Regarding to variability and its control we will go through some “laws” that will affect production flow.
Capacity can never be fully utilized. The closer we get the 100 % utilization the more WIP we end up and cycle time will also increase. Because practical systems will always have some variability with 100% utilization we end up in following treadmill;
- Capacity is defined by removing common outages like breakdowns, breaks and setup times.
- This calculated capacity will be scheduled 100 %
- Sooner or later the variability in the system will cause the bottleneck workstation to overload.
- Works are initiated but not finished so WIP increases
- Because capacity is fully utilized we cannot increase production, queues won’t be caught up and CT increases.
- Works will be delayed
- Customers are unsatisfied
- Eventually directors start to do something
- Overwork, extra shifts or subcontractors are allowed “one time only”
- This leads to increase in capacity and in example of increasing shifts from 2 to 3 utilization drops from 100 % -> 67 %.
- WIP decreases, cycle times normalize and customers are satisfied again.
- Everyone lets out a sigh of relief and wonder how things got out of control. They swear blind not to let this happen again.
- Everything starts over.
The closer 100 % utilization we get the bigger WIP and longer cycle times we have. Near the 100 % utilization they literally explode exponentially. The bigger variation (CV) we have the bigger problems.
First we have to define some concepts. Batch means both production batches and transfer batches. Production batches can be divided further into sequential production and simultaneous production.
Sequential production is the amount of transfer batches that is produced at the workstation before we change into different product model. This is called sequential production because products are produced sequentially one by one.
In simultaneous production workstation can have multiple works under production simultaneously (“real batch production”). Both production processes are similar in the operative point of view.
In order to decrease the utilization simultaneous production should always run at full batches. But if workstation is not the bottleneck it is not necessarily to decrease utilization and it may be reasonable to produce smaller batches in order to decrease cycle time (CT).
Transfer batch is the amount of products, “basket”, that is filled before it is moved to the next station. The smaller the transfer batch the bigger cycle time because parts need to wait for the transfer batch to fill up. However with small transfer batches we end up with more handling.
Example with forklift we can easily transfer all products made in one shift where with hand truck pallet we would have to transfer products 30 times per shift even though we would decrease transfer batch significantly into 3 % (example from 100 to 3 parts batch).
Products movement is entirely own process and production batch and transfer batch need not to be the same. Actually small transfer batches tend to decrease cycle time. Especially if workstations are close each other and setup times are long it is reasonable to keep production batches big and transfer batches small.
Production batch in simultaneous production
Lets take an example of oven that has capacity of 100 products. Products need to be held there for 24 hours and previous workstations has an output rate of 1 product per hour (24 products per day). Clearly it is not reasonable to put only 1 product in the oven. Secondly if we would put there 24 products we would have an capacity of product per hour which is the same than input rate. Utilization would be 100 % which isn’t be reasonable and possible in the long term.
Like previously told, we cannot schedule production for 100 % utilization so the only possibility is to have at least 25 products in the oven at once. Another option would be to take the whole queue in front of the oven (or maximum capacity 100 whichever is the limit). But if the output of previous station is bursty, in the other words its production time has high variability (HV), we should always wait at least the minimum batch of 25 products.
In the book Factory Physic Wallace Hopp tells enjoyable example of bad production planning. Production was organized into cells so that all the drilling, polishing and other work phases for raw metal were able to be done in one hour. This cellular production had dropped production time from multiple days into one hour.
Hopp and his visit group discovered that parts were waiting transfer after the production process in the batches of 10 000 units! So the first part processed had to wait 9999 other parts to finish before its transfer. Capacity of the line was 100 parts per hour so the queue time for the first part was 100 hours.
This way the good intention to decrease WIP and cycle time was nowhere near its meanings. Hopp says that he has never seen any production line to be as close the worst case scenario as this line was.
In this example decreasing the transfer batch into one tenth, 100 products, would have meant product to be transferred only every 10 hours. If transfer is made less often than once a shift it can’t be too much trouble but would have decreased cycle time and WIP a lot.