Raised-Panel Wainscoting for a Spiral Stairwell

true frame-and-panel wainscoting with concave panels rising up a spiral staircase. Woodweb's top trimwork minds chime in on this one for a top-flight discussion of a very tricky topic. July 9, 2007

Can anyone describe for me how raised panel wainscoting winding up a spiral stairwell is made? In other words, typical raised panel detail, but concave to apply to the wall, and slanted at the pitch of the stairs such that the rails are parallel with the slope and the stiles are plumb. Does anybody know of someone who provides these panels in paint grade?

Forum Responses
(Architectural Woodworking Forum)
From contributor A:
Do you have plans that show the details? I am sure there is more than just that going up the stairs that needs to match. Also, what is your time-frame, and location if you want it installed?

From contributor B:
This is a great challenge for architectural woodworkers. It involves helical, curve, angle, and springback. You need to build a curved wall clamping jig that duplicates the radius and rise and run of the helical stair wall that you are working on. The panels can be 3/4 MDF with kerfs in the back so it can be bent to the wall. You would need to make solid wood profiled edges for the panels. The top and bottom pieces of the panels need to be strips glued up at a curve before milling profile. Compensate for springback. You can measure the springback after gluing up the first one. The kerfed panels will have to be cut into parallelograms. The cut angle equals the hypotenuse line from the stair, rise and run triangle. I would glue the wood pieces on that go around the panels, while the panel is held to the wall curve. Then run the panels on the shaper for the profiles.

The wainscot top and bottom rails are going to be helical shaped, about 8 to 10 feet long I suppose. They need to be glued laminations, on your curved clamping jig that matches the stairwell. Your strips have to be cut at radius first, before twisting into the helical shape for glue up. Should make for an interesting setup on the shaper or router for the profile on the rails.

From the original questioner:
This is about the difficulty I figured. Plans call for four flights of this helical raised panel wainscot detail, and most likely two or three other wainscoted rooms that are either circular (in plan view) or elliptical. I'm in the Boston area.

No useful plans exist yet at this level, but the client provided photos from a house that has a similar stairwell. I have license to redesign to make things more practical, if that's even possible. Paul is also correct that it's not quite as simple as the wainscot matching the stairs. At the landings the wainscot must transition from level to the helix, and we at least want to investigate the extra work and cost involved in having panels that would do this transition within the panel - very challenging. Thanks for your responses.

From contributor C:
A simple solution which I have used in this situation on paint grade work, is to build the panel center in place by applying layers of thin plywood, and utilizing a high quality rubber flex molding in the shape of a raised panel edge

From contributor D:
Another alternative not mentioned would be to segment the rail sections. Not as graceful and clean as true laminated components, but adequate in limited circumstances.With regard to the transition between the stair and landing wainscots, the natural transition between the two paneling types is at the plumb cut on each end of the stringer. We rarely produce a panel that bridges both sections as the appearance can look awkward. If I can be of more assistance, feel free to email.

Click here for full size image

From contributor E:
To contributor D: Nice staircase. Are the raised panels flat faced or are they curved?

From contributor F:
This is historically one of the zeniths of the woodworker's skill. It can be done with ticky-tacky and cheap this and that (and will show accordingly), or be a true point of pride for owner, maker and designer. Be sure to either hire someone who has done this work before (preferably more than twice), or follow the clear advice of someone who is willing to share his expertise with you. I have done this work about once a year for the last 25, and know of only one other shop in the area that has ever done this successfully. I have seen some horrendous attempts made, to the detriment of the craft.

From contributor B:
Looking at the curved stair photo, I see faceted curve wainscot. The panels and rails are straight. This is not nearly as difficult as making curved rails and panels.

From contributor D:
To contributor E: The panels are flat, as are the top and bottom rails (short sections). The joints between the segments occurred at the junction of "1/2" stiles and the segmented rails. The cap was a radius pc. Thank you for the compliment.

To the original questioner: I agree with your assertion that this type of wainscot project demands the highest level of fine woodworking skills. I would add that a comprehensive knowledge of stair building would also be required. A close proximity to the job site would also be ideal. As with every project, we (the client, designer and tradesperson) tend to have a preconceived image in our head of the finished product. This image is typically idealized to the nth degree, representing only the finest in materials, crafted with only the highest of woodworking skills on the best equipment. The fruition of this ideal image would test the skills of the most experienced woodworker, be quite costly and require a large time investment to realize fine and lasting results. Your work appears to attest that it can be done, but I would argue that it’s a rare occurrence for most tradespersons or shops.

