[peteeng]

Hi All - I am starting on the design of a full sheet router and want to build the gantry longwise not trans. This is to allow easier loading of the sheet. This means the gantry will be 2.9 to 3.0m long. There are lots of industrial machines with gantries this wide. So I'm wondering what size gantry section should be used. Plus I want to build it using aluminium/timber laminates or some exotic organic spaceframe if it proves its worthwhile. So I started with a 150x150x10mm steel SHS as a benchmark. There is a 200x10mm aluminium extrusion I want to use which makes the gantry 200mm wide. I modelled a 150x150x10 shs section 3m wide plus the proposed Lamio gantry. The steel one weighs 104kg the Lamio weighs 97kg. I have recently made a small scale lamio gantry for a small router called Frankie. It used 3mm aluminium sheet and 30mm pine as the laminates. So far its proved to be damp and stiff. I suspect that the 150x150 is too small for a 2.9m span. I will research whats out there. If anyone has a suggestion from experience I'd appreciate the input.

So to the results: I modeled Lamio and the SHS in Fusion then brought those into simsolid. I have a Z axis that is "rigid" ie it can't deflect so the only deflection is the gantry. I set up the restraints and loads so the moments applied to lamio and the 150SHS where the same. In the push dirn Lamio is stiffer mainly due to its a solid so does not lozenge. Lamio 147um (70%) SHS 211um. Trans dirn Lamio 10um (125%) SHS 8um and the plunge dirn Lamio 146um (131%) and SHS 111um. So overall similiar but can change Lamio to catch up. I suspect something like 200x200SHS will be needed. A little more research on what's out there is needed... Peter

[peteeng]

Hi All & Sundry - Ran another study where the lamio was plated on back and front as well and the SHS I sized up to 200x200x8. I also capped the end of the SHS which made a very big difference. So guys cap your gantries!!

Plunge Lamio 115um capped 46um and uncapped 67um
Push Lamio 123um capped 91um and uncapped 164um
Trans Lamio 9um capped 6um and uncapped 6um

So take away is cap those open ended gantries!! this will also remove some of the acoustics that they make as well.... steel looks good again. Peter

weight of steel shs 144kg
weight of lamio 104kg

[jaguar36]

I would be very wary trusting any simsolid results, I've had it given different results for the same model on two different computers. For simple beam calcs like that you're better off just doing it by hand. Are you constrained volumetricly at all? Why not go with an even larger cross section wood beam with some steel sheets on the outside? The wood would prevent stability problems, while the steel will give you a large I.

[peteeng]

Hello Jag - I have had long discussions with simsolid techs about their results and have resolved those issues. Use high refinement, then details and 5 adaptions and the results are good. Simsolid does pass all the NAFEMS benchmarks so must be OK. For my commercial work there is no other software like it that I know of. Especially when it comes to analysing welded and bolted structures which I do a lot of. 100's of bolts in std FEM is not a commercial proposition... or waiting for days for a result unless I get access to a supercomputer...

The reason for me using aluminium on the outside of the timber is that I can machine that on my router. If I use steel I would have the cladding laser cut to size and have all the holes cut as well. It would be 3mm thick steel zinc plated. I use that quite a bit. Currently I'm not concerned about the materials, I want to figure out how stiff a 3m gantry has to be... I'm currently thinking 250mm high and 200mm wide. But this conflicts with my std Z axis assembly logic so will need to rethink that. This is because my std assm has the rails on top of the gantry. For a shallow gantry that's Ok but for a 250mm high gantry this means the Z has to "reach" down 250mm before its doing anything. So the rails may have to go to the front.

The image shows my std saddle arrangement. It's easy to assemble and check and its very stiff. Whether round or square rail is used doesn't matter. The gantry in the photos is a stack of plywood and 2mm aluminum sheet laminates. Very stiff and damp and was easy to machine in my router. Peter

[jaguar36]

Hello Jag - I have had long discussions with simsolid techs about their results and have resolved those issues. Use high refinement, then details and 5 adaptions and the results are good. Simsolid does pass all the NAFEMS benchmarks so must be OK. For my commercial work there is no other software like it that I know of. Especially when it comes to analysing welded and bolted structures which I do a lot of. 100's of bolts in std FEM is not a commercial proposition... or waiting for days for a result unless I get access to a supercomputer...

