In logistics, there has been a long-standing debate regarding the superiority of various flow logics. This applies to the areas of supply, production, and distribution alike.
In each of these areas, “pull” and “push” concepts are applied. Practical implementations of these concepts take various forms, ranging from the location of schedules for material streams, inventory concentration points, to the method of initiating and controlling flow volumes. In the area of production, this is also a key issue that influences the characteristics of production flow streams. A sound choice of concept and the creation of detailed solutions that implement them in practice require a full understanding of their advantages and limitations.
“Push” vs. “pull” – flow simulations
One way to gain a fuller understanding of the nature of “pull” and “push” concepts and to capture the practical differences between them is to conduct flow simulations. The simulations performed covered three variants of production flow control: the “push” concept, the “pull” concept, and the hybrid “pull/push” concept. Both the execution technology (sequence of operations), the layout of workstations, and their potential (unit capacity) were the same for all simulations conducted. In all simulations, the same production program was implemented, covering three different product ranges. In the individual simulations, the key difference was the adopted method (location) of production stream scheduling. Its implementation was consistent with the simulated management concepts (Figure 1). For the system based on the MRP concept (“push” logic), this was a production schedule located at the beginning of the production process, at the point where material is issued to workstations. Subsequently, the material was “pushed” through successive workstations until the final product, according to FIFO priority. For the simulation reflecting JIT concepts (“pull” logic), the production schedule was located at the end of the production process, at the point of quality acceptance of finished goods. Up the production stream, demand was transmitted according to the Kanban card signal, up to the material issuance point.
For the system based on the TOC concept (“pull/push” hybrid logic), the schedule had two locations: the first concerned capacity constraint (“bottleneck”), and the second – the point of issuing material for production. The constraint’s work schedule was optimized to minimize setup time losses for this resource. Issuing material for production (implementation of the second schedule) was synchronized with the flow of products from operations at the critical resource. Through the remaining workstations, a “push” flow was implemented according to FIFO priority.
Comparative analysis of “push” vs. “pull” – results obtained
The first results analyzed are the quantity of products manufactured broken down by individual minutes of the simulation (Figure 2). In the case of the “MRP” simulation, it is worth noting the “filling” of the production system with work-in-progress only after the 6th minute of the simulation. The first “flow” of products occurred in the 7th minute of the simulation. In subsequent minutes, the flow of products occurred at a level of 5 or 6 units, thus reaching a stable value. Since the flow occurred in “waves” equal or close to the batch size, in certain time units it was zero. Such fluctuations result from the nature of flow by production batch and the adopted time measurement unit equal to one minute. In the unit where the flow was zero, the result achieved was often the effect of the “established” production takt. The flow of parts often occurred right at the beginning of the next time period. Using a smaller time unit for measuring the flow of parts would, of course, make this phenomenon more visible. However, it should be stated that the characteristic (and therefore expected by observers) feature of “push” flow in production batches was confirmed.
The article comes from the magazine “Logistyka” 5/2007.