I would also argue that attaining a “true point of pride for owner, maker and designer” can be realized on many different levels dependent upon the consensus formed by all parties involved at the beginning of the project. Using “ticky tack” materials and less than true manufacturing processes may be perfectly acceptable to achieve the desired effect. If the end result is a horrendous example of the craft, yet the client is satisfied, we’ve effectively completed our obligation to the project. Our obligation to the trade is personally formed. Some have none, others overflow with pride.

Limited skills, poor workmanship and lousy design abound in the end products of our industry. Examples are everywhere. These instances should be analyzed in context and compared to established norms to assess their validity as a standard practice. Most do not live up to our standards. Others however, provide innovative solutions to complex projects (radius stair work). While these different ways of doing things may fall far short of our preconceived and many times subjective standards, dismissing them as somehow not worthy stifles innovation and creativity.

To contributor B: Nope, not nearly as difficult as true radius work, but the project met the client’s budget, their expectations and lastly, we did very well. Am I crazy about the look? No, not really, but the client was satisfied when complete. Would we or have we done it the same way again? No on both counts. The photo above simply illustrates an alternative process to those presented earlier in this thread. Its applicability would depend upon many factors

From the original questioner:
Thanks for all your input. And a special thanks to contributor E who gave me some great advice today about (vertical) stave-coopered construction for the panels using custom made concave stave stock.

First, I appreciate the segmented approach and think it definitely has its place. It's done very well. However, this particular client seems to have bottomless pockets and probably seeks a “true point of pride for owner, maker and designer” as contributor D puts it. So, flex molding and in-place lamination is out. (I can't figure out how the latter would even be possible.)

My current thinking is to follow contributor E’s advice and have cupped staves milled from custom moulder knives that I'd draw in AutoCAD. Joining the cupped staves together in a form would make a smoothly curved trapezoidal blank for a panel. My sense (pending pricing approval) is that a 5-axis CNC would be the way to go to trim the panels to size and apply the panel raise. I have a math background and a fair knowledge of Autocad so I think I might be able to interface between the site and the CNC shop. Alternatively I could run it on a shaper, but I'm thinking the CNC is the way to go to get perfect geometry, and would be even more helpful for the transition portions at the landings. My next thought is choice of panel material. I think poplar or MDF, the latter being good for stability which is critical.

Another thought - since I have some license to determine the stile/rail profile, panel raise profile, and top rail cap, perhaps I have the CNC mill the stair skirt board, base cap, bottom rail, panel, top rail, stiles, and top rail cap (chair-rail) all in the same slab - disadvantage being that seams would run continuously top to bottom and if it shows through the paint it would be a give-away that it's not built in the traditional method. Assembly and continuity at "joints" would be as good as possible however, and if it's 100% MDF there's little chance of movement in this 100 year old building. (Oh yeah, forgot to mention that it's an existing stairwell. I will be checking radius consistency very carefully.) Panels could be made with staggered joints to mitigate this telegraphing joint issue, but unless they're staggered by a pretty large dimension it might not be worth it. Plus it would seem to limit the max size of a particular panel as imposed by the CNC bed size.

In case I didn't mention this, this detail runs over the five storey height of the 10' diameter spiral staircase and is also found in two or three elliptical rooms elsewhere in the house. I think it's time for a beer.

From contributor E:

Be careful with MDF. I've worked on (and owned) many historic buildings and can tell you first hand that original style materials is a key issue in many owners' minds. I'm not against MDF. I've worked with it a lot and have even used it on some architectural details in my own home. If I'm working on what I would consider a historic building though I would most likely not incorporate it into the project.

From the original questioner:
I agree - in any case I would always describe to the client all the possibilities and the pros and cons. The MDF vs. poplar question would be discussed. Sometimes I think the composite material is actually superior to solid wood, and paint grade panels are one such case.

From contributor G:
If you plan on using contributor E’s stave method wouldn't that eliminate the MDF paneling anyway? I am very interested in this thread. My builder submitted bids for a rather intricate house that has a spherical staircase. We would have the opportunity to do the raised panel wainscoting. If we get this house, I will be flat out for the next 2 years. The crazy thing about the bid was the drawings. The plans were one of the best sets I've seen, except for the interiors.

The interiors were only concept sketches, virtually no details. I would describe them as quick sketches that were done on graph paper, photocopied 3 times, faxed once, and photocopied several more times. I love those cost/plus projects with an almost unlimited budget.

From contributor H:
"My current thinking is to follow contributor E’s advice and have cupped staves milled from custom moulder knives that I'd draw in AutoCAD. Joining the cupped staves together in a form would make a smoothly curved trapezoidal blank for a panel."

