Mario Vanhoucke

A genetic algorithm for the preemptive and non-preemptive multi-mode resource-constrained project scheduling problem

In this paper we present a genetic algorithm for the multi-mode resource-constrained project scheduling problem (MRCPSP), in which multiple execution modes are available for each of the activities of the project. We also introduce the preemptive extension of the problem which allows activity splitting (P-MRCPSP). To solve the problem, we apply a bi-population genetic algorithm, which makes use of two separate populations and extend the serial schedule generation scheme by introducing a mode improvement procedure. We evaluate the impact of preemption on the quality of the schedule and present detailed comparative computational results for the MRCPSP, which reveal that our procedure is amongst the most competitive algorithms.

A hybrid scatter search/electromagnetism meta-heuristic for project scheduling

In the last few decades, several effective algorithms for solving the resource-constrained project scheduling problem have been proposed. However, the challenging nature of this problem, summarised in its strongly NP-hard status, restricts the effectiveness of exact optimisation to relatively small instances. In this paper, we present a new meta-heuristic for this problem, able to provide near-optimal heuristic solutions. The procedure combines elements from scatter search, a generic population-based evolutionary search method, and a recently introduced heuristic method for the optimisation of unconstrained continuous functions based on an analogy with electromagnetism theory, hereafter referred to as the electromagnetism meta-heuristic. We present computational experiments on standard benchmark datasets, compare the results with current state-ofthe- art heuristics, and show that the procedure is capable of producing consistently good results for challenging instances of the resource-constrained project scheduling problem. We also demonstrate that the algorithm outperforms state-of-the-art existing heuristics.

RanGen: A Random Network Generator for Activity-on-the-Node Networks

In this paper, we describe RanGen, a random network generator for generating activity-on-the-node networks and accompanying data for different classes of project scheduling problems. The objective is to construct random networks which satisfy preset values of the parameters used to control the hardness of a problem instance. Both parameters which are related to the network topology and resource-related parameters are implemented. The network generator meets the shortcomings of former network generators since it employs a wide range of different parameters which have been shown to serve as possible predictors of the hardness of different project scheduling problems. Some of them have been implemented in former network generators while others have not.

A Decomposition-Based Genetic Algorithm for the Resource-Constrained Project-Scheduling Problem

In the last few decades, the resource-constrained project-scheduling problem has become a popular problem type in operations research. However, due to its strongly NP-hard status, the effectiveness of exact optimisation procedures is restricted to relatively small instances. In this paper, we present a new genetic algorithm (GA) for this problem that is able to provide near-optimal heuristic solutions. This GA procedure has been extended by a so-called decomposition-based genetic algorithm (DBGA) that iteratively solves subparts of the project. We present computational experiments on two data sets. The first benchmark set is used to illustrate the performance of both the GA and the DBGA. The second set is used to compare the results with current state-of-the-art heuristics and to show that the procedure is capable of producing consistently good results for challenging problem instances. We illustrate that the GA outperforms all state-of-the-art heuristics and that the DBGA further improves the performance of the GA.

A simulation and evaluation of earned value metrics to forecast the project duration

In this paper, we extensively review and evaluate earned value (EV)-based methods to forecast the total project duration. EV systems have been set up to deal with the complex task of controlling and adjusting the baseline project schedule during execution, taking into account project scope, timed delivery and total project budget. Although EV systems have been proven to provide reliable estimates for the follow-up of cost performance within our project assumptions, they often fail to predict the total duration of the project. We present an extensive simulation study where we carefully control the level of uncertainty in the project, the influence of the project network structure on the accuracy of the forecasts and the time horizon where the EV-based measures provide accurate and reliable results. We assume a project setting where project activities and precedence relations are known in advance and do not consider fundamentally unforeseeable events and/or unknown interactions among various actions that might cause entirely unexpected effects in different project parts. This is the first study that investigates the potential of a recently developed method, the earned schedule method, which improves the connection between EV metrics and the project duration forecasts.

An evaluation of the adequacy of project network generators with systematically sampled networks

This paper evaluates and compares different network generators to generate project scheduling problem instances based on indicators measuring the topological network structure. We review six topological network indicators in order to describe the detailed structure of a project network. These indicators were originally developed by [L.V. Tavares, J.A. Ferreira and J.S. Coelho, The risk of delay of a project in terms of the morphology of its network, European Journal of Operational Research 119 (1999), 510–537] and have been modified, or sometimes completely replaced, by alternative indicators to describe the network topology. The contribution of this paper is twofold. Firstly, we generate a large amount of different networks with four project network generators. Our general conclusions are that none of the network generators are able to capture the complete feasible domain of all networks. Additionally, each network generator covers its own network-specific domain and, consequently, contributes to the generation of data sets. Secondly, we perform computational results on the well-known resource-constrained project scheduling problem to prove that our indicators are reliable and have significant, predictive power to serve as complexity indicators.

