Author |
Message |
Gianfranco Prota (gianf)
New member Username: gianf
Post Number: 2 Registered: 12-2008
| Posted on Saturday, March 14, 2009 - 07:28 pm: | |
Dear Mr Nelson, I have got a new problem. I have a box 2D (a rectangle) with the wall bottom hot and the wall upper cold. Inside there is air that moves through the difference of temperature and inclination of the box. The box is shift upwards of 5 degree compared to the previous position whenever flexpde finds the solution of the equations: vx: dens*[(vx)*dx(vx)+(vy)*dy(vx)]+dx(p) - Fx - visc* del2(vx)= 0 vy: dens*[(vx)*dx(vy)+(vy)*dy(vy)]+dy(p) - Fy - visc* del2(vy)= 0 p: del2(p)- dy(Fy)-dx(Fx)- 1e4* div(v)= 0 Tp: -cond*del2(Tp)+ rcp*[vx*dx(Tp)+ vy*dy(Tp)]=0 With: Fx= dens*beta*g*cos(b)*(Tp-Tm) Fy= dens*beta*g*sin(b)*(Tp-Tm) b=(PI/2)-(stage-1)*increment increment=5*PI/180 I want to assign a “errlim” for each stage to reduce the time of analysis (more than 12 hours) without losing its accuracy. For example I want assign errlim=5e-3 for first 8 stages and errlim=1e-2 for other 11 stages. How can I do? I post the script where I don’t know assign the “errlim” for each stage but a only “errlim” for all stages (5e-3). Regards, Gianfranco Prota
|
Robert G. Nelson (rgnelson)
Moderator Username: rgnelson
Post Number: 1222 Registered: 06-2003
| Posted on Saturday, March 14, 2009 - 07:42 pm: | |
You can assign a stage-dependent value to ERRLIM. All the problems in the version 5 "Samples | Misc | Accuracy" folder use STAGED definitions, eg, ERRLIM = STAGED (1e-2,1e-3,1e-4,1e-5) Other definitions are possible, such as ERRLIM = 10^(-2-STAGE/5); The referenced samples are found in "Samples | Usage | Accuracy" in the version 6 distribution.
|
|