1. Eric van der Star

    Very interesting article! Thanks!

    I’m an engineering student researching the mechanical properties of materials manufactured by FDM.

    Some results actually surprise me, showing that understanding the behavior of FDM printed objects is quite complex. I expect that the effect of layer height is different when printed at 90% infill, because a smaller layer height will create less voids.

  2. In your Key findings table, you have some duplicate entries with different sets of Xs in the columns. For example, 70% infill and 0.2 layer height is showing twice; one with all 4 columns checked and the other with only Strength and Low Cost. There are a few duplicates in there. Can you please double check that table?

    • 3D Matter

      Hi Michael,
      The way we set up this table is the other way around: you choose what requirements you would like to have (e.g. I would like strength and quality), and we suggest settings for infill% and layer height (90% infill and 0.15 layer height). We have no duplicate on the left-hand side of the table in the combinations of requirements, but there can be duplicates on the right-hand side if a pair of settings (such as 70% and 0.2 layer height) makes sense in several cases.
      Hope this helps!

  3. Extremely interesting study, and thank you for this. I print my cameras exclusively in ABS as PLA deforms in heat, such as a car dashboard. Apart from naturally wanting to see these results replicated for ABS, I understand that repeating these tests for different speed/extruder temperatures would be extraordinarily time consuming, but perhaps you could confine such tests to the Z dimension to reduce the time required. Layer bonding is extremely important – because my cameras are tall, if I don’t get the extrusion temperature high enough, they can crack midbody from internal pressure gradients. A study in vertical layer bonding would be of value to many.

    • 3D Matter

      Hi Clint,
      Thank you for your message. We are currently testing some ABS filament suppliers so check back in a few weeks and you will get some data on this material. But we have not planned to replicate the study on the influence of parameters for ABS yet (in particular we didn’t plan to make the Z-axis comparison again). We wish we could do a lot more but as you said, it costs us a lot of time and money to do this type of study! However, if you want something tailored to your needs, even a small study, let us know at arthur@my3dmatter.com and maybe we can work something out.

  4. Dean Du Bois

    Do you have a theory why the 45 degree diagonal strength was 50MPa vs 45MPa X or Y axis prints? It would be interesting to see what happens with a Delta printer design (like Rostock).

    Do you plan to measure other materials (ABS, Nylon, etc.)?

    • 3D Matter

      The point was to assess the difference between the Z-axis and the X/Y plane, so we didn’t really look into the difference between 45 degrees and X or Y directions. Will look into it though and let you know if I get an answer!
      We are planning to test other materials indeed: ABS will be part of our next study, and nylon and other thermoplastics will be part of a subsequent one.

    • Alan Gallery

      I have seen at least one other test where 45 degree 100% infill proved to be the strongest. The test piece was a beam and not a direct tensile test but the material was PLA and so relatively brittle compared to ABS etc. In my view the 45 degree infill pattern allows the stresses in the crossed layers to be in indirect tension and shear in a balanced way between the layers. If this is so then less brittle materials would benefit more from 45 degree infill pattern.

  5. Andrew Pilgrim

    Thoroughly enjoyed this! Thanks for the research. A lot of this is intuitive and is somewhat witness able through lots of printing, however seeing the numbers and having someone perform a few good experiments is wonderful. Glad to have found the site through this study. Bookmarked!

  6. Fantastic work! This is the most detailed study I’ve yet seen on this subject!
    Actually, I’ve been working on some mechanical testing of my own. I’ve designed an open source (partly 3d printed) testing machine which I will use to determine various other 3d printed characteristics. I plan on using mostly ABS specimens so it will be interesting to compare notes.

    • 3D Matter

      Hi Michael, we would be interested to know more about your test machine when you complete it. Keep us in the loop!

  7. Eric Lani


    I am doing similar research using ABS and some nylon with infused fiber. I was wondering what ASTM standard you used. I a currently using D0638. I had to make some significant alterations to the standard, due to equipment limitations and to encompass my study’s scope.

