Accuracy of SketchUp

I’ve been using SketchUp to design boat hulls. First I model the hull, then I copy the panels and flatten them. These flattened panels can then be drawn onto plywood and, when folded together they recreate the hull. Frames drawings are created by intersecting with the hull at the desired locations.
The image attached shows a typical hull.
Someone has just built this hull from my drawings and they tell me that the bottom panel is drawn 1-1/2" too short and the frames are 1/8" too short.
I checked my SketchUp model, and the Layout drawings and I cannot find an error. My first check was to take the joining edge lines off the bottom and side panels and check the length using ‘entity info’. They are both the same length, so when the panels are joined, they should match in length.
I think he’s made an error in cutting out the panel. He tells me that there’s a ‘glitch’ in SketchUp.
Has anyone encountered any similar ‘glitch’ in SketchUp or can throw any light on this?
Please note that I find it hard to believe that there’s an issue with SketchUp, it’s by far my most favourite software.

Hard to know what has gone wrong without knowing what you supplied to the boat builder. The drawing you show has no dimensions on it.

I design buildings not boats, but my drawings would have dimensions for setting out. LO will produce those for you. If you don’t have annotated dimensions, the builder will have to rely on scaling a drawing. In construction, we generally don’t allow that!

As Simon says, it’s kind of hard to tell anything from your image.

Did you draw the original hull in SketchUp? How did you develop the flat panels from the original drawing of the hull? How did you create the plans from the panels you drew? How did you do the dimensioning?

Have you double checked that the dimensions with another tool besides SketchUp? Maybe make a paper model to scale. You should only need to make a few frames and the the panels.

I use SketchUp for a number of things and I do find it can be very accurate.

You say you compared the sum of edge lengths of the “curve” of the side panel with that of the corresponding bottom panel. And that they both match. Not surprising but it doesn’t prove that a the panels will match the real 3D model. At this time they aren’t curves anymore. Increase the number of edges and the total lenghts will also increase. An infinite number of segments would be representing a real model.

Look at it this way. compare two arcs (either same radii or different radii) that are the same in length.
Now explode both. Their lengths (now edges) will only match when both arcs were the same in the first place.

If you would like one of the forum members to give you an exact answer wher things might have gone wrong, please share (part of) the model to look into it in detail.

p.s. was this the first person who used your drawings to build the hull?

You have a 3d model that needs to be flattened into 2d. How did you accomplish that?

If flattening worked correctly, how do you know if you printed to scale? Where you in parallel projection? Did you use layout?

There are numerous forum topics with modelers having print to scale issues. Search for “print scale”.

I second this question. Also, how do you account for the thickness and the stretch of the plywood as it is bent around the frames? (I ask because I have wondered about this when considering designing a boat hull)

Thanks to everyone for their feedback.
For those who requested - I’ve attached the model file and a pdf of the Layout drawing.
To elaborate, I started to use AutoCad 25 years ago (I’m an architect). Since than I’ve never had drawing accuracy issues from contractors, whereas before the days of CAD, it was possible to make setting out errors simply by making an adding up mistake. In fact CAD has effectively stopped contractors making accuracy accusations / claims, because computers don’t make those sort of mistakes.
As with AutoCAD, I’ve always assumed that SketchUP is accurate too, there are a lot of architects using it (including me).
The person who made the hull claims that the top edge of the bottom panel is 1-1/2" shorter than the bottom edge of the top panel after he’s joined the panels, he also claims that the frames (which I created by intersecting flat planes with the model) don’t fit… But he’s also said “I am always suspect of computer generated dimensions.”. This rang an alarm bell, I think he’s made a mistake, in the old days he’d tell the draughtsman it was their fault and he’s now trying to use his suspicion of computer generated images to pass the blame … he seems to have missed the fact that we now live in a world where just about everything manufactured is from computer generated dimensions.
I create the flattened panels by exporting the hull panels plus a drawing of a rectangle to a 3rd party flattening program, I then flatten and import. Usually the flattened panels are not to scale, so I rescale the flattened panels, and the rectangle I exported, to be the same as the rectangle that I drew before exporting. I’ve checked this for accuracy, all of the edge lengths on the model and flattened panels are the same, and all of the frame edge lengths are the same as the meeting edge lengths on the flattened panels.
To Jesse_S, : I tested accounting for thickness this morning by using Fredo’s ‘Joint Push Pull’ plugin. It’s not an issue here, the inner edges are 185-3/16", the outer edges, 185-1/4" - I can’t imagine that this 1/16" difference would cause any issue and certainly not a 1-1/2" issue.
Yesterday I printed off the panel drawings on card, and made a model - everything fits.
So, to conclude, I’m facing an accusation from someone (who is suspect of computer generated dimensions), that my drawings are not correct. I’ve checked the model against the flattened panels, plus I’ve made a model and I can’t find any errors, so he suggests that SketchUp isn’t accurate. I don’t believe that this is the case, so my post is really to find out if anyone else has had accuracy issues with SketchUp. I very much doubt that anyone has, but I feel duty bound to ask.
Thanks to all and sorry for rambling on :slight_smile: Andrew

test panels SKF 20 Feb 16.pdf (92.7 KB)test panels SKF 20 Feb 16.pdf (92.7 KB)

Sorry, the SketchUp file is here…

test panels SKF 20 Feb 16.skp (783.8 KB)

Andrew, I expect you are correct and he has made the error. What did you supply to him as plans? Was it a table of offsets? Maybe a drawing with stations and offsets from a datum line?

