TY - JOUR
T1 - Scheduling multiple, resource-constrained, iterative, product development projects with genetic algorithms
AU - Yassine, Ali A.
AU - Mostafa, Omar
AU - Browning, Tyson R.
N1 - Publisher Copyright:
© 2017 Elsevier Ltd
PY - 2017/5/1
Y1 - 2017/5/1
N2 - Many product development (PD) projects rely on a common pool of scarce resources. In addition to resource constraints, there are precedence constraints among activities within each project. Beyond the feed-forward dependencies among activities, in PD projects it is common for feedback dependencies to exist that can result in activity rework or iteration. In such a multi-project, resource-constrained, iterative environment, this paper proposes two new genetic algorithm (GA) approaches for scheduling project activities. The objective is to minimize the overall duration of the portfolio of PD projects. These proposed GAs are tested on sample scheduling problems with and without stochastic feedback. We show that these algorithms provide quick convergence to a globally optimal solution. Furthermore, we conducted a comparative analysis of the proposed GAs with 31 published priority rules (PRs), using test problems generated to the specifications of project, activity, and resource-related characteristics such as network density (complexity), resource distribution, resource contention, and rework probability (amount of iteration). The GAs performed better than the PRs as each of these factors increased. We close the paper by providing managers with a decision matrix showing when it is best to use the published PRs and when it is best to use the GAs.
AB - Many product development (PD) projects rely on a common pool of scarce resources. In addition to resource constraints, there are precedence constraints among activities within each project. Beyond the feed-forward dependencies among activities, in PD projects it is common for feedback dependencies to exist that can result in activity rework or iteration. In such a multi-project, resource-constrained, iterative environment, this paper proposes two new genetic algorithm (GA) approaches for scheduling project activities. The objective is to minimize the overall duration of the portfolio of PD projects. These proposed GAs are tested on sample scheduling problems with and without stochastic feedback. We show that these algorithms provide quick convergence to a globally optimal solution. Furthermore, we conducted a comparative analysis of the proposed GAs with 31 published priority rules (PRs), using test problems generated to the specifications of project, activity, and resource-related characteristics such as network density (complexity), resource distribution, resource contention, and rework probability (amount of iteration). The GAs performed better than the PRs as each of these factors increased. We close the paper by providing managers with a decision matrix showing when it is best to use the published PRs and when it is best to use the GAs.
KW - Design Structure Matrix (DSM)
KW - Genetic Algorithm (GA)
KW - Iteration
KW - RCMPSP with Feedback (RCMPSPwF)
KW - Resource-Constrained Multi-Project Scheduling Problem (RCMPSP)
KW - Resource-Constrained Project Scheduling Problem (RCPSP)
KW - Rework
UR - http://www.scopus.com/inward/record.url?scp=85014855870&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=85014855870&partnerID=8YFLogxK
U2 - 10.1016/j.cie.2017.03.001
DO - 10.1016/j.cie.2017.03.001
M3 - Article
AN - SCOPUS:85014855870
SN - 0360-8352
VL - 107
SP - 39
EP - 56
JO - Computers and Industrial Engineering
JF - Computers and Industrial Engineering
ER -