engleski

Proactive-reactive project scheduling with flexibility and quality requirements

Due to the increasing complexity of real projects and well documented problems with efficiency and performance of current scheduling procedures, there has been growing research into stochastic project scheduling. Most of the procedures for stochastic scheduling focus only on regular performance measures, i.e. non-decreasing functions with increasing time, such as the makespan and project tardiness. Such procedures aim to optimize the compliance with the due dates. As a consequence of increases in projects’ complexity, not many of them are executed completely in-house and the synchronization between project collaborators is very important. Some project activities are outsourced and many activities require resources that need to be acquired from suppliers during the project run-time. Every increased synchronization effort due to sudden changes, such as longer than anticipated activity duration, results in increased project costs and delays that pose risk to the project success. Baseline schedules are used as a mean of synchronization between the project collaborators. Proactive schedules make for good baseline schedules as they have built-in desired amount of safety, making schedule feasible in a wide range of events. Considerable research was conducted on stability measure which aims at creating proactive baseline schedules that in expectation deviate the least from actually realized schedule. There have been attempts at rescheduling the baseline schedule during the project execution in order to increase its degree of proactivity while using the schedule stability measure, but only with moderate success. Stability measure, apparently, introduces rigid constraints that disable changes to the baseline schedule except during the activity start. Such changes could reduce future rescheduling costs and are commonly heuristically used by project managers. This dissertation presents a new family of robustness measures, a Cost-based flexibility (CBF) which introduces dependency of rescheduling costs on the temporal distance of the baseline schedule change. A mathematical model of our problem of interest was created which describes the need for collaborators synchronization and aims to comply with the due dates. As objective functions containing CBF are not necessarily regular and the existing theory does not cover that case, theoretical foundations are laid out. Exact solving procedure, as well as the two heuristic procedures are presented. Developed computational prototype was used to obtain experimental data regarding performances of the two heuristic procedures in comparison to the selected (currently best performing) benchmark originally constructed to work with the stability measure. The results show that our algorithms substantially outperform the benchmark algorithm. In conclusion, CBF measure, unlike stability measure, enables proactive rescheduling.

project scheduling ; risk ; proactive reactive scheduling ; proactive rescheduling ;

nije evidentirano