[SaiSW]

Voted for car with higher safety rating but just the mandatory safety features. But the truth is, the market has developed so fast that you can now get cars that have both, XUV 300 is an example.

But yeah, if I absolutely had to choose between the two options, then higher safety rating anyday.

[ashishk29]

Originally Posted by ron178

That is correct, but it is designed for the test and not to be "stronger overall".

We'll have to agree to disagree. These cars have proven in real world accidents that their "theoretical" safety ratings do lead to occupants coming out alive even after severe accidents. Some of these accidents involving rolling over,

other vehicles literally climbing over them

, and many other non-front-ODB-crash like scenarios. If it were "designed for the test" those cars would not have had much success in saving people in actual accidents which tend to vary more in severity and situation. OR, this could also imply that GNCAP testing does cover enough to correlate with real world safety. Otherwise there would not have been a correlation.

Originally Posted by ron178

A major limitation of this test is that energy dissipation takes place via the right front longitudinal (steel beam behind the bumper) which is engaged by the protruding 'bumper element' on the barrier as it would by a partner car in a car-to-car crash. By extension the test could help in other crashes where the longitudinal is engaged, that is why underride prevention devices for example are positioned so low to the ground. But crashes where the longitudinals are not engaged have to be accounted for separately. This happens in unprotected underrides and the IIHS' small overlap test. In underrides there is little a car can do. In the IIHS' small overlap test manufacturers use an upper frame rail above the wheel arch use the door sill for energy dissipation.[source] There are many cases of cars that performed equally well in the IIHS' moderate overlap (offset deformable barrier) test but performed very differently in the small overlap test just because they weren't designed for it. (see Camry vs Accord)

I understand what you're saying, but the MoRTH report provides an in-depth look into the scenarios in question as well. Like I mentioned, 43% of all accidents where a car was the victim, were caused by another car. Compared to that, trucks, lorries, and buses have a combined figure of 25% which is pretty much half. Thus making car-to-car crashes easily the most likely scenario. And hence my assertion that the front ODB test does in fact cover the most likely scenario of an accident.

Originally Posted by ron178

They used to look for it after stripping the car after the test but it's not part of the score directly.

Ah. Curious. Ideally it should be.

Originally Posted by ron178

They're a small team, they make mistakes sometimes. Even the GNCAP (same team). There are quite a few more. They've made the same mistake with the Indian first-gen i10 for example.

Oh yeah. They indeed did! Good catch.

Originally Posted by ron178

I didn't mean to portray that as a bad thing, I'm sorry if it came out that way. This is something not limited to the XUV700: even for the XUV300, to quote the GNCAP's secretary-general, "you could replace the door skins and drive the car as new" in reference to the structural side impact performance. They even used the ESC/pedestrian car to rerun the side impact on the passenger-side to look for asymmetrical performance (seen in some Latin NCAP tests) and the XUV300 performed similarly. Regardless of the reason, I can only applaud that.

Oh Yeah. No harm no foul. My perspective was more along the lines of, I don't care why they make safer cars, as long as they make them for Indian market too. Good excerpt on the 300/700. It's nice to see them have this strong of a conviction.( This also kind of is what I meant when I said that "the entire car is well built, not just front ODB worthy").
Add to that the leaks that I read about how the 700 was made 5 star worthy: Apparently they had procured all the leading competitor cars(Safari, Gloster/Hector etc) and crash tested them, and used that as a lesson to improve their own safety. I believe this was shared somewhere on the 700 thread.

[Rohan265]

Originally Posted by ashishk29

Excerpt from the above study (which seems to have been performed in 2017, and uses crash data from up to 2016)
"Percentage*distribution*of*917*cars*by*impact*loc ation:
The*impact*location*was*analysed*by*considering*th e*third*character*of*the*CDC*[5].
The*third*character*denotes*the*General*Area*of*Da mage*(or*direct*impact*location)*for*the*impact,*a nd*64%*of*the*cars*analysed*had*their*impact*locat ion*as*“F‐Front”."