    I am comparing infill settings, build time, material used and build orientation to strength. This info will hopefully help me find the best cost to strength ratio. The odd thing that I have found in my initial data processing is that Young’s Modulus changes drastically from low infill setting to a solid part.

    I am currently processing raw data and will soon move onto statistical analysis. Please contact me if you’d like a summary of the research once that is done.


    Eric Lani

    • 3D Matter

      Hi Eric,
      We based our procedure on the ISO 527-1 standard, which I believe is the European equivalent of D0638. We also made significant changes to the standard because we needed it to be adapted to 3D printing.
      I am a little surprised that your Young Modulus would change if you adjust it to the infill. As we point out in our study, the Young Modulus was constant
      for all infills once we divided the raw data by the infill % (except at very low percentages because then the shell’s influence is non-negligible, which skews the calculation).
      We would be very happy to get a summary of your research once you’re done!

      • Chow Loong Jin


        Is there a reason you didn’t correct for the shell’s area when correcting for infill percentage? As far as I can tell, it ought to be a simple constant for that test.

        Kind regards,
        Loong Jin

        • 3D Matter

          Yes we could have corrected by subtracting the shell’s area and then dividing by infill %, and we would have certainly found the Young Modulus to be constant even at low infill%.
          We thought this was maybe more complicated to explain to the reader, whereas emphasizing that the shell becomes important at low infill % was a good point to make.

  8. LenW

    This is great, and thank you for sharing. In order to be able to actually use this data, can you tell us the following:

    – What model Makerbot was this performed on?
    – Which specific ABS and PLA filaments were used?
    – Where was the testing done and the specifics of how it was tested (i.e. the equipment, etc.)?
    – Who are you and what institution you are affiliated with?

    Again. thank you for sharing all of your hard work. These questions are not to criticize or denigrate what you have shared but to be able to understand the experimental method used.

    • 3D Matter

      Hi there,
      We used a Makerbot 5th Gen to perform all the tests. The tests were conducted exclusively on PLA, with one French filament supplier and one Chinese filament supplier. We tested the specimens on an Instron Universal Testing Machine. 3D Matter is a startup and is not affiliated with any institution. If you are interested in knowing more about us, please contact us directly at arthur@my3dmatter.com
      Hope this helps!

  9. […] 3D Matter, in particolare, ha appena pubblicato uno studio fatto in collaborazione con ParisTech Arts et Métiers, concentrandosi principalmente su come – usando una stampante MakerBot Replicator con PLA a 60 mm/s – variabili quali la percentuale di riempimento, l’altezza dello strato e la geometria di riempimento influenzino la qualità, i costi e la rapidità di stampa; una ricerca che ha come scopo quello di ottimizzare il risultato in base alle esigenze specifiche dell’utente. I risultati dello studio sono visibili nel grafico che trovate qui sotto, ma l’intero processo che ha portato a queste conclusioni è stato descritto nei dettagli e si potete consultarlo qui. […]

  10. Chow Loong Jin

    Could you include information on shell thickness for your layer height test please? I believe that strength is more likely controlled by the W/H ratio of the individual shell extrusion lines, number of shells, and total shell thickness rather than layer height.

    The problem is that when you vary the layer height, you can’t keep the other variables I mentioned constant. At the very least, either the W/H ratio or total shell thickness will have to change with the layer height.

    • 3D Matter

      Hi Loong Jin,
      You are making a very good point. We kept the number of shells constant across layer heights, so I agree with you that shell thickness is 2x bigger for 0.2mm than it is for 0.1mm. This can partially explain the increase in strength as layer height increase from 0.1mm to 0.2mm, but if it was the only factor then the tensile strength would keep increasing from 0.2mm to 0.3mm.
      From a practical standpoint, we think that users usually set the number of shells once and then vary the layer height, but don’t necessarily adjust the number of shells based on the layer height they pick. So this test may represent what happens in real life better, but I will keep your comment in mind in future studies.

