Scheduling Performance¶

Enhancements have been made to the scheduling algorithms within the DSE to improve performance. This has focused on two areas: tightly constrained schedules and schedules with large numbers of activities per route.

Tightly Constrained Schedules¶

The target here was to improve the performance of schedules where most activities are constrained either to a very restricted time slot (or even a fixed time), or to a small subset of resources, or both.

The primary focus for improvements has been on the core optimisation engine, to ensure that it makes more intelligent decisions about what changes to a schedule should be attempted. In addition, there have been changes to make the optimisation handle hard minimum separation incentives more effectively.

Performance comparisons have been carried out for a sample problem of this type, and the results show a very significant improvement from the 6.3 version to the 6.4 version. In fact the 6.4 version consistently produced a better plan after 2 minutes than the 6.3 version achieved in 30 minutes. The graph below shows the plan margin over time for the two versions.

High Volume Routes¶

The target here was to improve the performance of schedules where many activities could be scheduled in each route. This would be the case for example with very high density, short duration activities, where it may be possible to carry out several hundred activities in a single shift.

The focus for improvements has been firstly on enhancements to the existing optimisation process used by the DSE and secondly on removing some inefficiencies with handling large volume routes and high density schedules.

Performance comparisons have been carried out for a sample problem of this type, and again the results show a significant improvement from the 6.3 version to the 6.4 version. The 6.4 version consistently produced a better plan after 2 minutes than the 6.3 version achieved in 15 minutes. The graph below shows the plan margin over time for the two versions.

Documentation¶

See the Architecture and Sizing Guide for full details of DSE performance benchmarking and system sizing information.