I've never been involved in a project quite as ambitious as this, but I've been following this discussion with interest. The preceding line of thought confuses me. If you were to use solid material for the panels, why would you go to the trouble of milling cupped staves to glue up the blanks? On the outside of a 10' staircase, the concavity of a 4" wide stave would only be about 1/32" deep. What's the point, if you're just prepping a blank that will be machined again after glueup? Or do I misunderstand?

From contributor F:
To the original questioner: It sounds as if you have a nice one coming together. I agree that the radius may not warrant cut staves, since the deflection from flat is so little. Even though this is paint work, the wood grain can show in certain lights (and should, in my opinion), and the narrow verticals could show negatively - especially if this stuff moves any now or later, and the joints telegraph. A panel of plywood will also prevent panel movement and paint cracks.

We have laid up thin plies of plywood on a form and vacuumed to the radius (plus, for springback), then sized the panel, then added twisted panel raise to the top and bottom, mitering to the straight, flat verticals. Sounds intimidating, but works out well. We can make the twisted panel raise and retainer (panel) molds on the shaper in lots of pieces, jig them up for mitering, and it goes quickly. You can also do the transition from curved wall to flat, raked to level with the above.

I don't speak CNC, so can't advise you there. I do find that the old school works very well - very intuitive - especially with good communication drawings (CAD), and 21st century tools (shapers and feeders and vacuum bags) and techniques.

From contributor B:
To contributor G: Is that really a hemispherical stair? Imagine building a stair that goes up a dome. With the rise of the stair all being the same, the run of the steps and radius of rail would change with each step up. Is it a helical or freeform curve - no standard geometry shape?

From contributor G:
To contributor B: I made a mistake. I meant true helical. I hadn't gotten my fix of coffee this morning. The spherical would indeed be challenging at the least.

From the original questioner:
I've visited the job site and am attaching pictures of areas to be wainscoted. This first picture is of the front foyer. All walls including going up the stairs want to have the raised panel wainscot up to about 32" off the floor - matching a detail found in an adjacent room. The wainscot is supposed to flow in the gentle curve to follow the existing baseboard. I'm assuming the 32" height will be in the spiral staircase as well. Pictures of that will follow.

Click here for full size image

From the original questioner:
View from the 5th floor landing down to the bottom. A domed sky light above my head provides the natural light. Sorry about the soft focus. I've measured the stairwell diameter as varying between 97" and 99" - tighter than I expected. You can see the baseboard gradually changes pitch at each landing and the helical raised panels are supposed to follow this form.

Click here for full size image

From the original questioner:
One of three partially elliptical rooms also to be r.p. wainscoted. Taking all three together - the front stairs has the panels warping in the vertical dimension, this room has it warping in the level dimension, and the helix has it warping in both dimensions. It's like a full kindergarten through Ph.D. education in the craft!

Click here for full size image

From the original questioner:
To contributor G: I fired up AutoCAD and assuming 98" diameter stairwell and 5" staves, the max deviation from flat at the center of each stave is .066 - just over 1/16". This is small but seems easily large enough to appear faceted when painted. I could use narrower staves to smooth it further, but then I'm dealing with more parts in assembly and more joints to move and crack. My thought about MDF is that I can just as easily stave MDF instead of solid wood and so ensure zero movement in the panel. In my experience MDF glued edge to edge with a spline or a few biscuits is totally stable and I suspect even less likely to crack than poplar, which itself should be very good over the years. My thinking with MDF staves is that I could go even wider than 4"-5" -- as wide as a moulder and the thickness of the raw stock permits. Fewer joints still, and faster assembly. Using MDF is a departure from the true traditional method, but I'm not at all sure there's any REAL quality sacrifice. In any case, I would do as contributor E suggests and discuss all these questions with the client.

Getting back to contributor H’s question of the necessity of milling cupped staves: Last we had custom knives made it wasn't very expensive - certainly a drop in the bucket on a project of this scale. My hunch is that the benefit of getting to the point of having perfectly smooth curved panel blanks ASAP is worth this cost of the custom moulder run (unless I really underestimate the cost of the custom run).

I'm not a CNC guy either, but have talked to a few who have underscored the importance of reducing the machining time for the ball mill to hog out material to define the large-area curve of the panel field. With the custom knives, this step is already done. All the CNC has to do is cut the profile and the panel raise. Or at least this is my thinking at the moment.

To contributor F: What you describe does sound intimidating but I'm sure like most things, it's pretty easy in hindsight. I'm still concerned about using a lamination for the panels because I would be afraid of the panel raise shallow angle cut through the thin layers. I guess I'm not so worried about delamination, but telegraphing seems inevitable and that sort of telegraphing would bother me a lot more than a telegraph that reveals stave construction, which is almost quaint. But again, maybe I'm making too much of that issue and it just wouldn't be a problem. Of course I have no idea what this CNC work would cost, so if the price is sky high all of what I've written might be moot.

