Rigid flexible coupling simulation of vehicle

[Nuaaer]

Hi, guys!

I am studying the influence of the bending and torsional stiffness of BIW on the handling. The process is as follows:

(1) Obtain the mnf file of the BIW by HyperMesh (CMS);

At this stage, to simplify the model, I removed the seats, doors and interior. In order not to change the mass center and moment of inertia of the whole vehicle, I used concentrated Conm2 at the mass center of each deleted part, and use rbe3 to connect them with the surrounding elements.

（2）Buliding the rigid multibody dynamic model in Adams/car;

(3) Replace the Chassis Part (rigid) with the mnf file of the BIW to obtain the rigid-flexible coupling multibody dynamic model;

I've verified that it was successful to simulate for the handling. The output curves of the rigid-flexible coupling model were very close to that of the rigid body model. However, there are huge errors when I use the same model to simulate for the ride. I guess that the method of obtaining the mnf file is problematic.

The following is the output result of a ride comfort simulation test under the condition of deceleration belt:

​

​

Figure1. seat rail bolt x acceleration

​

​

Figure2. seat rail bolt z acceleration

Note: Red-- the rigid-flexible coupling multibody dynamic model;

Blue--the rigid multibody dynamic model;

pink--the test data;

The measuring point is located at the bolt hole connecting the seat and the guide rail. For the ride simulation, the output curves is good in x direction (the direction of the vehicle). But in z direction being vertical to the ground, when the front wheel passes through the deceleration belt, the longitudinal acceleration of the rigid-flexible coupling multi-body dynamic model is too large. And the trend also seems to be wrong.

Please help me! It is appreciated for any reply!

 

[GregLocock]

2 mm/s/s is ridiculous.

It looks from the X result that you have a different speed to the original test, judging by the delay in the peaks at 1.6 seconds. What tire model are you using? What is the road profile?

It is very hard to understand how the Z result can be so bad when the X is so good.

Have you run a K&C or SPMM or whatever to confirm that your flex model is any good? Ride predictions are very difficult to get right, as you need a good tire model (Ftire for example) if there is any significant amplitude. You might well need frequency dependent bushing models, and a better shock absorber model than just F versus v. In ADAMS you should always start with quasi static rig tests before dynamic tests.



Cheers

Greg Locock


New here? Try reading these, they might help FAQ731-376

http://eng-tips.com/market.cfm?

 

[Nuaaer]

Thanks for your reply!
Please ignore the unit (mm/sec**2). Here the unit of the acceleration is g. As you said, the difference of longitudinal acceleration in time is due to different speed. Therefore, from the peak value, the output curve of the longitudinal acceleration is successful. However, the output curve of vertical acceleration is bad.
Here is the road profile describing passing through speed bumps.

$---------------------------------------------------------------------MDI_HEADER
[MDI_HEADER]
FILE_TYPE = 'rdf'
FILE_VERSION = 5.00
FILE_FORMAT = 'ASCII'
(COMMENTS)
{comment_string)
'Big bump for testing durability tire.'
$--------------------------------------------------------------------------units
[UNITS]
LENGTH = 'mm'
FORCE = 'newton'
ANGLE = 'radians'
MASS = 'kg'
TIME = 'sec'
$---------------------------------------------------------------------definition
[MODEL]
METHOD = '3D'
$-------------------------------------------------------------------------offset
[OFFSET]
X = 0.0
Y = 0.0
Z = 0.0
PSI = 0.0
THETA = 0.0
PHI = 0.0
$--------------------------------------------------------------------------nodes
[NODES]
NUMBER_OF_NODES = 18
{ node x_value y_value z_value }
1 5000 2000 0
2 5000 -2000 0
3 -10000 2000 0
4 -10000 -2000 0
5 -10000 2000 9
6 -10000 -2000 9
7 -10100 2000 40
8 -10100 -2000 40
9 -10172 2000 50
10 -10172 -2000 50
11 -10244 2000 40
12 -10244 -2000 40
13 -10344 2000 9
14 -10344 -2000 9
15 -10344 2000 0
16 -10344 -2000 0
17 -50000 2000 0
18 -50000 -2000 0
$-----------------------------------------------------------------------elements
[ELEMENTS]
NUMBER_OF_ELEMENTS = 16
{ node_1 node_2 node_3 mu }
1 2 3 0.9
3 2 4 0.9
3 4 5 0.9
5 4 6 0.9
5 6 7 0.9
7 6 8 0.9
7 8 9 0.9
9 8 10 0.9
9 10 11 0.9
11 10 12 0.9
11 12 13 0.9
13 12 14 0.9
13 14 15 0.9
15 14 16 0.9
15 16 17 0.9
17 16 18 0.9


The tire model is "FTire_T18_23555R18_220kpa.tir". I have not run a K&C or SPMM or whatever to confirm that my flex model is good. So how to confirm the flex model is good? Please tell me more about the verification steps.

 

[GregLocock]

You take your real car, put it on a K&C rig, get a whole bunch of results.

https://www.morsemeasurements.com/what-is-kc-testing/

.Then you run the same test in adams, rigid, and look at the differences, and adjust your model. Then you put the flex frame in and hopefully you can take some of those adjustments back out.

Cheers

Greg Locock


New here? Try reading these, they might help FAQ731-376

http://eng-tips.com/market.cfm?

 

[GregLocock]

Incidentally, you should use the same speed as the real test and adjust the road profile file until you get matching peaks.

Cheers

Greg Locock


New here? Try reading these, they might help FAQ731-376

http://eng-tips.com/market.cfm?

 

[Nuaaer]

Hello GregLocock!
Thanks for your patience!
My idea of verifying the rigid-flexible coupling model (including flex biw) is as follows:
(1)use the flex body to replace the rigid chassis part, set up the speed and road profile, and output the curves of the accelerations(The solid red line in the figure above represents the curve.).
(2)In ADAMS/car, there is an option to treat a flexible body as a rigid body. When this option is selected, the flexible biw is the same as the rigid body. I set up the same speed and road profile, and output the curves of the accelerations(The blue dash line in the figure above represents the curve.).
In this way, the influence of other variables is excluded, and only the factor of BIW flexibility is considered. As can be seen from Figure 1, the longitudinal acceleration curves of the two are very close. However, the vertical acceleration curves of the two are bad.
In addition, I import the test values under the same road properties(there are some differences in speed.), which the pink dash line represents. If ignoring the delay in time, the blue curve is close with the pink curve in the longitudinal and vertical direction.
In conclusion, the rigid model (treating the flexible body as a rigid body) is ok. There may be some problems in the MNF file of flexible body. But I don't have any idea to look it up. Hope you can help me solve it.
Regards!
Zaily

 

[GregLocock]

I understand what you are doing, but long experience with ADAMS (19 years) tells me that you need to start with statics before dynamics.

Cheers

Greg Locock


New here? Try reading these, they might help FAQ731-376

http://eng-tips.com/market.cfm?