3D Scan point cloud to surface.

[Anonymous]

Good day,

We are busy with a project for design of underground structures for a mine.  As part of this we had a 3D scan done and were given the point cloud data in a Recap format, i.e. *.rcp file.  I can attach this easily to either AutoCAD or Civil 3D and view it, rotate and all the nice visual things...

However been trying to create a surface from this in Civil 3D and I get an absolute wonderful look floor area, but somehow the walls and ceiling of the chamber (underground in a mine) is ignored.  Obviously I am missing out on something and our local reseller takes too long to get back to me.

Anyone care to take a stab at enlightening me please - I will appreciate it.  By the way, even though I am relatively expert in AutoCAD, Civil 3D is new to me, read total novice.

Software: Windows 10 Pro, AEC 2020, Product Design & Manufacturing 2020

Hardware: HP Z Book, Nvidia Graphics, enough RAM..

Regards

Kobus

[Pointdump]

Kobus,
No software can make a TIN Surface with 2 points at same elevation, so it's not possible to model walls or ceilings in Civil 3D. As for what to do about this situation, I don't know. Maybe someone will respond with a brilliant solution.
Dave

[Anonymous]

Hi Dave,

Thats the thing, the 2 points might have the same x and y but with a different z.  I have even tried to trim the cloud in sections that only contains the single sections, i.e. floor section, then the back wall, and lastly the side wall.  I am starting to feel like there is a wall specifically built for my head today....

Anyway, thanks for the response.  next time you're in South Africa, stop for a beer.

regards

Kobus

[rl_jackson]

You might want to look at Revit as a solution. I believe it will also hand cloud data, but can do vertical walls that your requesting. The solution may also be to use a combination of both both C3D and Revit as your ultimate solution.

[TerryDotson]

---

... the 2 points might have the same x and y but with a different z.

Software could evaluate your points and when duplicate X/Y pairs are found it could jiggle the X/Y an insignificant amount (like 0.01) so that when traditional TIN tools process them it would not consider them in exactly the same location.

However, since your data is in an undocumented proprietary file format, you would need to export it before such a tool would work.  Right now with RCP/RCS files you're limited to what Autodesk will allow you to do with it.

[Anonymous]

Thanks Terry,

Basically what I take from this discussion thus far is that we will have to invest in software to assist in this process as the basic Autodesk capability cannot cover our requirements.  I think most of these functions are geared towards scans having been done either in an open sky area (outside of buildings or infrastructure) or the purpose of the scan was really just to check physical fit so there is little need for a way to interact from a 3D modelling perspective - but I will happily be corrected if my perception is wrong.

Anyway, thanks for the assistance to you and Dave, it is much appreciated.

Regards

Kobus

[TerryDotson]

It actually goes back farther than Autodesk.  The TIN Surface was designed for 2.5d points where they are collected from the top of a single surface (either by aerial lidar or traditional surveying).

[Anonymous]

Hi Terry,

Point taken, I am learning new things on this topic on a daily basis.  Are you working in this type of environment as a rule?  Might want to tap a bit into your experience if possible...

regards

Kobus

[ChicagoLooper]

@Anonymous  wrote

I get an absolute wonderful look floor area,

Would it possible to post a screenshot the surface you able to get? The surface of the wonderful floor?

[TerryDotson]

---

Are you working in this type of environment as a rule?  Might want to tap a bit into your experience ...

---

Currently everything we do is basically 2.5d.  I used to do a lot of work in the underground coal mining industry which used a procedure known as room and pillar mining.

While the walls were not a vertically perfect as those above, we calculated volume and tonnage based simply on area and average coal thickness, etc.  Of course there were actual truck tonnages to compare it to.

If I were trying to model something like this in true 3D, I would look more toward meshing software.  While (basic) AutoCAD has some meshing and solids capabilities, it usually breaks down quickly when the data burden is large.

[Anonymous]

@Anonymousmay I ask what it is you are ultimately trying to achieve here. Are you trying to model the whole mine or just a small extract of the point cloud? Are you just looking for profile/sections of the existing mine? Are you trying to create a structure that will fit the interior profile of the floor, walls, ceiling or just one of these? Are you looking for quantities between your structure and the scanned elements?

How you might be able to do this and whether it would fit your purposes depends entirely on what you are trying to achieve and how you want to present your data.

