WSF Density/Fragility/Brittleness issue

This issue has been going on for a couple months now. Every so often I'll get a model, usually a large one, which will have 20%, or sometimes worse, less mass than the identical part properly manufactured. I actually discovered this somewhat by accident because, upon receiving said model, I noticed it was physically much lighter than it should be. While that in and of itself doesn't particularly bother me, what is a problem is this coincides with the parts becoming incredibly soft and brittle, to the point where I can cut the parts with my finger nail or use my bare hands to tear 2.5mm thick sheets of the plastic irrespective of layer boundaries. Far from being white strong and flexible, this is white soft and brittle. I discussed it with a very nice Shapeways rep. Andrew, who issued re-prints (some of which were also affected) until I got as many functional examples of the model as I had paid for, but was unable to provide any insight into the cause of the problem or what Shapeways was doing to resolve it.

Fast forward to a few days ago, I get yet another model in which has this issue. The 1mm wires, which themselves were only a few millimeters long, have completely disintegrated (we're talking significantly weaker/more fragile than Alumide), and what is left of them can be cut off at the base with your nail. The parts are soft, weak, brittle, etc.; they're totally useless. A correct part weighs just over 1 gram. These clock in at about 0.8 grams. I should add that I work in mechanical parts, rather than art, so the models have to do a little more than just show up mostly intact.

My big concern here is that the sales which are direct through the SW marketplace aren't QCed by me (us, the designers), and shapeways doesn't appear willing or able to do the necessary quality control to screen out these faulty prints. As a result, faulty products must inevitably be creeping out into circulation, and what will our mutual customers think? I'm perfectly happy playing the re-print game until my orders come out right, but ELUs won't know to do that nor will they likely have other correctly manufactured examples to compare against in the first place.


Has anyone else been experiencing this issue? Does Shapeways have any insight into what is causing this issue? Any timeline for actually fixing this issue?

I've seen this once in a while but more often during the Christmas rush season for whatever reasons. Someone would need to investigate if it's source material issues, machine settings, or problems in a portion of a tray (humidity variations, weight settling, etc.).

Another disconcerting aspect of this is if it leads to failures in polished white strong and flexible even though it's the fault of the printing process and not the polishing process. Total quality control implies controlled windows on all aspects of production and handling, including physical measurements and analysis of standard manufacturing test parts included in each tray. It helps to close the feedback loop.

Such large density/strength variations point towards a defect in the basic sintering process that should easily be identified through the use of a simple tensile tester that can be had for under $1,000. But that also means additional testing functions and data entry that needs to be performed by Shapeways staff. Not sure if they already do those basics or not. It might be cheaper for them to simply take the hit by sending out reprints if someone complains, seeing as a smaller percentage of customers are in the "mechanical performance" camp.

Nylon melts at 428 °F (220 °C)

As a batch of WSF is produced, the plastic is heated to within a couple of degrees of the melting point of the nylon.. let's say 425°F.
The laser then heats up the plastic the remaining few degrees so as to solidify the plastic together.

This arrangement allows for a less powerful laser, and even though it involves a lot of electrical power for the heater, it's a considerable savings over the amount of energy (cost) to run a laser powerful enough to melt the nylon from room temperature.

The trouble with this arrangement... if the heater 'fails' and doesn't heat the build high enough, then the laser just barely heats the plastic above the melting point, and in a lot of spots, does NOT melt the plastic fully. Such a situation would cause what you're seeing.. the plastic is barely joined together (not fully fused) and it would make sense that it could crumble.

There is also the situation where once the plastic has been heated, it's thermal properties change. Any reclaimed plastic used (or the percentage of reclaimed plastic used) affects what temperature the batch needs to be run at.

This is all about a failing machine, and lack of Quality Control. Remember.. Shapeways has two of their own production facilities that are operated by different teams of people, and they might even contract out some of the production to third party vendors if they get seriously behind. It's always hard to control what sub-contractors do.

I'm not making any form of excuses nor assigning any blame here.. I'm only trying to describe (what I understand of) what could have gone wrong.

