H.M.S. Atalanta - Drafting my own plans

[Ben752]

Good day everyone,

I've recently moved to Edinburgh for work. My personal effects are somewhere in the middle of the atlantic (above water hopefully).  This has given me time to focus on planning my first scratch build, the H.M.S. Atalanta.

 

I've selected this ship due to the wealth knowledge in the TFFM series, availability of contemporary plans from Admiralty models, historical plans and many build logs.  The hope is that with all of this information, it will give me enough information to stumble through the creation of plans using CAD and construction of the ship to a high level of quality.

I'm using Fusion 360 as my CAD program due to it's excellent price (free for hobbyist), professional quality and integrated t-spline, anlysis tools, parametric modeling features and CAM support.

 

I have made several false starts on the plans as Fusion is stew relatively new to me. One of the trickiest things is getting a good scan of the draughts into fusion and scaled.  When you can zoom in to miniscule details of the draughts it becomes apparent how warped and skewed they are.

One technique that I've found useful in fusion, is that when using the "attach canvas" that if you first create a component,  you can then duplicate the component and translate it.  This is quite welcomed after you go through the laborious process of aligning and rescaling the draught.

 

To start off, I've made a considerable effort to follow the practices outlined in Steele.  The dimensions have been sourced through a contract I found for the HMS Hornet at the RMG and the TFFM books.  When there is a discrepancy I've sided with the TFFM as the books are my guide.  My effort is to not trace but rather draft using a combination of traditional methods and 3d approaches.

 

Given this, the sheer plan is first up.  In Fusion, I've created "primary" sketches for each major major plan with the exception of more detailed aspects (following the order of Steel lends itself well to this approach).  For more detailed areas, I then create a separate sketch and project or intersect the references needed.  Fusion prefers this approach as it runs faster and easier to apply constraints and such.  For starters, here are my sheer sketches.

 

Below is the sketch in edit mode so the dimensions are visible (when zoomed it the dimensions are more manageable).

 

I've found using custom parameters exceptionally useful for cataloging scantlings and the source reference material I used.

 

 

 

 

[cog]

Interesting, I'll be looking over your shouolder to see how you use the CAD software

[tlevine]

For obvious reasons I will be following this with a lot if interest.  I am particularly interested in aberrations in the draughts.

[Ben752]

On 5/30/2018 at 6:47 PM, tlevine said:

For obvious reasons I will be following this with a lot if interest.  I am particularly interested in aberrations in the draughts.

These have really thrown me off.  I started tinkering with this Fusion model about 5 months ago and while I was able to make some progress in the drafting, I would hit a spot where dimensions would be way off and I"d start second guessing everything.

 

My natural inclination is to stop and review my work, and years of technology work has taught me to exhaust all explanations of issues using my own work before second guessing the work of others.  However, I have found that the body plan on the TFFM plans appears to be inconsistent in some areas.  Last week the use of the diagonals clicked and made a strong case the plan does seem to be off (incidentally the station on the designed and built draughts are very close to where i've drawn the curve).

 

I will say that the HMS Hornet contract I found in the RMG library (outside of the plans collection) has been very useful in providing a 3rd party dimension reference. It contradicts some of the dimensions in TFFM in ways that appear to be consistent with the draughts.   For those that are interested, http://collections.rmg.co.uk/archive/objects/512680.html#uaX1DK6V18Qgr5bl.99 the hornet contract is in SPB/27

 

This is not a criticism of TFFM at all!

[Ben752]

When I initially started this project, I started down the path creating a 3d model in Fusion following the order of the book.  It quickly became apparent that this strategy is not an optimal way of working.  However, the work below on the keel is fairly simple and was able to be salvaged before I shited to following a construction order closer to what is described in Steel.

The process used is as follows:

 

1.  Under a new component create separate sketches for the fore, aft and a middle timber of the keel.

 

 

2.  The top plane worked well to construct the aft and middle sketches as  it lends itself well to a extrude along the Y axis.

 

Because the mid keel components are repeated, I repeated the component using the rectangular pattern feature.  This gives me a reference edge to project in the fore timber sketch with the added bonus of propagating tweaks forward.

 

 

3.  On the fore timber, I constructed the sketch using the left side plane as it allows for projecting the arcs of the stem to model the the boxing.  Whenever possible, I"ve used projections off of one of my "master" sketches to allow for propagation of changes to the bodies that model the timbers.

 

 

4.  To create the simplified boxing joint I created the sketch for the lower stem on the left plane, extruded on one face left face to 1/2 the thickness and used the combine/cut option on the fore keel.   Then did the same on the other side but make the fore keel the cutting tool.

[Mark P]

Hi Ben;

 

I am not sure if you are doing this for ease of draughting,  to be altered later,  but the joints in your keel are what is called a 'half-lap' joint.  This is much weaker than the joint which was actually used in a keel,  the 'scarph' joint,  where the cut was only one third into the timber at the shoulder. 

 

I suspect that as you have obviously read the TFFM volumes,  you are aware of this,  but I thought it best to be sure. 

 

All the best,

 

Mark P

[Ben752]

Thank you for pointing this out.  I plan to go back and alter them but wanted to keep it simple as I work through the various components. 

[Ben752]

After a few transatlantic flights, I've managed to the roughly sketch body plan across the relevant planes set along station lines.  I ended up going back to the patch workspace instead of t-splines as the ability to edit edit and adjust far out-weights the smoothing ability.

 

I've ended up using projected points from the relevant half-breadth and sheer sketches combined with construction lines to establish bounds of the geometry.  Unfortunately it looks a little noisy in the screen capture.

 

 

I tried using splines but quickly became frustrated and went back to using arcs and tangent constraints.  Originally I used 4-5 arc segments for some but once I discovered the zebra analysis and repeatedly adjusted the curves for fairness, it became apparent that as few arcs as possible will result in improved fairness.

 

 

 

While still can be improved, if I keep continuing at this point i'll be spinning my wheels until I'm able to model other areas.  Which brings me to how to bridge the stern post, station 20 up through the wing transom.

 

I created construction planes using the sheer plan for the bottom of the wing transom, transom #1 - #4, and one in between the keel and the bottom of #4.

 

I then created sketches on these planes.  This adds some complexity as projections will be skewed when looking down towards the top of the keel.  I added the various reference points and intersected the stern post at each sketch.  However, a rough sketch of the filling transoms (?) profiles appear way off from the TFFM half breadth plans.

 

I've struggled making the 15" square at head of the stern post fit the rest of the plans.   TFFM states 1' 3" on pg. 41, pg. 64.    However, the plans seem to agree with about 12".   Additionally, the contract I have for the Hornet states:

 

"The stern post to be of good wound? oak tim of the best kind free from defects. sq. at head 12 1/2 (which is t run up to bolt in the Lua? deck beam)"

 

Which leads credence to the ~12" dimension. 

 

Does anyone have thoughts on regarding this?

 

This screen shot is taken top down which means the profile lines for the filling transoms are project per the angle of they're drawn on.  

 

 

 

 

 

 

 

 

 

[wrkempson]

It seems to me that a measurement taken from the plans should trump one taken from elsewhere.  A contract measurement might trump the plan but only if the contract is for the specific vessel at hand (ie, not from a sister ship).  My observation indicates that the lines on the plans are about 1/4" wide or so, so a variance of 1/2" is understandable in this case.

 

I love the zebra analysis.  The curvature of the rabbet in the area of the fore foot seems to be off, but that is due to the program, not your work.

 

Wayne

[druxey]

"The stern post to be of good wound? oak tim of the best kind free from defects. sq. at head 12 1/2 (which is t run up to bolt in the Lua? deck beam)"

 

This should probably read: "The stern post to be of good sound oak timber of the best kind free from defects." The contract and 'as built' plans trump any other reference! There were variances with any Swan class ship. Your filling transoms, provided they fair properly (check using both buttock lines and proof diagonals) will be confirmed correct.

 

The head of the stern post in Atalanta appears to be at the level of the top of the wing transom (ZAZ 4485), as there is no indication of a gudgeon strap on the rudder any higher. It is hard to tell, because of other overlaid detail.

[Ben752]

Thank you druxey. 

 

I revisited the the sternpost and found that the roughly matching the rate of concave curvature inward on the the half breadth plan suggests the head would be approximately where you said it should be.  Reconciling the different plans and sources certainly adds a wrinkle to selecting the right path when they don't appear to agree!

 

[Jack H]

On 2018/7/7 at 6:34 PM, Ben752 said:

I've struggled making the 15" square at head of the stern post fit the rest of the plans.   TFFM states 1' 3" on pg. 41, pg. 64.    However, the plans seem to agree with about 12".   Additionally, the contract I have for the Hornet states:

 

"The stern post to be of good wound? oak tim of the best kind free from defects. sq. at head 12 1/2 (which is t run up to bolt in the Lua? deck beam)"

 

Which leads credence to the ~12" dimension. 

 

Does anyone have thoughts on regarding this?

Hi ben752

On NMM original drawing(Atalanta J4428), the head of sternpost is not marked clearly(maybe), if it happens to be under the lower edge of gundeck,see following pic

The data of steel' is 1 ' 2",and the 1' 2 1/2 "is the measured value on the graph- to the inside of the" cover ".1’ 1 1/2" is the measured value of the head on graph--But I  adopted the extension line, so there will be some deviations.I used 1' 3" because I didn't go deep into it.

 

I suggest you need to prepare a data sheet, Just as Mr. David suggested me before.

 

In addition, I am very curious, how do you determine the center of the arc of the stempost,

 

 

 

I found the data in the data sheet about stem center."Above the upper edge of the rabbet"-13' 7 1/2",but I could not locate the horizontal positioning.

Thanks.

 

HJX

 

Edited July 10, 2018 by hjx

[druxey]

Sometimes the stem curve is made of two different radii, which complicates things! If you are lucky, you can see the prick-mark left by the draftsman's compass when he drew the arcs for the stem and rabbet. Take a line vertically up from the point where the straight part of the rabbet begins to curve and look carefully. 

 

The other method is to make an arc of the specified radius (13' 7 1/2" in your example) and put the center at different points along the curve and strike several arcs. The intersection of those arcs will locate the center of the curve.

[xken]

Ben752

On 5/28/2018 at 3:52 AM, Ben752 said:

One of the trickiest things is getting a good scan of the draughts into fusion and scaled.  When you can zoom in to miniscule details of the draughts it becomes apparent how warped and skewed they are.

I just was reading your post and just wish to make a suggestion for future reference for yourself and others. Scanners and copiers really do distort images; for better results use a camera telephoto lens and stand back and zoom in on a plan detail, the lens greatly reduces distortion, then import the image to trace more accurately.

keep up the great efforts!  Ken

[wrkempson]

If your program has this tool, construct a circle from three points along the curve of the stem.  I would use a portion of the curve that does not intersect the base line since if there is more than the one arc, the second arc will be found at the base line.  The center of the circle will of course be the center of the arc.  The same end is accomplished geometrically by placing three points on the stem arc.  Join the bottom and middle points with a line.  Draw a perpendicular line to this line that bisects the line.   Repeat the process for the middle and upper points.  The intersection of the two bisecting perpendiculars will give you the center of the stem arc.  If your heart is pure you will find the artist's pin prick nearby.  It may not be useful, however.

 

I have yet to find a table of scantlings helpful in drawing the stem curve.  I have always had to find the center myself.  By and by, I am thinking we are talking about the arc of the rabbet which often is the after face of the stem.  Be careful when then drawing the forward face of the stem since that arc does not necessarily share the same center.  The forward face arc center may have to be found separately.  

 

Also, the stem arc can be two (as per Druxey) or as many as three arcs.  This is known by observation.  

 

I don't know if this helps, or even if I am on topic.  In my defense I will say I enjoy drawing out the stem arc(s).  Of course, the curvature of the cutwater is even more a thing of beauty.

 

Wayne

[Ben752]

20 minutes ago, wrkempson said:

If your program has this tool, construct a circle from three points along the curve of the stem.  I would use a portion of the curve that does not intersect the base line since if there is more than the one arc, the second arc will be found at the base line.  The center of the circle will of course be the center of the arc.  The same end is accomplished geometrically by placing three points on the stem arc.  Join the bottom and middle points with a line.  Draw a perpendicular line to this line that bisects the line.   Repeat the process for the middle and upper points.  The intersection of the two bisecting perpendiculars will give you the center of the stem arc.  If your heart is pure you will find the artist's pin prick nearby.  It may not be useful, however.