We did some pretty extensive evaluations of SimSolid a few years ago after it was acquired by Altair and while it was very fast it had a ton of accuracy issues, including solutions not converging with increasing adaptions, different results on different PCs, and solutions not matching other FEA/Empirical solutions. We worked alot with the Altair folks to resolve some of these but never was able to satisfactorily. It's probably pretty good for this sort of stuff though , where you don't need a particularly accurate answer I guess. I regularly model 1000 fasteners in large Abaqus or Nastran FEMs and generally don't have runtime issues, even running it locally.

The reason for me using aluminium on the outside of the timber is that I can machine that on my router. If I use steel I would have the cladding laser cut to size and have all the holes cut as well. It would be 3mm thick steel zinc plated. I use that quite a bit. Currently I'm not concerned about the materials, I want to figure out how stiff a 3m gantry has to be...

What are you planning on cutting with it? How fast do you want cut it? For me, my constraint is generally space and money and I want it to be as stiff as possible. I'll then make due with whatever stiffness I can get. You can almost always reduce your cutting forces by taking shallower cuts, using smaller bits, etc. To get the stiffest cross section, you want the the most material as far away from the centroid as possible. So figure out your volume constraints, push all the stiff material to the outside. Steel is cheaper, but Aluminum is easier. Fill the center with either wood or epoxy granite (for better vibration damping) if you can handle the weight.

If you're happy with the stiffness of the one thats picture, use that to figure out how stiff you want it to be. Deflection is PL^3/48EI, so if that's a 1m gantry and you're looking at a 3m gantry you'll need 9x the EI. Conveniently, I is also based on the height cubed, so you can just scale everything up proportionally.

[ardenum2]

I regularly model 1000 fasteners in large Abaqus or Nastran FEMs and generally don't have runtime issues, even running it locally.

are you running nastran with a mortals hardware? I merge all my solids and just meshing the thing takes hours, when I run it half the time nx just straight up crashes or takes days to solve. I'm very surprised because I can import huge meshes as convergent body and use it to cut out parts from a solid, takes a minute or two, millions of polygons, but whenever meshing for simulation, it's a disaster.

I tried simsolid and it did the whole thing in minutes, sad to think it's inaccurate though the result was within 10% of nastrans.

[jaguar36]

are you running nastran with a mortals hardware? I merge all my solids and just meshing the thing takes hours, when I run it half the time nx just straight up crashes or takes days to solve. I'm very surprised because I can import huge meshes as convergent body and use it to cut out parts from a solid, takes a minute or two, millions of polygons, but whenever meshing for simulation, it's a disaster.

I tried simsolid and it did the whole thing in minutes, sad to think it's inaccurate though the result was within 10% of nastrans.

Nothing fancy hardware wise, I am using MSC Patran/Nastran not NX though. Never had an issue meshing reasonably complicated parts. These days I'll generally use Abaqus for detailed models though.

[peteeng]

Hi All - I think I'm at the corner of not going the long gantry way. Scoot is in its 5th generation and has resolved the gantry/saddle/Z axis/production issues quite well for a 1220mm wide sheet. So if I go the conventional way, I need to design good columns vs the very long gantry. Plus this gives 3 sided access to the working area vs one side if I use a high rail and wide machine. This does mean the rails and drives are in the dust but that's the trade off. I did more FE work on it last night trying to retain the current saddle design and it got messy.

Re: Simsolid - I think its solver has slowed down a little in the last couple of years. The large trailers I do initially solved in minutes now its more like 20-30mins but I do use refined settings. If I run non linear with separating contacts to check if connections gap then it could take 2 hrs. I think they have changed their algorithms due to the feedback. So I think they have addressed the accuracy. They must have coped a lot of flack about this.... Peter

edit 1 - following Jags comment of manually scaling up the current gantry here it is, Seems 160x160 may do it... seems small hmmm. someone check the math I'm known to be wrong... Opps just checked and there is an error but not significant... I wanted it 100mm max deep so maybe 200mmwide x 100mm deep may do it...

edit 2 - This machine is intended for cutting plywood and structural flat timbers semi commercially. The client wants to cut fast and 16mm ply on one pass. Scoot can do that so I think I'd like to cut at 6-8m/min. constraints are always $$$, space and time. Plus for me I want to commercialise the design so it has to be able to be productionised (if thats a a word) and it has to be scalable in production. I can do all sorts of fancy things but generally they don't scale if I have to build one per week. Thats why I'm particularly interested in cold casting, gives me the scalability...

