Lean Production of Raised Panel Doors

A discussion of tricks for evening out workflow, such as: to speed up the whole job, move labor to the bottleneck from downstream activities. June 28, 2013

I am the production/project manager of a 20 man custom shop in western North Carolina. I have been trying to implement some lean practices in our day to day operation. I'm looking for some good ideas for raised panel door production. I am finding that we can't get away from working in large batches, because of gluing up panels and wide belt sanding.

Forum Responses
(Cabinetmaking Forum)
From contributor T:
We recently reconfigured our door making cell in a way that it could be staffed by one person or could accommodate up to five people. Three of these positions have been simplified to where they could be staffed by the amount of training day labor would require.

The way to start is to list all distinct processes and time them. The goal is to have each process take the same amount of time. In some cases this will involve having more than one person on a station. To make this possible, the station has to be designed in a way that accommodates more than one person (think Lazy Susan, etc.).

Understand balance for staff sizes of 1-2-3 or 4. Just like soccer is played differently if you have three people on a team, the strategy will change if you have 6 or 9. You want your door department to be able to fluidly move from 1 to 4 people as demand requires or resources are available.

Not sure why batch size is a consideration at the widebelt sander or when panels are being glued up. I can understand batching for grain match or material yield, but after the material has been allocated, the batch size can certainly ebb and flow for all subsequent operations.

From contributor D:
"Not sure why batch size is a consideration at the widebelt sander."

Because running one door through the machine, resetting it, and running the other side is not very efficient when you have 100 doors to sand.

From contributor T:
Being efficient at the wide belt sander is probably a phantom efficiency. You could likely cut widebelt sanding time to zero and still not make a difference in how long it takes these guys to get a door onto a cabinet or onto a truck.

You can't get a door into the finishing booth until it's been widebelt sanded. You can't widebelt sand it until the panel has been glued into a frame. The last operation is the bottleneck. You want to feed some work to the last operation as soon as you can in the work cycle.

From contributor D:
Phantom efficiency? I want all my guys in all our operations to be as efficient as possible, and wide-belt sanding is an essential step. If you make any other assumptions about my operation, you are doing just that, making assumptions.

From contributor D:
"The last operation is the bottleneck. You want to feed some work to the last operation as soon as you can in the work cycle."

And that statement is a thing of beauty. It is a part of my DNA the second I walk onto the floor, so I would have to agree without reservation. But it still makes sense for us to batch through the wide-belt, panel shapers, moulders, and edgers, things that still, in spite of our best efforts, have a distinct and costly setup.

From contributor T:
One way to solve the problem is with bigger batch sizes. Another way might be to focus on setup reduction. The key, in any case, is to understand what the logic is for any particular batch size and see if you can make that reason go away.

The guys in the door department still cannot build their second door until they build their first one and they need a raised panel to build their first one.

If your organization has enough work to constantly staff all stations, then maybe efficiency at any particular station is paramount. Usually this is not the case. Your shop may be different.

From contributor D:
We staff most stations full time, but that doesn't change the fact that a small shop will need to process in batches that are dictated by fairly basic factors, such as tooling change-overs.

From contributor D:
We don't find this really practical, but I do like the idea of cutting stile and rails for the first door, then handing that to the shaper guy, then he hands those to the assembler.

From contributor D:
Contributor T, you say some very interesting things. For a small shop, I think this is true: "after the material has been allocated then the batch size can certainly ebb and flow for all subsequent operations."

To make the different processes match in time duration seems like a real challenge, especially with varying skill levels of different operators, but this scares me just a little: "Three of these positions have been simplified to where they could be staffed by the amount of training day labor would require."

What operations have you simplified so much that you can put a day laborer there and still get a consistently quality product?

From contributor L:
For any kind of repetitive work I try to reduce setup time or better yet, eliminate it altogether. If you are making doors using shapers, dedicate a shaper and feed to each function. If change overs are required, make setup fixtures to position fence and cutter head height if your machine doesn't have readouts. Keep a production sample at each machine as a comparator for the first part of a run. The only part requiring widebelting is the panel. The frame should be assembled dead on so only finish orbit sanding needs to be done, either hand or machine. The face of the frame parts should be nearly ripple free if it comes off a well setup molder or planer. Get or make a profile sander for the raised panel profiles or if your volume warrants a shape and sand.

From contributor T:
You can also minimize setup time for cope/stick shapers by stacking cutterheads.

We can move from Ovolo profile to Shaker profile in under 3 minutes without running any test pieces to see if the male/female patterns align. We do this by adding an adjustment shim to the copesled and stick shaper bed. Like contributor L, we also use a positive fixture to position the powerfeeder height when we make the change over.

