|Home » Forums » Business » Message||Login|
You are not logged in. Consider these WOODWEB Member advantages:
Measuring a shops capacity for growth5/25
How is this done? (I know it's a biggy)
This was touched on in another thread and perhaps skimmed over, but it stayed on my mind.
So lets just imagine a shop has been running on cruise control but gets a sudden influx of orders. How do you know where overselling lies?
There are a couple of approaches.
To my mind, unless the shop is very technologically advanced with accurate definition of processes and routings, the best way is to apply experience.
If you have good analysis of process and accurate routings for all jobs (in process and firm orders), you can use that to determine your constraint. Then, based on the throughput at that machine/workstation figure out if you can meet the necessary deadlines.
Kay so say finishing is the bottleneck. You figure out you can get another shift or another booth or whatever.
Makes sense for sure--but then how "in the clear" are you? Do you just figure out the rest as you go along? If you double your paint shop capacity what does it mean in terms of the shop's capacity?
I'd imagine you displace the bottle neck, and end up in a bit of a chase-the-bottleneck kinda game?
Theory of Constraints [Eliyahu M. Goldratt]
I agree with David Waldmann's approach to a large extent. It is great when his approach results in the identification of a single constraint in the system. However, I am not sure it can always happen for every system very clearly. If you know:
(1) For each order, all the operations involved, estimated hours and resource requirements of operations and precedence relations among operations
then it is possible to do beneficial analysis for many systems (without any assumption on the existence of single constraint) using good software tools. I have been promoting software tools for scientific analysis and planning of production systems. Nowadays, the powerful personal computers will do all the necessary number crunching in a fraction of a second. I regularly demonstrate such tools over web.
Th use of intelligent software tools for meaningful analysis and planning of production systems is a weird concept for many people in industry although those tools can facilitate fast and extensive what-if analysis and proactive capacity planning. How to perform what-if analysis of dynamic production requires human thought process even though the computer does all the number crunching in no time.
"I'd imagine you displace the bottle neck, and end up in a bit of a chase-the-bottleneck kinda game?"
Come on, Mel. You read "The Goal"...
Of course that's what happens. It's never as simple as the example, but the principle holds. You need to figure out how to apply it to your unique situation (and modify as things change).
Note: as shown in "The Goal", sometimes being "more efficient" does not move you towards your goal. If you could double your throughput simply by doubling your finishing by adding a shift even at 1.5 times current labor cost, I guarantee that profits will go up. Why? Because your overhead $ remains the same, but they are distributed over double the sales, effectively cutting them in half as a percentage. This is the same reason that paying overtime is often not an additional cost, but rather a profit maker.
Our moulder is constrained by the tailer, because he must do quality checks and defecting on the fly. But if we increase the feed speed by 20% (still maintaining acceptable quality) by adding 50% more labor with a helper, it allows us to increase throughput by 20% with very little added cost. Compare that to buying a second moulder and the space to put it in, along with two more guys to run it.
I'm hard of understanding sometimes ;)
So I did learn about TOC in school, and seen it applied in mega-manufacturing, with fancy custom software as Prasad mentions. And read The Goal.
I get the idea for sure--but not really fully reaching conclusions on applying it prehemptively to a custom residential shop that is virtually dataless.
It seems people get an order increase and dive in full tilt and adjust as SHTF. But if one wants to get a general idea of problems to come before they come, I'm wondering how it would go.
A very quick metric is dollars per employee. Calculate the existing value for a shop. Double the expected output. How many more employees or employee equivalents required? Another way is man hours per dollar output. Again, very easy to calculate based on history, and gives you a rough estimate of what is required to increase output at current efficiency levels.
If you are wondering how easy it is to transform a shop that currently collects no data into a lean, efficient, scalable operation, that depends on the caliber of the management and the pool of available labor, and the ramp-up period required to train new workers. (Hint: outsourcing is probably a lot easier, but also requires management horsepower to keep on top of everything.) If current management collects no data, that's a big red flag that they won't be capable of managing growth.
Mel, you asked, " But if one wants to get a general idea of problems to come before they come, I'm wondering how it would go.
For this purpose, I suggested what-if analysis which requires reliable prediction of work progress over time for given resource availability. I am not aware of any reference to what-if analysis and work progress prediction in TOC or Lean.
Custom production systems can do better when they have what-if analysis capability.
The difference between a small company and a large company is how well organized it is.
Not in the sense of Lean or TOC or TQM but in the sense of hats being worn.
There are certain functions that have to be done no matter what if the business is going to survive. This subject is covered in the E-myth.
They are the business has to be put there in the first place, equipment, shop location, some workers etc. Following that you have to have sales, accounting, production, quality control, marketing and pr, and planning.
Growth occurs as those things are put in place. Hint it requires that the big kahuna realize that he cannot do everything and that the growth of his company is dependent oh the establishment of people performing these functions.
There are subdivisions of these main divisions one of which is training, which can best be placed under quality control. Notice the more successful companies do plenty of this.
Paul-- agreed. If it was my shop, I'd have data. Watched the old shop deal with a new contract for 500 homes in addition to their regular work, without an action plan. I will never understand why anyone would do that to themselves!
As for converting a shop to Lean-- I'm totally over that one. It ain't happening. Nor is my plan for inside sales in a local cabinet shop.
This question had popped up with my best candidate, and I still like the question, although the endeavor is being dropped. It's too bad, but the area is a little behind the times perhaps!