So even this study agrees that 64% of all crashes were frontal impacts of some variant.

However it also says that
"The*probability*of*a*car*impacting*a*truck/bus,*a*motorised*two-wheeler*or*a*fixed*object*is*far*greater*in*India* than*the*probability*of*a*car*impacting*another*ca r."

This has since changed. If you check my original post for the MoRTH stats from 2019, to quote myself,
"16% of all accidents involve cars/taxis/lmv, and of that 43% are caused by another car/taxis/lmv."

The percentage has changed vastly since then. About half of the accidents are now cars impacting other cars. So this finding from the 2017 study does not hold entirely true.

I would like to butt in here as I noticed that you are relying on crash data. First, you are quoting the paper out of context. Yes, the paper does say that 64% of all car crashes involve the general area of damage as the Front. However, to bring the sample as close as possible to the EURO NCAP tests, the authors have also added another selection criteria of the Direction of Force of the impact to be as close to 0 degrees as possible (an important factor that many don't consider when talking about crash tests or even airbag deployment). The final applicable sample is 255 cars or less than 30% of all passenger cars.

Second, I would take the MoRTH data with a pinch, or rather, a couple of handfuls of salt. The reason behind me saying this is because I have seen the data and collection process of both RASSI data (the underlying data for the aforementioned paper) and MoRTH data. To put it mildly, the whole data collection process behind MoRTH data is deeply flawed. There are numerous holes in the whole process. Most of these shortcomings are hidden. But, some of them are seen even in the final report. For e.g. you have repeatedly refered that 43% of car crashes have other cars as the "crime vehicle". However, the same table also shows that around 13-15% of car, bus and truck fatalities each have two-wheelers as the impacting vehicles, which doesn't make much sense. The best way to interpret this data would be that 13-15% of all car/truck/bus fatalities have two-wheelers as the collision partners, but it is not necessary that an impact with the two-wheelers caused the fatalities. Ideally, this understanding should apply to the entire table. So, it may be possible that 43% of car fatalities have other car involvements, but the car impact may not be cause of fatal injuries. There are many other shortcomings in the data such as run off road and fixed object/overturn being considered as different collision types which again doesn't make sense as just running off road does not cause injuries by itself. That vehicle has to hit something/rollover to cause injuries.

My point is that, eventhough the paper is a bit older (2017 vs 2019) than MoRTH data, the technical paper's underlying data is far far more reliable than MoRTH data. Also, crash trends do not change so rapidly within a space of 2 years.

Originally Posted by ashishk29

Are you suggesting that the stronger materials here are somehow cognizant of how they're being impacted? No. Strong materials will be strong no matter how the crash is. If it's designed to take a certain amount of impact, and the body is designed to distribute the force in a certain way, it always will. Which means that the standards will work even in other varieties of crashes. It's not binary that if a car is does well in frontal impact tests it will no do well in other tests at all. Materials and chassis design don't work that way.

I agree that the chassis is designed to distribute crash forces along a certain way (called Load Paths). However, this doesn't mean that the load paths come into play for all crashes. The Direction of Force concept discussed in the earlier paper plays perhaps the most important role in crashes. Crash tests cover perperndicular impacts with a tolerance of 10 degrees on either side. However, most crash forces in the real world are angled beyond the tested parameters. Such crash forces activate only a part of the load paths or miss them entirely. The IIHS small overlap test is a perfect demonstration of what happens when a narrow, offset impact bypasses the car's energy-absorbing members.
Also, a single chassis is comprised of materials of different strength. Although unlikely, it is possible that a chassis can be designed to absorb only frontal forces effectively, but has parts of less-than-desirable strength for side forces. That is a car should also be tested for side impacts. But, that does not necessarily mean that the vehicle will do well in all side impacts, especially those that are not perpendicular.

