Loft puts creases in my model

[Anonymous]

Hello and thanks for reading!

I am working on a project right now where I am going to attempt to make an RC car. My favorite car is the Audi TT, so my plan has been to model the TT body, model the components I'm going to put in the car, model some mounts to attach everything together, then get the body and mounts 3D printed. 

I found some (vaugely) dimensioned drawings of the TT online, and I've created a point cloud of locations that have to match up to ensure that I'm putting features in the correct spot. Once I place the point with the manual position entry I then try to trace over feature lines on the front, side, and top view, then use the curve intersect to generate 3D sketches of key features. 

In the attachments, "Hood Lower" shows the bottom edge of the hood, "Hood Upper" shows the top edge of the hood. "Hood Rails" are the longitudinal features that I use as rails for a loft command. I loft from the lower to upper edge, generate the hood, then I would like to thicken the hood to generate a volume to be 3D printed. Those images are in the .zipped file. 

My problem comes when I loft. Everything looks *fantastic* except where it meets the outboard corners on the upper end of the hood. A deep crease develops from that spot and fades gradually down the hood. I've indicated the creases with arrows in the "TT Hood Creases" image. 


I've tried everything I can think of to get these to go away. I can't just generate an infinite number of rails and cross-sections as all I have to go on is the outline of the hood. I've tried rounding the corners a little, I've tried splitting the hood into two pieces, I've tried generating the corners separately, and I can't get anything to make the creases go away. I'm wondering if there's an option set somewhere that is doing this. 

The part is attached as "TT Hood.ipt". 

Thanks again for reading!

Chuck Saunders

[JDMather]

I noticed that your Top Sketch is not constrained, not symmetrical and poorly done.

You might read this document http://home.pct.edu/~jmather/SkillsUSA%20University.pdf

I noticed that you are using more points in your splines than needed.
I noticed that your Front Sketch is not symmetrical or making obvious use of symmetry about the origin.

I noticed that none of your 3D sketches are symmetrical.

I recommend starting over.

[JDMather]

---

@Anonymous wrote:

I've tried everything I can think of to get these to go away. I can't just generate an infinite number of rails and cross-sections ....

---

Try the tutorials in my signature.
Fewer points and cross-sections actually returm smoother curves (let Inventor do the interpolation between points.

This car was done by one of my students in Inventor many years ago (6 or 7 I think).

http://usa.autodesk.com/adsk/servlet/item?linkID=10914938&id=9729496&siteID=123112#

[Anonymous]

I recreated the entire hood from scratch. The only points I used are the ones that define the 4 corners of the hood, the corners of the two decorative \\ // lines on the hood, the centerline points, and one point only on the occasion when the previously mentioned points did not define a curve. Where a curve was not absolutely necessary I didn't add one, just for the sake of simplicity.

I used the symmetric contraint to make points symmetric about the CL where possible, but for whatever reason when I was working on the front it kept moving the reference points instead of the to-be-positioned points, regardless of the order in which the points were clicked, so I simply mirrored the reference points about the CL.

I still get the exact same creases in the exact same locations.

My comment about the infinite cross-sections was referring to attempting to make ribs, as in boat construction. I was thinking that if I could make a lot of ribs that I could better define how I want the surface to look, but I'm limited in the data available about the car, so I can essentially only make what I have already made. 

[JDMather]

Far better start than I am used to seeing here - good job so far.

I only looked at the first sketch and it looks to be much better (although yours wasn't fully constrained - I did go in and dimension it to make sure it was symmetrical).

On curvy parts like this it is often better to view as a sheet of material from which the part will be trimmed.  Trying to force bends into tight corners often gets the results you see.

Model the "natural" curvature beyond the bounds of the finished part and then trim to size.

I think if you do that in at least the area shown in red you will get much better results (I would probably do all the way around.

This takes some imagination as you have to figure out what isn't there and get it to form what you want as the trimmed result.

I don't think I have a ready example of this that I can release publically (I might have an example that I share privately).  I think Ed Eaton covers this topic in his Curvy Stuff tutorials (for a different SoftWare, but the same principles), search Google on Ed Eaton Curvy Stuff DiMonte Group.

[Anonymous]

I was considering that I may have to model the entire car body as one solid and then cut the hood, quarter panels, doors, etc. from the model, but I think that will be considerably more involved than making the pieces one part at a time. I do think that the parts will mesh much better in the end, though.

I got a bachelor's in ME from Virginia Tech (and am a grad student there now), but somehow went through the pipeline without ever having taken a CAD course. I've had a lot of courses require CAD knowledge, but everything I know is from trial-and-error and looking around for internet tutorials. I didn't know if there was an option or anything similar that would change the way Inventor tried to estimate my surfaces, so asking on this forum if such an option existed was my last step before attempting to model the car as one piece and cutting the parts out from that. 

I do appreciate your input, and I will surely be making use of the symmetric constraint and will be using far fewer points for the remainder of the car. I'll post the finished product here when I'm done, but that may be weeks to a couple months from now, as this is a personal project I'm working on outside of my coursework and research. 

Thanks again!

Chuck Saunders

[JDMather]

---

@Anonymous wrote:

I got a bachelor's in ME from Virginia Tech (and am a grad student there now), but somehow went through the pipeline without ever having taken a CAD course.... but everything I know is from trial-and-error and looking around for internet tutorials.

---

A fellow Hokie (Class of '93).
You can model it part by part as multi-body solids and they will all match that way.
Doing cars in Inventor is not particularly easy, but hasn't stopped me from trying.

Alias might be a more logical choice as that is what the automotive companies use.

There are several tutorials (including a simplified toy car) in my signature, but unfortunately most of them are getting rather old now. The Vacuum tutorial is fairly recent.

If your group of grad students would like a guest lecture - I pass through Blacksburg from time-to-time.

I just realized from the image I posted that you might have thought  I was suggesting modeling the windshield to get the hood.

That is not what I intended - I was showing the area of the sketch I woudl close loop to get the hood and then trim the surface to match the current sketch.