Brakes in Automation versus Brakes IRL

First of all I want to preface all of this by saying that there is always the chance I am just completely insane

:crazy_face: :upside_down_face:

About two weeks ago, I was testing one of my creations in BeamNG – specifically the Everette Vancouver GLR8 – and I noticed something strange. After several laps around both the Automation test track and Hirochi, the brakes on my regular street car were still performing admirably. This was rather contrary to what I would expect because most regular cars will boil their brakes after 5 or 6 good stops.

To cut a long story short, I got sucked down a rabbit hole investigating brakes and my conclusion is this:

Brakes in Automation do not supply nearly enough force but are also FAR FAR FAR too resistant to brake fade.


How did I even start thinking this?

Well its like this. The first thing I suspected was either

  • My own tuning
  • Or a glitch in the exporter

So the first sanity check I did was load up a couple native BeamNG cars and hoon them around the track for a while. Specifically, I used the Ibishu Pessima, the ETK 3000ix (BWM E30 clone) and the Bruckel Moonhawk V8 Sport. Sure enough, these cars boiled their brakes after about 5-10 laps. I also did a rapid succession panic stop test – 100 km/hr to 0 – whereupon I found all the cars boiled their brakes after about 6 panic stops. My Vancouver – I lost count at 12 and was probably up to 30 before it even started hinting at boiling brakes.

Next up, I considered brake spec. My Vancouver was using 315 mm twin piston vented front discs and 275 mm single piston solid disc rears with a 35 pad. Perhaps slightly overspec’d considering it is a 1999 car but nothing especially fishy here. So, this is sounding like an exporter bug. Except Automation’s own brake fade tests also showed virtually no fade…

Hmmm.

Things got a little bit more suspicious when I tried a car by @Marcus_gt500 from the Automation BeamNG user repository. As Marcus_gt500 spec’d it, the car stops from 100 km/hr in something like 95 meters! Lolwut?

I tried tuning it up a bit but was highly unsuccessful. Even with 15 inch wheels, 300 mm drums, and full race pads, I best stopping distance I got out of that car was about 73 meters. And yet, even with regular spec pads, the brakes again showed very little fade according to Automation.

Hmmm. :thinking:

This is when I got sucked down the rabbit hole.


The Rabbit Hole

I started looking up old brake test videos and performance reviews. There were a couple I found particularly intriguing. The first one that really caught my attention was this:

In summary, that video was a Chevrolet promotional video from 1936. At 6:34 in the video they quoted the stop time of a 1936 Chevrolet from 60 mph (~100 km/hr) to 0 at 4 seconds. Now if we remember our physics, distance traveled under constant acceleration is 0.5 * a * t^2, so if we assume 1G deceleration, then the car goes about 80 meters in the 4 seconds it takes to stop. EDIT: I realized this morning I am a dunce and have all the numbers to calculate the exact stopping distance - 60 mph is 28 m/s so 28 m/s divide by 4 seconds means a deceleration of 7 meters per second. Plug that back into our formula and we get… 56 meters! Also even assuming worst case number firing on Chevy’s part ie 4.9 seconds, we still get a stopping distance of 68 meters. Now I realize that is only an estimate making some perhaps generous assumptions but In 25 years, I would expect that braking technology would have improved at least enough for Marcus_gt500’s 1960 land barge to stop in less distance than that WITHOUT the use of full race pads. (Also, if you look at the design of the brakes in the video, they are single leading shoe on both front and rear)

I also found some old performance reviews of specifically a 1967 Ford Mustang GT:

And a 1967 Chevrolet Camaro SS 350:

The Camaro stops from 80 mph (130 km/hr) in 280 ft (85 m) and the Mustang stops from the same speed in 312 ft (95 m). And again that is from 80 mph (130 km/hr) not 60 mph. So if we do the maths, we find that these cars stop from 100 km/hr in about 48 and 53 meters respectively. And these figures line up well with other cars of their type. Even Ferraris of the same era had about the same stopping distances.

