This study addresses the resource leveling problem (RLP) under a fixed project duration, with the aim of reducing the maximum daily total resource demand (R max ) and improving fluctuations in the daily total resource usage profile. In this context, the grey wolf optimizer (GWO) algorithm was implemented in a Python-based environment and tested on Harris’s 11-activity benchmark network. The analyses were based on activity durations, precedence relationships, and daily renewable resource requirements, and resource leveling performance was evaluated under two different scenarios by taking the critical path method (CPM)-based early-start schedule as the reference. The first scenario is the no-split case, in which activities are executed continuously once started, whereas the second is the split-enabled case, in which activities can be fragmented within their feasible time windows. Performance evaluation was conducted using peak reduction (PR%), which indicates the change in peak resource demand, and resource leveling efficiency (RLE), which was defined based on the total absolute deviation of the resource usage profile around the mean. The findings show that the R max value was reduced from 16 to 10 in both scenarios, corresponding to a PR% value of 37.50%. In addition, the RLE value was obtained as 1.64 for the no-split scenario and 2.14 for the split-enabled scenario, while the corresponding improvement rates were calculated as 39.16% and 53.19%, respectively. The results reveal that the split-enabled scenario provides a more balanced daily resource usage profile at the same peak resource demand level. Overall, the study demonstrates the applicability of GWO to the fixed-duration RLP and presents a holistic evaluation framework that jointly considers PR% and RLE.
Bayram Ali Temel (Thu,) studied this question.