Originally Posted by ashishk29

These cars have proven in real world accidents that their "theoretical" safety ratings do lead to occupants coming out alive even after severe accidents. Some of these accidents involving rolling over, other vehicles literally climbing over them, and many other non-front-ODB-crash like scenarios. If it were "designed for the test" those cars would not have had much success in saving people in actual accidents which tend to vary more in severity and situation. OR, this could also imply that GNCAP testing does cover enough to correlate with real world safety. Otherwise there would not have been a correlation.

Yes, GNCAP testing does correlate to real world safety. But the tests do not cover enough crashes (that is what the research paper concludes). I have seen this personally in multiple crashes. The Army truck case shared shows that only the front of the car has underriden the truck. Tbh the damage that you see is quite common in many such collisions, even with un-rated cars. The truck has such a high clearance that it just scraped over the car. The car occupants were lucky that the truck did not impact the passenger cabin. But there are many such cases in which 5-star safety didnt amount to much because the car is not tested for such impacts:

https://www.youtube.com/watch?v=IS4HLBjm9_c

NCAP tests are the best grading system we currently have for a car's safety. But, a GNCAP score does not translate to the desired safety performance in most real-world crashes. Just the ones the crash tests cover (with some tolerance). NCAP is effective in other countries because the tests use that country's crash data and are designed to represent the crash data as close as possible. We are importing tests which may not even apply to our crashes. The only way to confirm is to improve our crash data collection method. Hopefully, the IRAD system will gives us better data in some years.

[Gypsian]

I vote for the latter. The safety features are not fail proof but the structural inbuilt safe guards and airbags are more or less fail proof , of course at decent speeds. I still can’t trust in features like ADAS or sleep alert etc.

[ashishk29]

Originally Posted by Rohan265

Second, I would take the MoRTH data with a pinch, or rather, a couple of handfuls of salt. The reason behind me saying this is because I have seen the data and collection process of both RASSI data (the underlying data for the aforementioned paper) and MoRTH data. To put it mildly, the whole data collection process behind MoRTH data is deeply flawed. There are numerous holes in the whole process. Most of these shortcomings are hidden. But, some of them are seen even in the final report. For e.g. you have repeatedly refered that 43% of car crashes have other cars as the "crime vehicle". However, the same table also shows that around 13-15% of car, bus and truck fatalities each have two-wheelers as the impacting vehicles, which doesn't make much sense. The best way to interpret this data would be that 13-15% of all car/truck/bus fatalities have two-wheelers as the collision partners, but it is not necessary that an impact with the two-wheelers caused the fatalities. Ideally, this understanding should apply to the entire table. So, it may be possible that 43% of car fatalities have other car involvements, but the car impact may not be cause of fatal injuries. There are many other shortcomings in the data such as run off road and fixed object/overturn being considered as different collision types which again doesn't make sense as just running off road does not cause injuries by itself. That vehicle has to hit something/rollover to cause injuries.

My point is that, even though the paper is a bit older (2017 vs 2019) than MoRTH data, the technical paper's underlying data is far far more reliable than MoRTH data. Also, crash trends do not change so rapidly within a space of 2 years.

I will take a pinch of salt. That only means that I will be open to accepting that the figures quoted may be +/- 5% in some areas. They do also admit that they have no data for 16% of total accidents.

Also, you claim that RASSI is unreliable. But the 2017 paper also uses that as a source.
Quote:
"The RASSI in‐depth crash database is used for analysis."

So according to your own assertion that
"I have seen the data and collection process of both RASSI data (the underlying data for the aforementioned paper) and MoRTH data. To put it mildly, the whole data collection process behind MoRTH data is deeply flawed. There are numerous holes in the whole process. Most of these shortcomings are hidden."

we must also disregard the 2017 paper since it relies on the same data source. Please help me understand this dichotomy you have presented.