Now I realize these are schporty cars and both models would have had front disc brakes. But even taking this into account, a Mustang / Camaro like sport sedan I made in Automation:

Cannot achieve the same stopping distances without either using twin piston disc brakes (uncommon in the 1960s except on limousines and exotics) or very aggressive brake pads:

But these are also smaller cars from the 1960s. What about land barges? Well, this is where I found the smoking gun so to speak that really solidified by my conclusion:

Although this performance review was conducted in the mid 1990s, the 1961 Impala SS 409 in question was a faithful stock build. If you download the original PDF article, you will find it even had the factory 11 inch (275 mm) drum brakes, front and rear! How did its stop test go? 253 ft to stop from 70 mph. Or 77 meters to stop from ~115 km/hr. Which – do the maths – translates to a 100 km/hr stopping distance of… 57 meters!

How many 1960s land barges can do that in Automation even with full race pads? I think we all know the answer to that - 0.

So what does this all mean then?

Brakes in Automation do not supply nearly enough force!


What about that brake fade?

The aforementioned Impala’s performance raises a particularly interesting question. If a 1960s land barge can get into the 50 meter club with drum brakes only, a feat we have always needed disc brakes for in Automation, then why the hell did we ever need discs in real life? Well, remember how I was talking about how the brakes wouldn’t fade on my Automation cars in both Automation and BeamNG?

So after browsing numerous car forums and reading various opinions, they all start to say essentially the same thing. In those old cars with drum brakes, if you really stand on them, you can lock wheels (going to point 1 - brakes don’t supply enough force in Automation) and you will be able to stop the car quickly. But you will only be able to stop the car quickly ONCE!

Drum brakes don’t cool nearly as well as disc brakes so once you panic stop with drum brakes, they fade BADLY! Hell, just go watch the most recent episode of Roadkill. Their 1966? Impala made to the top of Pikes Peak, Colorado but they couldn’t get back down. Why? Because 4-wheel drum brakes meant they boiled and faded nigh instantly.

The point here is this: with drum brakes, you actually get pretty decent stopping power. But you get two, maybe three, good hard stops out of them in a typical street car and then you’re sliding into the next county.

Even with disc brakes, cars will still brake fade. In my own car, a 2006 Scion xB, I can confirm that it will brake fade severely after the 5th or so hard stop from 60 mph. And yet many of my cars in Automation with conservative tunes exhibit zero brake fade both according to Automation’s calculations and BeamNG’s simulation. They often need multiple decades of panic stops before hinting at brake fade.

At very least, this indicates to me that brake fade is not handled correctly in exporter.


How I think this should be addressed

I my opinion, brakes in Automation are both overpowered and underpowered at the same time. On one hand, they don’t supply nearly enough force (especially in the early years) but on the other hand are impervious to fade.

One thing I had considered for why brakes are so shit in early years in Automation was that many cars up until the 1970s, had no brake booster meaning the pedal was VERY firm and it required significant effort from the driver to stop the car. Now, while I can see this as affecting drivability and comfort, it does NOT cap the absolute maximum force that the brakes can exert; that is something only the hydraulic system can do. I mean heck, the hydraulic system could be designed to support an arbitrary amount of pressure and by extension and arbitrary amount of force. The problem becomes engineering such a system.

This actually points to another weird thing I noticed. For any given brake configuration, maximum brake force doesn’t scale with the size of the vehicle which to me seems wrong because many larger cars do employ higher pressure hydraulic systems to extract more braking force without requiring different components. I realize that getting into hydraulics and system operation is a whole boatload of complexity I don’t think most people want to deal with so I would suggest either:

  • Raising the maximum brake force provided by in particular drum brakes to better reflect reality

OR

  • Incorporate a “Maximum Brake Force” slider which can raise the maximum force but lengthens engineering time, much like how the “Brake Airflow” slider works

Furthermore I would also suggest:

  • Power brakes should be added as an additional driver aide before anti-lock brakes (ABS), unlocking in about 1960 and should provide a buff to safety, drivability, and comfort
  • Brakes need to be a LOT more prone to fade, ESPECIALLY DRUM BRAKES!