Prasad-- the manufacturers I once worked for had what if analysis. Powerful tool indeed! How many cabinet shops use your software out of curiosity?
Pat, indeed.... And sometimes the big kahuna with many illsuited hats also has family members running around with equally ill suited hats. It's not a pretty runway...
Lol... and I lost my only hat somewhere!
Only seven or eight wood working shops bought our scheduling software. A lot of people in this industry felt that even Schedlyzer Lite, the lighter version of our scheduling software, is too powerful for their simple requirements and needs daily effort to update job status information.
The usage of a powerful, scientific scheduling tool for analysis and planning of production systems is still uncommon. Lean and TOC methodologies do not encourage their usage.
Such tools can serve managers as useful decision support tools in complex custom production.
My exposure to such software was for a very large manufacturer--and it was the only way to achieve JIT at that scale.
That said still trying to apply learned principles unto smaller scales, sometimes with difficulty. How are those few shops that did adapt it do?
The philosophy, principles, practices and methods of Lean are certainly useful to manufacturing industries but in my view they are not truly adequate for managing many complex, custom manufacturing systems which are not necessarily large in size and revenue. The principle of continuous improvement and abundant optimism about Lean advantages can obscure the true limits of Lean in complex systems. Those systems may involve (1) orders with different quantities, priorities, due dates, routings, process & resource requirements, precedence relations among operations, etc. and (2) resources which may be multi-functional machines and multi-skilled workers with different calendars. Toyota does not have a need to develop right methods for resolving some major issues which are characteristic to complex custom manufacturing systems.
I am afraid my response to your query may appear as sales pitch. Our clients are able to confidently fix right start times for jobs / projects under the influence of many variables in the system and have good confidence in projected completion times of jobs. This confidence is very valuable to them. Moreover, they are able to do reliable what-if analysis with respect to any changes in the system.
Our business may operate a bit different than many on here. It seems that most shops are custom cabinet shops. Obviously, you don't start a job until you have it.
However, we do a combination of make-to-order and make-to-stock. We can create forecasts for the M2S products, but they will never be right (even less right than the weather forecast). So we leave "holes" in the schedule for these recurring orders to satisfy inventory levels.
Add on to that, our yields vary from job to job based on the vagaries of using natural wood. We can literally be off by +/- 20% from run to run using the same material from the same mill. Software only does what it's programmed/told to do. If you don't know what to tell it to do, it isn't going to be able to give you an answer.
Case in point. We had an order in process today for a M2S item. I think it was for maybe 500LF. At noon today I got an order for 1000LF. If we had scheduled the next 4 weeks according to a lead time as orders came in I'd have to tell him to wait a month or start putting out fires. This happens (usually less dramatic, but sometimes more) on a daily basis. Sure, we could up our levels of finished goods, but we have somewhere around 1200 items. I'm not going to stock the levels necessary to always have everything on hand to ship within a one month lead time.
Moral of the story is, you need to know your business. What works for you? What doesn't? All businesses are the same to a very great extent, but it's the little things that make them unique.
"Toyota does not have a need to develop right methods for resolving some major issues which are characteristic to complex custom manufacturing systems"
Maybe that says you should strive for less customization or at least parametric customization?
"(2) resources which may be multi-functional machines and multi-skilled workers with different calendars."
This very much caught my eye. From a programing sort of perspective, how is is this done?
Like for example take a table saw, is it say cut up in "upright cut, cross cut, dado, miter cut"? Or how about the employee? How is that cut up in skills? And if so, what sort of skill labels?
Also--how much data entry is required to set this sort of system up? Is it someone's full time job? Or is it something a journeyman or foreman does? And at what scale does such a system make sense? How easy is it to use on a one off project?
Sorry, it's a lot of questions, you just got my curiosity going :)
(Disclaimer--I'm a totally cashless nobody that's just really nerdy)
What you are inquiring about resources is done in our solution by creating appropriate resource groups that may or may not overlap. The groups can be formed by functions of machines, skills of workers and types of tools. Creating resource groups and specifying resource requirements of operations in terms of groups rather than individual resources is the fundamental and unique aspect of our solution (which also handles multi-resource requirements of operations) and this is where many of our prospects ignore the solution in spite of our offer of full help although it will enable them to generate optimal production schedules. This kind of work which is mostly done as part of one-time software configuration is not fully understood and appreciated by many people in industries. I think it is a headache for them. You would not see such pain in Lean and TOC.
Experienced people on the shop floor assign resources to operations in real time in a rational manner. We often encounter hurdles when we try to capture that knowledge and pass it on to computer through a scheduling model so that the computer can make such decisions in advance while generating a schedule. The difficulty of using our tool does not depend on the number of projects to be scheduled.
I can explain our scheduling paradigm in full details over a web meeting if you are curious about it. I have no problem if you reveal all its weaknesses on this forum later. I can freely answer all your questions during the web meeting. You do not have to be our prospective client for this purpose. Your curiosity is good enough for me.
Totally in! Though I should warn you that real time may be a slow pace--I've had to look up about 4 things from your last entry to understand it (no programming background, just worked a lot with IT for change implementation).
First question of the bat: if your resource group(clustering?) nodes are company dependent? And if so, that means you need to extract nodes, and strength of relationships, from users, per company? Is that the initial painful part?
Anyway I'll pm you for web meet details--thanks for indulging the curious :)