Workforce schedule optimization is not compromization

One of the inputs we spend some time is demonstrating just how big a problem optimizing a workforce schedule is. In fact it is such a big problem and so expensive to achieve, examples are very hard to find – except for the most trivial scheduling problem. So why does everybody seem to offer this as a standard feature in workforce scheduling. Well in the main it is just a lazy use of marketing words. It sounds compelling, exciting and you know ‘you only get what you pay for’. That’s why you need very deep pockets, and a great deal of time on your hands to have a better than even chance to achieve that optimized workforce schedule.

An optimized schedule is the best possible solution available. First you need know all the factors that will define the pathways to be optimized to reach a defined goal. The more factors the bigger the problem to be solved. Even a handful of factors can generate a problem space that is measured in orders of magnitude. Put another way, defining the optimization model can be harder than doing the scheduling in the first place. Second, because there may be more than one solution you need to know all the solutions that are possible. These kinds of problem can take a very large computer a very long time to do this.

Another problem is we may decide what an optimum model is for staff headcounts and how they are distributed over a time range. Alternatively there may be a series of desirable goals for a staff day-on day-off working pattern. As more constraints are added the problem becomes easier to work out but the net result is what we considered optimum for one category is ruled out as new constraints for another category of information is added.

For example, the following goals ‘had to be’ achieved for a clients workforce deployment strategy. Not because they were desirable, but they had negotiated and signed off union contracts before realizing whether it was even possible:

  • between 4 and 6 consecutive work days
  • between2 and 3 consecutive rest days
  • an exact number of days off in a pay period
  • one weekend off and one weekend working in 4
  • and at least one weekend day off in 3
  • reduced staff at weekends

This occupied an HR team for a period a little over 7 months with no result. Using an intelligent agent designed to understand among other things the problem space of weeks in the context of week days and weekends completed the problem in 35 minutes. Out of a problem space of unknown size 4,712 candidates were identified as possible solutions. Only 13 of those solutions succeeded for further consideration.

Contraint scheduling can be contrasted with heuristic scheduling which promises to provide very good solutions most of the time. And a lot quicker and a lot more cheaply but that can be discussed another day.

Two things in conclusion.

  • Don’t agree to something you don’t understand in the context of workforce schedules, you will invariably underestimate the problem; and
  • When it comes optimization you probably are not getting what you pay for.

For more information about you staff deployment strategies contact Group Senior, Workforce Scheduding at Intellicate

Clip to Evernote
Posted in Shift Patterns, Workforce Factors Tagged with: , , , ,
0 comments on “Workforce schedule optimization is not compromization
1 Pings/Trackbacks for "Workforce schedule optimization is not compromization"
  1. […] For sure, there are such things as ‘friendly’ and ‘unfriendly’ shift patterns. This research concludes the optimal shift pattern required to maximise reproductive potential is yet to be established. In fact, when we start considering the word ”optimal”, or “optimization”, for any shift pattern we are in “very hard problem” territory. […]

Leave a Reply

Your email address will not be published. Required fields are marked *

*