Output of Modal Analysis

Dear All,
I face a problem where the output at modal analysis is not significant. so i try modeling it with Autopipe and Caepipe and the result from those two program is what i expected, and the result from both is almost the same.
So why only CAESAR that gives different result?
Example: with same model the output of natural frequency is almost the same for the three of them:

CAESAR II gives 14.78 Hz
Caepipe gives 12.54 Hz
Autopipe gives 12.32 Hz

But the mode shape is different from CAESAR, example i take only in x direction :

CAESAR II gives -0.00145 (in mode mass normalize)
Caepipe gives -10.865 mm
Autopipe gives -9.247 mm

I try to read the user guide and technical reference but it seem i can't found the answer of it.
please enlighten me.....

There is only one explanation - the models are different.

Send your "c 2" file in to techsupport@coade.com so we can see what you're talking about.

Dear Richard,
The model is the same. the way i modellingit in CAESAR that the way i modelling it in Caepipe and Autopipe. I think the different should not be so far just like between the result from Caepipe compare to the result from Autopipe. But the result in CAESAR is to much different.

My question if, i set the unit in SI (mm), does the output in Mode Mass Normalized is also in mm ? or it is using a scale factor and need to convert it again to mm maybe ?

The units don't matter, that will be taken care of.

I want to see your model:
- Do you have rigid elements? If so, c2 determines the stiffness differently from the other two programs you sited.
- Do you have non-linear boundary conditions? If so, maybe you linearized things differently in the c2 dynamics differently.


As I said before, you should get the same result, and if you don't, the models are different.

Check the "Unity Normalized" report, not the "Mass Normalized" report. There are no absolute magnitudes units (or units) to these mode shapes. They are relative positions of the pipe under no load.
We are confident that C2 produces proper results. Like Rich says, send your data if you can't find what input to fix.

Dear Richard,
I think you are right is about the rigid element. Yes, i have Rigid element in this model. but can the output be so different like that, it is almost 100 % different in mode shape result, but strange the other result is 85% the same. I try to attach my model, but it my office is close due to Christmas. If i get the model, i will send it to you. but meanwhile can you suggest me anything to do with the input?

Different programs handle rigid elements differently. In c2, the stiffness of a rigid element is determined by holding the ID constant and increasing the thickness by a factor of 10. The result is that an 8" rigid is much stiffer than a 4" rigid, but much less stiff than a 12" rigid.

Other programs simply multiply E by 10 or 100, so all rigids have pretty much the same stiffness.

While we feel our approach is closer to reality (and therefore more accurate), it really shouldn't make a difference unless the system is significantly sensitive to the rigid element.

Test this out yourself - remove the rigid and see if the three programs agree. Build the model up from there.

Dear Richard,
I have do like you say, i put away the rigid element in three of those program, but it still have the same result where the two program (Caepipe & Autopipe) give result that close each other. meanwhile CAESAR II in static case give the result that almost the same like the two other program, but in dynamic (modal) the frekueny is different, and the important thing that CAESAr always give the Mode mass normalized and Mode Unity Normalized below one (1) millimeter, but the two other program gives result of modal analysis in displacement of mode shape above five (5) millimeter, just like the first that i told in my post.

Send the model.

Dear Dave, what did u mean with Relative Position under no load. So where i should look to mode shape displacement just like Caepipe and Autopipe gives.

I do not know at what you are looking in the CAEPIPE or AUTOPIPE output.
A mode shape is just that, a shape that the system may take under no load. You do this by solving Ma+Kx=0. Given that response is harmonic, then let x=A*sin(wt). Acceleration is second derivative with respect to time of x, or a=-(w^2)A*sin(wt), or a=-(w^2)x.
Ma+Kx=M*[-(w^2)x]+Kx=0
[K-(w^2)M]x=0
So either [K-(w^2)M]=0 or x=0. The interesting solution is [K-(w^2)M]=0. The (w^2)'s are the eigenvalues that satisfy this relationship. The natural frequencies are those w's divided by 2*pi. For each of these eigenvalues there is a matching distorted shape of the system that can be achieved with no load. These mode shapes have no defined magnitude, only a relative position with respect to all other nodes, as no load was needed to set these positions. When you apply transient loads, each mode will influence the total transient response of the system - that is modal analysis.

What i need to see is the displacement that the Mode shape describe....in the other 2 programs, you will see a mode shape and its absolute magnitude displacement....
Thanks.

The displacement that you see at Mode shape are normalized values.
Without a load you dont have a real displacement result.
Only the relation between nodes displacement is true and dont change under load.
Mode shape show you the shape from the pipe for you visuell dynamic feeling.

... and in c2 you can see that "normalized" mode shape in two ways:

a) you can look at the report "Modes Unity Normalized", or
b) you can look at the graphic "Animation"

It mean that i will not get the absolut value of displacement in modal analysis (only the pictures of the system shape in each of its natural frequency) until i put a load or continue to harmonic analysis or other dynamic analysis (depend on the case). Ok then, but Once it is mentioned in coade seminar notes that this mode shape is ratios of the displacements at various degree of freedom in each mode, is it this what unity normalized represent of?