[peteeng]

Hi All - I looked at the 1500mm wide gantry again. I've had it in my head to use a 200x10mm extrusion as the basis. 100mm deep is the max I want to go with the current saddle. So I did a quick manual calc and the 200x100mm works really well. So I followed that up with some FE and it agrees. So its over twice the stiffness of my current steel gantry. Its heavier but i can fix that. There is a 150mmx10 extrusion that maybe the go. So now to look at columns and rail configs... Peter

[jaguar36]

Hi All - I think I'm at the corner of not going the long gantry way. Scoot is in its 5th generation and has resolved the gantry/saddle/Z axis/production issues quite well for a 1220mm wide sheet. So if I go the conventional way, I need to design good columns vs the very long gantry. Plus this gives 3 sided access to the working area vs one side if I use a high rail and wide machine. This does mean the rails and drives are in the dust but that's the trade off. I did more FE work on it last night trying to retain the current saddle design and it got messy.

Re: Simsolid - I think its solver has slowed down a little in the last couple of years. The large trailers I do initially solved in minutes now its more like 20-30mins but I do use refined settings. If I run non linear with separating contacts to check if connections gap then it could take 2 hrs. I think they have changed their algorithms due to the feedback. So I think they have addressed the accuracy. They must have coped a lot of flack about this.... Peter

edit 1 - following Jags comment of manually scaling up the current gantry here it is, Seems 160x160 may do it... seems small hmmm. someone check the math I'm known to be wrong... Opps just checked and there is an error but not significant... I wanted it 100mm max deep so maybe 200mmwide x 100mm deep may do it...

edit 2 - This machine is intended for cutting plywood and structural flat timbers semi commercially. The client wants to cut fast and 16mm ply on one pass. Scoot can do that so I think I'd like to cut at 6-8m/min. constraints are always $$$, space and time. Plus for me I want to commercialise the design so it has to be able to be productionised (if thats a a word) and it has to be scalable in production. I can do all sorts of fancy things but generally they don't scale if I have to build one per week. Thats why I'm particularly interested in cold casting, gives me the scalability...

Well your picture doesn't match your calcs, but also you say assume 90x90x3, but then your Ixx calc you use a 4mm wall thickness (1/12*82*82^3), so your Ixx should be 1,318,572 mm^4. That makes your I1= 9.5e6. On the 160x160x5mm your calcs are right, so 12e6>9.5e6. You could drop it to 4mm wall thickness and still be at 10.1e6 mm^4.

Not sure what bit sizes you guys use over there, but if you use a 3/8" bit (roughly 10mm) at 16mm deep at 200ipm (5m/min)you get a cutting force of ~12lbs. On the 160x106x5 thats 0.027mm deflection (assuming aluminum). Thats like 21% of your .129mm chip, seems like kinda alot, particularly if you start getting any vibrations.

[pippin88]

3m long gantry is a long span.

Better to go short gantry and use the money and effort saved on other parts of the machine and on gear to load sheets.

[ardenum2]

Not sure what bit sizes you guys use over there, but if you use a 3/8" bit (roughly 10mm) at 16mm deep at 200ipm (5m/min)you get a cutting force of ~12lbs. On the 160x106x5 thats 0.027mm deflection (assuming aluminum). Thats like 21% of your .129mm chip, seems like kinda alot, particularly if you start getting any vibrations.

do you use any charts or calculators for actual endmill load or you do it by hand? This is one piece of the pie I'm still missing...

[peteeng]

Hi Ard - have a look at millalzer gives lots of tool info and not expensive.

https://millalyzer.lophostrix.com/ Peter

Hi Jag - well spotted and thanks for checking. My current Lamio design is 2.5x stiffer then the 90x90x4 benchmark. So I'm happy with that and can tune that up now I have a direction. The general layout will be like attached. Open for easy 3 sided access. I shall cast a flat laminating table from self levelling grout to make the gantries. This will give me some low risk experience with casting a large part in grout. Although I don't see it being a problem. I'd like to cast the saddle in Eng-grout. Need a better word for the stuff. So I think this thread is near its conclusion. When I get the GA going for Lanky I'll start a new thread... Peter

[jaguar36]

do you use any charts or calculators for actual endmill load or you do it by hand? This is one piece of the pie I'm still missing...

I use FSWizard, it's a free app. Quite handy for feeds and speeds and seems reasonably accurate for cutting forces based on some rough measurements I've taken. I was actually thinking I'd put some strain gauges on my next build, but I don't think the strains would ever be high enough to give good readings.