3 ms·
I am worried that cars are designed to pass a few really specific tests well, for example 'head on at 40 mph', 'side impact at 30 mph', but then perform much wo
by londons_explore 4y ago
I am worried that cars are designed to pass a few really specific tests well, for example 'head on at 40 mph', 'side impact at 30 mph', but then perform much worse in a 'typical' crash, for example '37 degrees, offset by 6 inches, at 38 mph'.
I'd like to see the test methodology changed to correct this. Car manufacturers should be required to simulate what would happen in say 10,000 crash cases - with a wide variety of angles, loads, speeds, passenger types, offsets, etc. The car manufacturer should then present all those results to the NHTSA, who will then pick a few at random to test on a real car to validate the simulations.
The safety rating would be some average of performance of all the 10,000 simulated test cases, with a scaling penalty applied if the simulations didn't accurately reflect reality.
- bombcar 4y ago> I am worried that cars are designed to pass a few really specific tests well, for example 'head on at 40 mph', 'side impact at 30 mph', but then perform much worse in a 'typical' crash, for example '37 degrees, offset by 6 inches, at 38 mph This is literally the case: https://www.caranddriver.com/news/a40824939/iihs-side-impact-crash-test-sedans-wagons-results/ https://www.caranddriver.com/news/a40824939/iihs-side-impact... (and an earlier one that was added and IIRC only Volvo did well, because Volvo was testing it before it was required: https://www.caranddriver.com/news/a18737558/iihs-adds-new-frontal-crash-test-most-cars-expected-to-perform-poorly/ https://www.caranddriver.com/news/a18737558/iihs-adds-new-fr... ) I agree that digital simulation should become an important part - including aftermath simulation from actual wrecks.
- dont__panic 4y agoI wish companies had to test new cars in collisions against the smallest, most vulnerable vehicles on the road. Cars exist in a shared environment; if Ford creates a new, taller, heavier F-250, it shouldn't be allowed on the road unless there's a reasonable chance that it won't brutally crush anyone driving, say, a 2005 Honda Civic sedan around in the event of a collision.
- lamontcg 4y agoThis was my first thought about the sex differences in accidents which is that they might reflect the average vehicle weight being driven.
- Beltalowda 4y agoAlso: against pedestrians and cyclists. In fact, I'll argue that from a moral point of view it's more important to ensure the safety of people who choose to not operate a vehicle vs. those that did. Both are important of course, but this would be correct priorities. In reality the priorities are of course rather different.
- dont__panic 4y agoAgreed. The larger and larger vehicles of today are so much more dangerous for pedestrians, but they make me wonder: has any country ever prioritized pedestrian safety in car design? Feels like the sort of thing that manufacturers ought to do, but I'm not aware of it ever happening. Maybe there are regulators I don't know about who oversee this?
- necovek 4y agoEuropean tests specifically test for pedestrian collisions: this is why we've got softer, crumpling bonnets and bumpers in newer cars, for instance. I would be surprised if other test agencies like the US one did not do the same. Cyclists are definitely harder because they are similarly unprotected human bodies travelling at higher speeds and with more chances of different collisions (i.e. it's very rare for a pedestrian to hit a car on the side, whereas I am guessing that's more likely for a cyclist). In general, I would give equal priority to testing for everyone, iow, prioritise by the numbers getting injured/killed. Btw, cyclists also make a choice of operating a vehicle (motorcyclists are similarly vulnerable because the characters of their vehicles are similar).
- tfvlrue 4y agoAlong those same lines, I found this an interesting read: https://danluu.com/car-safety/ https://danluu.com/car-safety/
- hulitu 4y ago> Car manufacturers should be required to simulate what would happen in say 10,000 crash cases - with a wide variety of angles, loads, speeds, passenger types, offsets, etc And how will you compare the results ? (apples with pears, SUV 40km/h 2 tonnes with small car 20km/h at 37 degree 1 ton )
- londons_explore 4y agoWell every car would go through the same 10,000 simulated tests... And then they're comparable.