cancel
Showing results for 
Search instead for 
Did you mean: 
cancel
Showing results for 
Search instead for 
Did you mean: 

Chemeng Problem: Floating Point Error Runge-Kutta 4th order

Highlighted
Newbie

Chemeng Problem: Floating Point Error Runge-Kutta 4th order

Hello! I am a newbie to this community . I would really appreciate any help with my problem.

First of all, i was asked to apply the Runge-Kutta 4th order method to solve this problem.

I managed to get something out of this, but when i change the value of the rate constant (k), for valutes >2, i get a floating point error.

My aim is to get 9 different k (rate constants), ranging from 1 to 10, in order to get (via the Runge-Kutta) the C values of my components (i have named them x and y) . Then the C-t curves will be compared with my experimental curve.

However, whenever the value of k is greater than 2, i get a floating point error . Any advice?

Any suggestions or corrections would really be appreciated.

I attach you the mathcad file:

3 REPLIES 3
Highlighted

Re: Chemeng Problem: Floating Point Error Runge-Kutta 4th order

You algorithm doesn't seem to converge.

For k:=3 its the iteration step at i=48 which throws the error.

x.48=y.48=5.89*^10^291. Your functions f multiplies x and y and the result would exceed Mathcads limit of approx. 10^307, so you get the error.

You can insert a trace command in your program and toggle debug (menu/tools) to watch the variables in your program.

trace1.png

trace2.png

Highlighted

Re: Chemeng Problem: Floating Point Error Runge-Kutta 4th order

Mathematic is not mathematic without animation

http://communities.ptc.com/videos/1471

Highlighted

Re: Chemeng Problem: Floating Point Error Runge-Kutta 4th order

Well guys, thanks for the posts! I found the problem. It was a problem at the mass balances (should have a minus not a plus in front of k). The solutions i had with k=1 and k=2 were unrealistic in a chemical way (CNaOH was valued way greater than it should have been), σο i rechecked the problem.

Thanks for your help Werner! One word (not converge) gave me a serious hint!

Have a nice day!

Announcements