Yet another 'adhesion' post but more specific: *Designing* to reduce warpage?

[rtrski]

OK, context: A250, supplied coated springy bed. I do clean (Isopropyl 90% wipe with a paper towel so I have a little more fiber/friction than just a cotton ball to rub it off good). Printing PLA, various brands, various colors, and yes I know material matters. (I do not dry or dessicate mine, shoot me.) I know how to calibrate and always do a skirt of 3-4 lines so I can eyeball the squish and adjust the Z offset up or down if I think its a little wrong, and do tend toward more than less squish.

Parts adhere very well overall, rarely get an accidental release, usually do have to wait for cooling and pry gently with the putty knife and/or flex a few times to even lift a starting edge to be able to release the rest of the way. (Let parts cool first else the putty knife can gouge the bottom, esp with that kind of silky/pearlescent material that has a lot of extra goop [highly technical term] mixed in. All that said, large, mostly rectangular, flat bottomed parts do tend to warp up on at least one corner during say a 5-8 hour print, and I want to correct that. This is long after first layer deposition.

I’ve read the typical anti-warp advice and it mostly is taken into account by the A250 (with enclosure).

• Heated bed, clean, enclosed.
• Pre-heat the bed for more even heating, and to warm up the enclosed environment.
• Don’t use excessive part cooling (not a problem with the stock A250 head…it doesn’t cool that well at all heh).
• Rounding my corners (6-8mm radius on big parts).
• Only 2 bottom skin layers, low infill percentage. But I do want at least 3 top layers and side layers for rigidity, and inevitably that is going to still concentrate cooling stress on corners and want to lift them as the whole wall perimeter tries to contract.

The only thing I am not doing is using some sort of solution, gluestick, hairspray etc to promote further adhesion, or a full raft which is a pain in the [censored] to clean off the print bottom. I feel that cooling stress is going to overcome any adhesion not strong enough to make part removal severely unpleasant, and wonder if there’s a better way by design to reduce it (as rounded corners help, but do not prevent yet).

I am not a mechanical / thermal engineer. Any in the house that can comment?

Example idea, this is a topographical base for a small metal model with downloaded NASA mars data. I’m trying next little corner ‘feet’ with chamfered connections to the intended part, built into the design, kind of like a selective support or brim but hopefully easily knifed off. This is still just an adhesion trick really though, not dissimilar from a standard brim, or raft, or using some sort of goop…not design to mitigate stress directly. Would features like 1mm wide and tall slots in some sort of radial shape in the base (like quarter circles radiating from the nonexistent ideal pointed corner) designed into the bottom help to break up or release the stress concentration as the part cools?

[brent113]

The Cura draft shield could be used to minimize airflow around the perimeter, which would reduce thermal gradients and should help prevent warping to some degree.

My gut reaction to slots in the base is: not worth it. Possible benefit, extremely model dependent, solves a problem that shouldn’t exist in the first place.

The time spent modelling those features would be better spent making any required changes to the build platform to even out thermal gradients: put a small heater in the enclosure, add insulation under the heated bed, etc.

Just my 2 cents.

[eh9]

@rtrski

I am not a mechanical / thermal engineer

I’m not really either, at least not professionally, but I can comment.

FDM printing shares a fair bit in common with casting and welding processes. You have, in all cases, a molten substance that takes form as a solid product. In all cases you also have induced strains because of non-uniform cooling. Anybody who’s done even the smallest welding project knows that you have to plan for warpage because the forces exerted by the weld bead as it cools. Substances cool off from the outside in, which is obvious, but also the crux of the issue. In the typical case, you have at some point a solid exterior and a molten interior. The exterior solidifies at a size determined by the volume of the molten interior, but as the interior solidifies, it’s also shrinking because it’s cooling off. (This is because the thermal coefficient of expansion of materials is almost always positive, with a very few notable exceptions.) The final product has induced stresses; the exterior will generally be in compression and the interior in tension. This is how tempered glass product is made; the skin, under compression, is resistant to cracks, but if a crack does propagate, it’s going to release a bunch of energy stored in the mechanical strain. Only when those forces are in perfect balance do you not get warpage. The practical question is whether the warpage will be acceptable or not.