Please read my earlier response in this thread: #32142 - 22/12/09 08:56 AM.

Mode SHAPE is a system charateristic, no load is required.

Calculated pipe POSITION is (dynamic) response to an applied (transient) load.

Dave - How do I navigate to response #32142. I am sorry I have not visited the forum in a while to remember all the navigation tools.

We're now on page 2 in this thread. Note the indicator in the top right corner "Page 2 of 2" Response #32142 is on Page 1. Just click on the 1 in that indictor to go to page 1.

Dear Dave and Richard,

With your above discussion on modal analysis, please solve the following queries,

I am doing a vibration check for a slug flow line using static equivalent approach (DLF=2) and modal analysis. I applied the slug force on every elbow of the piping after the source of slug force (Say after PCV). My vibration frequency limit is 5Hz and above. For static case, all is fine. When i check for modal analysis, the starting frequecny is less than 1Hz for friction stiffness is 0 and 2Hz for friction stiffness is 1000 (as per Mr. Dave instruction), in both the cases i could not reach the frequency atleast 4Hz. In my model there are many places where the rigidty is less, to increase the rigidity i have to increase the number of supports and restraints.

My query are

1.In the first modal analysis, my starting frequency is 2 Hz (friction stiffness = 1000), so to increase the frequency to 5 Hz, can i able to find a point in the Dynamic output result where the frequency of the paticular node is less, so that i can increase the rigidity of that paticular portion of piping. (To be clear in my query, there are many portion of piping system where the rigidity is less due to non-availability of support location, so using the dynamic analysis - Frequency result, can i able to point out, in which point or part of the piping portion the natural frequency is less).

2. In my case, the strum check failure occurs, can i ignore it or not? If not how to rectify it?

3. Under the control parameters i selected the load case for non-linear restraint as Max. Operating condition with slug forces. Is the frequency shown in the dynamic output consider those slug forces or not?

Please clarify.....

1. The mode shape will illustrate locations where there is significant deflection. These are good locations for added restraint.

2. Increase the frequency cutoff. You may also address this by increasing precision on the Advanced tab.

3. Your modal results do not show response to the slug. But the restraint configuration considers the slug loads you ran in your static analysis. I would run the modal analysis based on support configuration both with and without the slug just to see if things change much.

Dear Dave:

Thanks for your reply.

Regarding my 3rd query and your reply of it, i applied the forces on all elbow of piping system after the source of slug flow (i.e.) downstream of PCV in my case. The way i defined the slug forces are such that the forces are not cancelling each other case (i.e. i used the available 9 forces defining provisions in the input spread sheet on each elbow so that one force shall not cancel other force) and forces that cancel each other.

When i started the modal analysis, i checked the 1st mode natural frequency (5.2 Hz) of piping system with the load case of slug forces on all elbows at a time. When i changed the load case to only one force at a time, the first natural freq. is 3.6 Hz. So i conclude the slug force has an effect on the natural frequency of the system to a great extent and while doing modal analysis for checking the frequency it is best to check on all load cases the 1st natural frequency is more than the required one.

Finally, in modal analysis, there is input sheet for lumped mass analysis, under what case, we must use the lumped masses. Please clarify....

Thankyou...

Dear Sillyman,
I somewhat disagree with your statement "So i conclude the slug force has an effect on the natural frequency of the system...".
1. Slug force will not affect the natural frequency of the piping system. The support configuration will.
2. Check out and compare the nonlinear supports in static analysis (gaps, liftoff) in both the cases. Dynamic analysis will not consider non linear effects, ie. In dynamic analysis the inactive support will be removed (support that is lifted or gap that is not closed).

Use lumped mass in case:
1. A mass that you have not included in static case.
2. A concentrated load that will have major impact on the dynamic output.
3. Refer Caesar technical reference manual for detail.

Regards.

Dear Singh,

I apologise for the word i used as "load case" in modal analysis. Actually i mean for static load case for non-linear restraint status. Till date there is no accurate solution to calculate the dynamic effect of non-linear restraint status, so we change it to linear and we select one of the static load case to carry on.

For a complex piping system with a chance of more non-linear restraints. In selecting the static load case for it, when i select the sustained case the 1st natural freq is say 4.5 Hz, when i selct the operating case without slug force the 1st natural freq is say 5.3 Hz, with slug force the freq is 3.1 Hz. So for non-linear conditions the natural freq is varying to great extent. My client requirement for 1st natural freq is more than 5Hz.

Any how we managed to keep the natural freq more than 5Hz in all static load cases by adding additonal restraints.

Regarding lumped mass, as per your reply, it is not generally recommended to exclude the concentrated mass in the static case.

Can you please explain with an expample for lumped mass usage.

Thank you.

Dear sillyman,
Sometimes omitting couple of flanges (eg. flanges on rack lines designed on modular concept) will not alter the flexibility analysis to a great extent. However, it may not be same in dynamic analysis.

While modeling a pipe to pipe support in caesar, if you like to include weight of supported pipe into the supporting pipe (that you have excluded in static analysis) go for lumped mass.

Regards,