Polylines are non-coplanar

[BigDumbWeirdo]

I've been getting this error on and off for about a year now, and I'm getting real sick of constantly erasing and re-drawing objects. So I'm wondering if anyone else has encountered this.

I sometimes will try to join two polylines (LWPOLYLINES) and it won't work. If I then try to fillet them with a radius of zero, I get the error that the two entities are non-coplanar. But they are. I've tried changing the elevation of both to something else (like 1) with no luck. I've tried the _FLATTEN express tool, with no luck. I've tried exploding, setting all the z-points to zero and then trying again, either by filleting with radius 0, or by converting them back to polylines and joining. Also no luck. I've tried purging (including purging DGN linetypes), auditing and even copying the objects into a new drawing.

The only thing I've been able to do that works is erase the object and re-draw it. For some reason, even if I put points at the ends of the line (or ends and middle for arcs), erase it then redraw using those points, it works. So the points are definitely coming in at z=0. I just can't figure out what's going on. One thing I've noticed is that this only ever seems to happen with drawings we get from outside our office. It never seems to happen on drawings made here.

Any help would be greatly appreciated.

[rkmcswain]

Make sure ELEVATION = 0.

If you draw a polyline (2D) and your first picked point has an elevation (even 0.00000001), then the whole polyline will have this elevation.

Lines can have different elevations on each endpoint.

Hope that helps some.

[steve216586]

" So the points are definitely coming in at z=0."

Not necessarily...

Lets say in the original file you get from the other office, there is a vertical line from 0,0,0 to 0,2,0.0000000001.

If your precision setting is .0000, and the original is .0000000000, and you place a POINT at each end of the line, your Points will be placed at the ends of the line at 0.0000,0.0000,0.0000 and 0.0000,2.0000,0.0000. When you redraw the line using those points it will be coplanar in the Z axis at elevation 0.0000.

Not coplanar if one point is at elevation 0.0000000001. 

[artc2]

Another thing to check is if the two polylines have inverted normals from each other. For example, one could be <0,0,1> and the other could be <0,0,-1>. Some functionality in Acad will treat them as coplanar and some won't.

[leeminardi]

Polylines store with them the UCS that was active when they were created.  If different UCSs were used for the two lines you may get the non-coplanar error message.

Assuming that you want all your geometry to lie on the WCS XY plane, set the UCS to World then give the list command.  It will show a statement something like "Extrusion direction relative to UCS: X=   0.3420  Y=   0.0000  Z=  -0.9397" for lines that were created when the World UCS was not active even if the line lies exactly on the world XY plane.  If you do not see "Extrusion direction..." for a selected line then it was create on the true XY plane.  

This problem is a holdover from the intial AutoCAD release in the 70's when it was considered a 2 1/2 D system. That is you had full control in two dimensions and limted control in the third.

I think the easiest solution is to set the UCS to World and trace over all relevant geometry using OSNAP on a separate layer.

~Lee

[gds1234 · Accepted solution]

I've found that when I have this problem (often) I create a new ucs at an object snap where the lines and splines appear to join, then check the location of the snap or node first on one line or spline then the other. Positioning the ucs here ensures that one node is 0,0,0 and it is easy to detect a difference at the other objects snap or node. Often this is 0.00000001 or slightly more. I've tried setting tolerances higher, using grid snap and only snapping to object snap points but I still find most of my time is used investigating why commands like join, region, boundary loft etc do not work and this is the only way I've found to progress.

It is also frustrating to draw an object including a spline with three non linear points to be told the object is non plannar. I assume the spline curves in 3 dimensions but I can not understand why this should happen if the spline is defined with only one start and on end point.

[BigDumbWeirdo]

Thanks for all the advice, but as I mentioned above, I've already tried setting a new elevation on the lines, which would get around the problem of the difference in elevations being less than my precision (which is turned all the way up, anyways).

Hell, I even tried a lisp routine to round off any floating point errors, but it didn't help either. I presume because there were no floating point errors.

I will definitely try the trick with the UCS, however. I'll mark as a solution if it works tomorrow.

[leeminardi]

Here's a much easier solution.  Since polylines store the coordinate system with them and lines do not, simply explode the polylines into lines , add the fillet then use pedit if you want to convert them back into polylines.

~Lee

[BigDumbWeirdo]

Here's a much easier solution.  Since polylines store the coordinate system with them and lines do not, simply explode the polylines into lines , add the fillet then use pedit if you want to convert them back into polylines.

~Lee

Yeah, I tried that too, and they still won't join. The two points where they meet have the same coordinates, but I'm still told they're non-coplanar.


I tried the UCS trick and that worked. It seems like there might be some floating point errors involved, but I'll be damned if I can think of a way to fix it that can be automated and easily repeated. If I do, I'll post it here. In the meantime, thanks for all the help, guys!

[leeminardi]

Using UNITS change the displayed precision to the maximum number of decimal places (0.00000000).  Using OSNAP END and the DIST command find the distnce from the end of one of the lines (not polylines) to the end of the adjacent lines.  It should be 0.00000000. Also use LIST to check that all the z values are 0.00000000 with the world UCS active.

