Errors in Plans errors in copying plans

[capnharv2]

I'm still playing around with transferring the NMM plans of the Ardent Class 64s to Turbocad (v17). Thanks, in part, to the folks on this forum, I've learned a lot and have made a lot of progress. But, I've stumbled across problems that have left me stumped.

 

I had a digital copy of the plans made (I don't know the process they used to make them), so it's been pretty straightforward in tracing the profile. When I start tracing the body plan, I start having problems. The first thing I found was that the body plan is offset vertically from the profile. It's about 10" (when I trace the plans at full scale). It took a while till I figured this out, but I've managed to work thru that issue. Now, I'm seeing angular errors between the body plan and profile. They are off by .5 to .8 degrees (depending on the line I'm measuring). There's also some curvature to the straight lines from the copying process, but I think I know how to deal with that. I'm finding that TurboCad doesn't like to rotate images in less than 1 degree increments (I've tried increasing the angular precision from 2 to 4, but it still won't let me fine tune the angle).

 

I plowed ahead and started the half-breadth plan from tracings of the body plan and the grid of the profile (if for no other reason than practice). The half breadths are way off. They follow the basic shape of original plans, but are offset from the original half breadths in one direction at one point (by about 10") and the other direction at some other point, and spot on in few places.

 

 I don't mind redoing the lines if I've made a mistake (I've redone them 4 times already), but I'm not sure how to resolve planss where the orthagonal views aren't orthagonal, and my program doesn't seem to be capable of correcting this.

 

Any pointer on how to proceed?

 

Thanks,

 

Harvey

[Doreltomin]

Hello Harvey,

 

Digital scans of large scale plans are made with a machine which carries the format of the paper back and forth between two rollers while an optical head reads the plan in continuous rows and puts them together. This only works well if the paper is perfectly straight, flat and and of the same thickness all over. Old plans are just the opposite of that. Due to being stored, often rolled, for long times, they are full of wrinkles, creases and irregularities. Besides, this "roller" system often gives an overall dimensional distortion over its lenght because it is calculated for an "average" thickness and it may be that the plan is much different of that.  Last, but not least, paper is a material sensible to moisture. It lengthens or shrinks depending of the air humidity. So while it may have been drawn in perfect scale, as the general level of humidity increased or decreased, so did the plan. Even worse, because the material itself is far for being uniform in its mass, certain parts of the drawing shrinked more, the others less.  All this cause unwanted, but sometimes heavy distorsions of the scanned image. It is precisely what you see in your plan. 

 

My suggestion is either that before importing the scanned image into TurboCAD you have to make it as clear and straight as possible. There are even unexpensive image softwares like the "Paint" which comes as a pack with Windows which you can use for that. You can even cut the plan into several slices and put them manually to fit the vertical.

 

Another option is that you can work with your TurboCAD (I haven't worked with that, but I know how to use AutoCAD, it may be fairly similar) and put your drawing on a working layer on top with a number of vertical guidelines at proper angles and distances, with the scanned drawing on another layer as a background and then as your drawing goes on, move the background slightly for each part to make the drawing in correct shape and dimensions. 

 

Hope I made myself understood, good luck with your drawing and show us what you did! 

Best wishes,

Edited January 7, 2014 by Doreltomin

[wrkempson]

Harvey, it sounds like you are encountering all the problems associated with the NMM plans.  You are right to press ahead.

 

Good new, though.  You won't be using the half breadth lines anyway so it is only for practice.

 

Maybe you can think of it this way.  The top of the keel is a straight line of known length which you draw (not trace) on top of the NMM plan.  The perpendiculars are at now places and a known distance apart so you draw (not trace) those as well.  You have already noticed that a perpendicular on the plan is not, well, perpendicular but is off for the reasons already described regarding the aging of the paper.  Now you have the three most important lines from the profile plan.  The same thing is done for the body plan for the center, top of keel and breadth lines (none of which are straight or perpendicular on the plan, but you draw them so as best you can).  Again, these have known dimensions that you reproduce.  At this point you place vertical perpendiculars on top of the station lines on the profile plan as best you can.  At this point you have almost all the lines needed for making a set of waterlines.  Just add the stem and stern by tracing along with the rabbet, and draw you own waterlines on the profile (but your new lines can be parallel to the keel and uniformly distanced). To the body plan add the traced body lines.  From this you make a set of waterlines that may look just awful.  But you fair up the water lines.  Then you get rid of the original station lines on both the profile and body plan.  Place your own set of station lines (more accurately placed) on the profile plan and then use the faired waterlines and the profile station lines to make new body lines.  At this point the body lines look a bit off so you fair them as best you can.  Now you get rid of the water lines so you can draw up an new set of waterlines that are in harmony with the body lines.  These you fair again with the body lines until you are happy with your results.  As you can see this is hardly "tracing" the plan anymore.

