Beam.NG Tire behavior is weird

[Akricius]

I know the new update rolled back the tire grip level's to previous values and that the behavior is supposed to be more accurate but in practice it has gotten worse. I tested two identical cars, one exported before and one after the update. The wth the new mechanics handles like it's on ice. Other cars that had plenty of grip before are the same. Main issue is terminal oversteer in cars that make absolutely no sense to oversteer and overall tendency to fishtail. If the car has a weight distribution of 54/46, 185/75 tires front and rear, very little rear sway bar and only as much rear spring and damp rate as necessary.

[Killrob]

That is a very high tire profile, those probably would not handle well IRL either depending on how heavy the vehicle is. Try out 55 profile tires for example and see if that makes it handle significantly better. Also, what compound are you using?

[Akricius]

Originally posted by Killrob:

That is a very high tire profile, those probably would not handle well IRL either depending on how heavy the vehicle is. Try out 55 profile tires for example and see if that makes it handle significantly better. Also, what compound are you using?

It's a replica of Opel Ascona D, 185/75/13R are the standard size option. It's medium compound, I always assume that sports compound is reserved for sports cars and wouldn't make much sense for a juiced-up economy car in the late 70's. I am aware it will not "handle well" with those tires. The thing is I'd expect terminal understeer for which the suspension tune is setup rather then oversteer. Plus this issue is present on all cars. I just chose what i thought to be the best example. The rear end gives out very quickly especially at lower speeds it seems. Taking a 45 degree turn at 60 km/h shouldn't result in a strong fishtail.

[Slim Jim]

That doesn't sound like a tire issue. That sounds like a weight distribution / suspension tuning issue. Can you show us what you're working with in Automation (weight distribution stats and suspension settings) and in BeamNG (weight distribution app on LEVEL ground).

[Akricius]

Originally posted by Slim Jim:

That doesn't sound like a tire issue. That sounds like a weight distribution / suspension tuning issue. Can you show us what you're working with in Automation (weight distribution stats and suspension settings) and in BeamNG (weight distribution app on LEVEL ground).

Sure, so in Automation the F/R is
53.1/46.9

And in BeamNG it's
FL28%, FR27%
RL23%, RR22%

The suspension values go as follows:
F / R
Camber -0.05 / -0.35
Toe in 0.10 / 0.15
Springs 3.65 / 4.25
Dampers 2.25 / 2.40
Sway Bars 2100 / 600

Supension height and travel is 299.0 mm and 89.8.
Car weighs 1066kg.

I also tried running the car with BeamNg tires and behavior was very similiar although less snappy, so you are most likely right. However I really don't know what else to try, I already fiddled with spring and damper values, and most importantly sway bars but bar straight up removing rear sway bars and pusing the engine far to the front I couldn't really fix it.

[Slim Jim]

The first thing that I notice is the front weight distribution in Beam is higher than it is in Automation. This is a known bug and is definitely contributing to the handling characteristics you describe. You can modify the weight distribution in Beam by adjusting the "Engine Offset Y" property in the Vehicle Config -> Tuning menu so that it matches the same weight distribution in Automation (moving more weight towards the rear).

Unfortunately, I can't interpret your spring and damper values as you've presented them. I would need the spring values in Hz (the actual ride frequencies 1.2, 1.5, 1.76 Hz etc.) and the damper values in coefficients (0.2, 0.32, 0.46 etc.) to be able to help you tune them to more realistic values.

We can attack the sway bars and ride height after we figure out the above. But at the very least, modifying the weight distribution in Beam using "Engine Offset Y" should get you a little bit closer to the expected behavior that Automation is displaying.

[Akricius]

Originally posted by Slim Jim:

The first thing that I notice is the front weight distribution in Beam is higher than it is in Automation. This is a known bug and is definitely contributing to the handling characteristics you describe. You can modify the weight distribution in Beam by adjusting the "Engine Offset Y" property in the Vehicle Config -> Tuning menu so that it matches the same weight distribution in Automation (moving more weight towards the rear).

Unfortunately, I can't interpret your spring and damper values as you've presented them. I would need the spring values in Hz (the actual ride frequencies 1.2, 1.5, 1.76 Hz etc.) and the damper values in coefficients (0.2, 0.32, 0.46 etc.) to be able to help you tune them to more realistic values.