 

I have yet to find a table of scantlings helpful in drawing the stem curve.  I have always had to find the center myself.  By and by, I am thinking we are talking about the arc of the rabbet which often is the after face of the stem.  Be careful when then drawing the forward face of the stem since that arc does not necessarily share the same center.  The forward face arc center may have to be found separately.  

 

Also, the stem arc can be two (as per Druxey) or as many as three arcs.  This is known by observation.  

 

I don't know if this helps, or even if I am on topic.  In my defense I will say I enjoy drawing out the stem arc(s).  Of course, the curvature of the cutwater is even more a thing of beauty.

 

Wayne

 

This is great info.   I've was going back and forth of using two different radii on this.  Seeing this confirms it's possible that might be the case.

[rshousha]

As a daily user of Solidworks, I would like to encourage you to go back and give the splines a second chance. It takes a while to figure out how they work but, once you get a handle on them, you will be much happier than using arcs. 

 

Rick 

[wrkempson]

Good observation, Rick.  When you speak of splines are you including Bezier curves in that category?  Frankly, while I use Beziers quite often, I cannot think of a time when I have used a traditional spline.  The control handles on Beziers make them a very powerful way to create a curve.  With that said, and with my completely amateur status noted:

 

The "arc or spline" question might have the age old answer: it depends.  For earlier plans arcs reflect the original practice; so if you want to mimic the old ways then an arc is your friend.  On the other hand, later plans I suspect made more use of ship's curves for which Beziers are a good substitute (vis-a-vis a collection of tangent arcs).  Using arcs is not as well adapted to 3D modeling since they complicate the need for uniform node counts, etc.  Beziers, as pointed out, can produce all kinds of accurate curves.  

 

There is something satisfying to me in drawing out lines with just arcs and a straight edge.  But for actually getting the work done splines (Beziers for me) are great.

 

So, as one who started out using Beziers (splines, if you will), and then learned to use acrs, I suggest we have both in the tool box.

 

Wayne

[Ben752]

One of the limitations of fusion is that the spline support is far from great compared to others out there.  They should be launching proper CV curves within the next year or so.  They have recently published a post showing some of the upcoming capability, which should put them more inline with that of solidworks.

[rtwpsom2]

Solidworks just implemented true bezier splines a couple years ago.  I am still not used to them, I tend to still rely on control point splines which are little more than the computers best guess for how to curve the spline between control points. I have played around with the newer "style" splines as they are called, but I still find myself heavily relying on the older tool. A symptom of my old age and older ways, I presume.

[Ben752]

Turns out they just launched CV splines (control vertices), 

 

https://www.autodesk.com/products/fusion-360/blog/july-26-2018-product-update-whats-new/

 

 

Edited July 28, 2018 by Ben752

[wrkempson]

I am not a mathematician nor do I play one on TV, but I find the visual explanation of Bezier curves to be quite interesting.  The math behind it is barely in my range of comprehension.  This link gives a picture of how linear, quadratic, cubic and quartic curves  are generated.  This is a very amateur understanding, but it takes something of the mystery out of how the spline is drawn and what is happening as it is adjusted.

 

https://www.jasondavies.com/animated-bezier/

 

Also, the Wikipedia article on Beziers has some interesting animations about two thirds of the way down in the article.

 

https://en.wikipedia.org/wiki/Bézier_curve#Constructing_B.C3.A9zier_curves

 

There are some other basic articles explaining Beziers curves and thus indicating how splines are generated.  As far as I can tell, a spline by control points and a Bezier are generated with the same equations.  None of this changes how we do things, but I thought it useful to visualize how the computer is doing things.

 

Wayne

[Maury S]

Wayne, While we're on it, in the lofting process you've described in your practum, should we be using splines by "Control Points" or "Fit Points"?  It seems to me that if we have to go back and adjust waterlines and body plans that do not appear to be fair, we need to use Fit Points.  Am I correct here?

Maury

[wrkempson]

You ready for this bit of wisdom?  It depends.  I used something that TurboCAD calls Beziers which is fit points plus control handles for every point.  I have more trouble with the other versions of the spline (by control or by fit points).  I find "fit points" easier to use than "control points" and Bezier easiest among the three.

