Hi, I wanna ask for opinions/suggestions to reduce the percentage error between experimental and simulation results in the ANSYS workbench ( static structural).
First, let me explain my research.
I have cubic lattice structure samples (60mm x 60 mm x 60 mm) made from ABS thermoplastic material using FDM additive manufacturing. These samples were compressed using INSTRON 5585 universal compression test machine. The data from the test were analyzed to get the stiffness, young modulus, yield stress, etc. However, the results from simulation using ANSYS static structural were nowhere near the experimental results. For example, to get the stiffness, I divided the exerted force to the deformation and I get +82% error (simulation >> experimental). for young modulus, I divided the equivalent stress to strain and get -234% error (simulation >> experimental). I suspect these as the factors for the error
1. The samples made from FDM AM have too many manufacturing defects (interlayer failure+ porosity), yet to try using other AM techniques.
2. During the experiment, the area used was 60 mm x 60 mm as I cannot specify exactly the contact area (much less contact area) for the lattice to the bottom platen as the machine only asked me to put in the width and length of the sample. where probably (I'm not sure) the area used in the calculation in ANSYS is the actual contact area to bottom platen.
3. The structure is simply too complex.
I've tried using actual density and young modulus I get from experimental in engineering data and use force distribution plate to the top and bottom of the samples in ANSYS environment but not much changes.
also for case #1 is there any way I can render the geometry in ANSYS to have layers and voids as in actual samples?
It is not quite clear to me what you want to do. Are you interested in the properties of the beams in your spatial frame, or are you seeking the properties of a homogenized continuum model? In the latter case, I can highly recommend the nice theory of cellular solids by Lorna J. Gibson, which she started in her Phd thesis supervised by Prof. Ashby. It is simple and gives good results and insights. You may find it in one of their books: Gibson, L.J., Ashby, M.F. (1997) Cellular Solids: Structure and Properties. 2nd ed., Cambridge University Press. https://doi.org/10.1017/CBO9781139878326
I would like to point out some stuff. First of all, why are not you using the data collected from the machine with the possibly known contact area instead of 60x60mm? Since, you are assuming a much larger area, naturally you are getting much lower stiffness in experiment.
Defects can definitely modulate overall simulation vs exp but it does not really explain that larger variation.
May be you need to revisit your experimental data first and then work on simulation to validate your model.
Happy Researching!!
In case you use non-linear material properties, try to use C3D10 instead of C3D4 elements.
Manfred Staat Thank you for answering my question. I have designed the lattice structure in multiple design parameter combinations (a total of 9 individual designs with different structure arrangements). Hence I want to see the effect of these arrangements on mechanical properties for example stiffness, young modulus, yield stress, and the derived of these properties such as the specific strength or specific stiffness.
so I think my case is the latter the which is the properties of a homogenized continuum model. I have conducted experimental work and these structure arrangements have effects on the aforementioned properties. The same can be said for simulated results, the arrangements have influences on the properties.
However, my problem is the experimental results and simulated results have huge errors.
Thank you for your reference suggestion. The book is actually my main reference and had helped me a lot as u said.
Soud Choudhury thank u for answering my question. perhaps my question was not clear, I apologize for that.
let me elaborate, during the compression test, the machine only allows me to specify the area of the base of my sample, for example as my samples are closer to the rectangular area (cubic model with 60 mm on each side). The setting (refer to attached picture; taken from online source so not my actual dimension) for the specimen in the machine only allows for the width (60 mm) and thickness (60 mm) of the model.
I'm also confused as to why so much error, this led me to my many assumptions which include maybe in the simulation, ANSYS used the exact area of the lattice struts' legs contact area but this is not possible in the experimental setting like I mentioned.
Matin Afshar thank you for your suggestion. However I'm not sure where do i go, is it ANSYS setting? or ABAQUS?
- Badges
- Science topic