ode15s DAE problem with the algebraic equation

Question

Iasonas Ioannou 2016-9-22

0
链接

此问题的直接链接

https://ww2.mathworks.cn/matlabcentral/answers/304099-ode15s-dae-problem-with-the-algebraic-equation

Hello every one. My name is Iasonas.

I currently trying to solve a DAE problem. I have to solve 4*n differential equations and one algebraic equation. The simulation will compared with some experimental results and because the algebraic equation is for a specific variable of an experiment at first I used a polynomial from my experimental results in order to be sure that everything else is correct before i continue.

And all was great. When i use the theoretical equation instead of the fitted results the simulation continue until the half time of the period in question and then i get an error unable to meet integration tolerance. I would like to mention that the result are correct until the error.

When i tried to find any bug I saw that the residual equation had oscillations on the time that the ode solver stops and with the adaptive smaller step of the solver the oscillations are greater ( + - oscillations). I tried a bigger time step, and i debugged all of my code..

I am trying to understand how i could correct it. I was thinking just not to use the ode15s and make my own function using crank- nicolson method so i will eliminate the oscillations. The problem is transient and the boundary conditions and variables of change, and I think that is the reason of making my system stiff after many iterations of the solver. If I said something that is incorrect or if anyone could help I will be thankful

Regards Iasonas.