I am new to Pro Mechanica and the Simulation Creo modules.
I really appreciate if someone could walk me through the basic steps to apply the following simulation to analyze a basic enclosure.
How to apply three shocks of 15 G force, 75ms half-sine shock pulses to an enclosure?
Solved! Go to Solution.
Kevin
To help you to begin to unpick the suggested technique, Consider only the 'dynamic time' part of the suggested technique for now.
There are 4 key work flow steps:
1. Carry out a modal analysis (guess the number of modes, start with 10,say)
2. Using the results from the modal analysis output (1. above), 'identify' the structure using a 'dynamic shock' analysis. We must ask for the mass participation factors [MPF's] as an output in the analysis definition and they appear in the .rpt file (I think). The MPF's must sum to 80% or more.
3. If the MPF's do not sum to >=80% repeat. goto 1. requesting more modes. repeat 2. to confirm we are now >=80% total MPF. Repeat as required. (care, brief comment on chasing irrelevant information below)
4. When sum of MPF's >=80%, the structure sufficiently 'identified', carry out the dynamic time study using the half sine F(t). Don't forget to ask for measures. 'Raw' output from a dynamic time analysis is minimal and you will be easily disappointed.
Practical pointer. When running shock/time studies the user has the choice of using results from a previously run modal study. It is clear that by doing this time can be saved as the steps 2 and 4 both require the results from the same modal analysis as an input. Running the modal study 'on the fly' means it is run twice taking longer. BUT if you want to reuse the modal study results to save time you must not change your model at all; not even by a datum point. If you do you will get a message saying something along the lines of 'the model has changed, can't use the modal study results'
Hope this helps
Regards
Charles
Aside:
'Pure shock' is an impulse; (Dirac function, http://web.mit.edu/2.14/www/Handouts/Convolution.pdf) a zero time length infinite amplitude event and exites all frequencies equally and is mathematically useful for 'system identification'. A structure has natural frequencies and it is only these that repsond to this shock. We get the impulse response 'PSD' graph as an output.
The half sine function you have is a short time based event, which we engineers tend to refer to as shock. This is why I proposed you consider dynamic time first.
More useful links ...
Alternative words describing use of mechnica for dynamic analysis here, just another presentation of the process ...
Useful for understanding modal mass participation:
http://www.vibrationdata.com/tutorials/ModalMass.pdf
Ideally we want modal mpf >=80% in relevant directions. By relevant I mean lateral modes may not be of interest and therefore we need to be careful that we don't chase a 'global' MPF >=80% as a higher mode may be in an irrelevant direction (so mode shape is importannt). How the model is constrained influences this.
To go the frequency domain route by Industry standard spectra 'from a book', this is the sort of thing I was thinking of ...
http://shop.bsigroup.com/ProductDetail/?pid=000000000000218537
Some more learning resources
NAFEMS also have a code independent web-based dynamic analysis course that lends substantial (though more general) background.
http://www.nafems.org/e-learning/all_courses/el_dynamic/
Lots of useful backround here too
http://www.bksv.co.uk/Library/Search%20Resources
Why >=80% mpf? One could state authoritatively "It's an industry standard" ... Start here
Hi Kevin
This attachment is from some releases back, but should get you through the first steps.
There's much more nice that you can do, but first steps count.
Regards
Erik
Hi,
I think this may be a dynamic time analysis.
Perhaps use the function definition (adjusting frequency and amplitude to suit):
This is from suggested technique;
http://support.ptc.com/cs/cs_26/howto/mst2477/mst2477.htm
Regards
Hi Charles,
Any thoughts on how to generate approximately 20 data points for graphing a half-sine function with 75ms duration and 15g amplitude. I like the data points in the foam of frequency vs g.
Regards,
Kevin
Kevin,
The method above is time domain, if I understand your question you want the same info but in the frequency domain. It would be g^2/Hz on the y-axis.
2 methods.
