Following the debate about all the different scheduling approaches it is inevitable that you are faced with two kinds of conceptuality: “flow shop scheduling” vs. “job shop scheduling”. The final intention of this blog is to give an understanding that the scheduling challenges and approaches of both worlds are so different, and why this is the case.
Therefore I first point out the different characteristics of flow shops and job shops as such. Based on this I derive the differences that come along with this in terms of scheduling. Finally, I give an outlook on what impact the differences have on tools to productively schedule flow shops and job shops.
The first look into the current debate gives you a rough – because short – definition of both kinds.
In a flow shop, the manufacturing process follows a fixed linear structure. That means that all orders need to be manufactured in the same way on the same machines.
In a job shop, the routing of each job can be individual. That means that all orders (potentially) need to be manufactured differently on the same machines or a certain part of the same machines.
This gives a good first impression of the significant diversity of both manufacturing environments.
But in my eyes the best way to fully comprise the diversity, it is very useful to further characterize a flow shop and a job shop based on common criteria of the manufacturing industry.
To keep this as easy to understand as possible in the following I refer to the most extreme forms of flow shops and job shops possible.
In terms of a flow shop, this would be a manufacturing site with 100% standardization operated in an assembly production line. On the other hand, a job shop would be a business with 100% customization with a typical batch size of 1, which implies that every finished product is unique.
The graph below sums up the diversity of flow shops vs. job shops:
As we now have an understanding of the significant differences of both manufacturing environments the next step is to evaluate what impact the diversity has on scheduling.
Or in other words:
If we translate these characteristics into scheduling requirements it is obvious that flow shop scheduling and job shop scheduling also differ significantly. Let's start again with handy summaries:
As the required data is not volatile and the planning environment is very stable and not dynamic, reliable input data can be generated. This is true for the demand as well as for the duration of operations.
In the case of an assembly production line, the challenge of scheduling can even be boiled down to optimizing the cycle time of how the line moves forward. In general, the output of flow shop scheduling is when to operate which task in which batch size.
To calculate such a schedule mathematical heuristics (like a genetic algorithm or simulated annealing) are very appropriate and common approaches to find the optimum solution for the given planning field. The respective - centrally determined - occupancy plan is then the firm instruction for the shop floor.
As the planning environment of job shops is not stable at all but very dynamic the estimation of planned duration comes along with a high degree of fluctuation. Furthermore, the demand is not calculable upfront as it is 100% customer order-driven.
So mathematical approaches are not appropriate as the high inaccuracy of the data would only pretend to deal with an optimum. Also, the ongoing need to reschedule due to daily incidents makes the flexibility to easily deal with this dynamic more important than the calculation of an alleged optimum.
The consequence is: for job shops visual scheduling approaches with automated scheduling support are most appropriate to answer the dynamic and productively generate schedules that then serve as a guideline for execution.
This again gives a good overview of the significant diversity of the scheduling challenges of both manufacturing environments. To work out the individual differences even more precisely, I now analyze the following based on common planning parameters.
To wrap up the scheduling requirements:
The planning environment of a flow shop can be designated to be more favorable due to the stability of the decision-making field. On the other hand, the planning environment of a job shop comes along with more dynamic and more complexity.
As written above flow shop scheduling can be described in mathematic models, that – fortunately – are solvable with modern technology. So flow shop scheduling tools consist of a heuristic engine that is capable to consider all flow shop parameters and then calculate the optimum of the given input.
Specifying a scheduling tool for job shops is not as favorable and finding an appropriate scheduling solution can be a herculean task.
But because of the knowledge gained from this blog post, the following four requirements must be met to successfully to face the special challenges of the job shop environment:
And, if you still do not know how and where to start: Here is a blog post how to successfully start with job shop scheduling.
In summary, it can be said that flow shops differ so significantly from job shops that one can even speak of diametrical opposites. This contrast is not only noticeable in the comparison of characteristics, but also concerning the planning parameters.
Also it becomes clear that the requirements of job shop scheduling are much more extensive and complex to handle. While flow shop scheduling can be done by a mathematical algorithm, job shop scheduling essentially requires to consider the human factor.