Search

A new study analyses the optimal management of an ecological-economic systems, focusing on resource allocation in the aquatic system Chesapeake Bay.

Operations research is applied mathematics which can aid in solving business operational and management problems. It is a business strategy which can be used to determine how best to allocate resources such as money, parts, transportation and labor, among others. Such approach works by dissecting the problem and finding new and different perspectives of looking at the details of the matter.

The material flow process in the dashboard area and the mechanic area of an automated automobile assembly facility was characterized as part of a cost benefit analysis of several dynamic dispatching rules. Consequently, simulation experiments under varying production conditions were performed using WITNESS. The results demonstrated the superiority of the matching rule over the push rule with respect to work-in-process minimization.

The application of an intelligent control system based on Benders decomposition and importance sampling to a large-scale real-world problem with uncertainty is discussed. The problem, which involves the short-term control of a hydroelectric power plant, is modeled as a multi-stage stochastic linear program to address its combinatorial complexity. Consequently, adaptive control via automatic rule generation, rule modification and learning is achieved.

The potential demand for SNCF's planned Autoroute Ferroviaire rail service from diverted road and ferry traffic was investigated. To estimate the potential traffic, a multinomial logit model for optimizing the cost per journey and the use of time with respect to the volume and the shortest route was employed. The results indicated that the rail offering, which will connect France with Italy, Spain and the UK, will be economically viable despite its large start-up costs.

The Housing Analysis Decision Technology System (HADTS) was developed to help the US Army in making housing decisions related to onpost construction or offpost leases. To help the army provide housing to personnel within a 30-mile radius of a military base, HADTS includes an executive information system, a geographical information system, and an enhanced version of the Army's Housing Analysis System decision support tool.

Two case studies demonstrating the application of mixed-integer linear programming to environmental integrated production and recycling planning problems in Germany are discussed. The first involves the recycling of waste from dismantled buildings in the Upper Rhine Valley region. The second involves the Strategic Recycling Management System based on multistage capacitated warehouse location problems for recycling the byproducts of iron and steel mills.

The Daewoo Shipbuilding Scheduling (DAS) system for integrated production planning and scheduling was jointly developed by the shipbuilder and Korea Advanced Institute of Science and Technology. Consisting of several subsystems, it is based on a combination of operations research and artificial intelligence methods such as hierarchical architecture, constrained directed graph search and neural network estimation.

Computational experiments were performed to schedule the assignment of locomotives to train segments for a large railroad network. The nonlinear integer programming problem, which was characterized as a multi-commodity network flow problem and decomposed into many subproblems, was solved using a branch-and-bound procedure involving Dantzwig-Wolfe decomposition. The results indicated an improvement of seven percent over existing solutions using actual data from Canadian National North America.

CIBA Ltd. operations researchers employed a linear programming approach to help the company's crop protection division minimize its production operating costs. Their investigation looked at volume-dependent and volume-independent costs associated with production, storage and transportation. The solution, which was obtained by modifying the production- and packaging-line product assignments and reducing the number of formulation lines, led to 25% savings in annual production costs.

An environmental integrated operative planning model (OPM) was developed for ICI Inc.'s AMV process of ammonia synthesis. The nonlinear optimization model, which was constructed using the ASPEN PLUS process simulator, takes into consideration emission taxes as well as the byproducts and residues while generating input-output relations using nonlinear thermodynamic formulas. Its results, which demonstrate higher accuracy over other linear programming models, indicated the feasibility of enhancing the operating margin of ammonia production using the AMV process.

Operations research (OR) methodologies continue to be applied to solve real-world problems faced by the telecommunications, energy, transportation, health care and manufacturing industries. Moreover, OR is helping organizations to make better decisions and find better solutions to production planning, budgeting, scheduling, risk management and other manufacturing-related problems. This may be seen in a review of past winners of the Franz Edelman Award and the INFORMS Prize.