My next concern is the helical rails and warped stiles. It sounds like I should talk more with contributor E about the details of making the rails. If anybody has a clue about what work like this goes for in a top-tier housing market I'd love to know. My boss has approached me about taking this on and at the moment I think it would take me days just to figure my costs, and another few days of meditation and soul searching to estimate my risk/reward and opportunity cost. I'd love to hear from anyone who has done this sort of work and could suggest some ball-park figures for a fair market value.

I forgot to explain that I have a complicated relationship with the person I refer to as boss. I work for him - though indirectly, through a foreman, on a part time basis and also have my own custom cabinetry side business and am independently insured. This sort of work I would only do on a contract basis, so I do have to generate one.

From contributor I:
Wish I could work on this project - these are the ones that are so much fun you find yourself thinking about it after you leave the shop. Just a thought but have you considered doing a bent lamination for the panels out of say three layers of 1/4" MDF? Could be vacuum pressed pretty easily to a form in 4' x 8' pieces of the right radius and yield 6 - 9 panel blanks per glue up depending on the width and height of the panels. Advantage would be no joints in the flush face of the panel to possibly telegraph. Plus the problem I see with using stave construction with MDF is that you are going to be cutting through the tempered face of the MDF in the center of each stave and exposing the core, so that the face of each panel will alternate between tempered and untempered vertical sections. In my experience, these flush transitions in MDF texture are very difficult to paint without some ghosting through the finish, even after double priming and sanding.

From the original questioner:
To contributor I: That's a good point about the tempered vs non-tempered in the MDF panel fields. I had just sort of dismissed that as a problem but hearing you say it makes me reconsider. Of course I've never done anything like this with MDF, but in my work as a finish carpenter I often expose MDF core which then gets painted, usually just an edge. I've used everything from joint compound to bondo to harden the core and it works but you have to be careful. In your scenario just the panel raise would be untempered, but that's the case with any mixed wood/MDF center-panel paint-grade cabinet door. Maybe I shy away from laminations because of (1) not having done it, (2) lack of equipment, (3) springback (4) fear of delamination in a way that would be horrendous looking and nearly impossible to fix. (5) the prospect of the worst kind of telegraphing in the panel raise which would be a dead give-away how the panel was made.

Your comments have me thinking more about making the panels out of solid poplar now. Maybe 4" staves and primed with a good oil base primer on all sides. I'm guessing this is the way it would have been done 100 years ago. Cross grain movement in a 24" wide panel is something I'll lose sleep over, but the house has central AC and enormous thermal inertia, and I suppose I could consider using polyurethane construction adhesive to bond the panels to the wall surface. Does anybody know of a company or individual who has done exactly what I'm describing with panels that transition from helix to flat with varying radii of curvature?

From contributor J:
If I had this job, I would mill that camber in the face of poplar, the mill a large 1" diameter nose on one side, and flute on the other. This is a common canoe edge aligning technique, and glue them up with thicken epoxy, by making a clamping frame with the desired arc. If you leave a tab sticking up on one end of the frame to stop the first plank, then clamping the last plank, forces all of them into the form for perfect alignment. If you do this for the full plank length, then there will only be waste on the two end parts. I would vacuum bag thick veneers run diagonally around a form to get the helical arc/ twist.

I am wondering though - are you going to break out the plaster, and try to get your work behind the base, or do you just plan to set your work on top of the base in front of the plaster? I think I would lean toward the later. You would have to totally disassemble the whole stair to get behind the base, which is captive by the treads, which I am sure you have noticed have the balusters on their other end. It looks like a big bucket of worms to me.

Do you have the face of the frames and panels on the same plane as the base? I suppose you could make another base that goes over the existing base. If you cut off the nosing of the treads, then you only have to cut the rise/run cuts to get a decent caulkable fit.

From contributor F:
I referred to the lamination of the panels, and the subsequent application of a separate panel raise mold to make the panel raises. This eliminates any exposure of face laminations or vertical stave laminations.

From contributor I:
I just had another idea that might be a sort of compromise. If your panel raise profile incorporates a step right at the face, you could do the stave core out of MDF, then laminate a 1/8" MDF skin over it, the seam of which would line up with the step in your panel raise profile. This would give you the one piece face without any springback and without a lamination joint occurring somewhere at a very shallow angle in the middle of the profile. I would really try to figure out a way to do this in MDF - yes, solid is the way it has always been done, but the wood that was used to build houses of this vintage came out of huge slow growth trees with growth rings so tight you could barely see one from another. A great deal of it was quartersawn. Today’s lumber is nowhere near that quality and I believe it moves around a great deal more.