First you take your time signal and transform it using an FFT, create the appropriate freq domain info. Many browser apps exist, mathccad, matlab, ms excel spreadsheets, write your own ...
Second, this half sine shock approach is a standard 'test' in many industries. Can you find the info you need from a related EN, BS, DIN ect standard?
Atb
Thak you Eric Haenen and Charles Simpson for links to the tutorial. I am going to give it a try this week.
Hello Mark,
Did you manage to solve your task with the help of Eric and Charles?
Is there some feedback you want to share with us?
Thanks,
Gunter
Hi Gunter,
I found the tutorial was difficult to follow. The work flow is not intuitive for a new user like myself. I am surprised that there is only one example available of half-sine shock approach as is a standard test across many industries as Charles Simpson stated above. Is there more examples of half-sine input shock analysis?
Regards,
Kevin
Hello, Kevin,
Let me join you some links on the shocks.
Not always connected to ProMechanica.
Sorry, some are in French.
Best regards.
Denis.
http://moi101.free.fr/m%C3%A9ca/Dynamique-th%C3%A9orie%20classique%20du%20choc.pdf
http://ingenierie.he-arc.ch/sites/default/files/ssc_0.pdf
http://www.sandv.com/downloads/0906alex.pdf
http://www.signalysis.com/pdf/srs.pdf
http://www.hermes-science.com/mori/chocs.zip
http://www.qucosa.de/fileadmin/data/qucosa/documents/6840/Jakel_Schockanalysen.pdf
Thank you Denis. I feel like being back in college now!
Hello Kevin,
there are in fact only few, mostly rather old suggested techniques available on the PTC Technical Support web site. The Central Page article contains links to all the relevant ones:
CS106224 : Pro/Mechanica and Creo Simulate Central Resource Page
https://support.ptc.com/appserver/cs/view/solution.jsp?n=CS106224
The suggested techniques suggested by Erik and Charles, can be found on this list, too.
There are also a lot of free tutorials created by Product Management on the Learning Connector, here:
https://support.ptc.com/go/ptcuniversity-lex
However, of the 33 Simulation tutorials there was none specific for dynamic shock analysis
Kevin
To help you to begin to unpick the suggested technique, Consider only the 'dynamic time' part of the suggested technique for now.
There are 4 key work flow steps:
1. Carry out a modal analysis (guess the number of modes, start with 10,say)
2. Using the results from the modal analysis output (1. above), 'identify' the structure using a 'dynamic shock' analysis. We must ask for the mass participation factors [MPF's] as an output in the analysis definition and they appear in the .rpt file (I think). The MPF's must sum to 80% or more.
3. If the MPF's do not sum to >=80% repeat. goto 1. requesting more modes. repeat 2. to confirm we are now >=80% total MPF. Repeat as required. (care, brief comment on chasing irrelevant information below)
4. When sum of MPF's >=80%, the structure sufficiently 'identified', carry out the dynamic time study using the half sine F(t). Don't forget to ask for measures. 'Raw' output from a dynamic time analysis is minimal and you will be easily disappointed.
Practical pointer. When running shock/time studies the user has the choice of using results from a previously run modal study. It is clear that by doing this time can be saved as the steps 2 and 4 both require the results from the same modal analysis as an input. Running the modal study 'on the fly' means it is run twice taking longer. BUT if you want to reuse the modal study results to save time you must not change your model at all; not even by a datum point. If you do you will get a message saying something along the lines of 'the model has changed, can't use the modal study results'
Hope this helps
Regards
Charles
Aside:
'Pure shock' is an impulse; (Dirac function, http://web.mit.edu/2.14/www/Handouts/Convolution.pdf) a zero time length infinite amplitude event and exites all frequencies equally and is mathematically useful for 'system identification'. A structure has natural frequencies and it is only these that repsond to this shock. We get the impulse response 'PSD' graph as an output.
The half sine function you have is a short time based event, which we engineers tend to refer to as shock. This is why I proposed you consider dynamic time first.
More useful links ...
