Discussion Closed This discussion was created more than 6 months ago and has been closed. To start a new discussion with a link back to this one, click here.
Convolution integral
Posted Nov 5, 2009, 11:43 a.m. EST 3 Replies
Please login with a confirmed email address before reporting spam
Hi, this is the first time i write in this forum but i work with comsol multiphysics since his name was Femlab, and precisely I use the pde mode apllication to solve nonlinear static and time depending problems. Right now I'm trying to implement a nonlinear time depending problem where I need to use a forcing term that is a convolution integral (in time domain) of the indipendent variables and/or their time derivatives. Does someone know if it's easily possible to do that directly with comsol commands? Maybe I need to modify the code by using comsol script? Or perhaps I definitely may not do that?
Hope someone could give me some good answer
Bye
Andrea
Hope someone could give me some good answer
Bye
Andrea
3 Replies Last Post Nov 10, 2009, 6:03 a.m. EST
Please login with a confirmed email address before reporting spam
Posted:
2 decades ago
Hi
Well if you are a long term user of COMSOL, what about passing by more often, and perhaps share a couple of goodies that you have elborated ;).
Concerning your question I have some problems to follow, can you manage to write out the equations then we would better understand exactly what you are trying to calculate.
For me with script, COMSOL can do about everything one can imagine, only within the GUI's its more restrained, but still.
I know that with the dest() operator, you can achieve interesting integrals in the space domain, but you want it in he time domain.
In any case, if its very important and urgent for you I would use the comsol support, if so pls report back here also, it could well help us others too.
Ivar
Well if you are a long term user of COMSOL, what about passing by more often, and perhaps share a couple of goodies that you have elborated ;).
Concerning your question I have some problems to follow, can you manage to write out the equations then we would better understand exactly what you are trying to calculate.
For me with script, COMSOL can do about everything one can imagine, only within the GUI's its more restrained, but still.
I know that with the dest() operator, you can achieve interesting integrals in the space domain, but you want it in he time domain.
In any case, if its very important and urgent for you I would use the comsol support, if so pls report back here also, it could well help us others too.
Ivar
Please login with a confirmed email address before reporting spam
Posted:
2 decades ago
Hi Mr. KJELBERG,
I did and I'm still doing some work with comsol for my master degree first and right now for my PhD, all concerning the nonlinear modeling of bridges (suspensions bridges and arch bridges) and the analisys of their response under wind loads. I'm trying to implement just an aeroelastic time depending force in the equation of motion of the bridge.
The contest in comsol is PDE mode, general form, in 1D space dimension, let u(x,t) be one of the independent variables and f(t) a known function of my problem. What I want to implement in the F term of the field equation ea*du^2/dt^2+da*du/dt+div(G)=F is
F= int_0^t[(du(x,s)/ds)f(t-s)]ds where "s" is an auxiliary time variable
so a convolutional integral.
I was thinking to bypass this problem by introducing a costraint equation which solution of is just that integral. I'm not sure if I could use the Global equation tool or i better define a new independent variable and set the constraint by the new generated field equation (I used this trick other times). I've not tryed yet this idea...for sure I'm gonna do it on monday when back at work at my University.
I know me too that I can use the dest() operator but just for space integral....and I also read that exist the at() operator but it works just on postprocessing mode.....I was also thinking to split the time analisys and restart it evaluating the convolution integral with this command and updating the equations....and for that reason i thought to use comsol script....
I will work on that!!!
However thank you for the answer
best regards
Andrea
I did and I'm still doing some work with comsol for my master degree first and right now for my PhD, all concerning the nonlinear modeling of bridges (suspensions bridges and arch bridges) and the analisys of their response under wind loads. I'm trying to implement just an aeroelastic time depending force in the equation of motion of the bridge.
The contest in comsol is PDE mode, general form, in 1D space dimension, let u(x,t) be one of the independent variables and f(t) a known function of my problem. What I want to implement in the F term of the field equation ea*du^2/dt^2+da*du/dt+div(G)=F is
F= int_0^t[(du(x,s)/ds)f(t-s)]ds where "s" is an auxiliary time variable
so a convolutional integral.
I was thinking to bypass this problem by introducing a costraint equation which solution of is just that integral. I'm not sure if I could use the Global equation tool or i better define a new independent variable and set the constraint by the new generated field equation (I used this trick other times). I've not tryed yet this idea...for sure I'm gonna do it on monday when back at work at my University.
I know me too that I can use the dest() operator but just for space integral....and I also read that exist the at() operator but it works just on postprocessing mode.....I was also thinking to split the time analisys and restart it evaluating the convolution integral with this command and updating the equations....and for that reason i thought to use comsol script....
I will work on that!!!
However thank you for the answer
best regards
Andrea
Please login with a confirmed email address before reporting spam
Posted:
2 decades ago
Hi Andrea,
At the moment I was trying to implement a 1D model in PDE mode with as source term the convolution of the variable u(x,t) and a function g(t).
It is not possible to acces the variable u(x,t) at the previous time step while solving, then the only way, as you said, is to use a script to stop the simulation, calculate the convolution integral and to update the equation .
I'll try to do something like that and, if you like to, we could keep in touch and try to find the best solution for that problem.
Regards
Alberto
At the moment I was trying to implement a 1D model in PDE mode with as source term the convolution of the variable u(x,t) and a function g(t).
It is not possible to acces the variable u(x,t) at the previous time step while solving, then the only way, as you said, is to use a script to stop the simulation, calculate the convolution integral and to update the equation .
I'll try to do something like that and, if you like to, we could keep in touch and try to find the best solution for that problem.
Regards
Alberto