If you are trying to create a structure within the mine then as others have suggested, maybe your workflow should be in Revit rather than C3D.

C3D is not really built for modelling enclosed objects as the surfaces just don't work that way. A TIN surface can't be vertical or overlap itself. You could though break things down to approximate the point cloud using C3D objects.

If C3D was a necessity then one example, if the side walls are near vertical, would be to scrub the walls from the point cloud and then split the floor and ceiling into two separate point clouds. You can then import the .rcp files and create a surface from a point cloud for the floor and one for ceiling separately. The walls could then be added by creating feature lines down the edges of the floor areas, applying floor surface levels to them and using a corridor to create a near vertical grade up to the ceiling surface. You could then run an alignment down the middle and get profiles/sections as needed.

Point clouds are usually very detailed and so a surface from a point cloud can be unmanageable without using the simply surface command to drastically reduce the number of points.

Another option would be to apply the same process, but instead of stripping the walls from the point cloud and splitting it into floor and ceiling, you could add points manually by snapping to the points in the point cloud. Add as many as needed to approximate the floor and ceiling areas. Create a point group for ceiling points and another for floor points. Then create a surface from each point group, clean them up and then add the walls as above. This could be useful for modelling just a section of the point cloud instead of the whole thing.

There may be other ways but as mentioned previously it really comes down to what you need to do with it and how accurate the representation needs to be.

Regards,

Peter

[Anonymous]

My apologies, when it was not what was required I deleted it.  I will see when time is not an issue anymore, if I can replicate that.

regards

Kobus

[Anonymous]

Hi Peter,

We have an underground chamber that was blasted out where an ore pass enters from one level up where there is currently active mining.  We're busy with designing a structure that will take the feed from this ore pass, through a surge bin, then a vibrating grizzly then through a jaw crusher onto a short conveyor.  The under size rock from the grizzly also feeds onto the conveyor and the conveyor feeds into another ore pass to a lower level where it is then conveyed to the skip to be hoisted to surface for further processing.

We do not have continuous access to the chamber due to ongoing mining and blasting so a survey would have taken too long and we then went the 3D scan route.  The idea is to have a point cloud of the chamber so we can place the ore pass entry correctly on our drawings through the use of a 3d surface.

Although we scanned about x = 50m x y = 30m and z = 16m high, we only need about half that.  We need the ceiling as well as we have an interface with that too, i.e. crawlbeams to be rockbolted, etc.

So after having gone through all the answers here I have come to the following conclusions:

1. Civil 3D is not the right tool for this.  We will explore Revit but at the moment I have nobody in my team with any Revit experience.  this is a solution for the next project.

2. I have now been creating slices through the cloud using Civil 3D, and then generating contour lines using the "Extract Section Lines" option. I made the slices 10 units (mm) thick, and changed the level by 500mm everytime which gives me enough detail for my purposes.

Anyway, everyone thanks for the help and advice.

Regards

Kobus

[Anonymous]

@Anonymous  you lost me at "where an ore pass enters from one level up". 🤣

From what you are saying I don't think C3D is the right choice here. There is always a way to approximate things but it's a broken workflow as you'd be forcing the software to do something that it's really not built for. You might be able to use it to solve a particular situation as you are doing now, but it's probably not a permanent solution going forward. Is there not specialist mining software for this sort of thing?

Regards,

Peter

[Anonymous]

Hi Peter,

Totally agree that Civil 3D is not right for the job.  So we will do it differently for any future projects.  

This is a screenshot of the Recap view of the point cloud, trimmed to only what we need.  As a point of reference to the scale centre bottom the scan caught one of the miners so you can see the chamber is quite big. To answer your other question - we only wanted to scan and model this chamber, not the complete mine.  This chamber is 40 minutes of walking from the hoist so its quite big.  Anyway, thanks for the interest shown, you are triggering thoughts about what we need to do in the future.  For now I am going to enjoy the last of my weekend.

Cheers

Kobus

[Anonymous]

@Anonymous  that looks like some interesting stuff. That chamber is f@ck!ng huge. It's bigger than most caves I've been to. I suppose my impression of underground mining is tight spaces that plant can only just fit through. I'm guessing the hole near the top right, probably the width and height of a house, is the ore pass that stumped me in your last post.

Scale is always a funny thing with large scale works as everything is so big that it's all in proportion. But put a man in the picture and it all makes sense.

Best of luck.

Regards,

Peter