Originally Posted by Rohan265

I agree that the chassis is designed to distribute crash forces along a certain way (called Load Paths). However, this doesn't mean that the load paths come into play for all crashes. The Direction of Force concept discussed in the earlier paper plays perhaps the most important role in crashes. Crash tests cover perperndicular impacts with a tolerance of 10 degrees on either side. However, most crash forces in the real world are angled beyond the tested parameters. Such crash forces activate only a part of the load paths or miss them entirely. The IIHS small overlap test is a perfect demonstration of what happens when a narrow, offset impact bypasses the car's energy-absorbing members.
Also, a single chassis is comprised of materials of different strength. Although unlikely, it is possible that a chassis can be designed to absorb only frontal forces effectively, but has parts of less-than-desirable strength for side forces. That is a car should also be tested for side impacts. But, that does not necessarily mean that the vehicle will do well in all side impacts, especially those that are not perpendicular.

I understand what you are saying. But it would be vastly infeasible to cocktail the car specifically such that it does extremely well in frontal impact tests while being weak in other aspects. Again, I am not saying that front ODB is the end-all test. But that is a test which does put a good amount of impact on a huge chunk of the car overall. Hence you can see so many bent A pillars, roof, door panels in unsafe cars.

Your assertion that "it is possible a chassis can be designed to absorb only frontal forces effectively but has less strength for side forces" is purely theoretical. In theory anything can be true. One might even theorise that only the right side of the cabin is reinforced since the test only hits that side. But realistically, this is not how it works. The chassis as a whole has to be designed to a certain strength. Also, our current 5 star rated cars did go through side impact test as well. While it was not rated individually, they did pass a UN95 side impact test as well.

Originally Posted by Rohan265

Yes, GNCAP testing does correlate to real world safety. But the tests do not cover enough crashes (that is what the research paper concludes). I have seen this personally in multiple crashes. The Army truck case shared shows that only the front of the car has underriden the truck. Tbh the damage that you see is quite common in many such collisions, even with un-rated cars. The truck has such a high clearance that it just scraped over the car. The car occupants were lucky that the truck did not impact the passenger cabin.

Sorry but the truck most certainly didn't just scrape over the car. The truck most certainly put a significant amount of weight ON the front of the car. The hood has buckled in. If it were a scrape like you said, it would have been essentially pushed on towards the front glass, while being crumpled upwards.

These images show that the hood bent inward, the front glass was shattered. And you can see that the car also has burst tyres. Meaning there was a tremendous amount of weight put on the car.



Compare this to an incident where a car just crashed into a truck which has a higher than usual barrier:

Originally Posted by Rohan265

But there are many such cases in which 5-star safety didnt amount to much because the car is not tested for such impacts:

https://www.Youtube.com/watch?v=IS4HLBjm9_c

NCAP tests are the best grading system we currently have for a car's safety. But, a GNCAP score does not translate to the desired safety performance in most real-world crashes. Just the ones the crash tests cover (with some tolerance). NCAP is effective in other countries because the tests use that country's crash data and are designed to represent the crash data as close as possible. We are importing tests which may not even apply to our crashes. The only way to confirm is to improve our crash data collection method. Hopefully, the IRAD system will gives us better data in some years.

There are also much more cases where 5 star safety did amount to much. While you say "car is not tested for such impacts", I say that you are skipping the human factor here. A car is only as safe as its driver. If a driver chooses to drive even an XUV7OO off a cliff there isn't much that can save them.

But we are talking about safety within the ideal usage bracket for any vehicle. Where GNCAP results do translate to real world safety.

Of course there are going to be crashes where the car will not be able to cope with the impact. I never said that a 5 star rated car is made of adamantium or something.

Your suggestion that "But, a GNCAP score does not translate to the desired safety performance in most real-world crashes." is based on basically denying the legitimacy of RASSI. And by that I mean refuting both, the 2017 and the 2019 papers. I don't. I did take the stats seriously, with some margins adjusted. So I do believe that while GNCAP is yet to improve and include more tests. But a 5 star rated car will still help you in the most likely road accident cases.

