As a side note, I am often asked why am I sticking with rF, GSC or Race07 (or with NKP or LFS) with "all that's going on with the next-gen sims".
That's a legitimate question, isn't it, all the hype and marketing machine-gunning must account for something and most simracers are facing it more and more. Why stick to "old" sims?
Hmmm...
Without going into details of how work is done inside a development studio, I'll say this again:
- even dev teams work within standard boundaries due to limitations imposed by knowledge, data, tools and time constraints;
- dev studios FREQUENTLY do not have the FULL info about cars (manufacturers often send simple brochures instead of relevant data) and they have even less data about tires. [Don't fall into the hype/marketing stunt of Greenaway that they (Turn10) will be revolutionizing the science of tires...]
So, much as I like the breathtakingly beautiful graphics of Project CARS, I inevitably focus on what really matters simulation wise: physics, mechanics, pure racing. Knowing the pitfals/shortcomings of the current development models, strategies and organization, knowing what I wrote above, it's totally impossible for me to accept the sophistication and novelty hype surrounding pCARS and AC.
Plus, after years of studying different physics models, I will comfortably settle for rF, GSC or Race07 for the next 2 to 3 years. The eye-candy or a ton of new, "never before experienced" FFB effects are pretty much irrelevant for me (fortunately, for others too). "Very sorry, hype-machine..."
Just by putting more "realistic" tire simulation, i noticed those weird "always go soft" suspension setups don't work anymore and they are starting to creep up closer to values i've been able to get my hands on. And i've never seen tire temperatures change so fast, [...].
Indeed.
The physics calibration model I advocate involves a full rework of cars.
FULL.
Most modders aren't prepared for that kind of work and the implications of it all, and mostly due to:
- by doing so, we enter totally uncharted territory. Unless you really, really know what you are doing and where you're heading, you'll be in for months of headaches and frustration.
- it's really, really a ton of work in all areas: chassis, aeros, tires. And that PER CAR.
- realistic simulations being increasingly more difficult the closer you are to the limits of car and tires. It takes time, persistence and intelligence (in experimenting with the setups best suited to you) to get the most out a simulated vehicle, to become faster. Not all simracers and their leagues are up for this challenge, imo.
When i got time, i'll get the tire model finished to next round of beta (the first one was basically alpha for tire model, with individual slip curves for all four compounds.. i still got a hell of a lot to do with aero, ... [...]
Eheh.
Well, my strategy (and the entire process) is actually pretty simple.
I start out with "Car basic data":
- mass, estimated CoG, weight distribution (front to back). A multi-block or integral-defined shape approach will give me inertias, aero coefficients (body), etc.
- fuel
- dimensions (length, width, height, track width)
Wheels, tires and brake disks:
- mass, dimensions. From this, inertias get calculated. In the absence of full data, one of the preset data models will yield proper results.
Wings and splitters:
- given width, chord, height, shape and placement, I model an envelope of minimum to max downforce achievable (i.e., from 1º to typical values/settings, anywhere from 10º to 17º, depending on the shape of the airfoil)
Diffusers:
- it's a little bit different and mostly based on data given to me by teams.
Tires:
- there's a lot of info out there about compounds, tests, etc. Some of it is in raw format, some of it comes from graphs which are a bit difficult to read from, some of it comes from books...and fortunately, some of it gets released by the likes of Michelin, Avon, Hoosier and Bridgestone. Getting to know someone within the industry is key to understanding how to translate all the info commonly available, all the data shared by teams, into practical models.
Telemetry:
- While some modders have been fortunate to work with RL teams (in spite of technical problems here and there) and indeed there's a lot of data from data acquisition systems we can use (within the confines of NDA's).
But occasionally, some files or data sheets surface here and there. I saw telemetry data from pre-season tests of a GT team competing on an european championship a couple of years ago, and that was posted on their website. A rear wing manufacturer (sorry, can't recall the name) posted some wind tunnel data on their latest products. The new Viper SRT aerodynamics devices team posted a study detailing some of their findings (and partly the process) on the new set of splitter and rear wing.
