Several commercial microcomputer project management software packages are tested for their ability to optimally schedule 110 projects in which early finish schedules have over-scheduled resources. Each examined package has the ability to remove the over-scheduled positions with resource leveling. It was found that no package consistently finds a schedule in which the project completion time is minimized. The best package obtains schedules for the 110 projects that average 5.03 percent longer than the optimal schedule of each; the weakest package obtains schedules that average 25.6 percent longer than optimal.
Commercial project management packages level resources that are over-scheduled to their earliest start schedule differently, seldom attaining the same or the shortest possible schedule. This article compares the schedules of 110 projects1 generated by 13 versions of seven commercial software packages: SuperProject Expert 1.0 and SuperProject 2.0, Timeline 2.0 and 4.0, Primavera 4.00. 4.1 and 5.0, Microsoft Project for Windows 1.0 and 3.0, Harvard Total Project Manager II and Harvard Project Manager 3.0, Pertmaster Advanced2, and Hornet2; with two academic procedures: Talbot's |8| optimizer, and Patterson's |4] heuristic.
All packages that use PERT or the Critical Path Method compute the same early and late start and finish times from the same project data'—unless resource leveling is required to remove over-scheduled conditions. Then, it is often very difficult to find the optimal (shortest feasible! schedule, and no commercial package examined consistently does so. Why? When two tasks require the same resource, it is not always obvious which should be scheduled first. Consider the five-task project depicted in Figure 1.
Project Duration » 5 Days
Project Duration » 5 Days
In the absence of resource constraints, this pro--ject can be completed in five days, and the critical, path is A-D-E. The total slack of Tasks A, D, and E is 0,\ and is 1 for Tasks B and C. (Total slack is the time which a task might be delayed beyond its early start^ without causing the project to be late. Wiest and Lev| |9| provide a clear description of this and other critu path basics.) But suppose the only available unit Resource R is required by Tasks B and D, which aj scheduled to be performed simultaneously in the eaj liest start schedule. To find the shortest schedule, it necessary to try B earlier, and also try D earlier. 0n| then is it clear to schedule B earlier than D to coi plete the project in six days. Both schedules are pro, vided in Figure 2.
An appealing way to resolve a two-task resou conflict is to schedule the task with less total sia earlier. This is what some commercial packages and it is very often the correct choice. But not alwa as demonstrated in Figure 2, in which Task B shoiw be performed first to shorten the project's duration even though it has more slack than D.
The difficulty of determining the optimal sched* ^ ule lies in the vast number of combinations of185^Sj and schedules. To be sure of obtaining an opt»11^ schedule, it is necessary to enumerate each option every conflict. If 20 two-task resource conflicts e» »
there are 1,048,576 schedules to enumerate.
Task B Scheduled Earlier than Task D A >| Project Duration = 6 Days B I-
Was this article helpful?
What you need to know about… Project Management Made Easy! Project management consists of more than just a large building project and can encompass small projects as well. No matter what the size of your project, you need to have some sort of project management. How you manage your project has everything to do with its outcome.