Chung-Li Tseng

Harmony search for generalized orienteering problem: best touring in China

In order to overcome the drawbacks of mathematical optimization techniques, soft computing algorithms have been vigorously introduced during the past decade. However, there are still some possibilities of devising new algorithms based on analogies with natural phenomena. A nature-inspired algorithm, mimicking the improvisation process of music players, has been recently developed and named Harmony Search (HS). The algorithm has been successfully applied to various engineering optimization problems. In this paper, the HS was applied to a TSP-like NP-hard Generalized Orienteering Problem (GOP) which is to find the utmost route under the total distance limit while satisfying multiple goals. Example area of the GOP is eastern part of China. The results of HS showed that the algorithm could find good solutions when compared to those of artificial neural network.

Short-Term Generation Asset Valuation: A Real Options Approach

This paper discusses using real options to value power plants with unit commitment constraints over a short-term period. We formulate the problem as a multistage stochastic problem and propose a solution procedure that integrates forward-moving Monte Carlo simulation with backward-moving dynamic programming. We assume that the power plant operator maximizes expected profit by deciding in each hour whether or not to run the unit, that a certain lead time for commitment and decommitment decisions is necessary to start up and shut down a unit, and that these commitment decisions, once made, are subject to physical constraints such as minimum uptime and downtime. We also account for the costs associated with starting up and shutting down a unit. Last, we assume that there are hourly markets for both electricity and the fuel used by the generator and that their prices follow Ito processes. Using numerical simulation, we show that failure to consider physical constraints may significantly overvalue a power plant.

Short-term generation asset valuation

We present a method for valuing a power plant over a short term period using Monte Carlo simulation. The power plant valuation problem is formulated as a multi stage stochastic problem. We assume there are hourly markets for both electricity and the fuel used by the generator, and their prices follow some Ito processes. At each hour, the power plant operator must decide to run or not to run the unit so as to maximize expected profit. A certain lead time for commitment decision is necessary to start up a unit. The commitment decision, once made, is subject to physical constraints such as minimum uptime and downtime constraints. The generators startup cost, is also taken into account in our model. The Monte Carlo method is employed not only in forward moving simulation, but also backward moving recursion of dynamic programming. We demonstrate through numerical tests how the physical constraints affect a power plant value.

A Framework Using Two-Factor Price Lattices for Generation Asset Valuation

In this paper, we use a real-options framework to value a power plant. The real option to commit or decommit a generating unit may be exercised on an hourly basis to maximize expected profit while subject to intertemporal operational constraints. The option-exercising process is modeled as a multistage stochastic problem. We develop a framework for generating discrete-time price lattices for two correlated Ito processes for electricity and fuel prices. We show that the proposed framework exceeds existing approaches in both lattice feasibility and computational efficiency. We prove that this framework guarantees existence of branching probabilities at all nodes and all stages of the lattice if the correlation between the two Ito processes is no greater than 4/√35 ≈ 0.676. With price evolution represented by a lattice, the valuation problem is solved using stochastic dynamic programming. We show how the obtained power plant value converges to the true expected value by refining the price lattice. Sensitivity analysis for the power plant value to changes of price parameters is also presented.

A transmission-constrained unit commitment method in power system scheduling

This paper presents a transmission-constrained unit commitment method using a Lagrangian relaxation approach. Based on a DC power flow model, the transmission constraints are formulated as linear constraints. The transmission constraints, as well as the demand and spinning reserve constraints, are relaxed by attaching Lagrange multipliers. A three-phase algorithmic scheme is devised including dual optimization, a feasibility phase and unit decommitment. A large-scale test problem with more than 2200 buses and 2500 transmission lines is tested along with other test problems. q 1999 Elsevier Science B.V. All rights reserved.

Solving the unit commitment problem by a unit decommitment method

In this paper, we present a unified decommitment method to solve the unit commitment problem. This method starts with a solution having all available units online at all hours in the planning horizon and determines an optimal strategy for decommitting units one at a time. We show that the proposed method may be viewed as an approximate implementation of the Lagrangian relaxation approach and that the number of iterations is bounded by the number of units. Numerical tests suggest that the proposed method is a reliable, efficient, and robust approach for solving the unit commitment problem.

Price-based adaptive spinning reserve requirements in power system scheduling

In a deregulated electricity market such as the California WEPEX, spinning reserves must be explicitly identified as an ancillary service and priced. Additionally, scheduling coordinators who match suppliers and demands may either self-provide spinning reserves, or rely on the Independent System Operator (ISO) to provide reserves at the spot price. The deregulated market structure makes explicit the implicit softness that has always been recognized in the reserve constraints: additional reserves may have value even when a minimum reserve requirement has been met. In this paper we formulate the spinning reserve requirement (SRR) as a function of the endogenously determined marginal values of reserves. The spinning reserve requirement depends, according to a non-increasing response function, on a price/value signal. We present three power system scheduling algorithms in which this price/value signal is updated at each iteration of a dual optimization. Game theory is used to interpret the proposed algorithms. Numerical test results are also presented.

Harmony search for structural design

Various algorithms have been developed and applied to structural optimization, in which cross-sectional areas of structure members are assumed to be continuous. In most cases of practical structure designs, however, decision variables (cross-sectional areas) are discrete. This paper proposes a combinatorial optimization model for structural design using a new nature-inspired algorithm, harmony search (HS). HS is also compared to genetic algorithms through a standard truss example. Numerical results reveal that the proposed HS is a powerful search algorithm for combinatorial structure optimization.

Trenchless Water Pipe Condition Assessment Using Artificial Neural Network

In order to assess the water pipe condition without excavating, artificial neural network (ANN) model was developed and applied to real-world case in South Korea. For the input in this ANN model, 11 factors such as (1) pipe material, (2) diameter, (3) pressure head, (4) inner coating, (5) outer coating, (6) electric recharge, (7) bedding condition, (8) age, (9) trench depth, (10) soil condition, and (11) number of road lanes were used; and, for the output, overall pipe condition index was derived based on 5 factors such as (1) outer corrosion, (2) crack, (3) pin hole, (4) inner corrosion, and (5) H-W C value. For the ANN computing, each factor was normalized into the range of 0 to 1. The ANN model could find better results than those of multiple regression model in terms of statistical correlation between observed and computed data.

A unit decommitment method in power system scheduling

This paper presents a unit decommitment method for power system scheduling. Given a feasible unit commitment, our algorithm determines an optimal strategy for decommitting overcommitted units based on dynamic programming. This method is being developed as a possible post-processing tool to improve the solution quality of the existing unit commitment algorithm used at PG&E. It can also be integrated into any other unit commitment method or used as a complete unit commitment algorithm in itself. The decommitment method can also be used as a tool to measure the solution quality of unit commitment algorithms. The proposed method maintains solution feasibility at all iterations. In this paper we prove that the number of iterations required by the method to terminate is bounded by the number of units. Numerical tests indicate that this decommitment method is computationally efficient and can improve scheduling significantly.