The main difference between all ANSYS flavors and COMSOL Multiphysics in general is that in COMSOL you can actually type in the PDEs you want to solve in the explicit mathematical symbols way and it automatically creates the weak form and discretizes the problem. You do not need to assign a specific element type like you do in ANSYS but only define the order of the shape function you want for the elements. For the case of structural mechanics in addition to selecting one of the existing (linear or non-linear) constitutive laws, you can also explicitly type the mathematical form of the constitutive equation you want and that's all it takes. This does not mean that the COMSOL solvers will always deal with these customizations but if you are experienced you can certainly achieve much more customization in that area. Thus you do not need to write a USERMAT routine as you need to do with ANSYS, compile it and link it with the ANSYS executable. I find COMSOL much more convenient in that regard than ANSY. However, ANSYS's solvers have proven to be MUCH more efficient than COMSOL's for me given the same geometry and mesh payloads. They both have scripting capability (APDL for ANSYS and Matlab for COSMOL) so that is no longer a significant difference. Also, the generic COMSOL module allows you to deal with much more physics than the basic ANSYS mechanical allows you to. For these reasons I use both.

Also depends on the graphical environment with which you are most comfortable.

The main difference between Ansys and Ansys workbench is Graphic User Interface. Ansys workbench offers better GUI than Ansys for beginners. About COMSOL I don't have any experience. Good luck !

Dear Awani Bhusan,

According to Mr Jawad Masood, the main difference between Ansys APDL and Ansys workbench is Graphic User Interface. In ANSYS APDL, you can write your code according your problem analysis and can changes data or changes analysis or changes meshing etc at any time as your requirements. According to me using ANSYS APDL is better than using ANSYS workbench. Thanks

I agree with Ranjan. But it take time to learn APDL while learning workbench is straight forward......

among the GUI, concerning the structural analysis, ANSYS Workbench provides less finite elements for your analysis than ANSYS APDL. This last has a very wide library.

to evaluate fracture parameters, you should use APDL, Ansys has suitable examples in fracture mechanics. please search in "Ansys helps" carefully By keywords such as j-integral and etc.you can start with "CINT" command. This command help you to evaluate several fracture parameters such as energy release rate,stress intensity factor, j-integral using different approaches. In addition, to make parametric study, it is more convenient to use macro(APDL code).

The main difference between all ANSYS flavors and COMSOL Multiphysics in general is that in COMSOL you can actually type in the PDEs you want to solve in the explicit mathematical symbols way and it automatically creates the weak form and discretizes the problem. You do not need to assign a specific element type like you do in ANSYS but only define the order of the shape function you want for the elements. For the case of structural mechanics in addition to selecting one of the existing (linear or non-linear) constitutive laws, you can also explicitly type the mathematical form of the constitutive equation you want and that's all it takes. This does not mean that the COMSOL solvers will always deal with these customizations but if you are experienced you can certainly achieve much more customization in that area. Thus you do not need to write a USERMAT routine as you need to do with ANSYS, compile it and link it with the ANSYS executable. I find COMSOL much more convenient in that regard than ANSY. However, ANSYS's solvers have proven to be MUCH more efficient than COMSOL's for me given the same geometry and mesh payloads. They both have scripting capability (APDL for ANSYS and Matlab for COSMOL) so that is no longer a significant difference. Also, the generic COMSOL module allows you to deal with much more physics than the basic ANSYS mechanical allows you to. For these reasons I use both.

@John Michopoulos, the question of which is the most efficient way to describe a multi physics problem in finite element terms is an interesting one. If we can agree that the end result is the same, then we are placed in the realm of convenience and level of detail in the GUI. I imagine that it also depends on the way that one is trained professionally. While I expect engineers to prefer the classic FEM ( read Ansys, Abaqus, Marc ) approach which allows them to think of the solution in broad system terms, while physicists and applied Mathematicians would prefer the PDE approach ( read COMSOL, Sfepy etc.) where they will feel that they are in better control of the mathematical translation of the physical problem. Since the computer is agnostic, it does not care which method it executes. The user however should be concerned with the verification of the program and the validation of the method. The validation will throw the user into a stochastic world.

for Fracture Mechanics Ansys 14.5 is better than COMSOL, as Ansys14.5 contains separate module FRACTURE which helps to introduce crack any where,also evaluates fracture parameters like SIF,J-integral

For any trouble in modelling with ANSYS, I would recommend you www.ansys.net orhttp://myweb.ncku.edu.tw/~hhlee/Myweb_at_NCKU/ANSYS14.html

Really interesting!!

The major problem with Ansys/workbench is that you have no options for selection of elements and its attributes such as full or reduced integration formulation as compared to Ansys Classic. Further, the loads/boundary conditions are only applied on geometric entities instead of finite elements or nodes in the Workbench. It only suits as a first check for the analysis of structures instead of detailed analyses.

Agreed. ANSYS APDL gives you lot of flexibility.

Well, i only use ANSYS FLUENT, and you have at your disposal a c compiler and a c/c++ interpreter. Hence you have two options in FLUENT:

1. Use predefined functions in udf.h and be able to define advection and diffusion equations, and whatever you can imagine such as source terms. You only have to define certain properties.

2. If you don't trust ANSYS solvers you can code your own c/c++ functions where you define partial differential equations making use of physical variables provided by FLUENT. (I don't recomend this because is a waste of time).

When it comes to simple mathematical domains, you may prefer to generate your own mathematical domain and program physical eqautions in FORTRAN, C++, Python, etc.

At this moment i don't know if other ANSYS component systems (APDL, CFX, mechanical) have similar capabilities.