One more thing - for the rails take a look at Ultraflex brand flexible trim. They sell an s4s stock that I have used for this type of thing. It profiles and paints really well and it will easily bend to this radius - last time I used it was on a 57" radius.

From contributor E:
Contributor I makes a good point regarding old growth lumber. I'll add to this that pre-1900 there wasn't the issue of central heat (or air conditioning later on) to affect wood movement.

Here's another point I've been pondering in this discussion. As most everyone here knows that a hand rail or skirt board traveling up a cylinder needs to twist to remain in a vertical orientation. However, I am visualizing these raised panels and I don't see a twist.
For example, if you took a large cylinder and cut a square out of the side you would have a curved square that matched the cylinder wall. If you then angled the top edge and bottom edge to match a rise/run situation, you would still have just a curved section of the cylinder with an angled top and bottom. It seems to me that the panel situation is simplified by the fact that there is no twist in its construction. Do others agree on this point?

From contributor J:
Yes, I agree with that, and have all along. However, any of the rails and parts that run around following the pitch will need to have that figured in rather than just a simple curve on a single axis. I don't understand why anyone is worried about glue-lines showing through paint on a solid stave glue-up. As long as the panels are kept to a reasonable width, with room to move, I don't think there is an issue. Just to clear up my suggestion to cut the nosing off of the treads to get a new base on top of the existing one, I was only talking about only 3/4" off of the end, or just enough so-as not to have to cope around the nose, and cove under it.

Judging from the number of people contributing to this thread, there seems to be plenty of people who would like to have the challenge that you are facing.

From the original questioner:
To contributor J: Your idea to bullnose/cove the staves is fantastic. This simultaneously increases the glue area by 30-40% (I'm guessing here), and, even more importantly, makes the quality of each glue joint higher on average because it becomes totally forgiving of misalignment during glue-up. Regarding the assembly form you describe, I had exactly that plan - without the tab though - that's a nice addition.

I will figure out the cross grain movement in a 24" wide poplar panel in Bruce Hoadley's book and will think more about the MDF/solid debate, and CL's idea about the 1/8" surface lamination. I too am not at all worried about stave telegraphing. As I said before, even if it happens (and I think it wouldn't happen generally), I think this is the nature of the medium, period. Embrace it!

I wonder if contributor J’s stave idea would help with the transition areas at the landings. At these points the panel simultaneously changes pitch and curvature. (I realized in talking with contributor E that the geometry of this ribbon in space is defined by only two variables: "r" - the radius of the stairwell, and, "s" - the slope at any given point along the helix. There is no twist therefore, as suggested.) Having confirmed the walls are reasonably consistent in radius, I think I can assume r to be fixed at any point away from a landing. Hopefully s is constant in the mid-flight areas too (will check). This should make a mid-flight panel from the first story interchangeable with a mid-flight panel on the 3rd, etc. The landings are very special though and while I could parameterize them by finding the r and s for incremental points along the length of the transition panel and interpolate the curve, I'm wondering if contributor J's idea for the bullnose staves might make these problem areas much, much easier. What if I use the wall itself as the form for the transition areas and glue sufficiently narrow, flat bullnose/cove staves to one another while pressed to the wall. I would wax the wall first, so I can remove and machine the panel after it's cured. The panel would be slightly faceted, but a sanding with a disc sander would smooth it out just fine at typical radii. A little bondo could be applied if not. Back for more later. Thanks to all of you. I will definitely keep you all in mind if I actually get to build this thing. I certainly can't do it alone.

From the original questioner:
To respond about the plane of the paneling surface, at the moment I feel strongly that I need to remove the plaster and work against the framing due to the issue of the existing baseboard, which I do NOT want to reinvent. I would first remove the existing base cap, and set it aside, assuming to replace it afterward. I would then strip the plaster from the top edge of the baseboard up to slightly above the panel top edge. I would hope the plaster/lath total thickness is about the thickness of the panel - say 3/4", so that the baseboard and basecap never notice that the tablecloth was pulled out from beneath them. On the other hand, if the plaster/lath is thinner than the panels, I could back-plane the cap at an angle so that its fattest point (at the bottom) is skewed in towards the wall. Anyhow, a minor detail which I hope to someday have the privilege to ponder.

So if the plaster is coplaner or very nearly coplaner with the panel surface, a helical version of a typical flat chair rail profile would bridge the gap between the ragged plaster edge and panel's top rail - a typical trim design for wainscot.

From contributor E:
Three things:
1. I would think twice about pulling plaster. At least from my historic preservation perspective I see the removal of existing material to be a serious negative tactic. Once removed original material cannot be replaced. Once covered the covering can be removed and the original material has a chance of being restored. My attitude is that an architectural component has been there for 100 years or more. If left alone it has a chance to still be there in another 100.

