How to simulate gunshot with FEMLAB

[lkjt]

I attached femlab document.

Why doesn't the ball(E1) go far away?

 

[Mech_Engineer]

I attached femlab document.

Why doesn't the ball(E1) go far away?

I don't have FEMLAB, but it could be that the problem is only solving for "small displacements." In ANSYS for example, you need to turn on large displacements for a problem like a gunshot. Are you trying to solve the trajectory of the bullet as well, or just the bullet leaving the barrel?

I would say that FE software is not a very good way to simulate the trajectory of a bullet, instead I would solve for its final speed after just leaving the barrel, and then use that velocity in MATHCAD or MATLAB to show its trajectory using the kinematic equations, taking air drag into account of course...

 

[PerennialII]

The attachment is still pending approval so still a bit hazy (anyone ?).

What physical aspect of a gunshot are you attempting to simulate, could you elaborate a bit.

 

[Integral]

I am reluctant to approve any attachment that I cannot view. I do not have femlab, so cannot verify the content. Guess I will aprove it without the verification. Open at your own risk.

 

[PerennialII]

It's a valid Femlab model file ... will return to the topic later if no one beats me to it .

 

[PerennialII]

Hi lkjt,

I looked at your model, your application mode & analysis type should be able to reproduce the type of result you're aiming at (transient dynamic one). Naturally there are lots of simplifications you've had to do and a thorough simulation of a gunshot is a really difficult task (like if would open the 'can' of accurate fluid-structure interaction modeling for a gunshot and add the acoustics problem to the mix), but looks like a good example to me. However, there are a couple of things which see potentially causing trouble.

1st you probably noticed the convergence issue - a believe it'll rectify itself pretty easily by controlling the time incrementation and specifying it according to the displacements the mesh experiences, so that the features of the solution can be traced accurately time-wise. Then there is the issue mech_engineer already touched, i.e. FEM is basically usually very lousy what comes to large trajectories, finite displacements combined with rigid body movement and so on (your analysis focuses on the early stages granted, but still compared to say, to a stress analysis or so, the displacements are really large). This isn't a theoretical limitation, rather how the numerical solution process plays itself out. But, since you're working with comsol/femlab, you're lucky (or you probably knew that already). If you've already gotten your model to work with your current settings ... great ... but I wouldn't be somewhat surprised if you could do so without using the adaptive and deformed mesh features of Femlab (meaning you'd get the bullet out of the barrel). It may be that h-adaptivity for example is in order in your analysis, but more likely using the arbitrary Lagrangian - Eulerian technique for mesh movement. Comsol (especially the latest update) has a really nice implementation of this technique which has as of late been introduced to a number of FE programs, and I believe it could be used nicely to solve your problem. Had you had any further luck with your model and in your mind does what I'm thinking make any sense (can help you with the buildup if needed) ?