*Dynamic time stepping algorithm 1

[jball1]

I am trying to validate a dynamic implicit analysis performed in a different software package by running the same analysis in Abaqus. The other software package uses a HHT time stepping algorithm with alpha = 0 (no numerical damping). In order to perform an apples to apples comparison, I ran a dynamic implicit analysis in Abaqus with the following parameters on *Dynamic:

*Dynamic,alpha=0, application=TRANSIENT FIDELITY, direct

I also ran an analysis with Abaqus’s default parameters for transient fidelity:

*Dynamic, application=TRANSIENT FIDELITY, direct

Strangely, the second analysis, which includes a small amount of numerical damping (alpha = -0.05) resulted in nearly identical results to the results from the other program, whereas the first analysis which I thought was an apples to apples comparison resulted in a difference in magnitude of about 20% for the variable of interest, and a phase difference as well.

This made me wonder. When you specify alpha, does the time stepping algorithm change from Hilber-Hughes-Taylor? Abaqus offers HHT-TF (alpha = -0.05) and HHT-MD (-0.333). What algorithm is used if you want an alpha that is not either -0.05 or -0.333?

 

[FEA way]

The value of alpha = -0.05 is just the default one for HHT-TF time integrator (which is default for transient fidelity application). It can be changed freely (in the range of 0 to -0.5) and these changes don't override the type of time integrator (HHT-TF is still used unless you modify it manually).

So for your first keyword (*Dynamic,alpha=0, application=TRANSIENT FIDELITY, direct) Abaqus uses HHT-TF with alpha=0 and for the second one (*Dynamic, application=TRANSIENT FIDELITY, direct) Abaqus uses HHT-TF again too but with default alpha = -0.05.

The difference in results may be caused by other parameters and algorithms used by default in Abaqus. These are explained in the "Implicit dynamic analysis using direct integration" chapter of software documentation.

 

[jball1]

Thanks for the reply. So would I be correct in saying that there is no difference between HHT-TF and HHT-MD, except for a difference in alpha (and beta and gamma, but those are calculated automatically from alpha)? So there would be no difference in the below two keywords?

*Dynamic, alpha=-0.3333, application=Transient Fidelity, direct

*Dynamic, application=Moderate Dissipation, direct, IMPACT = Average Time, Incrementation = Conservative

It is very strange to me that the addition of numerical damping in Abaqus gave me results nearly identical to the results from another program which specifies no numerical damping (and the results do not match when I have no numerical damping in Abaqus). I guess this will go down as a mystery for me.

 

[FEA way]

Indeed the only difference between HHT-TF and HHT-MD is in alpha, beta and gamma parameters. However, according to the documentation, automatic adjustment of beta and gamma for specified alpha refers to HHT-TF only. Also for HHT-MD the default value of alpha is -0.41421.

Thus the difference between these two keywords is that the first one will use alpha = -0.3333, beta = 0.4444222225 and gamma = 0.8333 whereas the second keyword will result in alpha = -0.41421, beta = 0.5 and gamma = 0.91421. These should be the only differences though.

 

[jball1]

Got it, thanks for the help!

 

[IceBreakerSours]

Sharing one somewhat of an abstract note related to terminology that might seem minor/boring/.. but turns out to be very important: Validation is used in the context of comparing numerical model prediction against reality (physical experiment). On the other hand, verification is what you do when you are comparing the numerical results against analytical (mathematical) results. So, strictly speaking, you are not doing either. Instead, as you allude to it, you are comparing two numerical predictions and you will never know which one is "right"/correct/accurate/.. - unless of course you have independent verification/validation for one of those numerical predictions.

If this stuff sounds intriguing, ASME has several verification and validation subcommittees that might be worth looking into.

*********************************************************
Are you new to this forum? If so, please read these FAQs:

http://www.eng-tips.com/faqs.cfm?fid=376

http://www.eng-tips.com/faqs.cfm?fid=1083