Erratic nodal acceleration

[serggg]

Hello all,

I'm working with a fairly complex model in OpenSees with thousands of degrees of freedom (not reproducible through text here) and I'm running a nonlinear time history analysis through the Uniform Excitation command, a Newmark integration scheme with parameters gamma=0.5 and beta=0.25 is used.

I wanted to get a peak floor acceleration profile but I've been finding an erratic output from the nodal absolute accelerations (attached figure https://ibb.co/JkJCWD0).

Interestingly enough, if I apply a simple moving average filter in Matlab it somehow corrects it, but not fully (https://ibb.co/WkzdsMQ).

The units of acceleration in the figure are in terms of "g", so it is fair to say that there is some numerical issue going on. This is not the case for every story or every node, only the roof, and stories where there is a lateral stiffness discontinuity (again, fairly complex building structure).

Has anyone come across something similar or know of a possible reason for this? I've been playing with the damping to see if it related to a higher mode but haven't had any luck, I'm using a Rayleigh damping model with about 2.5% damping ratio in the first modes of the structure, and to damp out higher modes I have about 25% for 20 Hz.

Thank you for your comments!

[mhscott]

It looks like damping has not been applied.

[serggg]

I have been applying rayleigh damping through the command:

rayleigh $alphaM $betaKcurr $betaKinit $betaKcomm

Using the value betaKcurr and betaKcomm as 0, thus just considering beta for the initial stiffness.

I'm inputting the command before defining the UniformaExcitation.

I'm hesitant to increase the damping ratios to values larger than 5% in periods between 0.2To and 2To, where To is the fundamental period after applying gravity loads. Right now I have the damping ratios for those modes set at 2%.

I've also done the exercise of recording the acceleration, velocity, and displacement response for a node that has this "erratic" behavior, and the issue seems to be only in the accelerations, the velocity and displacement do not show these oscillating peaks for the same node.

Thank you,

[mhscott]

You say your model is "fairly complex". Try a smaller, simpler model to ensure the results make sense.

[serggg]

Thanks for your advice Professor, in the past simpler models don't show the same blow up response using similar damping parameters.

Fortunately, the rest of the results in the complex model make sense, at the element and at the system level, displacements, forces, strains, stresses.

I will continue to investigate what may be causing this behavior in the acceleration outputs.

Thank you again for your time in responding and for your work in OpenSees