Saturday, July 23, 2011

Focus and Leverage Part 46

I mentioned in my last blog that the batch and queue production system is the worst possible scenario for a company, but that isn’t exactly true. If a company practices batch and store production, whereby instead of processing the material to the next process, they move the material to a storage location, then the cycle time becomes even more protracted!

Continuing on, suppose instead of producing material in batches, when a part is completed in one station, it moves immediately to the next station and then on to the next station until it is completed (Remember our four station process in the following figure)? This type of production is referred to as single piece or one piece flow. One piece flow refers then to the concept of moving one work piece at a time between individual work stations. One piece flow has several distinct benefits like keeping WIP to low levels, encouraging work balance and improved quality, but in a system like this, what happens to cycle time? Let’s take a look.
Let’s assume, as before, that the processing time for each work station is 1 minute. The first part is processed through work station A and takes 1 minute. The first part continues immediately to station B and simultaneously the next part enters station A. After 1 minute the first part continues to station C, the second part moves to station B and a new part enters station A. After another minute the first part moves to station D, the second part moves to station C, the third part moves to station B and a new part enters station A. After another minute, the first part has been completed in station D so that the total time the first part remained in the process was exactly 4 minutes. The throughput is 1 part every four minutes or 0.25 parts per minute (15 parts per hour).

The question is, “What would happen if we increased WIP to 2 instead of 1?” Once again we begin measuring as soon as the first part enters work station A. We know that this first part will take one minute before it is passed to station B. At the same time, the second part is introduced to station A. These two parts follow each other through the 4-station line, so both remain in the process for 4 minutes total. Therefore, this system produces 2 parts every 4 minutes or 0.50 parts per minute (30 parts per hour). But what happens when we increase the WIP even more?

The table below contains all WIP values from 1 to 10 and as you can see, there is an interesting phenomena or nuance that takes place in this process when the level of WIP reaches 5 parts. If we make the assumption that the process is full (one at each station), and each part is ready to be processed, then when the fifth part is introduced to work station A, it must wait until the station has finished processing the fourth part and it moves to station B. Therefore, the fifth part remains in the system for 5 minutes. Each time a part is completed, the next part is introduced at station A and waits one minute before it can proceed. Look at the column for cycle time. As long as the system has no more than four parts in it, the cycle time remains constant at 4 minutes. But as soon as the fifth part becomes part of the system, the cycle time begins to increase by one minute for each increase of 1 part as WIP.
This demonstrates that increasing WIP levels doesn’t result in a corresponding increase in cycle time until the process is full or until its critical WIP level has been reached. For this example, the critical WIP level is four parts. If WIP is increased beyond this critical WIP, parts simply stack up and wait to be processed, causing the cycle time to increase.

Equally interesting, however, is that reducing WIP levels below the critical WIP (i.e. in this example 1, 2 or 3 parts) results in a corresponding decrease in throughput! As you can see in the table, when WIP is at its critical WIP level of 4 (i.e. the system is full) the throughput is at its maximum value of 1 part per minute or 60 parts per hour. But when the WIP level is reduced to 3, throughput drops from 1 part per minute (60 parts per hour) to 0.75 parts per minute (45 parts per hour).

The following figures demonstrate the relationship between WIP and Throughput and WIP and Cycle Time. In this example, there is one WIP value that results in minimum cycle time and maximum throughput which is what we always want. Any less WIP and we lose throughput with no decrease in cycle time. Any more WIP and we increase cycle time with no increase in throughput. It is interesting to note that for a balanced line (i.e. all individual processing times are equal), the critical WIP level will always equal the number of process steps. For unbalanced lines, this is not the case.

What we’ve learned here is that too much WIP results in longer cycle time and increased holding costs while too little WIP results in decreased throughput and lost revenue. This means that there is an optimum amount of WIP (i.e. critical WIP) that should be in a system. Even though there are many “experts” who believe in the concept of “zero inventory,” we now know that WIP should never realistically be zero! In my next posting, we’ll take a look at what happens to WIP, Cycle Time and Throughput in an unbalanced line (i.e. when processing times are not equal).  This whole series of blogs is centered around the concept of factory physics.

Bob Sproull

No comments: