Last month, we began discussing the effect large swings in sales, either as a general trend, as the result of seasonality or as a result of changes in product mix, can have on your factory. This month, we will discuss specific solutions to these problems.
We have already stated that there are three main culprits in creating what I call temporal bottlenecks, or bottlenecks that occur over a short span of time on a piece of equipment that is not truly a bottleneck over the long term. These are:
- batch processing,
- order variation and
- process variation.
Let's look at each one and see what can be done to mitigate the problems they create:
Temporal bottlenecks and large batches
Large batches have always been a mainstay of the woodworking industry. We all recall with nostalgia those huge cuttings of years past, when you could spend days on a single model of product and weeks on a single series. Well, we all know those days are gone. In truth, they never were good for a company - it was simply that none of us knew any better, and we were all making the same mistakes, so we had a level field of competition.
Large batches moving through the factory create what is known as the "pig in the python" effect, because it resembles the bulge a pig would make if it were swallowed by a large snake.
This bulge moves through the factory, and as it does, it creates problems and a temporary surge in demand as it goes from station to station.
Another analogy would be like a tidal wave moving through the factory, and like a tidal wave, it leaves a vacuum in its wake. In other words, a large batch often commands so much attention that there is a void after it, before the production items come through - this amplifies the feast and famine effect in the factory.
So how do we go about fixing this problem?
Production batches and transfer batches
Traditionally, the number of units ordered became the minimum lot size moving through the factory. More often, these orders would be combined and then the total of all those orders would move through the factory as a group. The reasons for this were various - combining setups, trying to avoid duplicate paperwork, the perceived simplicity of only having one large load, etc.
However, there is no written law that requires the amount moved between machines - the transfer quantity - be the same as the production quantity. By breaking up huge 1,000 or more lot sizes into lots of 100 or less, many benefits can be realized.
- Individual parts flow through the system faster
- Material handling can often be simplified
- Product cycle times can be reduced
- Congestion is reduced
- Response is improved
Reducing transfer batches can do a lot of positive things for your factory. However, it should be emphasized that reducing transfer batches is part of an overall improvement strategy - not an end to itself or a magic bullet to solve all your problems.
In short, if huge batches are causing huge problems, reduce them!
Order variation has a number of sources - but we'll concentrate here on the normal variation that occurs in sales levels. How do we go about reducing the effect order variation has on our factories? One common solution is production leveling, and we discuss it here with a simple example:
Say that I have a factory making wooden letters. I'm one of several factories in the "Alphabet" Corporation so I'm only responsible for making letter "A" through letter "E." I consider myself lucky because I have two vowels in my product group, two of the best selling letters! However, I've noticed that I seem to have a lot of problems with my other letters - "B," "C" and "D." I always seem to have either way too many, or I'm out of these letters.
My factory can produce 100 letters a day, and my average weekly sales of these letters are:
Product Wkly. 4 Wk.
E 200 800
A 150 600
B 75 300
C 50 200
D 25 100
Right now, our factory runs on a four-week (20 day) cycle. So at any time, we are producing products to cover demand for the next four weeks. Therefore, our production cycle would be the same as shown in column three in the table. Since we group production runs (for the reasons discussed earlier), we run eight solid days on "E," six days on "A" and so on (see Chart 1).
From this chart, its easy to see why you tend to run out of the lower-volume products, since you spend almost three weeks running the high-volume products. By changing to weekly volumes, sales more closely track to inventory. This is also a simple example of production leveling, when the amounts running closely mirror sales levels.
With experience and improvements, the factory could later move to daily schedules, as changeovers and production methods improve.
Every job suffers some variation, even the most tightly controlled process. Throw in the addition of human beings, and a fair amount of variation is almost assured. The thing recognized by the Japanese, however, was that there is very real difference between having variations and creating defects. The goal is to reduce the amount of variation so that almost no defects occur. This is the basis of six-sigma production systems (see FDM December 2002 article for more on six-sigma).
How do you reduce variation? Myriad methods (in particular, Poka-Yoke) have been discussed in past articles as ways to reduce variation and improve quality. While an in-depth discussion of them all is beyond this article (they are all reviewed in the new second edition of A Better Way ), here is a summary checklist of the most common reasons of process variation, and some of the tools available to correct them:
Type of Variation Correction Tool(s)
- Part Poka-Yoke,
Omission Visual Mfg.
- Method Work
- Wait/ Cellular
Transport Time Manufacturing
- Raw Vendor Certification/
- Machine Total Productive
- Setup Quick Changeover
For every major problem, there is a tool in the manager's toolbox to fix it. However, in today's challenging economy, finding the right problem to fix is just as important as knowing how to fix it.
Have something to say? Share your thoughts with us in the comments below.