Well I appreciate the explanations. While I realize not everyone makes mechanical parts, I'd imagine that it'd be significantly more difficult to get decorative models to simply survive the cleaning and shipping process as well with this issue. *shrug*

Another possible scenario, which I may have witnessed, is that the machines use two lasers to print layers faster. (or so I'm told) A model I had seemed to experience density failures on only one side. Diagnostically, this suggests a laser fault. Of course who knows. All units of both runs of this model (the model is multiple iterations of the product) were faulty in this case. *sigh* Just another frustration.

If you look at the material pages about the only spec you'll see is for size accuracy. So unless the density problem puts that out of spec you're still in the decorative item domain even with a wimpy sintering.

That's a primary issue I see as well. Often quality control is often perceived as a money loser but in fact getting a grip on printer quality and consistency right from the start (in this case) is key for eliminating so many other issues down the line for both Shapeways and customers, saving everyone money and wasted effort. I just had a polished white SF model fail production and not get shipped but the same model in red SF and polished metallic plastic yielded 100%. Why? Who knows. About all customer service could say was that it was a mistake and references in the rejection notice to unattached photos showing the failures are also a mistake. There are apparently no photos. It might be from the same tray as those gears or wires.

Perhaps 3D printers and their capabilities are still cr*ppy and unreliable compared to all the hype. I'd like to think that's not true but perhaps it is. I can imagine a scenario where it is financially unfeasible to scrap every 3rd tray due to something like low density because the process cannot be sufficiently controlled with existing printers. It looks iffy but hey, ship it!

I suspect some beancounterisms at work, such as "see if we can extend the service or calibration intervals", "see if we can cut the preheating period", "see if we can make do with a cheaper supplier of nylon powder"... and I do not think their disclaimer about decorative use comes into play here, as
even the decorative value of a brittle or broken part is questionable at best.
On the topic of multiple lasers, I think it is the bigger EOS models that employ separate laser systems for the top and bottom half of the build volume,
mentioned somewhere on the material page as it can give rise to a visible step or groove on tall parts if the lasers are not perfectly aligned.

EDITED to add: It could also simply be size and statistics working against them - lets assume the early shapeways had 3 or 5 printers, some 10+ enthusiasts striving for technical perfection (and burning money all the while), while they must be at more like 20+ printers and 100+ employees now,
increasing the likelyhood of bad print runs (while earlier investors are probably getting more interested in ROI now)

It is funny you should say this, not only because of how untrue it is, but because it reminded me that they even have a density spec. Anyway, for your perusal, this is the datasheet Shapeways associates with WSF:
https://www.shapeways.com/rrstatic/material_docs/mds-strongf lex.pdf

As an aside, in this instance, even if the datasheet weren't provided, it could still be argued this is unreasonable because it is WSF that is significantly weaker and in some cases more brittle than FUD. Calling it then "white strong and flexible" at the very least could be seen as grossly disingenuous advertising even if you used some odd

Time for some napikin math. So if we apply Shapeway's density spec strictly to the model volume, we come up with 1.12g per part. Technically though, it should probably be higher than this because parts tend to grow when sintered. This means though that we're actually closer to 30% under density, so even models I've deemed of adequate durability may not be optimal.


You're raising good points though about cost cutting. I'm starting to wonder if whatever decision was made which lead to this problem coincided with their decision to basically scrap and crack down on Sprues, as what was once strong enough to hold the model together is now worthless. In old models, 1.5mm rods were fairly durable objects.

The 2008 vendor data sheet doesn't mean much if Shapeways has tweaked the process from what was used to generate the vendor data. The specs aren't even bracketed by any ranges. There's a lot of outdated information floating around on the website. New machines or the aging of machines and process drift caused by multiple factors over the span of 7 years could also be a cause of print variability. This highlights why recorded metrics for every print tray that can be tracked over time are so valuable. A simple and small test structure for density, shear strength, and tensile break point tests to name only a few could easily be included in each tray or in multiple locations/orientations within a tray. I have no idea to what extent Shapeways does this, if it all. Without being able to examine historical statistical data of physical parameters and correlating the data to other aspects of the process you're basically working with an open loop process - which is not a well controlled process.

As I recall late 2013ish was the first time I received a model (a 2013 WSF ornament which is why I remember) that had that same kind of low density chalky and brittle feel to it. I gave that one away as a present and kept the higher grade version from another order for myself.

