[WarrenPease]

Hello all,

I'm finally acquiring a 24"x36" granite surface plate, and I don't really care for the factory stand designs so I'm endeavoring to build my own. I've looked at the factory data sheet for building your own stand, and downloaded a CAD model for one of the correct size from McMaster (gotta love their CAD library), so I'm confident I'll have the 3 points in the right place (will verify before welding those bits) and the clearance for the "oops" corners will be correct and so on. What I'm not confident about is what Fusion's static analysis is telling me with regard to deflection.

I modeled the tubing as hollow with the correct wall thickness (3/16"), set the material data for A500 steel, simplified the casters as the fixed faces and put an 800 lb/f load distributed on the top face of the "granite". The plate itself weighs about 600 pounds so that gives me about 200 pounds of master squares, the subject material and so on. With 2" square tubing the maximum deflection at the single suspension point for the plate it shows 0.003392", and with 3" square tubing it shows 0.001178" at that same point.

I'm not an engineer I just jumped in over my head on the whole simulation thing, so do the results seem at least plausible and if so is ~3.4 thou deflection with the 2" tube well within reason or is 3" tube at ~1.2 thou more than 2x "better" and a definite advantage of some sort for measuring things or is deflection not a big deal as it doesn't affect the plane of the surface plate? Or did I botch the simulation and the numbers look bogus (a definite possibility, perhaps hard to assess from the information provided though)...

I would appreciate any feedback on how concerned I should be about the menacing red colored image and associated range of deflection for a surface plate stand application.

[peteeng]

Hi Warren - The red is just a colour does not mean danger Publish your fusion model here and I can check it (ie export it to a zip file and publish zip file here). It would be a better design to have a brace or use thicker tubing at the single support as it has twice the loading of the other end. The deflection does not really matter (as its static and once deflected does not change) as long as both ends deflect the same to keep the granite level. Peter

https://www.starrett.com/precision-g...on-granite-faq

[WarrenPease]

Hi Peter, I probably should have found that not needing to be level information on my own. My OCD of course wants the plate to be level all the time which I suppose I can accomplish by adjusting the caster levelers with all of the gear on the plate, but the only real drawback to being off by a few thousandths is apparently perhaps the ball bearing method of determining square / perpendicular might be a bit more challenging if the bearing keeps trying to run away from you.

With regard to keeping both cross supports equal in deflection, I agree that would be a better design but is it worth fretting over since the point loads involved in any given usage scenario are essentially arbitrary? Everything heavy could end up on one side or the other thereby making any attempt to balance deflection across both supports inadequate. It seems to me as long as the total deflection is "reasonable", level can be dialed in on a use by use basis and if being really level is critical to a given usage then care would need to be taken as loads shift around, and don't lean on the plate either.

Since I had the model of a factory stand, I cleaned it up and added the granite, set the material specs to A36 and ran a similar simulation to what I've done with my design. The max deflection result for the factory stand, again at that single support point, came in at 0.005096". Is it a valid engineering concept to then say "That must be considered a reasonable standard because that's what 'they' did"? If I set up for 2" square tubing with a 1/8" wall I end up with 0.00502" max deflection at the same point, and my safety factor is 12.93 where the factory stand safety factor is 8.43, assuming of course I set everything up correctly. I am tempted to say that's "good enough" since for example sticking a 1,000 pound lathe on the plate to measure something would just be a really bad idea so there's no good reason to account for more extreme loads.

I'm trying to develop a feel for what structural "good enough" is, which the experienced folk seem to just know. Otherwise I tend to way over-do it, increasing the cost and complexity, and if there's no tangible benefit involved then I'd have to say I was doing it wrong.

I exported my design to a .f3z file then renamed it to .zip so I could upload it. It appears to have worked. If you have any suggestions on how I can better utilize the static analysis I greatly appreciate learning from someone who knows what they're doing. It seems like it would be really easy to set up not based in reality constraints or incorrect application of a load and get an all clear result that was entirely misleading.

[peteeng]

Hi Warren - I agree that its no biggy having one side slightly different deflection to the other as its adjustable. Depends on if you have normal levels of designer OCD or extraordinary levels of OCD I could not open your file. One of them is a caster which opened fine. You need to export it as a fusion file or a step file. The error report says it references external files which I do not have . but I think you have the gist now so keep at it. Peter

[WarrenPease]

Hmmm... In Fusion I used the File->Export command, and chose Autodesk Fusion 360 Archive Files (*.f3d), then renamed that file to .zip so it'd upload. Do the forums support uploading .f3z files? I didn't see that extension on the allowable list of options. I assume you'd need to rename the .zip back to .f3d and then present it to Fusion?