A nonlinear multi-commodity network flow model and a branch-and-bound procedure based on Dantzig-Wolfe decomposition were employed to reformulate and solve France's crew pairing problem within a set partitioning problem, respectively. This approach, which used resource constrained shortest path subproblems to generate all feasible pairings, produced significant improvements over solutions generated by an expert system being used by the airline.

Computational experiments involving that of scheduling the shutdown of nuclear power plant for refueling and maintenance activities are discussed. Single-site and multi-site problems were considered under site and demand constraints. The problem, which becomes difficult to solve as the number of reactors increases, may be solved using a branch-and-bound search, a price-directed decomposition method and other mixed-integer programming techniques.

Sabre Decision Technologies' Crew_Solve optimization algorithm for solving American Airlines' crew pairing problem enabled the airline to minimize costs by improving crew utilization and reducing pay and credit. The procedure, which recast the problem as a set partitioning zero-one integer program, employs linear program relaxations to trim the search tree and a graph-based branching heuristic to a 15,000-column set partitioning subproblem.

Operations research has been implemented many times to solve large-scale real-world problems. Despite its reputation for being a largely academic exercise grounded in mathematics, it is currently being used to derive more efficient, environmentally friendly and feasible solutions to scheduling, planning and control problems in the nuclear power, telecommunications, airline, and shipbuilding industries.

A Lagrangian heuristic of complexity O(m.f) for solving the NP-complete set covering problem (LHSCP) is introduced. The heuristic employs standard Lagrangian duality, greedy heuristic, redundant covers and subgradient optimization techniques. According to test runs of 65 benchmark problems on a Sun Sparc station of the SUN-Pascal-coded heuristic, it is efficient for SCPs with low density and large number of variables.

An algorithm for solving the minmax Steiner tree problem in graphs (SPG) in O(/E/) time where E is the set of edges in the graph is introduced. The algorithm combines two existing procedures for solving the bottleneck problems of the shortest path and spanning tree types, respectively. Likewise, an O(n3) algorithm that obtains the partial sum shortest paths for an undirected graph and an O(/E? + n log n) algorithm for solving single-node two-sum shortest path problems are described.

A block sandwich algorithm for convex curve approximation that uses two convex piecewise linear functions is introduced. The procedure, which is illustrated by approximating the efficient frontier of a bi-criteria convex quadratic network program, features a quadratic convergence property which holds the total number of scalar optimization problems required to reach a specified error is less than the constant multiple of the square root of the inverse of the given error.

An interior point approach to solving linear programming relaxations of cutting problems attained through the application of branch and cut techniques for integer programming problems. The method involved fixing variables using the dual feasible solution and generating Gomory-like cutting planes and lift-and-project cutting planes from the objective function without resolving the relaxation to optimality.

The vector-valued SPV and SDV problems were defined as a pair of dual symmetric multiobjective programming problems under various convexity assumptions to fill a gap in the literature on symmetric duality in nonlinear programming. Their formulation resulted in the strong and weak duality theorems for pseudo-convex and pseudo-concave functions .

The topological connectedness of the set of efficient solutions to multiple criteria optimization problems defined using a linear programming formulation was investigated. Specifically, computational experiments were carried out to show an exemption to the connectedness of efficient solutions to the shortest path problem and to prove the non-connectedness of the spanning tree problem.

The application of the linear mean value cross decomposition method for solving linear programming problems to deriving lower bounds on the optimal objective function value of mixed and pure integer programming problems is considered. The results, which showed the method to be superior over the LP-relaxation and the Lagrangean relaxation in a branch-and-bound setting, affirmed its use for obtaining the Lagrangean dual bound, the bound of the Lagrangean dual of the LP-relaxation of the x-part and the LP-relaxation bound.

A Lagrangean decomposition method for solving the problem of scheduling a fleet of vehicles to transport items from one point to another at minimal cost is introduced. The solution approach, which reduces the problem into trip planning and trip assignment subproblems, derives a suboptimal solution via two implementations, a fixed charge transshipment heuristic and a transportation and scheduling heuristic.