Nobody says that having a 5 star car guarantees that you walk away from any and every accident. It just greatly raises the probability of you sustaining the least amount of injuries in a brutal accident. Beyond a certain limit every object has its max structural strength. No one can save one there.

[RubberGuru1113]

Originally Posted by ashishk29

Your assertion that "it is possible a chassis can be designed to absorb only frontal forces effectively but has less strength for side forces" is purely theoretical.

I'm sorry but this statement is not theoretical. Examples of cars with good front impact performance and poor side impact performance and vice versa were given on the same thread sometime back :-

Originally Posted by ron178

Sorry I stand by my statement that structural side impact performance is independent of frontal impact performance.

Attachment 2335688
2017 Ford Ka (Brazil) (also see: 2016 Peugeot 208)

Attachment 2335689
2017 Chevrolet Aveo (Mexico)

[itsmeparvathy9]

For people like me, who drive at controlled speeds in cities, I would prefer a car with more active safety features and more than 2 airbags as against a 5 -star NCAP rated car with just 2 airbags and few active safety features.

Moreover, a car is an experience, a bare bones car with 5-star safety features will not be the same as a feature filled car with 3-star safety features.

The market sentiment proves it, despite all claims of 4- star and 5-star safety in Mahindra and Tata Cars, Maruti Suzuki with its cars which are rated mediocre at best with respect to safety continues to rule the passenger car market mainly due to mileage it offers and service availability and now they are bringing in features too. Hyundai and KIA also stand 2nd and 5th respectively in the passenger car market due to some of these aspects, their cars are of course more feature loaded.

However, the Safety ratings have brought out better, more feature filled as well as safer cars from many manufacturers and manufacturers have now begun to understand that safe and practical cars are the ones that will sell, this being a major reason in the rise in the demand of many Tata and Mahindra cars.

So, if a car is safe and feature filled (if not fully feature loaded) it is good recipe for success in our market, Take the Tata Nexon for example.

[Rohan265]

Originally Posted by ashishk29

Also, you claim that RASSI is unreliable.

So according to your own assertion that
"I have seen the data and collection process of both RASSI data (the underlying data for the aforementioned paper) and MoRTH data. To put it mildly, the whole data collection process behind MoRTH data is deeply flawed. There are numerous holes in the whole process. Most of these shortcomings are hidden."

Please help me understand this dichotomy you have presented.

I apologize as I realised that my comments were not clearly framed.
I meant to say that RASSI data is far more reliable than MoRTH data. I would say that RASSI data is of the same standard as international databases. I would never claim it to be unreliable.
MoRTH data is lacking in many aspects and cannot be relied upon to make any inference.

Originally Posted by ashishk29

Sorry but the truck most certainly didn't just scrape over the car. The truck most certainly put a significant amount of weight ON the front of the car. The hood has buckled in. If it were a scrape like you said, it would have been essentially pushed on towards the front glass, while being crumpled upwards.

These images show that the hood bent inward, the front glass was shattered. And you can see that the car also has burst tyres. Meaning there was a tremendous amount of weight put on the car.

Compare this to an incident where a car just crashed into a truck which has a higher than usual barrier

I guess we have a different understanding of the same crash. I believe that the army truck had an angled impact with the truck. That is why the bumper is not pushed rearwards and upwards towards the windshield as we normally see in longitudinal underride crashes. The reason the left-side tyres deflated was because the angled impact caused the car to tip over a bit and put all the vehicle weight on the left-side tyres.
Again, my understanding of "merely scraping" is different as I see much more severe damage in underride crashes. I should not have used such terms as it leads to a different understanding.
Also, the Ecosport shown has had longitudinal underride damage. I don't have data on speeds but I feel that the Ecosport has done reasonably well to prevent any intrusions.

