Do CAD format Boat Plans still require Lofting?

Computer generated math surfaces are completely trusted in the auto industry to create the final dies and molds. My experience was that iteration between clay models and math surfaces is typically used to develop the preliminary surface from which the final math surface is developed. That was explained in the paragraph the one sentence quoted above was taken from.

 

If the computer surface is totally trusted, again, why use a clay model?

 

"Is that a question or a statement?"

The clay model is typically used along with computer models in developing the shape of the auto, much as a half model and/or paper and pencil were traditionally used in developing the shape of a boat.

The final refinement of the shape to fair the surface and create the math model used to build the tooling (dies, molds, etc) is done on a computer. This stage is analagous to the lofting stage in traditional boat building.

 

Clay models used in the automotive industry are cut buy a CNC machine, which regurgitates the code. The clay is manipulated by hand and the model is then laser scanned to generate a corrected code. The next stage is splashing this hand corrected model. I've actually seen this preformed on the latest models from Ford in recent weeks.

There are a lot of things to consider that have been only lightly touched on, in regard to computer generated and CNC cut pieces. It boils down to four basic error points - thermal displacement, tool and machine wear, machine tolerance and lastly tool error. All these things need to be considered. This doesn't include software assumptions and conventions, plus and probably most importantly the cost of accuracy, possibly more appropriately called an acceptable tolerance level.

Simply put, a bit cutting a part has some wander in it, the motor propelling a gantry, on which the router (or whatever) is mounted is assumed to be fairly accurate, the parts do move with environmental and tooling changes and thinking these don't affect accuracy is just a blind eye view.

Anyone that's ever had something machined at a machine shop as found this tolerance level, which is a few thousandths at the standard professional level. At the hobby level, 10 to 20 thousandths is typical. If you go to a cabinet shop, as previously mentioned in this thread, and have your station molds cut on their CNC machine, you can expect at least 10 thousandths errors, worn bits accounting for more errors and a machine that has significant tracking error. How much, for all the wear and errors added up, who knows, as it's application specific, but a hell of a lot more then those sitting behind a machine drawing things up, without sufficient building experience can possibly imagine.

How accurate is a plasma cutter? The tool path might be pretty good, but how about that jet of vaporized metal? Ever look at the edge of a waterjet cut part? So, how are you going to handle this issue. It's not enough to assume the software you're employing is accurate, it's just a computer generated image. You have to get results in reality, with perfect tools, that never wear, get hot or have flawed algorithms, plus the multitude of other things that can cause inaccuracies.

Now, cost enters the equation. Sure you can get below a single thousandth precision, but these machines are very costly and not commonly employed, cutting up foam blanks for someones bulkheads. Just too cost prohibitive and a typical example of reaching an acceptable tolerance level. Truly accurate software and tools aren't cheap. I don't know how many here, are willing to discuss how much they've got tied up in software, but I've got many thousands of dollars in it. How many have a 3D laser to check their work - have you any idea of spindle bearing accuracy or how many feed back loops are used to offset errors? How about the heat differentials in bearing and slide lubricants? Yeah, it can get real costly and complex, all simply because the human eye is very discerning.

 

Maybe the machine shops you use. Tolerance is directly related to the requirement and is typically stated on the drawings. It is a very sloppy fit if it's measured in 'a few thousandths' meaning 3 or more. That's more akin to a metal butcher with a file IMO.

Run of the mill tolerances for bearing fits run an order of magnitude better than you quote.

I'm a hobbyist metalworker and the day I can't hold my tolerances in metal to better than 0.002" is the day I become a wood butcher instead. If I have to, I can get to 0.0002".

Got no idea what acceptable tolerances are on CNC woodworking machines like routers but I wouldn't be surprised to hear it's 2 orders of magnitude looser than metalworking machine tools.

WRT CAD software etc I always keep firmly in mind that it was written by people like me. I lofted my boat the old fashioned way, full scale on sheets of plywood on the floor.

PDW

 

PDW, you're the exception to the "hobbyist" rule. Most machine shops will not offer more then 3 thousandths, unless you're willing to pay for the improved accuracy and they have the equipment. Yes, you're right a 3 thousandths variance is an "interference fit" for a pressed in bushing (for example), but fairly common, mostly from, heat displacement (much more then half the errors), tool and machine wear. Yes, you're correct in that wood working CNC routers are way off the mark, compared to metal machining accuracy. My point of the post, wasn't how fine you can hone a cylinder, but that there are lots of things to consider in accuracy and fairness, with cost usually driving an "acceptable tolerance level" being the major player.

When I do callouts on a set of boat plans, I offer 1/16" precision. The computer file is more precise then this, but people using a circular saw aren't. Most other designs use an 1/8" of precision in their callouts, maybe with a + sign to suggests it's a little fat. With this level of acceptable tolerances, you're going to need to spring a batten and have a look see, regardless of how accurate someone might be, sitting behind a machine generating pretty pictures.