Choice constraints and exclusion constraints were used to characterize and solve, via integer programming, NP-hard finite constraint satisfaction problems. The solution approach is illustrated through its applications to the n-size conjunctive normal form satisfaction problem, propositional satisfiability (SAT) problem, the quadratic assignment problem, the traveling salesman problem, and the minimal set covering problem.

The applicability of modififed genetic algorithms (GA) for solving multiobjective zero-one programming problems with fuzzy objective functions is considered as a means of taking into account the imprecise nature of a decision maker's judgments. The revised GAs, which do not use the penalty function for characterizing the constraints, were evaluated using the conjunctive function in each fitness value of individuals as the criterion.

A Lagrangean heuristic for solving capacitated facility location problems with staircase-like production costs is developed and compared with the ADD, convex piecewise linearization (CPL) and LAMPS heuristics. The heuristic employs Lagrangean relaxation and subgradient optimization to derive initial dual solutions as well as CPL of the staircase cost functions to derive upper and lower bounds on the optimal objective function value. It is also used as a base to build a primal heuristic.

The Heuristic Concentration methodology for enhancing the effectiveness of heuristic solutions to combinatorial optimization problems is introduced. The two-stage strategy, which is illustrated using the p-median problem, constructs a concentrated solution set (CS) during the first stage by exploiting information from multiple iterations of an interchange heuristic. It then derives the optimal solution from the restricted CS during the second stage using an exact procedure.

P-facility minsum location problems on a plane and on a network of rapid transit lines which may be utilized as a shortest-path alternative transportation system to minimize travel times and costs are considered. Specifically, the nonconvex optimization problem is reduced to Weber or multi-Weber problems which, in some cases, turn into combinatorial optimization problems or, if the gauge used to induce the metric was polyhedral, p-median problems.

Audit staff scheduling problems that take into account the overlapping relationships between audit activities and the changeover times and costs for the auditors are discussed. Specifically, the problem is characterized as an integer linear program (ILP) based on an activity network with lead/lag relationships to minimize the number of decision variables. The ILP model, which is implemented in a four-auditor two-engagement example, results in lower cost schedules and audits.

Lexicographic goal programming was employed to derive the weights of alternatives from inconsistent interval comparison matrices associated with the application of the Analytic Hierarchy Process to multicriteria decision making problems. In addition, a new algorithm for identifying and changing inconsistent bounds was developed and illustrated via several examples.

A basic cost minimization framework for determining the marginal costs and minimum cost capacity plans of a service provider offering deferred service and facing stochastic demand is developed based J.S. Boronico's reliability constrained capacity planning approach. The results show that changes in the demand distribution's variability influences the optimal capacity through the regular and overtime unit cost parameters associated with delivery of the service.

Depth-first search, Hill-climbing and dynamic programming methods were integrated into a hybrid algorithm for solving unconstrained two-dimensional guillotine cutting (UTDGC) problems. The new approach reduced the complexity of large problems for both the weighted and unweighted cases by restricting the depth of the developed tree and deriving lower and upper bounds for each subproblem. This was demonstrated in comparisons with algorithms developed by R. Morabito and colleagues and J.E. Beasley.

H.M. Kwon and D.S. Bai's two-stage screening procedure with the objective function of minimizing the expected total costs per item and the average outgoing quality is extended to situations where the performance variable is bounded by upper and lower specification limits. The procedure, which was modified for application to cases with known and unknown parameters, employs a variable to sequentially inspect the items and the rejected items during the first and second stages.

The influence of imprecise market information and network externalities on a firm's decision to switch between two substitutable, nonproprietary technological standards was characterized as a stochastic process. The results, which model the number and position of market equilibria in terms of the stochastic attractiveness of each standard, revealed that network externalities induce only a temporary lock-in to one of multiple equilibria in market share.