  11. Greg

    This is great research! I was wondering if you would be willing to share your .stl files for your tensile bars?


    Greg Smith

    • 3D Matter

      Hi Shubam, not sure what you are asking here but we use a universal testing machine to conduct tensile testing. We measure the max stress that a material can undergo before it breaks (by slowly pulling on it).

  12. Alexandre Araripe Cavalcante

    Very valuable work for we all 3D printer users. I am thinking in make some test by my own and would be glad if you could provide me the stl files of the specimens you use. Thank in advance.

  13. Rohti

    I wanted to know if you have created any comparison between the results of this test to the mechanical properties of the material manufactured using traditional manufacturing process. If yes, then it would be very helpful for me if you can share the information
    thank you.

    • 3D Matter

      Hi Rohti, we have not done the tensile testing ourselves on the PLA we are using because it would mean melting the filament and injection molding it (if we want the exact same PLA), but you can get a fairly good comparison to injection molding PLA by comparing it to the datasheets available online. For example, this grade of PLA is a very good comparison:

  14. Nice work guys! I have one commentary about the results of layer heights, because i can agree with your results.

    I think, at more layer height will be less interlayer lines that act like cracks that will making the piece more weak. That´s why a minor layer height, the parts will be more weak. I think this only apply to materials with very good interlayer adhesion.

    • 3D Matter

      Hi Jose, thank you for your input! We have conducted many more tests on the influence of layer height on other polymers and don’t always reach the same trendlines, maybe your suggestion that it depends on how good the interlayer adhesion is for a given polymer is something to explore…

  15. OlafM

    Slicer now supports 3D honeycomb. Will it change the findings of this study? Or do you expect it to still be equivalent to diagonal/linear?

    Also, while on XY I understand that diagonal and linear should be equivalent to HC, I would expect on Z a very different behaviour in compression. 3D honeycomb is meant to extend that advantage to also XY.

    • 3D Matter

      Since this study we have designed a new tensile specimen specifically designed to test patterns, and we will try to write soon a brief article about the new results we got and in particular we will include 3D honeycomb in the scope of the work. Meanwhile, you can see our updated results for PLA on our data tool Optimatter. Even the Free Version should enable you to look at the data for different patterns at low infill (set it to 10%-30% to get the biggest differences). 3D Honeycomb is not yet in OptiMatter though, but we will try to add it soon.

  16. Michael

    You didn’t say although I assume you used an 0.4mm nozzle. Can you confirm this as I’d expect the layer heights can strength to vary if this is different.

  17. Tracy

    Very interesting and nice work. Thanks.
    I have a question for you: what is your linear infill pattern? What is your slicing software? I used rectilinear in 0 degree, but I am encountering defects in the areas of changing width due to the nature of tensile bars. Any comments? Thanks.

  18. […] percentage infill. Remember those this is brand specific to some extent and the material as well. What is the influence of infill %, layer height and infill pattern on my 3D prints? ? 3D Matter I had 4 roles of petg remnants that I printed this spitfire this past week. Waiting on some […]

  19. Mario O.

    I’m an engineering student researching the mechanical properties of materials, such as PLA, CF-PLA and CNF-PLA, manufactured by FDM.
    I’m very interested in your work. I was wondering if you have a .PDF version of this publication. If you do, could you send it to my Email?


  20. Harri Laaksonen


    there were earlier in this discussion question about “how you measure the stress”. The dimension of the stress is N/mm2 (MPa).
    When using the tensile testing machine (pulling the test specimens) the data what you get is force and elongation (stretch, how much test specimen length has changed). In order to get stress you have to define the area which is under affect.

    What cross sectional area you use? Same for all eg. area of solid test specimen.
    In figure you have shown max stress vs. weight. From that figure I got that impression.


    • 3D Matter

      Yes we use area of solid test specimen, but we understand it can be misleading because the specimens are partially void. When it comes to FDM 3D printing, we are actually testing structures, not materials.

Leave a Reply

Your email address will not be published. Required fields are marked *