de contratação de energia elétrica no curto prazo para distribuidoras

In this paper is presented a new approach of an intelligent model for optimization under uncertainty to determine the best strategy of electricity trading in the short term (referring to A-1 and Adjustment auctions) for distribution companies. This model reproduces all the rules of purchase/sale of energy for a distribution company and the transfer of this cost to the final tariff of the consumers. The optimization process uses genetic algorithm, and seeks to minimize the cost associated with the purchase of energy, penalty for subcontracting and the cost of trade (purchase/sale) energy by the spot price. The optimal trading is obtained considering several load scenarios, obtained by Monte Carlo simulation, for a period of five years of analysis. The decisions of trading are taken in the first two years in that period. The evaluation of the model results is done by means of a combination between the expected value of the distribution of costs and the CVaR (Conditional Value at Risk), for the different load scenarios. The model also uses the PLD_robust, which seeks to minimize the exposure of the distribution company in the spot price. To illustrate the results of the proposed model, a study case based on realistic data is presented. The results obtained are compared to the results obtained with the trading of energy without using the optimization model presented in this paper. That comparison is done to verify how much the proposed method can be better than the solutions based on intuitive analysis. In addition, further analysis is performed by considering two mechanisms of compensation of the surpluses and deficits of contracts, named MCSD4% and MCSD_Ex-post, established by ANEEL to reduce the risks associated to the energy trading to the distribution companies.

Real Option Value Calculation by Monte Carlo Simulation and Approximation by Fuzzy Numbers and Genetic Algorithms

This chapter describes, in two parts, the methodology proposed for obtaining an approximation of the real option value and of the optimal decision rule for several project investment options by considering technical and market uncertainty.

High-Performance Processing for E&P

This chapter addresses the modeling and development of the modules that use distributed processing systems so as to achieve a high performance in processes for the optimization of alternatives [1] [2] [3] (GA, CA, Cooperative Coevolution Algorithm, Schedule Optimization), for distributed simulations of “random” alternatives, for case simulations and for distributed Monte Carlo simulations.