Scheduling high-mix, low-volume operations
Scheduling remains one of the most complicated, least predictable aspects of contract metal fabrication. But why is this exactly, and what can a shop do to change this? In this first article in a series, Dick Kallage uncovers the answers.
Of all the issues that I hear about from custom fabricators, the one mentioned most often is scheduling. The problem seems to be that what should be an orderly and deterministic process leading to highly predictable results in fact turns rapidly into something that is anything but that. Parts that were due on a given date somehow are late, while other parts due at a later date pop out of the process a week early.
From this comes a fur ball of activity: Supervisors and leads run around doing mini-expedites; sales and customer service folks go nuts, make up excuses to tell customers, and create daily priority lists; production meetings become intense, last for hours, and fail to accomplish anything except migraines.
And so it goes, week after week, month after month. The company is in official firefighting mode, when non-value-added activity rises dramatically, the waste is huge, and heroes who save the day are created, as are the villains who are assigned the blame for the mess du jour. Of course, the title of hero or villain is highly temporary. Next week the titles are often reversed.
From an outside perspective, the company is beginning to take on the infamous defining characteristic of the airline industry. Nobody’s happy—not the owners, not the employees, and certainly not the customers. This is a feat no sane person would want to duplicate.
I give many talks at FABTECH® on aspects of Practical Lean, a lean system developed specifically for high-mix, low-volume shops. I typically start by asking the audience to describe their usual daily operating mode. The vast majority respond with the words firefighting or even chaotic. When I ask why this is so, a similar majority will say either “poor scheduling” or “constant changes in priorities.” When I ask if anyone believes that these are root causes of the issue, few if any raise their hands.
This is enlightening. The fact is that few really have a grasp on the actual reasons that the firefighting is necessary, so they assign blame, without proof, to the things that appear most directly connected to the problem. This phenomenon can lead to some really bad, expensive decisions.
So how can you work around the firefight to put some order and predictability back into the process? A number of ad-hoc techniques commonly are employed. Schedulers, sick of getting blamed, might simply release orders sooner, if possible, or insert time buffers in the processes’ operations routings, or create inventories of commonly used cut blanks. Production managers will schedule overtime to “catch up” (though even with overtime, things never seem to get caught up completely). All of these tactics are valid and probably necessary to some extent on a short-term basis. But they are expensive Band-Aids, meaning less is better.
Back to the Basics
What is the purpose of scheduling? I think we would all agree that a schedule should create a time sequence of events that, to a very high degree of certainty, ensures a given part or assembly arrives at a given place and at a given time. The “given place” could be the customer’s dock or simply the next operation in the process. The “given time” for the customer is, of course, the time the product is due on its dock as agreed upon between the customer and supplier. Scheduling’s purpose is simple and obvious.
The task of scheduling is to release orders into the value-adding process such that, given the time it takes for the order to be processed through the operations routing, the order will appear at the output of the routing at the approximate time (not too soon and certainly not late) desired by the internal or external customer. To do this, the scheduler must offset the desired output time by the cumulative processing times to arrive at the order release time.
The scheduler must also be aware of the available capacity of each operation for a given time period. For example, if the lasers are already fully loaded for a certain day or week, then the scheduler must not schedule further work into that time “bucket” for the lasers, but push it to the next available bucket. This should have been communicated to the customer to establish an agreed upon completion date. This is the role of master scheduling, and most companies do this satisfactorily, notwithstanding the inherent tension between customers/customer service and operations to get things faster.
Sounds simple, right? Sometimes it is. Sometimes it is more complex, especially when purchased part lead-times, outsourced operations, or parallel paths in the routings come into play. Nonetheless, it is a deterministic process that can be detailed in a step-by-step, learnable set of procedures. It is not a random seat-of-the-pants thing. If schedulers do all this in a consistent, robust manner, then they have done their part in one of the operation’s fundamental tasks: Get the right material to the right place at the right time.
Great. So what goes wrong? Where in this orderly scheduling process does the fur ball happen?
Making a Fur Ball
To see what really goes wrong, let’s start with what a deterministic process is. In this context, it means that given a set of parameters for a process (capacity, capability, etc.) and the time sequence of inputs to that process, the output and its time sequence is entirely predictable—at least to a very high degree of certainty. Most important, it is predictable without ad-hoc, on-the-fly manipulation of the process.
The problem lies in the previous sentence. What if the value-adding process is not predictable? Then it’s not deterministic. Then we have to create a series of non-value-adding tasks to put it into an order such that we can say, “You’ll have your parts next Thursday.” (We think.)
This set of non-value adding tasks is itself not deterministic, at least in its frequency and intensity, and certainly not in its assurance of any given outcome. But, hell, at least we’re doing something! But here’s the reality: The tasks done to correct a nonpredictable process create the fur ball.
The first thing that can go wrong is that there was a scheduling error, though this is relatively rare. Typically, an order didn’t get released because of some communications breakdown between sales and master scheduling. Or an operation was inadvertently scheduled beyond its available capacity. By the way, these occurrences are the ones remembered forever by the production folks. These errors can be damaging, no doubt, especially when even just one of the operations is working at or near full capacity. But these types of errors cause only transient damage. The effect soon goes away, and the process recovers with little intervention beyond, say, some overtime.
The same can be said about the second thing that goes wrong: a schedule move-up and expedite by a key customer. But like the scheduling errors, this is usually transient. Its occurrence is not as rare as the scheduling error, but for most companies, it’s in the once-a-week category at most. (If schedule move-ups and expedites are daily events, that’s a different story. It does then, indeed, become a fur ball creator. But this is symptomatic of other problems.)
So what else goes wrong that could cause a constant state of firefighting? To answer this, we need to return to the task of scheduling and its input parameters, including the actual capacity of the operations in the value-adding process and the actual time required by a given job in a given operation. What if they’re wrong? What if schedulers deal with input numbers that don’t reflect reality?
If the capacity numbers are wrong, we have an instant fur ball that is not transient. It becomes a way of life. If there are time errors (there usually are), these can be transient if the errors in a string of successive jobs tend to be random and cancel each other out. However, if capacity numbers are biased toward better performance than is realizable, then they accumulate. You now have a capacity error, and that’s bad news.
In the next column, we’ll dig deeper into the fur ball and set some direction for detangling it.
The FABRICATOR is North America's leading magazine for the metal forming and fabricating industry. The magazine delivers the news, technical articles, and case histories that enable fabricators to do their jobs more efficiently. The FABRICATOR has served the industry since 1971.