About loads for testing a new vehicle

[pietro82]

Hi all,

It's well know loads on a vehicle are strongly dependent by vehicle parameters itself and they can be different markedly from a vehicle and a similar one (as example from the old mercedes c-class and the new one). It's also well known that measuring the loads on a prototype is time consuming. What is it commonly carried out from vehicle companies? Are testing loads measured anytime a new prototype is built in some conditions?, Are they computed from multibody simulations? Are they measured only once every a couple of new vehicle and correct somehow as example using the GCW?

thanks

 

[GregLocock]

Are testing loads measured anytime a new prototype is built in some conditions?
No, because the time pressure on prototpyes is high and collecting road loads is a time consuming and expensive task. one approach is to measure the road loads after the design is complete and check that the road loads that were used to design the vehicle were conservative.

Are they computed from multibody simulations?
This is the Holy Grail for durability. Unfortunately, given a known road surface profile and a typical vehicle MBD model it is not possible to predict the loads to sufficient accuracy. The problem is worst on the sort of events that comprise the greatest damage dose. We are getting there, for example I have included a cherry picked comparison between the real vibration spectrum measured on a car and the results from running an ADAMS model over the same measured road surface. If they were all of this standard then I might have some confidence in the whole procedure. Note the very restricted frequency range (I used a 1-6 hz bandpass filter in post processing), this has to be vastly increased to be useful for fatigue life predictions. I'd also add that looking at spectra is handy for corrletaion but of course is not directly related to fatigue calcs, if you were to try and predict fatigue you need a different approach entirely.

Are they measured only once every a couple of new vehicle and correct somehow as example using the GCW? Yes, that is a typical approach. It woud be wise to use an additional scaling factor to account for the uncertainity.









Cheers

Greg Locock


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

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

 

[pietro82]

Hi Greg,

Thanks for your reply.

No, because the time pressure on prototpyes is high and collecting road loads is a time consuming and expensive task. one approach is to measure the road loads after the design is complete and check that the road loads that were used to design the vehicle were conservative.

Is it done for any new vehicle or once every few releases of the new vehicle? What's happen if the design loads are not as conservative as requested?

 

[GregLocock]

Given the rate at which expectations for the NCAP crash tests are changing I'd say the structure is getting a significant revamp every 4 years or so, typically, which would probably lead to some road load acquisition and analysis.

If the measured road loads were found to be greater than those that were used in the design then a late design change could be made. Fixes might include: additional reinforcement, retune of suspension components such as shocks and jounce bumpers, or modifications to the allowable GVW of the vehicle. This breaks program discipline to some extent, as with all late changes extra oversight should be applied.

This all sounds a bit uncontrolled, but managing a few late changes is vastly preferable to the old way of doing it, where everything was changing right up to 6 months before Job 1.


Cheers

Greg Locock


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

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

 

[pietro82]

Thanks Greg...

 

[pietro82]

Hi all,

Here again on this topic. I have just read this SAE paper (

http://papers.sae.org/2003-01-3436/),

that is about the flow procces of developing new transmission. In it, there is the following sentence: "Component load-life relationships, i.e. component SN curves, must be developed for each component and failure mode.". For transmission there are many failures mode, since each one should be tested with two amplitude levels, that process is timeconsuming. Moreover for designing new transmissions the SN curve must be known, but it's not because the transmission is not tested. So how manunfactures usually deal with this issue?

Thanks

Cheers

Pietro

 

[CoryPad]

For powertrain applications, this is dealt with by time-consuming testing.

 

[GregLocock]

Automated rig tests are a couple of orders of magnitude less expensive than full vehicle testing, they can run 24/7, and in the csse of transmissions the failure modes are known well, so they can offset higher loads vs number of cycles quite accurately.

Cheers

Greg Locock


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

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

 

[BrianPetersen]

Seldom is a "new" transmission not designed based on known principles and components, even if it is a differing size or layout.

You know what the gears will take. You know what the clutch-packs will take. You know what the hydraulic actuators will take. You know what the bearings will take. Put it all together. Then test to verify that you didn't goof the design and to hopefully catch what was missed in design.

 

[pietro82]

Thanks to all, so in case of powertrain, how many test are usually carried out at the 4 square bogey? How many samples for each one?

 

[Comcokid]

Want to really test a vehicle? Ship it to a third-world country for severals months then examine it.

25 years ago one of the major school bus companies got a HUGE order for buses in the middle east. However, the way vehicles are used in other countries differs. The buses they shipped over broke down within weeks. Seems a vehicle built for 35 in here needs to carry 3x that number plus a few goats over roads that wouldn't be rated for a jeep in the US. In the end, this huge order put the company out of business and it was sold.

 

[GregLocock]

We call them Local Market Requirements. For example some American 4wd come with tires that are unsuitable for Australian usage, so after destroying 3 tires in 100 km on a dirt track they get towed to an outback tire shop, and 4 days later a set of Coopers or Toyos are flown in, and several thousand dollars later the happy owner sets off again on his trip he'll never forget.

Cheers

Greg Locock


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

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

 

[pietro82]

Hi Greg,

how do manunfacturers usually comply with this Local Market Requirements? Do they change the warranty specifications for each market?

Can also 4 post rigs run 24hours per day as transmission rigs?

 

[GregLocock]

LMR is internal terminology, for tests that have been found to be necessary by experience. They are no loegal requirements (for example the tires I mentioned were legal, but not satisfactory).

I hope you aren't trying to use a 4 poster for general durability testing.



Cheers

Greg Locock


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

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

 

[pietro82]

Ok thanks.

Why wouldn't you use a 4 poster for general durability testing?

 

[GregLocock]

Vertical loads are not enough to test the vehicle. Our durability rig uses 3 or 4 or more actuators at each wheelhub. A tire bouncing up and down on a platform is not teh same as one rolling along the road, never mind potholes.

Cheers

Greg Locock


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

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

 

[BrianE22]

Four posters are used more often for body testing. Spindle coupled is the better choice for suspension components testing.

 

[tbuelna]

pietro82- going back to one of your earlier posts regarding design and analysis of transmission components and S-N curves, the S-N curves represent statistical data for specific composition/quality/form/condition of metallic materials. Since this S-N data is compiled from testing a statistically relevant number of representative samples for the huge number of load cycles shown in the data, there is not a lot of S-N data available. If a transmission OEM decides to spend the large amount of money required to develop S-N data on metals they use in their products, they will likely keep the information confidential.

The cycles and loads used for a transmission component fatigue life analysis are based on a composite of several operating conditions, and there are many different analysis cases with different conditions performed for each component in the system. The components are designed by analysis, and testing is used to validate the analysis work. Since calculated fatigue life of transmission components is a statistical number, it is usually not practical, in terms of cost or schedule, to test a statistically valid number of transmissions for their design life.

 

[pietro82]

Greg, BrianE22,

thanks you're providing me very useful information. In our case, we want to test a not suspended vehicle with a spindle coupled rig, but at our company we call it improperly 4 post rig.

Tbuelna,

thanks for your reply. So how many transmission are usually tested? Only one?

 

[GregLocock]

I don't know what his answer will be but my guess, is no, they'll test at least 3 and then do a weibull analysis.

Cheers

Greg Locock


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

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