We can attack the sway bars and ride height after we figure out the above. But at the very least, modifying the weight distribution in Beam using "Engine Offset Y" should get you a little bit closer to the expected behavior that Automation is displaying.

Sorry, I should've realized.
Springs
F 1.69Hz
R 1.93Hz
Dampers
F 0.33
R 0.35

I tried to move the engine rearwards in BemaNG as you said and HOLY CRAP, a huge difference in the cars behavior. Much more predictable, actually understeers when pushed to the limit as it should with the tires.

[Slim Jim]

Indeed! Glad we're making progress. Now as for the spring rates and damping coefficients, the latter are probably in the ballpark but those springs are WAY too stiff. That's 997 Porsche 911 Turbo spring rates. There's no way that car had springs that stiff.

As a general rule for damping coefficients for replicating real life cars, I use: 0.2 for comfort, 0.3 for drivability, 0.4 for sporty cars, 0.5 for serious performance cars, 0.6 for medium downforce track-oriented cars, 0.7 for high downforce track-oriented cars.

I use these videos as reference to determine the optimal ride frequency for whatever car I'm building.
Stock ND Miata - 1.2hz front, 1.39hz rear
Various 996 and 997 Porsches
Various cars over time. Although the game says anything less than 1hz induces motion sickness, there were a number of cars with ride frequencies less than 1hz. As long as there is sufficient damping to compensate for the soft springs, you should be fine.
NB Miata ride frequencies
Various mid-90s to early 00s cars.

edit: Maybe try 1.15hz front and 1.32hz rear with the similar damping coefficients (0.33 ftont and 0.35 rear), then see how that exports. Softer springs will give you more weight transfer, which means more grip and should reduce the oversteer tendencies even more.

[Akricius]

Originally posted by Slim Jim:

Indeed! Glad we're making progress. Now as for the spring rates and damping coefficients, the latter are probably in the ballpark but those springs are WAY too stiff. That's 997 Porsche 911 Turbo spring rates. There's no way that car had springs that stiff.

As a general rule for damping coefficients for replicating real life cars, I use: 0.2 for comfort, 0.3 for drivability, 0.4 for sporty cars, 0.5 for serious performance cars, 0.6 for medium downforce track-oriented cars, 0.7 for high downforce track-oriented cars.

I use these videos as reference to determine the optimal ride frequency for whatever car I'm building.
Stock ND Miata - 1.2hz front, 1.39hz rear
Various 996 and 997 Porsches
Various cars over time. Although the game says anything less than 1hz induces motion sickness, there were a number of cars with ride frequencies less than 1hz. As long as there is sufficient damping to compensate for the soft springs, you should be fine.
NB Miata ride frequencies
Various mid-90s to early 00s cars.

edit: Maybe try 1.15hz front and 1.32hz rear with the similar damping coefficients (0.33 ftont and 0.35 rear), then see how that exports. Softer springs will give you more weight transfer, which means more grip and should reduce the oversteer tendencies even more.

I did exactly as you said and the difference was TREMENDOOUS, the car had more grip as you said but most importantly the movements were some much slower and predictable. I had a suspicion that it could be oversprung but never to that extended. I guess I had "trained" myself that the springs always had to have a higher frequency than the dampers and that they should be abouve 1.5 Hz. Don't know why but i guess it worked to some extent so I didn't think to change the obvious.

Thanks a lot for that. Combined with the new chagnes to the gearbox I feel like putting few more hunderd hours in.

No I guess the last thing would be the sway bars, what would be an acceptable amount of roll on a car this size and weight ?

[Slim Jim]

Awesome! Glad it's working out. I found this on the web as a general rule:

Large Passenger Cars: 7.0 degrees/g
Small Passenger Cars: 5.0 Degrees/g
Sports cars: 4.2 Degrees/g
High Performance Road Cars and Track Day Cars: 3.0 Degrees/g
Purpose Built Racing Cars: 1.5 Degrees/g

To which I would add an additional 1 degree for older cars. So an older large passenger car would be 8.0 degrees, an older small passenger car would be around 6 degrees, and older sports cars would be around 5.2 degrees etc.

For what you're making, I'd say somewhere between 6-7 degrees. The only road-legal car I've ever found roll angle data for is the Ferrari F50 -> which had 1.5 degrees of roll (just like a race car!).