Employees’ skills and tasks can be classified into several so-called skills. For each of these skills, separate turnover curve estimates are made. How can the available worker resources be used most efficiently to produce a shift plan that ideally covers these plan curves?

The area under the curve for each skill distribution, shows the total amount of resources they require over the plan period. Obviously, somehow the total monthly working time of each worker should be divided in proportion to the skill resource requirements. If this were done horizontally, i.e. the total number of hours were broken up in proportion to the skill resource needs, it could very easily be the case that each piece could not be produced as an integer multiple of the base time.

• For example, one piece would be 29 hours and you would get shift lengths greater than 38 but less than 48, i.e. fractional shift lengths.
• Instead, split the beams of the 8 hour shift vertically, i.e. sk1:sk2:sk3=50%:30%:20% with sk1 skill curve underflooring with only half a man (half man height) and sk2 with only 0.3 height.
• At the end of the parquetting, we sum up the pieces and round the results to a whole number.
• At the time of the operation, we implicitly assume that if the worker is in his workplace, e.g. because the start time in skill sk1 allowed it, he will also perform the tasks in skills sk2 and sk3 with 30% and 20% of his workforce, respectively.
• These percentage weights are an important guide for calculating global rates, but the specific daily curve values may differ significantly. For example, it may be that when sk1 curve is in an upward sloping phase during a shift start period, no calls from sk2 skill are expected, well the monthly resource requirements may be considerable. The intelligent parquet algorithm also takes these local deviations into account when generating a multi-skill shift plan.