 

You might try to draw the body lines as per the old draughtsmen, but I think you will find that every bit as disheartening as the above and, at the extremes of the hull, a bit daunting.

 

The suggestion about putting the NMM plan on a separate layer is a good one.  This is especially helpful when you stop using the plan and are working only with your own lines.  This point comes a lot sooner than most people think.

 

From what you have said, it sounds to me like you are on the right track.  Just don't think the NMM plan can be traced all that much.  It is very much the guide to drawing, not tracing.  Somewhere there may be a bit more information on the process I have tried to describe briefly.

 

I just took a minute to rotate lines in TC v.14 and had no trouble with a rotation of .01 degrees.  I'm not sure why you are having a problem with that.

 

Wayne

[capnharv2]

Thanks for the responses everybody!

 

Doreltomin, thank you for you explanation of the copy process. I remember some of that from doing drawings on the board, but we had dimesionally stable mylar copied at 1:1, not 200 year old drawings that had been copied and modified numerous times. I don't like breaking the profile up into chunks and re-assembling it. I fear I would create more errors than I would correct. I do have the image on a separate layer, and have been rotating and/or moving the image as required, based on a set of horizontal and vertical lines I have down.

 

Jud,I didn't understand your question. Was it addressed to Doreltomin or me?

 

Wayne, thanks again for your voice of reassurance. I was just starting to construct the waterlines from the traced station lines and waterlines. I didn't expect the constructed waterlines to follow the originals, but I was surprised how far off they were.

 

There was a time 20 years ago when I did a set of plans on the board for a small cargo vessel, using Deane's Doctrine of Naval Architecture. It was a fun and enlightening experience-too bad I've forgotten most of it

 

I'll try playing with small rotations again. A lot of these problems I attribute to my slow learning ability. I used to be able to jump from CADDS 3 to CADAM to CATIA V3 and V4 with no problem. Then V5 came along and wham!-I went from understanding a lot to understanding nothing. Since I never used Autocad, Turbocad has been that same type of stumbling block. I'll probably get there, but it will be a couple more years, not a couple more months, till I feel reasonably proficient at TC.

 

The other problem I struggle with is how much interpretation and/or interpolation of the plans is considered reasonable. I don't know what the building tolerances were in the 1700's, but many of my wooden boat builder friends are still building boats to .03/.06" tolerance on a 50 ft boat. With that in mind, when I need to move a line or curve 1 or 2 or 3 or 10 inches from where it was originally drawn, I question whether or not I'm on the right track. Of course, the completed model is a representation of the original ship, not an exact copy, that I'm building,so maybe I'm too picky for my own good.

 

I'll play with these ideas some more. I still have a lot to learn.

 

Thanks again everybody.

 

Harvey

[jud]

Jud,I didn't understand your question. Was it addressed to Doreltomin or me? To Doreltomin but anyone could have answered. My question was how are ships curves defined mathematically so arc lengths and intersections could be computed. We run into Spiral Curves with Highways and RR C/L's. They are a constantly changing radius curve used as a transition from a circular curve or a tangent into another curve, eases you into the turn rather than a sudden change of direction. Curves on ships are seldom on a constant radius along their length. Thinking about it my guess is that the curves were created graphically and all else was made to fit. Frames were probably laid out in the Loft using offset distances and an arc drawn between measured points. The dedicated software to do all that today did  not exist when the ships normally modeled were put together or designed. Just wondering how today's computer modelers lay out all those compound curves found  in a ships hull. I deleted the question because it didn't add to the intent of your thread.

jud

[capnharv2]

No problem. Thanks for the explanation. When I drafted my Speedwell, Deane mentioned some of the waterlines were defined by quadratic equations. I just spent some time looking for the origin of Copenhagen Ships Curves, but couldn't find anything.