Woof. Sorry for wall of text. But this issue has been bugging me for a while and I want to see what other people think.

And again, there is always the chance I am just completely insane so that being the case, prove me wrong. I am listening.

18 Likes

Good write up, I agree especially on drum brakes for heavier cars; and on the brake fade for drum brakes.

For lighter cars, I think - apart from the brake fade - the drums are fine. I’ve been driving a lot 1952 cars recently for reasons and the smaller ones do stop, even with non-race pads in Automation. So I think you are spot on in your analysis that the brake force doesn’t scale with the size of the vehicle, and this might have to be addressed.

Regarding disc brakes though, I notice a lot of brake fade on most cars unless race pads or heavy brake cooling. Usually around lap 2 of the Automation test circuit, the brakes are absolutely cooked.

I’de like to point out that most Disc brake front cars in the late 50’s and early 60’s were 2 and 4 piston calipers, not 1-piston sliding calipers.

1 Like

Great post kmBlaine!
Automation’s brakes have long bugged me as well, but I’ve never quite worked up the courage to address it.

Part of the problem is the brake sizes most people are using… in automation people size there brakes to avoid that little blue advisory… often using higher tech sooner, and bigger brakes than real life. The funny thing is the markets already don’t care that much… in fact most cars I’ve seen score better in there best markets if you pull the pads 5-10 more into comfort or down size the brakes… yes the stopping distance goes up, and yes you get the sportiness fade notification, but your still miles from the yellow warning, and unless your scoring for super\hyper or GT\GT premium the comfort, drivability and cost improvment is way more important than a few more meters should you try brake to a dead stop from modern highway speeds.

However regarding beamNG, even with peoples over-specified brakes, I still cooked the rotors till I got fade on most cars I’ve tested if I push from best times on back to back laps… my automation back straight braking will almost always be giving fade warnings and visible smoke after a few laps, and none of killrobs GETD cars that where not painfully slow, nor my sports car CSR 78 finalists where any different, so its not just my designs.

I do however agree about not giving enough force in automation… the whole idea of drum brakes is the have a self applying mechanism, hence the level of force needed to lock the wheels is set by the lever arm used, meaning you could have them lock up with extremely minor forces (or indeed pass a threshold where they do not release until the drum stops, even if the pedal is released). The reason very high self actuation is not generally done, is locking up is not the idea braking state (save for on dirt\snow, but I digress), and so providing less multiplication allows more user control… just like with power steering systems.

As for disk brakes, you don’t need higher pressures to get more force, just a higher ratio of piston area at the caliper vs the master cylinder. Bigger slave cylinders (in calipers) work equally well at adding area as more cylinders, and either will provide more force for the same total system pressure… and doesn’t really represent more engineering time or even much higher costs, instead just as with the drum brakes, a super soft pedal isn’t ideal for control, and adding more pedal travel beyond a point also makes the car less comfortable\drivable, as you end up lifting your leg a half a foot to get onto the pedal, and just as with a level, your still doing all the work, its just over a longer distance. All of this about calipers also applies to later car drum brake systems, which unlike early cable braking systems or truck air brakes (where the force actually comes from a spring in each brake, and the air only releases it), use essentially the same hydraulic master\slave setup as a disk, just with lower forces due to the drums self application.

So what does this mean for automation? I think fade is fine, once we have more force, you’ll use smaller brakes that will fade to realistic degree with no further changes. However I’d like to see more force… and good news, the devs have stated at a few occasions (including on killrobs GETD youtube series) they are toying with the idea of increasing force a lot, and increasing the threshold or reducing the penalty for too high force. However I honestly think for now it should be a pretty low priority… after all it only makes a impact for comparing automation stats to the real world, both the markets and other player cars already work perfectly fine if you just expect the number given in automation is going to be high… and get this, its not even the same result you’ll get in beamNG. Yep, once you export, you’ll actually stop in a different distance, and can lock wheels on cars that shouldn’t have enough force based off the numbers in automation.

