Capacity Planning for Complex Value Streams

Criteria and Problems of Capacity Planning  

People and machinery are needed in factories to produce products and process orders. From manufacturing to shipping, various steps are carried out in a value adding stream. These steps must mesh together like interlocking gears. If only one cog fits, orders cannot be properly processed and overloads and backlogs occur at other points. Such unforeseen bottlenecks can lead to substantial economic losses.

Sufficient capacity is critical for smooth operations in the value stream. It is not enough to plan each step in a process separately: Perhaps, production is equipped with efficient machinery and achieves top throughput rates, but this is ultimately of little use if the logistics is inadequately staffed to perform the picking. All process steps and their interactions must be taken into account to coordinate each capacity.

Another problem is created by short-term capacity planning. Bottlenecks occur that can only be partially alleviated or are associated with high costs, such as leasing expensive storage areas. For this reason, a forward-looking view is important so that countermeasures may be implemented in the early stages. Of course, a forward looking capacity planning also carries risks such as an uncertain order book in the future.

Auslastungsprognose
© ITWM
Sample capacity planning forecast using value flow simulation

Development of a Software Tool

The Optimization department develops a tool for proactive capacity planning that addresses these challenges through practical modeling. The trick is in having the appropriate level of detail in the mapping of the individual process steps. A detailed simulation of all operations is not practical as there are too many assumptions about the internal decision processes that affect the results. Similarly, an approximate forecast of capacities using average values only is too imprecise. Instead, each process step is parameterized with capacity relevant parameters.

Process related internal fluctuations are taken into account by stochastic distribution functions. The appropriate mix of data-driven and expert modeling is key, for example, representing highly complex production operations or hard to describe human resources.

A Monte Carlo simulation provides a sufficiently accurate, highly realistic capacity forecast with throughput times and flow rates. The user can measure the target values in “What-if” scenarios or explore different calibrations in parametric studies. The capacities and their effects on the value stream can be planned in a transparent and proactive manner.