On Maximizing the Net Present Value of a Project Under Renewable Resource Constraints

In this paper we study the resource-constrained project-scheduling problem with discounted cash flows. Each activity of this resource-constrained project-scheduling problem has certain resource requirements and a known deterministic cash flow that can be either positive or negative. Deterministic cash flows are assumed to occur over the duration of the activities. Progress payments and cash outflows occur at the completion of activities. The objective is to schedule the activities subject to a fixed deadline to maximize the net present value subject to the precedence and resource constraints. With these features the financial aspects of project management are taken into account.We introduce a depth-first branch-and-bound algorithm that makes use of extra precedence relations to resolve a number of resource conflicts and a fast recursive search algorithm for the max- npv problem to compute upper bounds. The recursive search algorithm exploits the idea that positive cash flows should be scheduled as early as possible while negative cash flows should be scheduled as late as possible within the precedence constraints. The procedure has been coded in Visual C++, Version 4.0 under Windows NT, and has been validated on two problem sets.

Multi-mode resource-constrained project scheduling using RCPSP and SAT solvers

This paper reports on a new solution approach for the well-known multi-mode resource-constrained project scheduling problem (MRCPSP). This problem type aims at the selection of a single activity mode from a set of available modes in order to construct a precedence and a (renewable and non-renewable) resource feasible project schedule with a minimal makespan. The problem type is known to be NP-hard and has been solved using various exact as well as (meta-)heuristic procedures. The new algorithm splits the problem type into a mode assignment and a single mode project scheduling step. The mode assignment step is solved by a satisfiability (SAT) problem solver and returns a feasible mode selection to the project scheduling step. The project scheduling step is solved using an efficient meta-heuristic procedure from literature to solve the resource-constrained project scheduling problem (RCPSP). However, unlike many traditional meta-heuristic methods in literature to solve the MRCPSP, the new approach executes these two steps in one run, relying on a single priority list. Straightforward adaptations to the pure SAT solver by using pseudo boolean non-renewable resource constraints has led to a high quality solution approach in a reasonable computational time. Computational results show that the procedure can report similar or sometimes even better solutions than found by other procedures in literature, although it often requires a higher CPU time.

WORK CONTINUITY CONSTRAINTS IN PROJECT SCHEDULING

Repetitive projects involve the repetition of activities along the stages of the project. Since the resources required to perform these activities move from one stage to the other, a main objective of scheduling these projects is to maintain the continuity of work of these resources so as to minimize the idle time of resources. This requirement, often referred to as work continuity constraints, involves a trade-off between total project duration and the resource idle time. The contribution of this paper is threefold. Firstly, we provide an extensive literature summary of the topic under study. Although most research papers deal with the scheduling of construction projects, we show that this can be extended to many other environments. Secondly, we propose an exact search procedure for scheduling repetitive projects with work continuity constraints. This algorithm iteratively shifts repeating activities further in time in order to decrease the resource idle time. We have embedded this recursive search procedure in a horizon-varying algorithm in order to detect the complete trade-off profile between resource idle time and project duration. The procedure has been coded in Visual C++ and has been validated on a randomly generated problem set. Finally, we illustrate the concepts on three examples. First, the use our new algorithm is illustrated on a small fictive problem example from literature. In a second example, we show that work continuity constraints involve a tradeoff between total project duration and the resource idle time. A last example describes the scheduling of a well-known real-life project that aims at the construction of a tunnel at the Westerschelde in the Netherlands.

An Exact Procedure for the Resource-Constrained Weighted Earliness–Tardiness Project Scheduling Problem

In this paper we study the resource-constrained project scheduling problem with weighted earliness–tardinesss penalty costs. Project activities are assumed to have a known deterministic due date, a unit earliness as well as a unit tardiness penalty cost and constant renewable resource requirements. The objective is to schedule the activities in order to minimize the total weighted earliness–tardinesss penalty cost of the project subject to the finish–start precedence constraints and the constant renewable resource availability constraints. With these features the problem becomes highly attractive in just-in-time environments.We introduce a depth-first branch-and-bound algorithm which makes use of extra precedence relations to resolve resource conflicts and relies on a fast recursive search algorithm for the unconstrained weighted earliness–tardinesss problem to compute lower bounds. The procedure has been coded in Visual C++, version 4.0 under Windows NT. Both the recursive search algorithm and the branch-and-bound procedure have been validated on a randomly generated problem set.