Originally Posted by ashishk29

While you say "car is not tested for such impacts", I say that you are skipping the human factor here.

Of course I am. The whole purpose behind crash testing vehicles is that "humans make mistakes and it is the duty of the vehicle to prevent any serious injuries". Crash tests cover a certain percentile of crashes due to human errors. The goal is to cover as many crashes as possible based on the real-world scenarios in that particular country.

Originally Posted by ashishk29

Your suggestion that "But, a GNCAP score does not translate to the desired safety performance in most real-world crashes." is based on basically denying the legitimacy of RASSI.

As clarified above, I am not denying RASSI's legitimacy. I can vouch for the data's reliability and hence I stand with the above statement.

Originally Posted by ashishk29

Nobody says that having a 5 star car guarantees that you walk away from any and every accident. It just greatly raises the probability of you sustaining the least amount of injuries in a brutal accident. Beyond a certain limit every object has its max structural strength. No one can save one there.

This is something we can agree on.

[prajwalmr62]

All of the collisions I've personally seen are because of driver error. No safe distance followed, overtaking on blind curve, too fast on wet road, zero situational awareness, no safe speed maintained, etc. People are very good at overestimating their skills, and easily blame it on some issue with road or car or something else. But whatever the reason, all these are driver errors nonetheless. And most of them could have been prevented easily if one of them followed defensive driving.

In a country like ours, where speed is less, but drivers are usually maniacs, we need restraint over our driving behavior, and it can only be achieved by ADAS and other host of similar features. They also cost less to manufacture in mass. We can avoid most accidents with driver aid features like ADAS and car state monitoring like TPMS.

A 5 star rated car of 700KG GW cannot hold up against 3 star rated car of 2000KG GW or a 10T truck. The "stronger" structure definition also changes as the speed changes. A car strong enough for collision at 60kmph is not guarantied to perform better at 70kmph.

If we go just by these rules, we will just end up with bigger cars on our roads like US and it is not going to be sustainable. And it doesn't reduce accidents or fatality rate either, because our issue is with the driver here, not the car.

[Livnletcarsliv]

I guess there is an ambiguity here.
Safety rating vs safety features vs build quality.

The safety rating is a post-production test and it is purely a test. It cannot be manipulated. Neither it can be customized/configured. It is the same across the brands and across models and across variants.

However, safety features and build quality differ.
Safety features differ for every model, every variant, and every brand.
On the other hand, build quality differs only brand and model-wise and not variant-wise.

Build quality is the basic DNA of the car itself. No matter what, build quality has to be top-notch. Eg - M&M, Tata, and VAG. Poor build quality examples - Maruti, Hyundai, Kia.
Safety features only minimize the impact and reduce the loss (vehicle damage and human injury).

There is no point in giving 100 safety features when the build quality of the car itself is not stable/good. On the contrary, the possibility of impact during a collision/accident being minimal is high in a better build quality car even with basic safety features. So it is always BETTER BUILD QUALITY than more safety features.

[Arun Varma]

[QUOTE And it doesn't reduce accidents or fatality rate either, because our issue is with the driver here, not the car.[/quote]

Absolutely agree.

It is indisputable that a safer (rated for build quality) car will result in better outcomes for any driver in all unfortunate situations. Having said that, other aspects of car ownership contribute to the driver's mindset leading to these unfortunate incidents. And in my mind, a car that provides a better overall ownership experience can contribute to a better driving experience and, thence, safer outcomes.

If the car one prefers, at the budget one has, is super safe but misses out on comforts, necessities, and conveniences, but one ends up choosing just for safety alone, is not likely to provide the overall positive ownership experience. This, in turn, can influence the negative outcomes.

Of course, having said all that, there are bad drivers who are bad irrespective of the ownership experience.

Again, a safe car is the best option for all of us, but we must not overlook the other aspects which provide for a more robust ownership experience.