If you think about it, marrying up cope to stick patterns is probably one of the most critical alignments you will ever do in your shop. Replicating crown molding can be a little bit tricky but the rubber meets the road when you can actually compare male to female and a misalignment will cause the door to twist.

This change over used to take our most experienced guy 45 minutes and about 10LF of scrap material to dial in. Moving from 45 minutes of high dollar labor to 5 minutes of anybody-can-do-it labor is a pretty big testimony for the reward from focusing on setup reduction.

Solving the problem with big batches just means you need more carts and more aisle space to navigate the carts. More work in process just equals more management.

From contributor Y:
Congratulations on going Lean. I would first be certain that my production staff fully understood Lean Manufacturing and that I have identified my primary bottleneck. Have you implemented a labeling and organization system?

Once those issues have been addressed think about:
- Does my stock replenishment methodology help reduce batch size?
- What metrics make you think your batch size is too large? So, for a given batch size determine the cost at each station as well as process cost (storage, movement, damaged or lost parts) for that batch size. Include the production station as well as its input and output buffers as one unit.
- As you make changes be sure to include the appropriate stakeholders in planning and design phases as early as possible.
- Look into change management techniques to help you implement the changes you want to make without negatively effecting the morale of your fellow cabinetmakers.
- Be sure to think about the system as well as the individual parts when you make changes.

From contributor K:
I don't know a thing about door production, but something you said caught my eye. The large batches and setup time on machines. I listened to Paul Akers and Art Byrne discuss machine setup times and they took a machine change from something like 12 hours to 5 minutes. Take a look at how you can solve that. A second machine or second set of stops on the same machine can save a lot of time and money if you can knock off a half an hour of setup 4 times a week.

From the original questioner:
First of all, thanks for the information - we have implemented much of it already. Some of the problems we are having have less to do with tooling setup (we are using stacked cutter heads) and more to do with operations that just take longer to complete. The amount of doors run through the system is directly related to the amount of cabinets for that job. Job has 8 rooms, four are alder stain and glazed with same door style, about 120 doors and drawer fronts, other 4 rooms are painted bathrooms, all four are different door styles with anywhere from 8 to 25 items. So usually we run the painted items first by room for finishing purposes, then go into the bigger job. We treat each finish color as a job.. Everything seems to move fairly smoothly through cutting parts and assembly and slows down to a crawling pace from wide belting to finish sanding. Any help with speeding this process up, apart from throwing labor at the finish sanding, would be appreciated.

From contributor T:
Sanding might be improved with just a change in lighting. In most shops the lights come from overhead. This tends to flatten the scratch pattern so that people oversand because that way they know everything was sanded.

Try this experiment: Take a flashlight and aim it directly at a door (90 from face of door) and see what the scratch pattern looks like. Take that same light and rake it across the face of door (i.e., flashlight is parallel with plane of door). In the second approach all of your scratches will become profoundly visible. In fact you will see that much of what you thought was sanded was still quite coarse.

This is a trick that photographers use to accentuate crevices and produce shadows. The models who appear to have perfect skin were photographed under a light source more like your overhead lights, big and diffuse.

If you configure your sanding station like a pocket screw face frame easel, it will be easier to see the scratch pattern. If the easel is configured like this and you use more than one grit of finish sandpaper, then operator 1 can do the 120 grit crossgrain, then slide the door down to the next guy with the 150 or 180.

The type of paper you use on the small sanders also makes a difference. Aluminum oxide is a real work horse but owing to the crystal structure of the stone tends to smash fiber. Silicon carbide crystals are sharp and tend to comb the fiber rather than smash it. Aluminum oxide at lower grits is probably adequate (and even desirable) for paint but you might want a different paper composition for your stain grade. If you smash rather than cut your wood fibers, your stain absorption will be uneven.

For some species of recalcitrant wood you might even want to consider misting the lumber face slightly and letting it dry. This will cause loose fibers to stand up when dry. They will be easier to cut off with sandpaper if they stand up from face of door.

From contributor K:
Another thought I had, if you are doing everything right and it is still a bottleneck, is an additional person might be your answer. If it takes 8 hours of labor to sand x number of doors, you can reduce the amount of time at a station by adding a person to it. It doesn't cost more, as you still have 8 hours of sanding, but it passes through that station in 4 hours with 2 people instead of 8 hours with 1 person.

From contributor T:
What contributor K is talking about is something called "leveling production." In a perfect world work would flow through every station in exactly the same amount of time. As he pointed out, if one of the stations takes twice as long as all the rest, then doubling the staff at this department gets back to that perfect flow.

You say that your problem bottlenecks from sanding forward. The other processes are the ones you have staff to draw from if you are deficit in sanding department. For this reason you should also experiment with leveling earlier processes. This effort will show you exactly where your surplus talent lives.