Broos Maenhout

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.

Branching strategies in a branch-and-price approach for a multiple objective nurse scheduling problem

The efficient management of nursing personnel is of critical importance in a hospital’s environment comprising a vast share of the hospital’s operational costs. The nurse scheduling process affects highly the nurses’ working conditions, which are strongly related to the provided quality of care. In this paper, we consider the rostering over a mid-term period that involves the construction of duty timetables for a set of heterogeneous nurses. In scheduling nursing personnel, the head nurse is typically confronted with various (conflicting) goals complying with different priority levels which represent the hospital’s policies and the nurses’ preferences. In constructing a nurse roster, nurses need to be assigned to shifts in order to maximize the quality of the constructed timetable satisfying the case-specific time related constraints imposed on the individual nurse schedules. Personnel rostering in healthcare institutions is a highly constrained and difficult problem to solve and is known to be NP-hard. In this paper, we present an exact branch-and-price algorithm for solving the nurse scheduling problem incorporating multiple objectives and discuss different branching and pruning strategies. Detailed computational results are presented comparing the proposed branching strategies and indicating the beneficial effect of various principles encouraging computational efficiency.

On the characterization and generation of nurse scheduling problem instances

Due to its complexity and relevance in practice, many different procedures have been proposed in the operations research literature to solve the well-known nurse scheduling problem (NSP). The NSP assigns nurses to shifts per day maximizing the overall quality of the roster while taking various constraints into account. The often highly case-specific workplace conditions in hospital departments have resulted in the development of dedicated (meta-)heuristics to find a workable schedule in an acceptable time limit. However, in spite of research community posing a growing need for benchmarking, these procedures lack any base for comparison. In this paper, we propose a range of complexity indicators which characterize nurse scheduling problem instances, and a problem generator in order to construct a comparative test framework for various solution procedures for the NSP. We show that the different complexity indicators for the NSP presented in this paper predict the computational effort of a particular NSP instance for a particular solution procedure. Moreover, the comparison of procedures and good predictions of their performance allow the a priori selection of the best solution procedure, based on the simple calculation of the indicators. Hence, with the developed NSP generator those indicators can facilitate the evaluation of existing and future research techniques. Tests on a simple IP model illustrate the use of the proposed indicators.

A hybrid scatter search heuristic for personalized crew rostering in the airline industry

The crew scheduling problem in the airline industry is extensively investigated in the operations research literature since efficient crew employment can drastically reduce operational costs of airline companies. Given the flight schedule of an airline company, crew scheduling is the process of assigning all necessary crew members in such a way that the airline is able to operate all its flights and constructing a roster line for each employee minimizing the corresponding overall cost for personnel. In this paper, we present a scatter search algorithm for the airline crew rostering problem. The objective is to assign a personalized roster to each crew member minimizing the overall operational costs while ensuring the social quality of the schedule. We combine different complementary meta-heuristic crew scheduling combination and improvement principles. Detailed computational experiments in a real-life problem environment are presented investigating all characteristics of the procedure. Moreover, we compare the proposed scatter search algorithm with optimal solutions obtained by an exact branch-and-price procedure and a steepest descent variable neighbourhood search.

An evolutionary approach for the nurse rerostering problem

The personnel scheduler constructs a deterministic personnel roster that determines the line-of-work for each personnel member. When unexpected events disrupt this roster, the feasibility needs to be restored by constructing a new workable roster. The scheduler must reassign the set of employees in order to cover the disrupted shift such that the staffing requirements and the time-related personnel constraints remain satisfied. In this paper, we propose an evolutionary meta-heuristic to solve the nurse rerostering problem. We show that the proposed procedure performs consistently well under many different circumstances. We test different optimisation strategies and compare our procedure with the existing literature on a dataset that is carefully designed in a controlled and varied way.

Comparison and hybridization of crossover operators for the nurse scheduling problem

Abstract In this paper, we present a hybrid genetic algorithm for the well-known nurse scheduling problem (NSP). The NSP involves the construction of roster schedules for nursing staff in order to maximize the quality of the roster schedule subject to various hard constraints. In the literature, several genetic algorithms have been proposed to solve the NSP under various assumptions. The contribution of this paper is twofold. First, we extensively compare the various crossover operators and test them on a standard dataset in a solitary approach. Second, we propose several options to hybridize the various crossover operators.

New computational results for the nurse scheduling problem: a scatter search algorithm

In this paper, we present a scatter search algorithm for the well-known nurse scheduling problem (NSP). This problem aims at the construction of roster schedules for nurses taking both hard and soft constraints into account. The objective is to minimize the total preference cost of the nurses and the total penalty cost from violations of the soft constraints. The problem is known to be NP-hard. The contribution of this paper is threefold. First, we are, to the best of our knowledge, the first to present a scatter search algorithm for the NSP. Second, we investigate two different types of solution combination methods in the scatter search framework, based on four different cost elements. Last, we present detailed computational experiments on a benchmark dataset presented recently, and solve these problem instances under different assumptions. We show that our procedure performs consistently well under many different circumstances, and hence, can be considered as robust against case-specific constraints.

An artificial immune system based approach for solving the nurse re-rostering problem

Personnel resources can introduce uncertainty in the operational processes. Constructed personnel rosters can be disrupted and render infeasible rosters. Feasibility has to be restored by adapting the original announced personnel rosters. In this paper, an Artificial Immune System for the nurse re-rostering problem is presented. The proposed algorithm uses problem-specific and even roster-specific mechanisms which are inspired on the vertebrate immune system. We observe the performance of the different algorithmic components and compare the proposed procedure with the existing literature.

The impact of incorporating nurse-specific characteristics in a cyclical scheduling approach

Nursing staff in various hospitals in Belgium are principally cyclically scheduled. The employed cyclic schedules embody, however, only a weak reflection of the ultimate nurse rosters constructed for a specific month. In this paper, we investigate the benefits of integrating nurse-specific characteristics in the cyclic scheduling approach. Moreover, we analyse to what extent these characteristics should be incorporated and compare this approach with a general and more robust cyclical scheduling approach and the flexible acyclical rostering of nursing personnel.

A scatter search for the extended resource renting problem

In this paper, the extended Resource Renting Problem (RRP/extended) is presented. The RRP/extended is a time-constrained project scheduling problem, in which the total project cost is minimised. In the RRP/extended, this total project cost is determined by a number of extra costs, which are defined in this paper. These costs are based on the costs that are used in the traditional Resource Renting Problem and the Total Adjustment Cost Problem. Therefore, the RRP/extended represents a union of these two problems. To solve the RRP/extended, a scatter search is developed. The building blocks of this scatter search are specifically designed for the RRP/extended. We introduce two crossovers and an improvement method. The efficiency of these building blocks will be shown in the paper. Furthermore, a sensitivity analysis is presented in which the five costs have diverse values.