I had a look at your SKP file. I increased Precision to 1/64".

It appears that you intend stations as shown below.

However the dimensions stations marked on the flattened panels are different.

If your plan marked out station 1 on the flattened panels as 33" from the bow point, and you gave offsets to those crossing points, there could be an issue since there’s more than 1/2 in. discrepancy in the location on the sheer strake panel. The distance between the tip of the panel and station one on the sheer strake should be longer than the distance to station 1 in the plan view but if your plan shows that distance as 33 in. there would be a problem. If he started measuring at the bow tip of the panel and placed the first station line at 33 in., he be more than 1/2 in. short by the time he got to the other end.

I wonder how he laid out the panels before cutting them. I would have located station 2 or 3 first and measured forward from there to locate the other stations and the ends of the panels.

I think if you want to prove the error is his, you need to make panels from the same documents you supplied to him using the same dimensions he had available. Unless you supplied him full size drawings from which he could lift the lines, printing out the the panels from the SketchUp model isn’t really a good test of the accuracy of the plans.

By the way, there is a difference between accuracy and precision. I would suggest that you model at higher precision than you intend to build at so that you can identify potential issues as shown above.

I will say that I’ve modeled some rather complicated parts to form a diverter for moving sand and gravel, which was then fabricated from steel plate. It took a lot of explaining how to cut the parts correctly from the scale drawing, and lots of showing them the model on the screen, but it fit perfectly, so I think SketchUp is accurate.

That said, I see @DaveR’s point, that if the drawings left any room for misalignment or mismeasurement, it probably happened. Happens to me more often than I’d like.

Lastly, have you ever modeled a Panga style hull?

[quote=“AndrewW, post:8, topic:20687”]
… so he suggests that SketchUp isn’t accurate. I don’t believe that this is the case, so my post is really to find out if anyone else has had accuracy issues with SketchUp. [/quote]

That isn’t (or at least should not) be an issue here, for SketchUp is accurate.
It’s the way you approach the 3D model with curved surfaces. SketchUp is a surface modeler but don’t forget that surfaces are segmented. So you need to be careful how to interpret this segmented surface. The simplest way to explain is to role out a cylinder flat on plywood or one with the same diameter but with 24 segments, the sketchup equivalent. Its perimeter will be shorter due to a smaller sum of the segments. Same applies to the bottom and side panels.
The only way to make accurate panel plans is to provide a list with X / Y values or by increasing the number of (virtual) frames to increase the number of triangles to be flattened. Infinite would be best but can’t be achieved in reality.

I have’t looked at your model (and the plans) yet but the complainer could still be right. That is if you included the wrong flat panel plans (instead of just X/Y data)

Hi Dave… Thanks for the feedback The lengths on the flattened panels will be different because they take account of the curve, i.e. the 33" bow to frame 1 length needs to be 33-1/32 on the curved edge. I didn’t dimension frame 1 on the flattened panels - maybe they built it incorrectly assuming that 33" was the flattened panel dimension. I’ll investigate further, thanks.

Thanks Jesse … I suspect he made either a misalignment or cutting error.

Hi Wo3Dan … Thanks I understand the segment issue - the segmentation is shown on the image below. But I can’t see how this should create a 1-1/2" error. The meeting edges of the flattened panels are the same length.

This might be true and isn’t that much of a surprise. But will the assembled hull from the given panels be able to fit around the frames with (by you specified) spacing? I’m still not certain that with the amount of triangles flattened you get enough accuracy for panels to represent the true shape.
“On”/at each flattened triangle there’s a slight difference between flattened face and “rolled out to flat” curved surface (the real situation). Adding up all these differences may surprise you in its value.
Don’t you agree?

p.s. you didn’t say anything about whether this person is the first one to be building this hull.
Is he/she?

If you increase the number of triangles you would then know that the lost in length is due to the flatten

That’s the whole idea but …
To convince oneself start doing so with a much simpler (less triangles) curved panel. and the increase the number. This one seems quite detailed already but still far from having infinite narrow triangles.

Ideal test to use the plugin Sub-d

Hi Wo3Dan and Kimberleydesigner

Thanks for the suggestion on increasing the segments. I did this as follows:

  1. I went to metric mm accuracy 0.1mm - seems more accurate than fraction inches.
  2. Note that only the first 1022mm is curved.
  3. The original curved part had 31 edges of varying lengths.
  4. I revised this curve to have 191 edges, most are 5.1mm long.
  5. The edge length on both the curved and flattened hull panels on the previous design and the revised design with increased segments is 4704.5mm. i.e., the same.

Thank you everyone for your feedback. At this stage I can only conclude that the builder made a setting out error.