2. I don't know that the transition from curved wall to flat wall at the landings is a big issue. Panel layout can be designed so that there is a stile between a curved wall panel and the following flat wall panel. As to the transition from downward travel to horizontal travel of the rail system that is a basic change not unlike any compound curve. See the attached photo of a set of 1/2-round casings in a curved wall. This would be a similar sort of construction project.

3. The bullnose self-centering stave scheme is a good one. I used a similar idea on a small production run family gift project this past holiday season. The eight sided box was able to be glued up with nothing but a couple rubber bands due to the self-aligning joints. While working on this it occurred to me that column makers must so something similar.

Click here for full size image

From contributor I:
I agree with contributor E regarding the plaster. In addition to the reason he mentioned, I would be willing to bet that the framing behind the plaster is not concentric to the stair at all. These old houses are almost always built with full size rough sawn lumber, and so the flatness of the wall where the trim elements meet it was dependent not on the framing, but on the plaster grounds that went around the doors, windows, and just behind the baseboards below the base cap. One way to deal with the wainscot/skirt plane issue would be to strip the base cap and discard it, and think of the skirt as your bottom rail instead of a baseboard. If there is enough room, you could jig up a router and profile the top edge of the skirt in place to match your other rails and stiles, and set your panels right into this groove. If that didn't work you could make up the stile/rail profile as a small moulding and nail it on as a replacement for the basecap you take off.

I also agree with contributor E that the panels should be laid out and divided equally from a sharply defined point where the radius begins to leave the flat plane of the landing walls. Thus there would only be two types of panels - flat ones and radius ones.

On the solid v mdf debate, the problem with solid isn't joint telegraphing, it's panel shrinkage. Even well dried stock is maybe 8-9% MC. Take a moisture meter and measure the MC of that skirtboard, which has reached equilibrium with the indoor space. I'll bet it's below 6%. When it's painted the joints between the panels and stiles/rails will be bridged by the finish, and when the panels shrink, that paintline will crack and pull paint on both sides of the joint. May take a year to do it, but it will happen and then you are faced with scraping all those joints and repainting. As an aside, this type of job to me demands an old school finish that fits the house, meaning every hairline crack caulked in and several coats of oil enamel, applied by some good painters who still remember how to use a brush and not just an airless.

Have fun with this one, it's a great project and you have the benefit of some great advice from all the other envious trim and shop guys like me who are watching this thread.

From the original questioner:
To contributor E: Your aversion to removing plaster hadn't occurred to me. In my job we're constantly modifying or demolishing 100 year old plaster and lath walls, joining them to modern blueboard/skim coat walls, shoring them up with large modern washer head screws, etc. etc. I guess we're not a very respectful bunch!

The problem is I just can't stomach the alternative which would seem to force reworking the baseboard. The recent wallpaper is getting stripped anyhow, which will probably pull off chunks of plaster which will then have to get skimmed again regardless. Besides, if in 100 years anyone seeing my wainscot laments the missing plaster and lath, I'll most definitely roll over in my grave.

About the transitions at landings - maybe I am overreacting. I was staring at my top view picture trying to discern how abrupt that cylinder-to-flat transition is, and where one might choose to place stiles which could eliminate the need for variable compound curved panels. The stiles might want to relate to the treads, perhaps one stile every other tread? What does everyone think?

To contributor E: If you were to make these helix rails, would you suggest milling them as you would standard five piece door rails? (with maybe an ogee edge and a groove for the tongue of the panel?) Is this too ambitious, particularly when taking into account the landings which would mean the rails would have to flair out to level along with the panel? I'm a little turned off by the idea of flex material for the rails, but that would make it far easier. What if the rails are milled square and a very small wood quarter round is applied at the frame/panel joint. This would get around the need to cope the profiles. A profile that small should easily bend to radii this large.

From the original questioner:
To contributor I: Okay, you're starting to bring me around. I agree about the likely framing situation and the plaster grounds which I was cursing not six hours ago working on another old house. And besides, stripping all that plaster would be miserable. I'll have to think about the look of omitting the basecap and forsaking the break in plane between baseboard and panel. It strikes me as a more simplified look and maybe not strictly in keeping with the rest of the house. I'll think about it though.

From the original questioner:
What is the consensus on the best way to do the panel raise on these curved blanks, keeping in mind that I don't have a shaper and have never used one. I could buy one and figure it out, but before taking that step would probably look more into the CNC method and some contraption/jig that might make it possible with a hand held router and a vertical panel raise bit. If the hand held method is even possible, it sounds a little scary with solid wood, but maybe not so bad with MDF, (or lightweight MDF, easier still). Does this warrant experiment or am I nuts?

