Wednesday, October 12, 2011

Following Mazerolle: Pavillion Carré Rampant Établi

Another in a series looking at French carpentry drawing from the 19th century masterwork by Louis Mazerolle, Traité Théorique Et Pratique De Charpente. Having crawled my way through the dormers in the book, I decided to take up where I left off at the end of the series on three-legged joiner's benches (<--link) or trépied établi. The word établi means workbench, and a établi de menuisier is a woodworker's bench. In the case we have here, is is a square pavilion (pavillion carré) on a sloping wall plate, which seems to be thought of as a bench, or rampant établi

Here's a perspective view of this sort of structure from the text:


It's meant to go on a stone wall structure.

Mazerolle shows another building on the same page which has the situation of a different sort of descending wall plate - a pair of hipped roofs atop a gable:


A most curious looking structure.

As you can see clearly from the first example, the wall plate slopes on all sides, and hence the overall plane of the slope travels on a diagonal, from non-adjacent corners. This is not the only case where one might have to deal with a wall having a titled plate line - another 19th century layout text by Delataille shows a couple of examples, the first with a tilt along just one orthogonal axis:


Like Mazerolle, Delataille shows these tilted plate buildings following on in terms of the lesson track, from odd 'joiner's benches' - like this one with 4 legs, Saint Andre's crosses out the wazoo, and a pronounced tilt:


Last, Delataille also shows an example with a plate tilted on a diagonal in relation to the plan, with a dormer tossed in for extra fun:


 Delataille's text has very abbreviated developed drawing of the pieces illustrated in perspective, however these are nearly indecipherable. To say they are cryptic would be a compliment. At least, they do not tend to encourage this writer to delve into them at this point in time.

In terms of other students of French carpentry over the years and what they have chosen to tackle, it isn't hard to find examples of this sloped plate problem within larger assemblies meant to demonstrate a certain school's technical repertoire - as this photo shows, if you look to the upper left side of the shot:


 That example shows a hipped roof with a ridge and the common rafters oriented square to the descending plate - and a Saint Andre's cross on another roof face for good measure.

Of course, a hipped roof with a tilted plate is but one step along the course of study - one could also have, say, conical roofs with curvilinear Saint Andre's Cross bracing, on tilted plates, like this example:


But, one thing at a time. I continue to work away at Mazerolle's drawings, which have their share of issues, as I've mentioned more than once in past postings. At the outset of the pavillion carré rampant établi exercise, Mazerolle makes the following remark:

"En pratique, les pavillions rampantes ne s'exécutent que rarement, c'est plutôt une étude de trait de charpente,..."


Rarement, indeed. So, this étude is all about studying what happens in a framed structure when the plate is tilted, another example serving to deepen a student's understanding of L'Art Du Trait (developed geometrical drawing for carpentry). I will say that while structures with tilted plates are pretty uncommon, I think if I were to see one, I would find it an utter delight, the kind of thing that would brighten up the landscape and bring a smile to my face. Those are the sorts of structures I wish I saw more of in the built environment - we've got quite enough big box stores and cookie-cutter McMansions already, don't you think? It would need to be done with sufficient tilt that it looked deliberate, if you know what I mean.

There is more than meets the eye upon initial inspection of this roof form though: in this example not only is the entire wall plate is tilted, but the interior purlins in the roof are also tilted just the same. Of course, those tilted purlins are sandwiched between principal rafters and common rafters, and are attached to diagonal truss struts on their back sides. The commons and principal rafters are aligned, in plan, more or less orthogonally to the descending wall plates.

At the outset, when considering the problem, I could see that some parts were going to have to be parallelogram-shaped sections, however I also know from past experience working problems in the text that the French carpentry preference seems to be the employment of square section timbers - to rotate a square section timber into a given roof plane and then make connecting parts fit it by cutting curious birdsmouth-like connections on their ends. There was plenty of that in the crazy Mazerolle sawhorse project from a year or so back. The French, at least as far as what Mazerolle (and As it have noticed in other works from that time) does, the upper surfaces of a hip rafter are backed in many cases, but otherwise the plan is to get square-cornered timbers to do the job everywhere, even if it makes for ungainly and ill-aligned constructions. But hey, beauty is in the eye of the beholder as they say.

I started drawing, and after oh, a day's work, things were not quite lining up as they were supposed to. That led to some erroneous conclusions on my part, and further steps down the wrong path, to the point of having the entire roof nearly complete. Then it became more obvious something was wrong. At one point it appeared that one of the fundamental aspects of the drawing, the diagonal tilting plane for the wall, and how it relates to heights along the walls themselves, was, uh, out of kilter. I communicated with my collaborator Tim Moore in France, who took a look and did a brief mock-up of the tilted plate, only to find that it did match the conditions indicated in the text.

I was baffled for a while, not knowing where things were going awry for me. I checked and double-checked my drawing, and while I could find nothing apparently wrong, the numbers still weren't adding up as they were supposed to. Then I decided to go back to a blank slate, as it were, and redraw the tilt of the plate from the beginning. When done, it conformed to the text's version! Upon further reflection, I realized that the way I had approached the titled plane at the outset of the drawing process, using a 3D rotate tool in SketchUp, had caused the discrepancies. Assumptions can be a killer sometimes! Nothing to do with SketchUp, just the logic of how I went about the drawing work, and the penalty paid for the wrong turn. Oh well, only 8 hours of drawing lost....