[rkmcswain]

lminardigmail wrote:

Using UNITS change the displayed precision to the maximum number of decimal places (0.00000000).

Or just use this simple lisp code to see the maximum precision, beyond 8 places

(rtos (getdist) 2 20)

[gds1234]

Yeah. I spend most of my time trying to work out why things work then
another identical procedure doesn't.

I have to investigate reducing the precision because it seems illogical that
if you set the precision to say 0.1 the vertices of splines are located at
an accuracy of 0.00000001. I thought that the vertices would be positioned
with the same accuracy. Unless of course I am not doing this correctly.

This is my first attempt at doing a large 3D drawing and I've been playing
with it for the last 12 months or so in my retirement. It's not work, It's
supposed to be pleasure but it can be a very frustrating process). It is a
model of a ship and the hull has posed some interesting challenges using
splines, surfaces and solids. I try and work through the problems by
investigating the many variables options available with the aim of
understanding them but at time I just do not get the logic. Happily when I
get pissed off I can shut down and go fishing.

One thing I'm going to check now is how I've set up my overall scale. I
prefer to draw in mm rather than metres and the length of the vessel is
134000 mm (not 134 m). Seems odd to need an accuracy ten billionth of a
millimetre for something that is that long. I'm sure there will be some
suggestion somewhere to suggest how to best set up this scenario. If I scale
the drawing down to metres the accuracy may be as much as 10 millionths of a
millimetre. Better still I might try scaling down to kilometres ( maybe even
light years)

I've given up on the Autocad help because the search function seems to be
useless, again probably some setting is not what it should be for my
requirements, so I rely on chats like this to progress.

Thanks for your comment.

Regards Grahame

[prh3355777]

I find the properties box works well.  If you don't care what the elevations of your plines are, select "all" and then choose pline on the properties entity type  pulldown.  Change the elevation value from Varies to 0.  Now all the plines can be fillet.

[GlennLaugesen]

The suggestion to explode the non-coplanar polylines actuallly can work (but usefuleness of the result may still vary) - but it requires another step.

The procedure is:-

1. Explode the non-coplanar polylines back to line segments.
2. Select all the resulting line segments and open the Properties panel.
3. Set the Start Z and End Z values to 0 (if these show up as 'varies', then you'll see why AutoCAD is complaining about non-coplanar lines).
4. Use the command PEDIT>Multiple followed by Join to re-combine the line segments into a polyline.

There may still be issues with the resulting polyline though even if AutoCAD happily created a single polyline from the resulting pile of lines.

For example, lets say that you wish to hatch the resulting polyline (if it is closed) - AutoCAD may allow you to select the closed polyline as a hatch boundary, but may refuse to display the resulting hatch when you try and preview the hatch or if you just accept the result and complete the hatch command.

This may be a result of using PEDIT>Multiple followed by the Join option to recreate the polyline as this allows the specification of a 'fuzz' precision which implies that the resulting polyline might be acceptable enough to have its closed property set to TRUE, but may not be good enough for the Hatch method to actually fill the resulting boundary.

I discovered this when trying to form up a hatched area from polylines that were originally created in Civil 3D from sections through surfaces so use this method with a bit of caveat emptor.

Regards,

Glenn

[blfreilich]

How on earth do lines become non coplaner when i am only working in 2d ????

[seattlejoe7]

I'm also having this problem (and I'm also working in 2D). In the past, I've found that one or the other polyline has somehow gotten an elevation besides "0" ... don't ask how, I don't know.

But now, this "Two entities are non coplanar" thing. Changing the elevation of both to "1" (just for kicks) doesn't do a thing.

[seattlejoe7]

Woohoo!!! Solution!!!

My overall UCS somehow got "out of whack" ... I fixed the UCS issue but the polylines still complained about being non coplanar ... but when I redrew the new polylines (with the correct UCS), the polyline fillet now works!

[leeminardi]

Set UCS to World

Go to top view.

Give FLATTEN command.

[blfreilich]

How can i set my drawing that it is only 2d ? Never should a line with an elevation in the z axis exist in my house ?

How do i set that ?!
🙂

Ben

[Kent1Cooper]

---

@blfreilich wrote:

How can i set my drawing that it is only 2d ? Never should a line with an elevation in the z axis exist in my house ?

How do i set that ?!
....

---

You can force all points that you get by Object-Snapping  to have a Z coordinate of zero by setting the OSNAPZ System Variable to 1 [on].  But that is only of any value if there are already  things in the drawing with non-zero Z coordinates to Osnap to -- if there are not, there should be no need to set it.  And that setting affects only Osnapped  locations -- you would still be able to get non-zero Z coordinates in things in other ways, for example if a location or a Move or Copy displacement is typed in  with a third coordinate number included, or if something gets Rotated or Moved or Copied under a non-World UCS.  I don't know of a way to prevent those possibilities, other than being careful and conscious in drawing.