Alternative words describing use of mechnica for dynamic analysis here, just another presentation of the process ...
Useful for understanding modal mass participation:
http://www.vibrationdata.com/tutorials/ModalMass.pdf
Ideally we want modal mpf >=80% in relevant directions. By relevant I mean lateral modes may not be of interest and therefore we need to be careful that we don't chase a 'global' MPF >=80% as a higher mode may be in an irrelevant direction (so mode shape is importannt). How the model is constrained influences this.
To go the frequency domain route by Industry standard spectra 'from a book', this is the sort of thing I was thinking of ...
http://shop.bsigroup.com/ProductDetail/?pid=000000000000218537
Some more learning resources
NAFEMS also have a code independent web-based dynamic analysis course that lends substantial (though more general) background.
http://www.nafems.org/e-learning/all_courses/el_dynamic/
Lots of useful backround here too
http://www.bksv.co.uk/Library/Search%20Resources
Why >=80% mpf? One could state authoritatively "It's an industry standard" ... Start here
Hello Kevin,
please consider coming back to the thread when you are done with your investigation and leave some feedback on which info from the posts was useful and how your investigation went in general. Maybe you found some additional information you want to share about the topic?
Also make use of the Correct Answer / Helpful Answer buttons, where it seems applicable.
This is helping others to get a quick overlook over a lengthy discussion like this one.
Thanks,
Gunter
Kevin,
You need to examine the mode shapes and decide which are relevant. The software can't make this decision for you. The '80% rule' is very flexible and is a cover all.
You also need to decide what useful bandwidth the exciting signal has. There's no point looking at modes at 1kHz if the if input signal rolls off (half power point) at 100Hz. Get this from the FT of the input half sine function. Try Theory of Vibrations , William Thomson.
An example of irrelevant missing mass : a large number of similar stiffening gussets say. Each one having a response at a higher frequency than anything that is relevant in the input signal. Each one ringing at a slightly different frequency. Because there are many, their cummulative contribution is not negligable. By examination, the engineer could ignore them with justification.
Another, consider a suspended walkway. It's has 1000 uprights holding on the hand rail. They will all ring in a frequency band that is way above the key damaging modes of a few Hz of the suspended platform where the exciting frequencies are a few Hz. (people walking, building movement due to wind loading) It is these key modes that are of interest not the individual ringing modes of each upright.
Every plug, bolt, gusset, bar will have their own mode and contributes modal mass and you can chase these to higher and higher frequencies where the power at these frequencies in the input signal is negligable. They can probably be dismissed for the modes of interest. Remember that their mass does play a part in lower modes so don't remove them without justification!
Ansys is great! Feedback from Ansys is more comprehensive. You get lots of buttons to click , you have many more decisions to make ... but s/w learning curve is steeper and longer, integration with CAD is nowhere near as good as Simulate and answers are, for usual purposes, the same. Both Simulate and Ansys use Lanczos methods to ensure modes are not missed.
The process one has to follow in Ansys is essentially the same as Simulate.
I thoroughly recommend the NAFEMS dynamics couse. The Q&A forum associated with the course can leave one a little frustrated but the content will answer all of your questions and more.
atb
Charles
I finally worked on the tutorial over the labor day weekend, with all your suggestions, support and related links. Thanks to Charles for laying out the general approach and all others: Erik, Atb, Dennis, Gunter……
1. 1. I started my modal analysis with 10 modes. After several increments from 10, 50, 100 and reached all the way up to 200 but my mass participation factor is still at only 30%. I am missing something here! Why not the software do the automatic increments of modes until it reach the requirement?
2. 2. Half-sine shock approach is a standard test throughout the industry, particularly in the defense. Promechanica may be improved with a half-sine function button with only input as the amplitude and the duration. Why manually enter the function and make an error, or trying to figure out what the function may be?
3. 3. Overall I found the work flow in ProMechanica was tedious, not intuitive. And lack more examples in shock and random vib.
4. I am curious how other software in the industry handles this type analysis, particularly ANSYS!