 

I've often had similar questions on lofting airplane wings.For a long time, airfoils were defined by NACA cross sections, but I don't know how the hows and whys of spanwise taper, or the transition of a fairing into the basic wing shape. I'm not an aero guy, and it was never a requirement to understand it for my work. But it's always been a passing curiousity, just like mathematical definitions of hull shapes.

 

Perhaps, for those of us who are not Naval Architects or Naval Historians, the history of mathematics in ship design (prior to CAD) might create an interesting thread.

 

Thanks for bringing it up.

 

Wayne, I tried the small angular rotation in TC and it worked like a champ. I suspect part of my problem is understanding how to adjust the image reference point. I'm still learning how to do that and use the snap feature better. I'll eventually get it, but it will take time. Thanks again for your help.

 

Harvey

[Doreltomin]

Hello Jud, 

 

Now that your question is back again, even if it might not add precisely to Harvey's thread, I can, with the due apologies to him, at least try an answer; although I am not by any means very good in maths. It is right that almost any curve may have a mathematical formule attached. I used to know how to calculate some simple basic formulae a long time ago, but now I believe for practical reasons today with our CAD-CAM's we can simply forget all that.

 

In simple days of lofting ships "the old way" starting with, say, 17th century, these lines were drawn using long flexible profiles of wood which were held in place using weights. I draw my ship plans even today using the same technique, only the wooden profil is kept in place with sewing pins nailed lightly into the drawing board. To be more specific, I use a software for my job as an architect, but when it comes to ships which is just a hobby, I love to do it by hand in the dear old way. 

 

Back in the 17th century they even had several kind of devices with more or less the likeness of a bow where the shipwright could tighten or loose several screws to obtain the required curve.These curves obtained from flexible profiles are called in today's language "splines" . AutoCAD has them as well as my professional software which is designed by a Munich based firm called Nemetschek. 

 

Basically, you can trace a spline using certain attach points and the spline takes its shape exactly in the same way as the wooden profile. Moreover, this is a curve which can be "adjusted" by tiny movements up or down of the attach points. Of course this curve may indeed have a mathematical formule attached, but since it is the software which "produces" it, who cares which is that?

 

The mathematical curve was useful in the old days because you had to use the formulae to replicate the curve in a bigger or smaller scale. Today the technology does that for us with practically no pain. Hope this helps... 

[WackoWolf]

What Doreltomin said is right on. In CAD software by using the spline command and features a person can smooth the trace out. Different software do it in different ways but it all comes down to the same thing which is basically using the spline command.

[jud]

Thanks Doreltomin; My General Cad Pro also has the Spine command. I do all of my curve compilations in a COGO program and export the results into Cad. The devices with the characteristic of a bow in the Loft makes sense. Thanks Again.

jud

[Don9of11]

I've had a little experience with this type of problem when I got plans for the USS Douglas Fox, DD-779 which I used to help build my Zellar's project. I took photographs myself from the original Booklet of General Plans which had been folded up in a cabinet drawer for who knows how long, so there were serious folds and bends to contend with. I used photoshop to stitch everything together,  I set up the distance between perpendiculars, frames and other ship particulars as best I could using guide lines. Even though we talking about different ships the process would be similar. Once I did that step for all the photos and different parts of the ship I sliced the images into smaller chunks to import them into my CAD program. Once I got them in the CAD program and despite all the work I put in to fair up everything, there were still areas where I had to make a decision about how the lines went. I prefer to try an fair the photos, scans or drawings in photoshop as opposed to doing it in the CAD program. I attached a few photos so you can see what I was dealing with. Hope this helps.

 

 

[Doreltomin]

Hello Don9of11 (That's a funny nick you have there!)

 

Thanks for posting your mosaic photo support, it illustrates perfectly what I wanted to say; which is, you don't need to put all the support picture under your vectorial plan at a time, you can put only the relevant part and draw it, then close and put in the next etc.

 

It's easier to do that if you intend to draw all the plan in vectorial, because the bitmap image plan actually uses a lot of resources for the drawing, has to 'map" every point of the image, while the vectorial only uses the coordinates of the start and the end points of a certain (straight) line. 

 

This way you will have a full plan free of any distorsions which you can later print and use for your modelling purposes. 

 

Best wishes,