Problem with Node Recorder

[GiuseppeDelGobbo]

I am having a problem with the node recorder. I am recording displacement at a node in one DOF. I specified the recorder time step to be 0.01. This works until t=1 sec, where the recorder time step randomly changes to 0.011. What would cause this to happen? The time history analysis I am conducting is done at an analysis time step of 0.001.

Recorder:
recorder Node -file $dataDir/Floor_Disp.txt -time -dT 0.01 -node 116 -dof 1 disp;

Sample output:
0.97 -4.28586
0.98 -4.26486
0.99 -4.23848
1 -4.20708
1.011 -4.16704
1.022 -4.12126
1.033 -4.06933
1.044 -4.01031

Thanks,

[skamalzare]

You need to use a "-" before your dT. The command will be:

recorder Node -file $dataDir/Floor_Disp.txt -time -dT 0.01 -node 116 -dof 1 disp

[GiuseppeDelGobbo]

thanks but I have the dash before dT so that is not cause of the problem (typed it in the forum post wrong).

[skamalzare]

If you look at the OpenSeesWiki for node recorder, you will see dT is the: "time interval for recording. will record when next step is $deltaT greater than last recorder step." It is not exactly dt by dt, it is one dt greater than the last one! Now, if you implement "variable transient" analyses and your solver fails to converge, it will reduce the time-step to meet convergence. That might be the reason behind your recorder time-steps. Apparently, after 1 sec, your solver has failed to converge, and has reduced the time-step to of 0.001 sec. The time 1.011 was the first analysis after time 1.00 sec with 0.01 sec greater time.
Another reason might be behind your input motion time-steps. If it contains variable time-steps, your solver will work with variable time-steps and records with $dT criteria you assigned to it!

[GiuseppeDelGobbo]

Hi,
I am not using variable time steps for my input motion or variable transient analysis so that is not contributing to the problem.

[fmk]

it looks like your steps are being recorded at .011 steps when you requested 0.01 steps .. the problem is roundoff .. .01 is a number that cannot be represented in a computer exactly, the dt the recorder is checking could be .010000000000......00001 .. try specifying 0.009999999999999999999 instead.

[GiuseppeDelGobbo]

Thanks for the help, roundoff was the problem.