Oh no doubt, certainly we're not seeing things which stand up to it. My argument is more from a commitment/marketing standpoint, where they committed to deliver a product which performs roughly within those parameters. That is all.

It would seem that sometimes competition is the only cure for such ails.

"Two parallel lasers permit volume rates well above the market standard" and "scan speed up to 2 x 6 m/s" sound like one laser takes care of roughly half of every layer... so there is a common zone where they have to work together to make the (big) parts continous. Any deviation in that common zone will cause artifacts.

Well, just got a reply finally from Shapeways CS. As usual, polite and agreeing to look into the issue, but apparently unaware the issue is continuing and offering no real insights on permanent resolution. And, also, implying that it is just me, or at least that it is fairly rare, that the material needs the material sintered correctly *sigh* How they clean some models this fragile, I have no idea. Like imagine trying to clean a Strandbeest with this improper sintering?
(ref. https://www.shapeways.com/shops/theojansen)
I do wish they allowed access to their technicians from time to time, or at least acknowledged them. I have a hard time imagining that, if I can tell the density issue just by picking up the part, they in all their handling, cleaning, packaging, etc they wouldn't notice and be able to screen it.

Lets be as optimistic about their intentions though as possible and say that they couldn't qualitatively assess models by handling them, and couldn't track down the source of the fault, so this persistent issue required an alternate means of being addressed. They have density specs written, and volume accompanies every model. It'd be trivial to give each model's production tag an estimated weight. Anything lets say 10% under optimal could be pulled aside for additional screening. This could not only prevent these incredibly fragile models getting into the hands of consumers, it could also help screen for missing parts or parts which are present that shouldn't be. I, for example, have surprisingly often gotten a little piece of something that clearly wasn't mine. Someone is going to be unhappy now, because they didn't get part of their little gizmo or widget. It wouldn't' be 100%, because lost small parts in large models or under-density prints with foreign parts in them would read as in-spec, but it'd help some.


Or, going back to what draw said, a tensile strength tester would be an option but a barcol impressor would probably be more pragmatic as a broader range of samples could be tested.

I'm told that brittle prints was a maintenance issue and should be resolved.

Well I've been told nothing, so that is news to me. Thank you for taking the time to respond though.

I can however take/post a video of a recent order where I'm able to use my fingertips to just tear apart a 2.5mm thick part if you like. Not the most severely fragile print I've ever gotten, but still unusable.

For anyone looking into this at the machine level it would probably be a good idea to determine whether this is resulting from tray to tray variability or for variability within a tray. That's where I would begin.

I have had this issue with Elasto Plastic. But, aside from whatever material any one is using, the QC between Shapeways and the designer needs to be addressed.

"Lets be as optimistic about their intentions though as possible and say that they couldn't qualitatively assess models by handling them...". You are being nice, but for Shapeways that would just be an excuse. I'm sure Shapeways is aware of all this, so I am looking for Shapeways to implement the tools/protocols to get the best product they/we can deliver to a/our customer.

Do these parts have poorer detail resolution than normal? I'm wondering if too much recycled resin might be the culprit here.

Steve Hunter

My new plan is to order verification models using ONLY the polished white strong and flexible. It does appear to be the most sensitive and touchy process (at the moment). It's probably better to determine success with this before wasting money on any other material. Maybe they need the extra experience with pWSF to solve these problems.

I had a problem with my most recent order that may be related to this. The parts themselves were not brittle over all but there appears to be what seems like a single micron layer that didn't fuse, so the model came broken in half completely!

I ordered 22 of this same exact part at once. 20 of them printed fine and 2 came broken completely in half, clear across the "Z plane", in different places.

The top one is the whole piece that printed properly while the bottom two are the ones that broke. The walls are 1mm thick and every other part of the model is 1mm thick or less - and yet, around the entire circumference of the thickest part of the part, it broke, just as a result of being transported.

I emailed customer service about the issue (along with some others) and they said my walls needed to be thickened. Clearly, they didn't look at the picture I sent or read the description of the issue.

Hey everyone. The issue was dirty mirrors in the printers. This causes the laser to lose focus and strength. The issue should be resolved for all future orders. If you have a faulty print, email service and we'll make it right. From what I've seen, the nylon powder penetrates the machinery. We're using these machines like they've never been used before, so we're learning how to maintain them with such high usage.