From contributor I:
If it was me I would use a vertical panel raise bit in the shaper. You would make a convex wood fence to shroud most of the bit and run the top and bottom edges of the panels face in. Running face in means there is no chance of ruining the panel by coming away from the fence or the table - if you do, you just run it again. This is always preferable to face up or face away from the fence in these custom situations where it is difficult to use a feeder. For the vertical edges you would need to make up another fence, this time straight but curving away from you so that the panel is run face in and tilted to the right. The radius of the curve in the straight fence would match the concave face of your panels, but you would have to experiment with how much it is tilted in relation to vertical, until you find the position at which your tongue is the right thickness, and your profile miters properly with the top and bottom profiles. You know what I mean? The more you tilt the panel into the bit, the tongue stays the same thickness, but the step in the profile at the face gets deeper. You would just have to play with it until it matched the top and bottom edges, which I think you would want to do first.

Sounds to me like you would really benefit from teaming up with a good shop, and splitting some of the project up between shopwork, sitework, measuring, installation and coordination. Preferably a small shop that does primarily custom work and not one of the larger shops that cranks out edgebanded particleboard all day.

From contributor K:
A couple of ideas occurred to me while reading these posts. Because there is very little cove to these panels, there are a couple of things that I would try. Glue them up flat and carve them on the CNC machine; the back side first while the face lays flat on the table, then cradle the panel in an MDF fixture (coved to match), and cut the face on the machine. The coving/carving on the back side can be quick and dirty because it doesn't show- the face side would be slower for a better cut. I think you could get a good enough finish without excess machine time that would clean up with a random orbit sander very nicely.

The trapezoidal shape of the panels could be cut on the machine at the same time- there is no need for 5-axis cutting because the curve is so shallow that the slight difference in angle at the edges will not matter. The raised part of the panel can be cut on the machine with ballmills or a raised panel cutter; the panel cutter may require an additional curved jig to allow the machine to approach at the correct angle - the top and bottom are done first as an arcing move, then the panel is held at the correct angle for a straight cut with the same cutter.

Alternatively, it seems to me that you could lay out a thin (1/4") template in a CAD program (or just mark it on the wall) and "unroll" it for CNC cutting- this template could be screwed/clamped to the panel blank to use a bearing/bushing/offset to guide a hand-held router for the panel raising.

The stiles and rails are straight-forward vacuum formed strips. I think you will find that a router with stile/rail bits will probably make these cuts in the blanks. Think about how the old guys did this before; planes with shaped soles. You may need to make a very slightly curved plate for your router for some of these operations. Of course, all the curved plates, cradles and jigs are made easily on the CNC machine, but you can do it all by hand if need be. I think that once you get grooved-in to this job, things will start to become clear to you, and it will be fun.

It is a little hard to explain some of these things in writing, but I can visualize how this would start to come together, and I think that a good CNC guy is going to be an ally here. I would lean towards poplar for this job, since it is traditional and machines and paints well. The concern about glue-lines is moot- there will be a lot of joint and glue lines just like elsewhere in the house, and they will blend in to the overall look; the old guys had to glue up panels too, once in a while. The panels should be floating in the frame to allow for movement, you can glue the center portion to the wall, but let the sides move to avoid problems with the panel glue joints.

From the original questioner:
To contributor I: Thanks, I finally "got it" regarding the shaper setup and completely understand the issues you describe. (I know what I'm doing as a finish carpenter but have never worked in a full cabinet shop so my shaper experience is limited to my router table.) Anyhow, I see this now as very straightforward once I have the setup tweaked just right. A 3hp router in my router table should be adequate for starters to test the process, and if needed I could make two passes, just shaving a little on the second pass for a cleaner cut. You're right - I could also talk to a shop I know nearby about running the raises for me on their shaper.

I got a call from Jess Coleman from James River Millwork today (who has done this type of work) and he advised against MDF for the panels due to durability concerns, particularly of those corners on the panel raise. We also talked about the panel movement in a stave panel and he felt that provided it's reasonably dry from milling to priming and installation, that some paint cracking is likely but that it shouldn't be terrible. Someone also suggested an epoxy sealer in place of or in addition to alkyd primer on all sides of the panel before install. I can't remember the numbers on moisture exclusion but I think epoxy is in the range of 90% with 100% solids, and alkyd enamel is more like 50-70% after a couple coats.