 

So I would say use what makes the most sense to you.  I would expect there to be a learning curve on using splines.  rtwpsom2 says he prefers the control point splines to Beziers,  I have the opposite preference, mostly on the basis of what I am most accustomed to using.   I like Beziers but have seen some lines drawn with fit points and control points that were just as good.  Use what falls most easily under your mouse.

 

There are folks on this forum who do CAD for a living, and I would greatly appreciate reading their insights on this.  My CAD work and knowledge is purely that of the interested amateur.

 

Lately, however, I have been trying to use arcs rather than splines.  Sometimes that works out, sometimes it may be more trouble than it's worth.

 

I hope this helps.

 

Wayne

Edited July 30, 2018 by wrkempson

[rshousha]

Hi, 

 

OK, as a professional user of Solidworks, and a daily user of splines, with all their features, what is a Bezier? 

 

[wrkempson]

See above post #22.  Does Solidworks have something called a Style Spline and is that the same thing as a Bezier?  I can spell Solidworks and that exhausts my knowledge of the program.

 

I may be confusing you by using the names given to spline tools in TurboCAD.  At this point, it seems to me that splines and their control points all use the same equations, but differ in how one manipulates the control points.  To be sure, I am not the person to opine here in that my acquaintance with mathematics falls in the interested hobbyist category.  There are videos on YouTube that explain Beziers and the history thereof using basic algebra.  That and a series of Google searches constitute the foundation of my awareness of the math involved.

 

Wayne

[rshousha]

Ahh, OK, I see. Indeed, there are two versions of the Spline tool in Solidworks (your spelling is excellent, by the way) for 2D drawing, and one of them probably has the algorithm you noted above. However, I would just not worry about that.  Since I am an older fellow, I do remember some interesting algorithms from my days in Electrical Engineering in the '80s, when Computer Engineering was the new kid on the block. Yes, I actually do know how Google finds things. 

 

I would say forget about all that. Just use whatever feature kind-of works and practice, practice, practice, until you beat it into submission. The power of current computers is so far above anything we actually see that thinking about how it is working will just slow you down. 

 

You should look at learning one of the CAD programs more like playing a musical instrument than learning a computer program. As with music, expect it to take 10,000 hours of conscious practice to learn. 

 

And I say this as an EE who spent years really tinkering with algorithms and programming. They are quite useless to me now. Now, I need to create content; get that creativity out from where it's been stifled for the last forty years. 

 

Regards, 

 

Rick 

 

 

 

 

 

 

 

 

Edited August 15, 2018 by rshousha

[rshousha]

Here's another point I just remembered as I'm working on my latest model. When you extrude your bulkheads you need to anticipate the faring process. In other words, the frames towards the bow will be sanded down in one direction while the frames towards the stern will be sanded down in the opposite direction. There are only a few frames in the middle which will be sanded down evenly. So, in order for the lines to remain true, you need to extrude the forward frames towards the bow and the rear frames towards the stern. Then, you simply fare in the direction of the planking but just up to the edge of the bulkhead. If you just leave that outer edge (the stern edge of the frames in the bow section and the forward edges of the frames in the rear section) you will be right at the lines you traced. 

 

If you extrude all the frames mid-plane, then your boat will be too small in the bow and the stern (unless, of course, you leave gaps and only fare half the bulkhead). 

 

OK, now who actually was able to follow that? 

 

Cheers, 

Rick 

 

[jud]

At scale, using a median that expands and contracts with humidity and age, who has the tools to measure such small differences? Be difficult at full scale using the materials at hand and the measuring tools of the day. With computers it is handy to have precise mathematical data so that not accounting for minor details does not slowly accumulate and distort the whole. When Surveying, I did my Comps. to ten thousands, drawings were shown to hundredths and to be truthful on the ground be lucky to hold tenths, when monuments were set, time and natural movement moves things around, so discovering later measurements does not match the record is normal and seldom caused by mistake or sloppy work.  If you have the time, tools and ability to take such things into account as the tapering of framing, great, just don't then forget that the outside of each frame is on a curve and not account for that also, most have little need for such detail to build dimensionally acceptable models at scale. The refinement of the data can get very detailed or it can be limited to actual useful data to accomplish the task at hand.

 

[BANYAN]

Very clear Rick

 

cheers

 

Pat