Hi

Beauty lies in the eye of the beholder. The same goes for FE software. 

The most trusted - the usual answer, the one I happen to know (= trust) the best. 

However, 'best' likely very much depends on what you want to do with the code. 

At the risk of attracting flack.

There are a few distinctions that can be made

• Large linear problem: Nastran, Permas and I believe Abaqus has moved in the direction also. High performance codes usually come with lots of clever solvers for speed and parallelism. 
• I do not know Nonlinear well but Abaqus used to rank high for general problems.
• Crash analysis is a separate case. 
• Integration for convenience: Ansys was early with integrated Pre/Solver/Post but this was in the 90's. Also, in my opinion, there can be too much integration. 
• Flexibility: Code Aster is open source, executes in Python and uses HDF5 to store data and has integrated Pre/Post. For problems where you want to make special things, I guess this combo and other open source codes are hard to beat. 
• Integration for wide scope: Many codes interact with CFD solvers, Multi Body Simulation and so on.
• Lowest cost: In this category, open source likely takes the game when it comes to cost as in spent money. When counting cost as speed required to complete a project, commercial codes usually win the day. 
• If you are into nasty geometry, meshing capabiility may matter more than solver. In other situations, solver speed may be the factor that wins the day. 
• Nowadys there are places where you can run simulation without even having to install the CAE code. Imagine that - in the old days, the workstation carried 1/3 of the cost, the software 1/3 and training 1/3. 

The list can be made longer and cover aspects I have not touched upon as I am sure other members of RG will point out. 

On a more practical note. When I first got started at university, I was using a commercial code without support. This was somewhat of an uphill climb. At my next job, I purchased a commercial code with support. I did not choose the best Pre/Post processor for two reasons, i) the seller was clever enough to make a good Pre/Solver/Post bundle price and ii) he was sitting across the street which made it very easy for me to get help. I knowingly did not choose the cutting edge tool of this time as the support provided by this vendor would have made getting started difficult for me. 

Nowadays, much of the support is made using email. A code with an active user forum may well be the better choice. 

Me - I have used Abaqus, Nastran and am currently using Code Aster. The latter as I run simulation on/off and as I mostly do odd things (i.e. no support available anyhow) and as Code Aster is a code with lots of clever functions in it. 

The bottom line - if you know your stuff, you should always be able to switch code. Sure, you have to throw in a few hundred work hours to get used to the new gear but you should try never to get stuck with a single code as it tends to narrow your mind. 

Sincerely

Claes

Dear Mostafa

ANSYS and ABAQUS are black box softwares, and they aren't open source. So I suggest to you a open source platform "OpenSees".

OpenSees is a software framework for developing applications to simulate the performance of structural and geotechnical systems subjected to earthquakes.

http://opensees.berkeley.edu/index.php

Claes Richard Fredö and Mehdi Moghaddasi, thanks for your valuable help