We are collecting frequently asked questions related to the mechanics of structure genome and SwiftComp using this Wiki Post.
Q: I recently read your interesting paper “Constitutive modeling for time- and temperature-dependent behavior of Composites, 10.1016/j.compositesb.2019.107726” that reinforces my knowledge on multiscale analysis of composite materials. However, I do not understand some details and want to learn more. Therefore, I am writing this discussion to ask for your direction. More details are attached below. Details-for-discussionn.pdf (94.7 KB)
A: In regards to your question, Eq. A1 is taken directly from the derivation of Prof. Schapery (see reference [11]) of my paper. The Beta (T, t) of the formula already accounts of the temperature and time dependencies of the stiffness matrix and the coefficients of thermal expansion, but the coefficients of thermal expansion are not explicitly written within the integral form of the thermoviscoelastic equation. In other words, the coefficients of thermal expansion are derived from the Beta (T, t) and the stiffness matrix once both have been computed. Eq. A1 is the most general way to write the thermovisoelastic equation. I am not familiar with the derivation that Chung et al. followed (or referenced) in their paper for Eq. A3. However, when the thermoviscoelastic equation is presented in the integral form with the coefficients of thermal expansion within the integral, it is common to assume that the coefficients of thermal expansion are only temperature dependent (i.e. independent of time). Therefore, as you said Eq. A1 and Eq. A3 are only equivalent if the coefficients of thermal expansion are time independent. Hence, my suggestion is to check what are the assumptions that Chung et al. followed (or referenced) in their paper as the time independency of the coefficients of thermal expansion may be an inherited assumption of Eq. A3.
A: When the composite is not orthotropic, you will have shear-extension couplings corresponding to these two values. We will update the user manual to reflect this.
Q: I saw your power point presentation on buckling of stiffened panels recently. I have a question that makes me very confused and I hope to get your reply. The ppt shows that SGs of different dimensions give the same stiffness matrix, the 2D SG is a surface without thickness, the 3D SG has length, width and height, how to set the length of the 3D SG so that the stiffness matrix is the same as that of the 2D SG.
A: The length in y1 direction of the 3D SG (red line) does not matter, it is uniform and can be arbitrary. if the length of such 3D SG becomes zero, it is equivalent to a 2D SG (in shell element), which will give you the same constitutive relations (6x6 plate stiffness). Actually, the 2D SG is the fundamental building block, or a unit cell, that contains all the necessary information for constitutive modeling in this case, the length (your red line) is only required for 3D SG. However, the 2D SG only works when the stiffeners are uniform in the longitudinal direction, if a cross-type stiffeners are present (stiffeners are pointing both directions), then SG has to be in 3D.
Q: It is learned from the forum that ABAQUS cannot use the beam stiffness matrix as direct input, and the corresponding terms must be extracted as the input of ABAQUS. Assuming that I have obtained the beam stiffness matrix, how should I operate the corresponding term? I have guessed a method. As shown in the figure below, input the equivalent attributes in the K. file into the spaces such as d1111 and d1122 one by one. How to set the section attributes? There is only equivalent stiffness in the K. file, and there is no information related to the section attributes.
A: You need to understand how Abaqus inputs (material properties and section attributes) used to compute the corresponding beam stiffness. Make them equal to each, then you can define the corresponding terms. If all you need is a beam analysis, I will suggest you to use GEBT instead. You can directly use the beam stiffness matrix from SwiftComp or VABS. We did something for extracting SwiftComp beam stiffness for Nastran beam elements, which can be found in the following paper.
Liu, X.; Gasco, F.; Yu, W.; Goodsell, J. and Rouf, K.: “Multiscale Analysis of Woven Composite Structures in MSC. Nastran,” Advances in Engineering Software, vol. 135, 2019, 102677.
To assign beam properties to beam elements in Abaqus, you need to do some editing in the input file (.inp). The way I know is to put beam properties in a separate ‘.bsp’ file and then include this file in the ‘.inp’ file.
In this ‘.inp’ file, you need to add the following:
*BEAM GENERAL SECTION, SECTION=MESHED
…
*INCLUDE, INPUT=jobname.bsp
where the ‘jobname.bsp’ contains the beam properties including EA, EI, GJ, etc. You can search ‘meshed beam cross-sections’ in the Abaqus documentation for more detailed information.
For more understanding of beam properties, refer to Yu, W.: “Inertial and Elastic Properties of General Composite Beams,” Composite Structures, Volume 352, 2025, 118690.