So I agree with you, but the effect small and isolated to only the Automation sandbox graphs\numbers right now.

4 Likes

I’d like to add an additional nitpick: Wheel camber has no effect on braking distances in Automation. It ought to, in most cases the tire is the limiter in braking distance in modern cars. I don’t consider it a high priority, and I haven’t done any testing to see how it plays out in BeamNG.

Oh I am well aware of that and it would be asinine to expect the numbers to be the same. First of all, we are talking about translating something between two totally different platforms, so there is room for error there. Second, BeamNG is a purpose-built physics simulator meaning we have to leave the idealized frictionless spheres in vacuum at the door and play by the rules of real-world physics meaning wildly imperfect test conditions. And third, there is also the human operators to account for, so yet more room for error.

What I mean to address though, is how the numbers in Automation can be so drastically different from what they should be, not how they differ between Beam and Automation. Their violation of reality grossly interferes with people’s expectations. So while the below is a completely valid point :wink: :

it leaves a lot of people scratching their heads and doing bizarre things to compensate for the perceived deficit. Basically, flawed design is flawed (to state the obvious); people shouldn’t be expected to change expectations that are based around a lifetime of observation of real cars to account for that.

Overall, though I tend to agree that overspecing brakes due to lack of force is definitely a contributor to the problem. Going back to my Mustang / Camaro like thing in the original post, you might notice it has 275 mm (11 in) drums while a real mustang had 240 mm (9.5 in) drums. Definitely I think that pad type is playing a big role in that too. I don’t think I have any cars, even modern ones, which use below a 25 pad despite that some of them probably should. And the reason is that if you don’t overspec the pad, then you have to overspec the brake in other, far more costly ways like adding pistons. If we consider how real life materials line up to Automation’s brake pad type:

image

Essentially, organic pads are never getting used because while they are nice and comfy, they have fuck all for brake force. So we overspec into the ceramic and semi-metallic region for regular cars, both of which have much better thermal characteristics and therefore they don’t fade as easily.

Anyways, long way of saying, yes I think overspeccing, particularly pad type, is really messing with brake fade calculations. Which I guess points to the simplest of all solutions - increase maximum brake force. Still, even on cars with “properly” spec’d brakes, I have observed an above average resistance to fade. I am not saying I can’t get them to fade; I am saying they can go a lot longer before they do.

Just as a side, one last thing here. What you are talking about with hydraulic systems is exactly what I mean but just didn’t word it correctly. There are parameters of the hydraulic system you can adjust in order to provide more force independent of the pads and rotors.

6 Likes

Good stuff! Thanks for the write-up and concrete examples. This stuff is very helpful.

A few weeks ago I noticed that problem too and looked into it. The brake force deficit you are rightly pointing out is a bug. In Automation we have implemented brakes via a general brake force progression that then is modified by brake type parameters. Thing is, the “early point” of the linear progression in brake force is (in the public version) / was (in our dev version) a factor 2x lower than it should be.

The brake force was set to progress from 20 to 80 between 1940 and 2020, instead of 40 to 80.
So that change will address most of the imbalance you see in brake force, as modern brakes seem to be fine.

The other brake parameters we have to tune things are:

  • Cooling Factor (affects how quickly the brake sheds heat)
  • Heat Capacity Factor (affects how much heat it tolerates and/or can store before fading per kg)
  • Effective Piston Count (affects comfort and stuff)
  • Base Area (directly affects brake force, scales with size)

Looking at the values they mostly seem to make sense, but once the change of the general brake force is there, it would be good to look into this again. If brake fade is too low overall, the first two parameters can be tweaked for the different brake types. Here a little table I made for you with the current data:

image

As you can see, it would be easy to tweak the numbers to get them closer to observed reality. Drums probably need to be nerfed more in Cooling Factor, and maybe a little in HC Factor. Modern ceramics seem to perform fine and I remember doing quite a bit of testing on those to get it right in Automation. They do however not port over to BeamNG correctly at the moment, keep that in mind :slight_smile: I think Beam things they are drums or something. xD

Cheers!

15 Likes