From there, it's simple:
- What are the maximum speeds on straight?
- What are the expected max cornering forces at 60 kph, at 100 kph, at 160 kph, at 200 kph?
- What's the expected braking distance (depending on speed)?
- What type of tire compounds to model? What tire widths?
Like in real life, tire compound and width will allow us to calculate the grip coeffs, load bearing capacity and expected slip angle envelope (for maximizing forces). From this, we get, depending on load, what forces the tire will develop.
That's the stepping stone of simulating a vehicle.
Then, if you know the engine output, you know how fast it can go. If you know how fast it can go, you can calculate how much downforce you will get from splitter+rear wing+body+diffuser at whatever the speed. If you know the downforce, you know the load PER TYRE, therefore you know the cornering forces developed for the car (depending on speed).
Some of the things we have to contend with afterwards are:
- weight distribution + aerodynamics CoP (we can work with splitter CoP + Body CoP + Diffuser CoP + Rear wing CoP, in order to understand the overall aerodynamics distribution and its effects on load distribution)
- springs: natural frequencies, etc
- dampers
- ARB's
- suspensions: geometry, component mass, component inertia, dimensions, etc
I do a somewhat generic treatment of engines (power output, shape of curves, heat generation, cooling, engine braking) and gear ratios.
What I never do:
-
tweak tires to match the aeros or tweak both to match the desired "feel" (there's a very well known mod group, with ties to several motorsports teams and personalities, that does this...). There is no feel involved. [
A somewhat known modder - he did, at some point, the physics of a well known GT league - sent me furious emails on how he could never mod by equations, that it destroyed all "feel"...]
-
use impossibly high grip coeffs (example: F1 tires level of grip - official estimates range from 2 to 2.2 - applied to GT and proto racing)
-
use unrealistic levels of downforce. 95% of modders do this to compensate for the lack of proper force generation
-
use street cars slip angle envelopes on GT and proto cars. 95% of modders do this because their cars don't stick to the road properly (whatever the physics engine) and when they snap, they snap with zero possibilities of recovery.
-
use low inertias. 98% of modders go by a simple "copy and paste" rule (if it works for that car, it should work for this), and as such errors related to inertias simply spread all over the modded content (and developer made content as well). Modders and devs use low inertias due to lack of information, carelessness or because it makes cars react faster to driver input. Problem is, cars are 950 to 1430 kg entities, they offer resistance to change of state. Without this "resistance", cars seem to be made of cardboard. The whole physical behavior goes down the drain due to this seemingly unimportant factor.
For me, by following the above, I stick to the rules of the real world. The physics engine, though not perfect, reacts very well to proper input. The cherryh on top of the cake is that more than a few simracers (especially the RL pros) enjoy and demand cars done this way.
Once i get to the next stage, you will be first i'll PM ,most likely before beta group (they are just too kind, hard to get negative feedback....and that's the only thing that counts..). But i think i need study the parameters all over a more. First tests were done with very simple values ( 1.0 where ever i could, nothing of that 1.234234 crap), just entered them in there for fun and the car just worked like one would expect.. Also what i learned from inertia, a lot of ingame models and even more so with modded content, they use high mass and low inertia (or vice versa), just putting those without any safety what so ever, just pure physics.. First test showed behavior and lap times that made sense from the first laps onwards. but i would surely need a well educated second opinion.
Sounds like a really good plan. You can PM me then.
I believe RBR used rFactor Pro up to 2009.. That should tell anyone what kind of physics engine we are tinkering with.. SimBin locked a lot of values and made their own FFB response, other that that, it should be pretty vanilla.
Indeed.
The thing that some people are so bent on ignoring (and making sure others ignore it too) is that, while not an engineering tool, rF/ISIMotor2 can be used as a high level tool for driver preparation. With proper setups (a 6 DOF motion platform, 3 screen split, full telemetry) drivers can begin their preparation for upcoming seasons without the need of expensive track testing or costly simulation-hours.
If it's good and credible/realistic enough for the pros, why not for the rest of us?