---

Hallo everyone,

I encounter a problem with my project.

Description on my project:

The object/structure is a car chassis with four legs at four corners which are fixed on the top of a steel desk. The desk simply stands on the ground without any fixation.

In ANSYS I try to make modal analysis of this chassis. Appropriate boundary conditions are expected.

At beginning, I only modeled the chassis without four supporting legs. Using DL, L, ALL,0 leads to higher resonance frequency than the experimental result. Then I attempts to add COMBIN14 to simulate the legs. The comeout was not satisfactory either.

Finally I added the four legs to the chassis in ANSYS, and constrant:
DL, bottom lines of the legs, ALL/UX,UY,UZ, 0
The result is still not good.

If anyone have any tips in this field, please inform me.

Thanks in advance,

JUN NI

Eine Antwort auf diesen Beitrag verfassen (mit Zitat/Zitat des Beitrags) IP

---

Hello Jun Ni,

question is - how far are the frequencies off test results (presuming verified, reliable test results) ?

That simulated frequency values are off a few % would be to be expected - depends on how well fixities in the model reflect those in the physical test setup, how complete stiffnesses/panels etc are modeled (compaired to structural members actually participating in prototyp setup), quality of discretization/mesh.

------------------
Gruesse,
Frank Exius
IFE Deutschland
Dienstleistung in ANSYS ife-ansys.de
Simulation Berechnung FEM Digital Prototyping

Identifikation bei Fragen an ANSYS Forum erwuenscht - Name, Firma/Bildungsstaette.
Identifikation im Beitrag, im persoenlichen Profil, oder in eMail an Moderator (IFE) - in letzterem Fall
bitte Thema Beitrag als Referenzpunkt angeben. Danke, Moderator ANSYS Forum.

Eine Antwort auf diesen Beitrag verfassen (mit Zitat/Zitat des Beitrags) IP

---

Hallo Frank Exius,

I have established couples of models with various boundary conditions. Different configuration leads to variant distance between the simulation outcome and experimental result including 10 resonance frequencies and corresponding mode shapes. Some of the simulations give good agreement over certain range of resonance frequencies, some of the simulations yield good mode shape. It's a pity to find one fitting the two criteria. I'd like to know whether you have experience with the similar experiment setup, and any suggestion on possible boundary conditions is welcome.

By the way, can you recommend me any software dedicated for experimental modal analysis?

Thanks for your further help.

Jun Ni

Eine Antwort auf diesen Beitrag verfassen (mit Zitat/Zitat des Beitrags) IP

---

Hello Jun Ni,

I do have experience in the sense that I've seen shake tables in operation (car industry), know about how a experimental modal analysis is done. Usually for multi-component products like car chassis, NVH evaluations are performed uttilizing FEA plus experimental methods. The FEA model must be tuned usually by way of experimental info, because BC's, damping incl. at interconnections etc aren't precisly known. Experimental approach needs points of measurement accessible to sensoric placement etc, benefits from refering with FEA results in turn. Together-they-hit-the-target approach.

For a software which is designed to tune-in-to-target between the two approaches, see http://www.download.ife-ansys.de/lms-fe+versuch-tune_nvh_model.pdf  and the site refered therein.

How much parts, how much shell elements, DOF's does your actual model consist of?

------------------
Gruesse,
Frank Exius
IFE Deutschland
Dienstleistung in ANSYS ife-ansys.de
Simulation Berechnung FEM Digital Prototyping

Identifikation bei Fragen an ANSYS Forum erwuenscht - Name, Firma/Bildungsstaette.
Identifikation im Beitrag, im persoenlichen Profil, oder in eMail an Moderator (IFE) - in letzterem Fall
bitte Thema Beitrag als Referenzpunkt angeben. Danke, Moderator ANSYS Forum.

Eine Antwort auf diesen Beitrag verfassen (mit Zitat/Zitat des Beitrags) IP

---

Thanks a lot for your tips and the activation links.

My purpose is to suppress the vibration generated at one corner of the chassis in the z axis, perpendicular to the structure. Without fixation, there is only one part of interest, the whole chassis structure. After meshing, which I did not pay much attention to, ANSYS yields around 4000 elements/nodes. I measured 20 evenly scattered points along the chassis driven by PRBS, with the help of two 1DOF accelerometers. I downloaded one free software for experimental modal analysis, which unfortunately seems to have some flaws. In order to garantee the resonance frequencies. I used chirp signal to excite the chassis and observed the struture to obtain the approx. resonance frequencies. Finally, I excited the chassis with one sine signal at one of resonance frquencies observed following the mentioned procedure, and then calculated the magnitude ratio and phase shift at all the concerned points on the chassis. However, my animation of mode shape at the specific frequency doesn't exactly match the courtpart from ANSYS. Frankly, the animation i programmed looks unreasonable at some points.
On the basis of my description, do you think my thread/procedure is correct? and how can I get better mode shape by means of experimental data.

Eine Antwort auf diesen Beitrag verfassen (mit Zitat/Zitat des Beitrags) IP

---

as I said a mixed approach using FEA and experimental evaluation may be used, to center in on relevant frequencies/shapes, of particular interest. A whole chassis meshed with 4000 Elements is rather coarse, esp. if higher modes are of interest. But even with a detailed mesh, dicrepancies are expected, between the 2x methods, because of the uncertainties mentioned.

There are several general questions to be checked before embarking on a approach: how much time available, how much does it cost, how much is it worth. If this is a project w/o tight time/budget constraints, you may follow the FEA/experimental tuning process. This might provide insight in a systematic manner, also in sight of  future investigations.

If a vibration problem, causing noise or drive comfort problem or the like, represents a urgent task with restrictions on time and other budget, I'd probably recommend the engineers approach (engineer in terms of the common English language interpretation, i.e. the machinist who fixes it): put some wisely placed lokal stiffening in - bead, corrugation, seam/fin were you detect unwanted vibrations, some resort to damping meadows, if the effect is more distributed.

------------------
Gruesse,
Frank Exius
IFE Deutschland
Dienstleistung in ANSYS ife-ansys.de
Simulation Berechnung FEM Digital Prototyping

Identifikation bei Fragen an ANSYS Forum erwuenscht - Name, Firma/Bildungsstaette.
Identifikation im Beitrag, im persoenlichen Profil, oder in eMail an Moderator (IFE) - in letzterem Fall
bitte Thema Beitrag als Referenzpunkt angeben. Danke, Moderator ANSYS Forum.

Eine Antwort auf diesen Beitrag verfassen (mit Zitat/Zitat des Beitrags) IP

---

Ich danke ihnen sehr für ihre Antwörter, und ich wünsche Ihnen eine Fröhliche Weihnachten und einen guten Rutsch ins neue Jahr 2005.

Jun Ni

Eine Antwort auf diesen Beitrag verfassen (mit Zitat/Zitat des Beitrags) IP

---

wuensche das Gleiche, inkl. Erfolg beim Projekt

------------------
Gruesse,
Frank Exius
IFE Deutschland
Dienstleistung in ANSYS ife-ansys.de
Simulation Berechnung FEM Digital Prototyping

Identifikation bei Fragen an ANSYS Forum erwuenscht - Name, Firma/Bildungsstaette.
Identifikation im Beitrag, im persoenlichen Profil, oder in eMail an Moderator (IFE) - in letzterem Fall
bitte Thema Beitrag als Referenzpunkt angeben. Danke, Moderator ANSYS Forum.

Eine Antwort auf diesen Beitrag verfassen (mit Zitat/Zitat des Beitrags) IP