For small parts, warpage generally don’t cause problems because the total forces aren’t large. The total forces are proportional to the size of the object, so it’s large parts that cause problems. There are two basic ways of dealing with it: clamping, and don’t make large parts.

In welding, the most common kind of clamping is a tack weld, because you don’t need to set up a clamp tool. Clamping isn’t typically used for FDM printing, but there’s no reason it couldn’t be. First print a base, clamp the base into a fixture, and then print the rest of the object directly onto the base. Honestly I don’t know why this isn’t done more often, but it does require a certain amount of extra tooling, particularly on the print bed to enable clamping.

In the don’t make large parts category, this shows up in welding as the intermittent weld bead. Long weld seams are usually welded up in sections, not from one end to the other. One section has time to cool off while another is being worked on. Doing this in FDM printing would require better slicing software. Base layers laid down should not be done continuously in any pattern, but rather broken up into patches that are laid down independently and filled in afterwards. Each individual patch would have cooled by the time it’s connected to its neighbors.

[rtrski]

Are you aware of any slicers that might offer applicable different patterns? I can’t help imagine that the the shrinkage ‘along’ a deposited line is a tiny bit bigger than ‘between’ (or that at least ‘between’ lines the contact is slightly smaller, so the stresses can simply be still there without the net part actually moving). Seems like a pretty simple tensor field map of the part shape in the XY dimension could put a radial tensor around sharp corners, then the layer instead of being just two alternating straight line maps could kind of do concentric circles toward those corners then swap further in.

But you definitely reminded me of the most obvious solution I should re-examine: just split up the part into two and thus reduce the area. But then I have to work up a nice alignable seam and possibly do some putty or primer work after assembly.

Thanks for the replies thus far!

[eh9]

@rtrski

Are you aware of any slicers that might offer applicable different patterns? […]
But you definitely reminded me of the most obvious solution I should re-examine: just split up the part into two and thus reduce the area

I don’t know of such slicer software, but I’m guessing you can get a lot of the way manually. Divide your design into three parts.

• Base A: A checkerboard pattern, white squares only. One or two filament layers thick. The adjacent corners should be separated by about 105% of the width of the filament, allowing it to be filled in next.
• Base B: The complement checkerboard pattern, black squares only. Same thickness as the A layer, because you’re going to print it at the same height. Each (interior) square has four corners, and you want to fill them with filament to make a solid piece with Base A. So have a pattern that prints half the square each time, so you get the head going twice in and twice out.
• Top: After you’ve printed both Base A and Base B, you should have a good adhesion layer. It won’t have the same kinds of internal stresses as a typical deposition pattern with long filament strings.

[rtrski]

Well, some might argue that it “should not be necessary” to do such extra processing.

But after trying it, it worked FABULOUSLY. No warpage at all, with no additional effort at promoting adhesion, just my usual half-assed alcohol wipedown.

I just drew 25mm, 26mm and 50mm and 51mm dia circles in my tool of choice, making 2 rings of 1mm thickness, centered at each nominal rectangular base corner. Swept 1mm thick vertically and subtracted from my base. Was just enough to break up the bottom pattern but didn’t require any support or (near as I can tell) impact the strength of the finished piece.

[eh9]

That’s excellent news. Thanks for trying out my idea. Could you post a photo of the base of your print? I’d like to see what kinds of artifacts it has. Raking illumination should bring out the detail.

[rtrski]

I forgot I also still had the ‘breakaway corner pads’ on it, as well as adding a perimeter break. You can see where my Xacto knife (dull blade) left slight marring around the corner radii. But I still think the basic theory of just breaking up the long lines in the pattern was the right effect.

This has had some other processing to it (been painted on top, and was set down somewhere dusty before I flipped it over) but you can clearly see the slots, and there’s no part of the bottom that has that kind of telltale “I peeled away” gloss shift you get when you look at a part that did actually warp on its own some. Sliced in Luban.

I’ll upload a finished product picture one of these days, once I get it wired into its new home.