I also appreciate an email from contributor F whose methods I initially dismissed way too quickly. I thought about his hybrid lamination/solid raise construction on my drive home tonight and reasoned my way through how he might do it with the jigs and fixtures which I can now sort of visualize. As he said at the time, it sounds complicated, but does seem to provide the best of both worlds, with a stable panel and solid edge. I'm curious how he attaches the mitered edges to the laminated panel and how easy it is to get the miter joint line to agree with the spring point of the panel raise. Also, how important he thinks this is assuming the miter joint might crack seasonally.

I'm see now the CNC might not be necessary at all for the sprial, though for the 2D curves in the front stairwell it might speed things up if only for templating. It would take me maybe an hour to draw that S-curve of the baseboard accurately in AutoCAD and all my frame and panel components could easily be derived from that curve.

From contributor L:
Great depth of knowledge posted here. I am tickled to see how the quality and correctness was developed and realized for all involved especially the readers. My first reaction was to do it first class or not at all. The difference in materials is minimal when compared to the entire job. Depending on the complexity of this project it could be low to middle six figures.

From contributor M:
I completed a double stacked freestanding freeform/helical curved stairway about a year and a half ago in Vail, Colorado. The homeowner desperately wanted the bottom of the staircases to be full paneled with bolection mould to match the extensive raised panel work throughout the home. Unfortunately, there was not enough time left to embark on such a project. And quite frankly, I had reservations about attempting such a complicated task without any seasoned help. Has anyone here ever tried this?

By the way, I'd steer you away from CNC unless you plan on rebuilding the house around the perfect parts generated by the CNC machine. I must add that I envy you! Projects like this are "Masterpiece" material by the most literal definition of the term. Will you be making the balustrades for this stairway also?

From the original questioner:
To contributor M: No balusters as far as I know. I've only been approached about the two partly-elliptical parlors and the hourglass shaped hallway that separates them (nice to have some convex curves too), the entire front foyer and hallway with straight-run stairway - and, it appears - only two flights of the spiral. These details were clarified for me today following a meeting between the GC and client. Apparently they might replace the treads on the spiral, which might mean work on the balustrade. Not sure. I can't imagine having time, though I've done balustrade work before and joined many custom shaped rail segments, etc.

I'm confused on recommending against the CNC. If I can confirm that a single panel shape is acceptable for everywhere except landings (and I hope I'm not too optimistic to think that), then uniformity is a good thing isn't it? I'll know more after I get my radius templates up against those walls to see what it does. I'll also measure average slope over many treads and compare that number to smaller scale slope measurements, like over various intervals of several risers. It seems to me this will tell me that the exposure on my bottom rail should remain consistent as you ascend. Should be close enough, and I suppose there's not much I can do if it's off, but I like to go overkill on the layout and templating stage to be aware of issues so I'm not surprised later.

Thanks to contributor E for the long talk this evening. He and I are currently leaning towards an MDF vacuum-pressed lamination sized and raised on the CNC (pending the pricing, due tomorrow). I hear that there are strong opinions among you, and most of them based on much more experience and broader knowledge than I have. The lack of consensus on stave vs. lamination vs. plywood/solid hybrid vs. MDF, etc. tells me there's more than one way. Maybe it comes down to contributor D’s point about presenting the options to the client and design/build team with as much objectivity as possible. With your help I can now do this, and also present an informed opinion for how I would do it given my resources and experience. All methods have downsides and are susceptible to failure in one way or another, so I'll choose what I feel is the least of evils while being realistic about what I can and cannot do in the given time frame at a given budget.

If this 29 year old client is fickle and backs out on the whole thing I'll be sure to belt him a few times for each of you, and a good kick while he's down, for me.

From contributor N:
One of the things you may need to look at with running your panels over a shaper is that the cutter, based on how tight the radius is, may rub on your panels at the outside of the cut. It’s hard to explain but the cutter edges hit the wood too soon and also too late. We have built a few different curved projects, some for very wealthy clients. My experience is that at some level it really is about the money. The kind of project you are getting involved in will be very expensive. Do yourself and your client a favor by really pounding that point home. Let them know that it will be more then they expect. Get a real good contract and a real good commitment from your clients. All of your great skills and effort will not be enough if the clients are not fully committed to this project. Additionally our experience with our in house CNC leads me to believe that utilizing that technology appropriately is really happening. Good luck with what looks like a great project.

The comments below were added after this Forum discussion was archived as a Knowledge Base article (add your comment).

Comment from contributor O:
I've done a lot of curved raise panel archways and stairways, and the best way for me was to glue 3 layers of 1/4" MDF strips together. The easier way to do any paint grade curved panel now is to use 3/4" AZEK. Cut your panel to size and then raise the panel with the detail and then apply to curve. AZEK will form to most radii without breaking. Too tight of a radius will break when you nail it.