My bad on the caster, I had the external link still in place, which I have now broken so I assume that would include the caster in the archive. I re-exported the .f3d archive, and had to rename it as .zip to get it to upload, and I also included an export as a STEP model, which I also had to rename from .step to .zip to get it to upload. Let me know if one of those works for you, or if you've got a better way to do this!

-W (not that one)

[peteeng]

Hi Warren - Don't change the extension of a file. Just place the file into a zip file. I think the cnczone is processing the file and changing it somehow due to the suffix... Peter

[WarrenPease]

Oh, good idea! Why didn't I think of that... Give this a try.

[peteeng]

Hi Warren - success I'll get back to you. Peter

[peteeng]

Hi Warren - I looked at your file and model and it looks good. I would include gravity as the granite is heavy and stiff. But for your purpose the model is fine. Attached are images showing the deflection of your model. Highly exaggerated for the viewers. Peter

The granite has an E=67000Mpa (sorry I have a metric head) which is about the same stiffness of aluminium so even though there is 800lbf on the table the granite stays "flat". If I was building the bench I'd put a 3rd longitudinal down the middle so the transverse with the single hard point had a bit more support. The exported file does not include the FEA results so I had to rerun the solver. No biggy. Good exercise so I know F360 files can be exported etc....

[WarrenPease]

Thank you very much Peter, I thought I had a handle on the simulation setup for something as simple as this, but without checking with someone in the know I never really knew if I could believe the results or not. For something larger and more serious I would have turned on gravity, but for the simple good enough / not good enough determination I was after here I figured I'd just include the weight of the granite in the load.

I'll put in a center longitudinal member and see what impact it has. I'll likely add a small transverse or two on the bottom to support a sheet metal bottom / shelf, and I thought about running a vertical down to a larger transverse below the upper two transverses, but that would muddy my fantastically complicated height adjustable sliding platform (drawers without sides that can be moved up and down on pins similar to kitchen cabinet shelves) design. Which is what led me to am I really going to gain anything by turning 5 thou deflection into 1 thou? Or a half a thou? I don't think I'd notice the difference, and the nice thing about doing it myself is if I do decide I want to change it I'll have a good reason why and hopefully a good idea of what I want to change.

I've still got a long way to go with all of this fabrication stuff, but hopefully I'll be able to keep learning from all of the projects and questions on the forums and end up being at least not dangerous.

Thanks again! And best of luck on all of your 2,471,816 projects.

-W

[routalot]

I may be missing the point here but unless you have something like a Faro arm standing on the floor beside the plate and are measuring an assortment of parts,does it really matter that much?If you use height gauges,squares and V blocks on the plate wouldn't you still be referencing the plate surface with all of them?

[WarrenPease]

I may be missing the point here but unless you have something like a Faro arm standing on the floor beside the plate and are measuring an assortment of parts,does it really matter that much?If you use height gauges,squares and V blocks on the plate wouldn't you still be referencing the plate surface with all of them?

That was one of my initial assumptions, I just wanted verification. Unless the slope on the plate was enough to make some of your round or air-glide instruments go walkabout, then no it doesn't matter if the plate is level. That's assuming you've utilized the 3-point mount so the stand/floor combination doesn't introduce any twist.

I could have kept adding steel and adding steel to the stand until I got practically no deflection, but one can never truly get no deflection so the question became how little deflection is good enough? Cuz uneven loading would flex one end more than the other perhaps, so you can't even guarantee a consistent deflection ratio. I didn't think it was a big deal to maintain perfect levelness, but I'm no expert so I wanted to make sure I wasn't missing something. Analyzing the factory provided stand and being slightly better than that, along with the Starrett document that states you do not need to be perfectly level, pretty much covers it. Unless you're using a Faro arm on a floor stand, that is

-W

[jaguar36]

Its always a good idea when doing a FEA to back it up with a hand calc. It's far to easy to fat finger an input or make another mistake otherwise.

In this case all you really need to do is calculate the deflection of the cross member and see if its giving a similar answer.

Assume the beam is simply supported, load in the center, max deflection is Fl^3/48EI, F=400lbs, L = 20, E = 3e7, I = .7523 (assume 2" box with 3/16 walls, no fillets) That gives about 0.003" of deflection on the cross member, which matches your model pretty well. You could take it a step further and calculate the deflection of the side beam as well.