 

Again. If there is 100% trust in the computer faring software, why use a clay model?

 

to see it.

Clay model is used to see the shape in full size real life perspective. They do use fancy 3d goggles and all that to minimize the need for clay work (I don't think clay work is cheap - the craftsmen are pricey and it takes time).

Yet they do it. I have a friend who works for motorycyle design and even after a very long design process (and quite near to the end) they do a clay model to confirm how it looks/works.

 

You can generate models with a variety of materials - print them molecule by molecule, CNC them from foam or other material, etc. The model material is of little significance. The need for a full size representation is usually just to sell it to who ever is paying for it, even it this might be corporate staff. Full size mockups and models also can serve several other functions (I know you know all this John), such as ergonomic and man access issues, compliance and interface with other physical items, testing, aesthetic acceptability, etc. It's a lot more then just wanting to look at something at scale. Last year I built a full size mockup of a cabin's interior spaces, cabinets, seating, berths, overhead, the whole works (plywood, cardboard, plaster and paint). In spite of my clients abilities to see wonderful walk through views on screen, he needed to feel the actual "fit" of these spaces, work out access to other areas for maintenance, storage, etc. The net results were several changes, to address what was obvious in the mockup, but less so on screen.

 

Wow PAR, based on that I would say that the chances that your client will be disappointed with the final product are small.

 

When doing a large project, such as a full up interior redo, a mockup is an insignificant cost in materials and effort. This was done on the inside of a early 60's "Connie" and I had maybe $60 bucks in materials, with a day's labor. Considering just the veneer costs involved on the project, I could eat a day or two in labor and materials and not get terribly upset about it, particularly if the client could sit on the new, open and explore the revisions, etc. It was used as all models and mockups are, to sell client and check access, compatibility, etc.

 

You are fortunate to work with such high budgets that let you do these size models.
Let me exaggerate a bit. Suppose your client wants to "feel" what he would feel having a beer on the deck of his boat, the sea breeze, the gentle rocking of the waves. Would you do a floating model?

 

I think that claiming "0.0002" as a common tolerance for a hobbyist is not realistic. A frame on a ship changes by much more than that from morning to afternoon. To attain tolerances of that magnitude the parts need to be in a controlled environment. Parts cut for welding need a sloppy fit for penetration of the welding material. That means that there are jigs to hold them which add to the variances. Only someone that only has academic knowledge would think that what you draw is what will get built. On the shop floor, parts and sections often get corrected to make them fair to the rest.

 

True, but a few thousansths of an inch precision is fine for a boat sized project, and better than most will do with hand tools. As to cost, i had all my frames / bulkheads CNC routed from divinycell foam core, about 50 8x4' sheets IIRC, it cost me $1000AUD - in the USA i would expect the figure to be around half that as our wages / costs are roughly double over here... saved me shed loads of labour not to mention how many measuring mistakes i could have made along the way...

 

You guys really need to get a clue WRT machine shop tolerances. I suggest you take your comments over to www.practicalmachinist.com and try them there. Be prepared for an interesting reaction.

Note carefully that I said *nothing* about *welding* tolerances. I was referring specifically and exclusively to PAR's comments on machine shop tolerances. To my mind, that means having stuff machined - you know, using metal lathes and milling machines etc.

WRT bearings, if you try to fit a roller or tapered bearing into a housing or onto a shaft with a 0.003" tolerance, let me say that it isn't going to last long if the housing is that much undersized or the shaft will flog badly. Just go and look at the specifications for bearings before you start arguing.

And, Gonzo, you also need to brush up on your reading comprehension as I did *not* say that 0.002" was a common tolerance for a hobbyist. I said that I was a hobbyist and I can, *if required*, hit that tolerance. Not the same thing at all. However, any half competent machinist can set up a shaft in a lathe with a total TIR of 0.001" or better. I can get to 0.0001" on my Colchester Chipmaster lathe and it was made in 1964.

Incidentally the plans (lines & offsets) for my boat were toleranced as per PAR's description. Tom Colvin points out that when dealing with 4mm plate, a 3mm error is huge so you have to loft, fair with a batten and take your frame templates etc from the lofted drawing. I'm pretty happy with the fit of the plates to my hull so the time spent lofting and making templates was very well spent and I'd do it again. I don't think I'd be at all happy with a set of Mylar (or worse, paper) templates and the only way I'd trust CNC cut steel or aluminium hull plates is if they were going to be laid & fit inside a very accurate female support frame as per some of the Van De Stadt designs. No way if it was plate over fabricated frames, a 3mm movement due to welding distortion is going to be bad news.

Old joke about precision fits & tolerances:

Instrument makers work to the nearest 0.0001".

Machinists work to the nearest 0.001".

Cabinet makers work to the nearest 1/16".

Carpenters work to the nearest 1/8".

Shipwrights work to the nearest ship.

PDW