VERICUT Users' Forum

Due to relentless spammers, we are no longer automatically accepting new forum registrations. If you wish to register for this forum, please send an e-mail to: info@cgtech.com

You are not logged in.

#1 2012-01-20 00:20:16

wheels of steel
Member
Registered: 2011-08-30
Posts: 11

head collisions on multi head machines

Hello all

My name is Peter.
I am the CNC programmer at ONESTEEL in Australia making railway products, ( wheels & axles )
we specialise in Vertical turning machines with multiple heads, usually two ( 4 axis ) but have three machines with Three heads (6 axis )...yes fun and games

Timing and collision detection is critical to our business, and is the core reason why we purchased Vericut. The cost to the business from head colission if the timing is out just a little is great, as maintenace has to repair bent dowels and usualy the loss of 2/3 of a shifts production.

Lets start with collision detection
We need to be able to detect collision between the two turrets and tooling when the head come close together. see image

[attachment=0:2edh1bjg]<!-- ia0 -->collision.jpg<!-- ia0 -->[/attachment:2edh1bjg]


I have been told by the builder of or machine model that this would slow down the simulation even more than it is now.
second, because we are turning inside of the wheel, it is detection collision with the tool holder and the stock, and slows down the simulation and creates a log of errors that aren't really collisions, our toolholders have the clearance to get into the tight radius.

so my questions are

can the detection be setup between tip and stock only, not the tool holder so we eliminate false collision reporting
can we set up detection between the tool holders on the two heads to detect head collisions....(critical)
can we set it up to choose one at a time, se we run through and check for collision with the tip and stock, then switch to tool holder and run again.

Any advise would be welcome

Thank you

Peter

Offline

#2 2012-01-20 06:54:58

paehv
Senior Member
From: Eindhoven, Netherlands
Registered: 2005-01-17
Posts: 167
Website

Re: head collisions on multi head machines

A slow performance would be better than a machine collision, right?
There are several options to speed up your simulation.
- Get the right hardware, Win7-64bit with loads of RAM and a top video card.
- Make sure you use a "reasonable" cutting tolerance (huge impact) (File > Properties > Cutting Tolerance) See http://www.cgtech.com/forum/viewtopic.p … 727&p=4215
- If you don't need the options (File Properties) "Model Export Cut Mode" & "Enable NC Program Review" switch them off.
- In your collision detection table, it's better to specify each component explicitly, rather than using the "sub components" options.(*)

(*) for detecting collisions between the tools in the turrets you will probably need the "sub components" setting on the two turrets.

On the tool holder collisions, I would take a look at your tool models. If Vericut reports a collision while in real life this doesn't occur, I would say that your tool model is incorrect.
Our company has large vertical lathes too and we take these holder collisions on inside diameters very seriously as this will indicate to the programmer, to use another tool, or to add an extra clearance to the tool holder.


Patrick Delisse
KMWE Aerospacehttps://www.kmwe.com
(Vericut V9.5, Siemens NX2206, Campost)

Offline

#3 2012-01-20 14:34:29

MustaphaC
VERICUT Specialist
From: Le Mans, France
Registered: 2004-10-14
Posts: 130
Website

Re: head collisions on multi head machines

Hi Peter,

As suggested by Patrick I would also keep the collision checking between holders and cut stock. If false collision errors occur you might have bad models for holders, try then running Polyfix tool in VERICUT.

I would also suggest you read that topic <!-- l --><a class="postlink-local" href="http://cgtech.com/forum/viewtopic.php?f=2&t=1337">viewtopic.php?f=2&t=1337</a><!-- l -->, it is about how to speed the simulation with collision detection by splitting the models used for machine components.

Another tip is to use "Nc Program Preview". This will process the nc program, make the collision detection, verify over travel limits, gouges in the part etc... whithout doing the material removal. Fixe all problems found during that pass if any then run the simulation with material removal.
Note that you can also use "Constant Gouge/Excess Check" during the "Nc Program Preview".


Mustapha

Offline

#4 2012-01-23 20:48:52

wheels of steel
Member
Registered: 2011-08-30
Posts: 11

Re: head collisions on multi head machines

Thank you both for your replies.

I will follow up on your suggetstions and see if they make an improvment.

Thanks

Peter

Offline

#5 2012-01-30 15:14:47

vhubbard
Beta User
From: Dallas, TX
Registered: 2004-11-11
Posts: 222
Website

Re: head collisions on multi head machines

Things that we see effect the speed of simulation. (We are on 7.1.5)  Roughly in order of importance.
1. cut tolerance.
2. collision detection.
3. Speed slider/skipping frames
4. review/preview
The rest we consider minor if you have adequate hardware.

A key in collision detection brought up in the VUE conference a few years ago was collision detection models and block size.  This is how I understand it.

Collision detection works in 2 layers.  1st fast, see if "bounding boxes" of the models collide, if they collide, do fine collision checking, check each internal triangle of the suspect .stl files for collision.  Since the fine collision checking checks each triangle in the .stl files, this could be significant to processing time.

What is a bounding box?   The limits of each model file made into a rectangle or box.  Lets take a sheet metal cover over a lathe, and it is a U shape covering most of the machine. It is 1 .stl file in the machine model. When used in collision checking, it will be in fine collision mode all the time since the "bounding box" of the cover would cover most of the lathe, always interfering with the stock ,cutters etc.  Vericut will not report an error, but processing time goes up doing the fine collision checks all the time.  Now if you take that same cover and make it 3 flat walls, 3 stl files instead of 1, then the bounding box of each of the separate models would better fit the outline of the sheet metal.  It would almost never be in fine collision checking mode since the bounding boxes are now far away from the stock, cutters etc. 

This bounding box collision method applies to fixtures, tools, everything that is checked for collision.  When making .stl files, think how it would look in "bounding boxes".   Breaking it up to separate pieces may be best.

The "bounding box" applies to models made within Vericut also.

Fine mode collision checking speed is driven by the number of triangles used to make the model.   More triangles, slower.   The number of triangles is related to the number of non-flat surfaces and tolerance used when the .stl files are created from a CAD system. The tighter the tolerance the more triangles needed for non flat surfaces and holes.   If it is not used in collision checking, you can make it pretty as you want.  When used in collision checking and it has non flat surfaces, open up the tolerance when creating the files as much as possible.

Example,  On one of our  5 axis machines a collision checked component is a U shaped yoke with fingers that taper down where the spindle housing connects to it.   We broke this up into 11 .stl files.  One for the base and 5 for each of the yoke's fingers.  This makes the "bounding boxes" follow the yoke's outline and taper.   Since our collision tolerance was going to be .1"/2.5mm, we created the .stl files with .025"/.6mm tolerance.    Before, when the yoke was 1 .stl file, anytime the head was at 90, the simulation slowed drastically.  After the change, no noticeable slowdown from collision checking.

Offline

Board footer