Time Discontinuity Error (stress rel...

Log Out | Topics | Search Moderators | Register | Edit Profile

Author Message   Shu-Ting Cheng (coolmouse) New member Username: coolmouse Post Number: 1 Registered: 03-2008 Posted on Sunday, March 09, 2008 - 11:33 pm:    1d-relaxation 1D-relaxation.pde (1.0 k) 1d-pre_01 1d-pre_01.tbl (1.7 k) initial value 1d-pre.pde (0.8 k) hello, I am trying to model the stress relaxation problem (as attached "1d-relaxation"). My situation is a line initially loaded by two ends with unit force equals to 3, and will generate a uniform strain(stress) on the line (as attached initial value). The stress will decay by the relaxation equation ast>0. When I started to run this profile, there came the time dicontinuity error. I have tried what it has suggeseted like adding initial value, increasing time step, or adding the HALT script, but they didn't work at all. As I thought, it should be a simple form wiht stress decayint with time. I don't know what is going on here. Please help me..^^"   Shu-Ting Cheng (coolmouse) New member Username: coolmouse Post Number: 2 Registered: 03-2008 Posted on Tuesday, March 11, 2008 - 04:44 am:    besides,I have tried the method "reducing errlim below 1e-8 gives erro...(2006.4.28)" you suggested to other users.I used the RAMP function and decreased the time step,but when it comes to the time range where the load should be applied, the same error message comes out or the load is not applied at the time interval! revised-profile 1D-relaxation-2.pde (1.1 k)   Robert G. Nelson (rgnelson) Moderator Username: rgnelson Post Number: 1081 Registered: 06-2003 Posted on Tuesday, March 11, 2008 - 02:54 pm:    The load (or Natural) boundary condition takes its meaning from the integration-by-parts of second-order spatial derivative terms (see Natural in the Help Index). Your equation has no second-order spatial derivatives, so the load BC is undefined. That is, the Load BC does not correspond to a surface term in the integration of the equation, so it merely becomes a source term. You can ask FlexPDE to integrate first-order spatial derivatives by use of the FIRSTPARTS selector. If you do this, the Load condition now has a defined meaning in the solution system, and the problem runs.   Shu-Ting Cheng (coolmouse) Junior Member Username: coolmouse Post Number: 3 Registered: 03-2008 Posted on Tuesday, March 11, 2008 - 11:57 pm:    Thank you so much.It would do me a great favor!