  # Dirk76

Members

9

## Everything posted by Dirk76

1. My goal is to have them in contact during second subcase. If the initial gap is 5 mm and during the first subcase I apply 100 N, the two plates shouldn't come in contact... this is why there's some mistake in the model. If you try to increase the gap, with that contact (or different, I tried various solutions) formulation, no matter what, the plates will come into contact. A model that work with these initial conditions should be made in a way that the gap only reduces of a finite quantity < initial gap, so that increasing the load in the following substeps the gap should reduce to 0. When you tried running this model, you obtained these results?
2. after the first iteration with 100 N, the two plates don't come in contact? then it's what I need, since everytime I apply different load values the two plates reach the initial gap. If it's the case, would you share the model? Thanks. that's because my main goal is to have the two plates in contact only after the second step of the analysis, so I need to set the problem in this way.
3. Hi all, I was running a model in Optistruct 13.0 dealing with a two step non linear quasi-static analysis of two plates coming into contact. In the first step my goal is to make sure the two plates don't come into contact; let's say my initial gap is 5 mm, one plate is pushed down by a force of 100 N, the other one is fixed at the base. At the end of this step the two plates should have minimum gap between them, in other words no load should be transferred from one to another. After the second step in which I apply a force of 1000 N, the two plates obviously come into contact, resulting in a load transfer to the spcf applied at the base of the fixed plate, since friction between the plates is activated. I tried using a gap element in order to simulate the problem but no matter how I change the values into the pgap card (for example I first put a value of Kb equal to 25 N/mm to make sure the pushed plate didn't reach 5 mm of displacement with a force of 100 N.... didn't work, then I put it equal to 0 and I obtained the same result ), at the end of the first step the gap of 5 mm is always reached, resulting in load transfer. How can I avoid this situation in order to make it work properly? I'll attach the .hm file so you can check it. Thanks for any replies, hope someone will help me with this matter! model_with_gap.hm
4. Hi Prakash, thanks for the reply. Yes, you'll find them attached to the post as well as the comparison of the results in terms of stress, displacements and constraining reactions. As you can see from first sight, the two models are practically correlated, the slightly differences are acceptable since we're comparing a 1D model with a 3D one. I think that the value of 1.020e+06, in the 3D, for MRx is due to the fact that the independent nodes have an offset of 10 mm from the central node on the neutral axis, for both the applied force and the constraint. In the 1D you have, instead, the correct value since you apply directly the force and the constraint on the nodes of the beam. If you try to make a simple calculation with pencil and paper, you'll find these values for the two models, meaning they work pretty well. model_1D.fem _odel.3D
5. Hi all, I was running optistruct 13.0 in order to make a correlation between a 1D and 3D model. First of all I simply created a fixed beam at one end, and the other end with a force of 1 kN. The outputs were very simple to obtain, indeed I could actually have pretty similar results in terms of stresses, displacements and constraining reactions. For the 3D the force and the constraint were applied using RBE2 elements, considering a node as independent and the ones on the cross-section of the beam as dependents, as you can see in the figures below I attached. Now the thing is, when I tried to make a correlation between a slightly different model than the one mentioned above, I couldn't obtain similar results, instead they were really different. I made a comparison between stresses, displacements and constraining reactions. The force applied was 1 kN as well, on the free end of the "T-structure". The section used for both 1D and 3D was an IPE360. For the beam of 1 m (the one where I applied the force) I used a freeze contact with the one of 2 m (fixed at both ends), in order to replicate the same condition of the 1D. The displacements were really different, almost 2 order of magnitudes of difference... the stresses obtained in the 1D was 4 times bigger than the one in the 3D and the constraining reactions (MRy) obtained didn't equal the real flexural moment generated by the force applied at the free end (it should equal 250000 N*mm as in the 1D, for a fixed-fixed condition). I attached all the comparisons and the .hm file if you wanna see the pre-process of the analysis. Could someone give me an explanation on this? Thanks in advance for any replies! model_1D.hm model_3D.hm
6. Thanks guys for the response. I need it because I wanted to see delamination, since the classical contact gap in HV doesn't work for me and I really don't know why... maybe my solver is too old. I think I have to do it in LS Prepost.
7. Hi all, for a certain purpose I do have to make a cross-section of a plate made up with shell elements. Now I know how to do it but I was thinking, is there a way to visualize the actual shell thickness in Hyperview? It is possible to do it? Thanks in advance.
8. Hi all, I'm encountering a problem in a low-velocity impact simulation with a composite plate. As you can see from the image below, at the beginning of the analysis, the impactor literally penetrates the plate but without effective failing of any elements. I experience this problem when I try to simulate the delamination, ply by ply (in my case 16 ply), using very thin shell elements (the thickness of each ply is approximately 0,23 mm). Indeed, after studying the contact problem in LS-DYNA, I figured out that, when very thin shell elements are used, the plate drastically fail when the impactor approaches it. For this reason I had to reduce the SLSFAC from 0.1 to 0.001 in *CONTROL_CONTACT in order to have reasonable results, because the impactor of only 1,5 kg and with an initial velocity of 6,5 m/s destroyed the plate in the very first milliseconds of the simulation without this change. But now, solved one problem (however I'm not really sure about this, maybe someone can explain me better this contact stiffness problem), I have another one I'm using a MAT22 model for the composite and a MAT20 for the impactor (made of steel). To simulate the delamination I chose the *CONTACT_AUTOMATIC_SURFACE_TO_SURFACE_TIEBREAK with 15 delamination model in between, penalty contact SOFT = 0 (justified by the fact each ply is identical) and OPTION = 8 for the failure criteria (I didn't know if PARAM = CCRIT for this option had to be used, so I left it blank). For the contact between impactor and the plate I chose *CONTACT_AUTOMATIC_SURFACE_TO_SURFACE with SOFT = 2 (because of the different materials and meshes), FS=FD=0,5 and with the impactor being the master and all 16 plies being the slaves. Any suggestions to solve this problem? Thanks in advance for any response. impatto.bmp
9. Hi all, I'm a Master Degree student in Mechanical Engineering and I'm currently working on the final thesis. I do have a problem that I want to share with you. The topic of my thesis is focused on low-velocity impact on a C/SiC plate, modelling in hypermesh and using LS-Dyna as solver. I chose, to model the composite, the *MAT_COMPOSITE_DAMAGE (Mat22) option and here's my problem. Since in the literature there's practically nothing about mechanical properties of C/SiC, I'm finding a lot of difficulties to estabilish the inputs of the Mat22. In fact I need, in the order, EA, EB, EC, PRBA, PRCA, PRCB, GAB, GBC, GCA, SC, XT, YT and YC. I thought, after a day trying to find some information on the mechanical properties, to proceed in a different way. The composite is made of a 2D woven carbon fiber, so we're talking about a transverse isotropic composite where the plane of the woven is the isotropic plane. For this reason EA=EB, XT=YT, and other parameters such as PRBA, GAB, SC and YC is possible, in a way or another, to find (even if I have to do semplifications for what concern the properties related to the volume % of fibers and constituent materials, since, like I said, there's nothing in the literature about these CMC). Now the question is, for the remaining parameters (EC, PRCA=PRCB, GBC=GCA) is correct to assume them to be that of the matrix (SiC)? I really need your help so thanks in advance for any response.
×
×
• Create New...