Practice, practice... I worked the redraw though for several hours and caught back up. Here's how that looked:


Another view:


The usual spaghetti nest. I hope is is clear how the purlins, colored brown and orbiting around the roof at about mid-height in the above picture, are tilted just like the wall/floor.

A closer in view reveals the interior bracing structure, tilted in plane with the floor and purlin ring, tying together all the principal rafters with king piece, an assembly which the French term an enrayure:



I was satisfied with my drawing result at that point. In order to achieve completion to that point, I had varied my approach from Mazerolle. He wanted, it appeared, all the parts to be square-cornered sections of wood. In some email with Tim Moore, he suggested that maybe it would be okay to use square rafters and principals, something I had more or less ruled out at an early stage. I wasn't even sure they could be configured to work, such that a purlin could be sandwiched between them. I did some exploration in that direction, and found that square-cornered rafters, both common and principals, could in fact be used. Trouble was, as I had surmised earlier, there would be downstream effects as a result.

Here's one of those downstream effects - note the fit of the principal rafter against the king piece:



The fact that is is skewed is not the end of the world, and there is enough of a broad surface on the face of the king piece. Okay, well, you might sat it is not all that noticeable, so perhaps it could be lived with.

Worse though is the fit between the truss struts, what the French call contre-fiche, and the underside of the principal rafters:


You can see that the end of the contre-fiche is not centered on the principal. The problem with off-centered pieces in truss structures, of which this is a very simple type, is that they distort the load path, which might tend over time to induce a certain amount of twisting to the principal rafter or other places. Maybe not so bad in this case, and the contre-fiche could be re-angled to be centered on the skewed-over principal, but there is that matter of load paths again. It's not an ideal arrangement as far as I'm concerned.

However if you take a look at how the horizontal member of the enrayure abuts the principal, one can see an even bigger disparity of fit:


Again, while the horizontal strut from the enrayure could be kicked over, and it appears in Mazerolle's drawing that this is intended, there is the load path issue. Multiply that by four principals, and I wasn't liking the way things sat. While the principals could be made square and the commons left as they were, I elected to erase all those changes and went back to the arrangement that had made sense to me in the beginning - virtually all the parts in the roof are parallelogram-shaped to conform to the various intersecting planes involved.

One exception are the purlins, which are quite clearly intended to be square sections, more or less, in the text's illustration. It's actually not possible, given other constraints in the drawing, to make them dimensionally-perfect squares in section, though they end up being pretty close. One area though where the insistence in using square-cornered material is leading to a slightly unattractive outcome though is at the underside of the hip rafters, where the purlins meet one another. Here's one such meeting:


And here's another:



All four are staggered from one another in regards to their top surface heights, though it is only apparent when you can compare them at the meeting points under the hips. The stagger between pieces is remediable, however it would mean that the top surfaces of the purlins would have to be decidedly non-square. I initially drew them that way and liked the look, and if I were to actually build a roof or structure like this I would opt for the purlins to have a more seamless fit by making their upper surfaces non-square to their sides, and they would fit much more nicely to one another under the hips.

The enrayure in my version of the drawing is composed entirely of parallelogram-shaped pieces, and I think the result is quite tidy:


All the pieces have co-planer surfaces and are centered upon the faces of all connecting members. If square-cornered pieces had been used, as the text shows, the result would have been, well, to choose a phrase, higgledy-piggledy. That is a word, isn't it?

The adjusted shapes of the principal rafters back to parallelogram-sections means that the contre-fiche pieces, themselves parallelograms in section, fit in a centered manner and the load path is directed as I think it should be:


Some final pictures of this tilt-o-whirl roof placed on a framed wall structure. Obviously, it needs the rafters extended to form eaves, and there are a few parts missing from the roof, but it conveys the idea and was as far as I decided to take it for the time-being:


An elevation:


Another elevation:


The drawing conveys certain lessons, some obvious and some hidden, and I think the main one related to establishing the tilted roof surface outline on the ground plan, and having a way of relating different locations on the roof in any elevation view of the two trusses. I'm not going to get into those particulars here today, but it was worth the walk down the path to see what Mazerolle was trying to show. I'm sure there may well have been a lesson I missed there, by choosing to construct something a bit different that what was shown, but I imagine I will be presented with that lesson sooner or later by way of another drawing exploration. I've got a long way to go yet.

The reason I focus such energy on studying drawing, is that if I can draw it, I can build it. I may never get a chance to build such a structure, however you never know what comes along in project-land, and sometimes a lesson learned in one application has unexpected uses elsewhere. Understanding this oddity may unlock the keys to other puzzles later on.

By studying these drawings, which believe me take a huge time and energy commitment, I am able in certain respect to get into the head of Mazerolle, understand more deeply the tradition of which he is renowned master, and hopefully come to understand the uniquely French approaches to timber carpentry, even if I don't always agree with what they are doing - or appear to be doing!

After all, traditional carpentry, as someone once remarked, boils down to just two things really:
  • putting lines on the wood
  • cutting where the lines are 
While I cannot claim mastery of either area, it is the putting of the lines on the wood which proves to be the far more difficult task of the two, and the route to gaining a deeper knowledge of that aspect of the work is one requiring thousands of hours of study, no way around it. It's an enjoyable trip so far, despite the frustrations and wrong turns which plague much of my experience.

Thanks for coming by the Carpentry Way.

No comments:

Post a Comment

All comments are moderated:

Anonymous comments will be deleted

Spam will be deleted

Comments containing links unrelated to blog content will be deleted