General Mechanical

General Mechanical

Issue Capturing 90% Modal Mass in 10,000 Modes for Spectrum PSD Analysis

    • mjoyce
      Subscriber
      For my research I am performing a random vibration analysis using the Spectrum PSD analysis. I am performing a Modal analysis and the first step is to make sure I include enough modes to capture at least 90% of the structure's mass in my modes in the x, y and z directions. The issue is that it takes over 10,000 modes to capture 90% of the mass for my model and the input for the Spectrum PSD analysis has a cap at 10,000 modes. I have seen commands where you can expand only the modes that contribute more to the mass, but I can not find a command to only use those modes in the Spectrum analysis so that I capture 90% of the mass with at most 10,000 modes. Below is part of my code of what I have so far where I execute the Modal and Spectrum analyses.nn/SOLnantype,modalnmodopt,lanb,60000nsolve nfinishnn/SOLnantype,modal,restartnmxpand,,,,yes,,,EFFMnmodseloption,0.9,0.9,0.9,no,no,nonsolvenfinishnn/SOLnantype,spectrnSPOPT,PSDn
    • BenjaminStarling
      Subscriber
      nSome structures do not require that the total modal mass captured be 100% or close to it. For example, a steel structure sitting atop a very stiff concrete foundation. The concrete foundation is present in the analysis to accurately capture the boundary conditions of the structure. If we are only interested in analysing the structure between 0-30 Hz, and the concrete foundation has no mode shapes before 1000Hz, there is no need to ensure we capture this mode shape, it will not be excited by such low frequencies. But because concrete foundations generally have a mass comparable to the overal steel structure, this will result in a considerable difference in the modal mass captured.nBeyond this, your code currently will not work. Try the following.n/SOLnmodopt,lanb,60000nmxpand,,,,,0.001,YES,,EFFM,nmodseloption,yes,yes,yes,yes,yes,yesnRestarting the modal analysis to select expansion options makes no sense in this context. The code above will only expand modes if they are greater than 0.1% of the total mass, therefore you should only end up with 1000 modes maximum. You may find that you will end up with extraordinarliy high frequencies which may be unnecessary for your spectrum analysis. You may also find that you still don't acheive 100% modal effective mass. nAnother issue not discussed, you may have some loose mesh somewhere, producing many modes with a very small mass participation.n
    • mjoyce
      Subscriber
      Hi BenjamiThank you for your response.nnFor the edited code you provided, in what location should the YES be in? It seems that a few things are in the wrong place for mxpand command so I just wanted to double check.nnSecond, won't this just expand some number of modes and the spectrum analysis will still try to use the 60,000 extracted modes? The issue that I am running into is that the spectrum analysis takes in the number of extracted modes, not expanded modes, so in this case 60,000. I understand that you can choose which modes to expand but it does not seem like I can chose which extracted modes will be used for the spectrum analysis.n
    • BenjaminStarling
      Subscriber
      nThe yes is currently in the MSUPkey position. this is the position that is required for future analysis that requires modal superposition. You can use Yes in the position you previously quoted, Elcalc, and this location, MSUPkey, defaults to yes. You will need to keep Elcalc set to yes, which I missed in my command.nIn regards to how this behaves when a subsequent analysis is run, I am not sure/familiar. An MSUP analysis uses the expansion results that are in the .mode file. This would imply that by limiting the modes that are expanded, that the MSUP analysis would use only these modes. I have never done this myself as I have never had a need to limit modes in such a way. I am not sure if a response spectrum analysis behaves the same way. My instinct is yes, that it will only use the modes written to the .mode file, which is controlled by the mxpand command, but I am not that familiar with response spectrum analysis. A quick test would be to run a simplified model, see if the mxpand command has an effect, and whether the spectrum response analysis uses the .mode file (try deleting the .mode file and see if it errors out).nIf this does not work, you will need to revisit the importance of acheiving high effective modal mass percentage.nn
Viewing 3 reply threads
  • You must be logged in to reply to this topic.