The weighted maximal clique problem (MCPb) without node weights, otherwise known as edge-weighted MCPb, is modeled anew by using additional variables to represent the set of nodes and natural variables to represent the set of edges. The reformulation facilitates the application of a projection method to the inequalities to obtain new families of facet-defining inequalities for polytope associated with natural variables and, eventually, derive, a strong cutting-plane procedure for solving MCPb.

The joint economic lot size problem with multiple price breaks is solved within a fuzzy mathematical programming framework. The problem involves computing the number of price breaks that a seller can offer, the quantity discount and order quantity at each price break that simultaneously optimizes the joint cost to the seller and buyer from the point of view of both the seller and the buyer.

An inventory policy for items which deteriorate at a constant rate is introduced. The model, which is based on T.L. Urban's 1992 model, relaxes the condition of zero inventory at the end of the inventory cycle. It controls replenishment of the items by assuming an inventory-level-dependent demand pattern followed by a constant demand rate for the items. The sensitivity of its decision variables to modifications in the parameter values is likewise examined.

The combined Shewhart-Cumulative Sum (CUSUM) quality control scheme was extended into a double CUSUM scheme for monitoring out-of-control states in a manufacturing process. The construction of the double CUSUM scheme was achieved by adapting the average run length to those of a set of locally optimal standard CUSUM strategies. This entailed the use of an extrapolation procedure for evaluating the run length distribution of the double CUSUM scheme.

A procedure for optimally exploiting the flexibility of group technology is introduced. The procedure includes a multiple-period planning model and a grouping algorithm that facilitates the reconfiguration of flexible manufacturing cells (FMC) through a decomposition method. This enables the best part-machine combination to be generated via the selection and grouping of parts into parts families and machines into smaller FMCs.

The problem of minimizing the number of setups is encountered as a subset of the one-dimensional cutting-stock, deckling or trim loss problem. Although formal constraints can be imposed, they do not guarantee the best solution but only provide the necessary conditions for accomplishing the task. Moreover, sufficient conditions can only be derived if all possible patterns are generated.

A procedure for determining the level of media expenditure that optimizes the rate of return of a fixed-budget advertising campaign with respect to advertising effectiveness is introduced. The approach, which looks at advertising as an investment instead of a cost within a media planning context, correlates the reach and effective reach functions with GRPs to show that a maximal return on investment may be obtained with less than 100% of the allocated budget.

A decomposition procedure for optimizing the quality of a multiple-stage manufacturing line with inspection stations is introduced. The model, which decomposes the system into two models representing no inspection between stages and 100% inspection after each stage, looks at process improvement as a step-wise yield improvement task that maximizes the probability of successfully processing an item.

The problem of optimally hedging output in the presence of output price uncertainty faced by an expected-utility-for-profit-maximizing manufacturer was examined within a theoretical and empirical framework. Specifically, the envelope theorem and indirect expected utility function were employed to derive estimating equations via Taylor series approximation. The model's application to cattle data, which indicated the decreasing absolute risk aversion of live cattle farmers, gave results that were consistent with theoretical predictions.

A sequential mixed-integer programming procedure for solving discontinuous tour scheduling problems expressed as general set-covering formulations is introduced. The heuristic is better than linear programming relaxation-based methods because it requires modest hardware and software capabilities and provides near-optimal or optimal solutions to full-time workforce and mixed-workforce problems.

A review of literature published between 1976 and 1996 shows the increasing application of risk analysis techniques to assess the quality and reliability of software during its development. The detection and modeling of software faults is important in light of the growing use of software probabilistic risk analysis programs safety critical situations in the nuclear power and chemical industries.

A structural analysis of the production possibility set under the generalized data envelopment analysis (DEA) model proposed by G. Yu, Q.L. Wei and P. Brockett was carried out. Subsequently, an efficient tree search procedure for deriving the maximal subsets of decision-making units that lie on the same DEA-efficient surfaces was developed. The results indicated the model's suitability in production efficiency analysis such as the analysis of Pareto solutions to multiple criteria programming problems.