11 replies to this topic

[Aren]

Hi there, guys.

Just before making questions, let me introduce a bit myself and the problem I'm facing.

For the last three years I've been a staff member of an amateur racing team that has took part in the 24 hours endurance race of Barcelona (it's being held in Circuit de Catalunya in mid-late September every year).

In 2004 we're going to participate driving a Ford Focus RS and the main aim of the whole team is, apart from finishing as high as possible ;), to think and act in a professional way to try to avoid some of the organization problems we had to deal with the past years.

One of the things to do is to have a software that would allow us to keep control of laptimes, use of tyres, brake discs, etc... (i.e., mechanical components of the car), availability of drivers, fuel consumptions and so on... both during tests prior to the race and during the race itself.

So, as an IT Engineer student, I'm the responsible of developing this software (I'm also going to use it as my tesis). And what's the problem? The fuel consumption calculations.

The team wants the software to tell us how much fuel is being used per lap so, knowing the fuel tank capacity and this consumption, we can predict when the driver needs to go to the refuelling area and we can define a proper strategy based on that info.

But the issue here is that the race rulebook forbids any telemetry between car and box so I have to make the calculations based on lap time and lap length. I know that just based on that info I will never be able to get an exact figure but does anyone in here ever faced a problem like this? How can I resolve the problem in order to get a quite accurate estimation? I know I can use other variables such as driver weight, fuel weight, car weight, time of the day/air temperature, but don't know really how they're going to influence in the overall result.

So could anybody give me a hint? Any info will be really welcomed

Thanks in advance, guys.

[MRC]

If the driver can know when to come in by himself then you could do the following.

Well, I don't know how much access you have to various parts of the car, but typically, a lot of systems just count the injector on time. So the ECU will know how long the injector has been on (added up from all the injection events) and you add in a static flow number. These two things will get you the amount of fuel consumed. There will be some errors, especially at low pulsewidths, and with tailbiting. However, you could just flow the injectors at each of these cases, and provide some correction factor.

If you don't have this much access to the EFI system, then you can get a fuel flow meter that will give off a certain number of pulses for a certain quantity of fuel. This would probably be even more accurate then the aforementioned way I discussed. But you have to buy the sensor and hook it up.

[Greg Locock]

What do you mean by "the race rulebook forbids any telemetry between car and box " ?

ie, what box?

Can you have an accurate fuel gauge in the car that the driver can read to you over the radio?

Can you run laps at the circuit you will be using beforehand? If so you could build a database of fuel used against lap time and vehicle weight and maybe wind speed and direction.

If not then you need a predictive model. Do you have an sfc map of your engine? Do you have a speed/time trace for the circuit in question?

I used to do the fuel consumption estimates for a large British manufacturer when it was still done by calculator - that's when I first wrote a useful Fortran program! (1979) So I may be able to give you a clue or two.

[uri]

Hi all,

I am working with Aren with this calculation so I will try to clarify some questions arised here.

First of all, we can't have access to the ECU, at least during the race because there is not allowed data transmission between the car and the pit. So the only solution is to have a database of laptimes and fuel consumption, at least this is the easiest way we have right now.

Having the driver telling us how much fuel remains in the car is not a correct way, because what we will try to do is to match the refuelling with the driver change (it's a 24 hours race). Two hours of driving per driver is the maximum permitted, and with the refuelling a lot of time is lost because it is done on an specific area outside the pits. And you have to move a lot of people when you change the driver, so it can't be decided just the lap before.

We don't know yet the fuel consumption of the car, but with the first test, we saw that probably the fuel tank will last less than two hours, so for the strategy is very important to know when the car should go to refuelling.

I hope I make the things more clear. Then we go to the problem we are facing.

What I have thoght to make the estimations is to have a database with laps and fuel consumption. But to problems arise here. First we don't know if we would be able to access the ECU to read the fuel consumed on an specific lap, so we will have to make also estimations when measuring during tests.

Firstly, during the race we will have at least two types of laps. I mean, a normal lap, and safety lap (when the safety enters the track, it can take two or three times more to make the lap than with a normal lap, so the fuel consumption is reduced drastically).

So the idea is to take an average laptime and an average fuel consumption, and make the estimations. Of course, we will validate it on the track before the race. But with this system we will have to work with great error margins.

Another option is to go to a predictive model as mentioned, but I don't know if we can get access to the data needed. I will check it out.

What do you think? Which is the easy and correct way to do this?

Thanks for your help.

[Greg Locock]

Do you have access to the track before the event? If so you can build a database by measuring the fuel consumption accurately for a lap at each speed. It sounds as though you'll have to take the instrumentation off for the race itself, but that will still give you the best possible data.

This would look like

Laptime(seconds) Fuel used (litres)

89 19.7
90 18.6
91 17.4
...
180 3


OK?

However, that is not very interesting. Much more interesting is to have to do it from scratch, in which case you need to know the curve of fuel consumption against speed for your engine at full throttle, among other things. Do you have that? Do you have the layout of the circuit?

[Yelnats]

Of couse Greg a table like that would be thrown off it it rains and it would be very difficult to predict the effects of 3/4 throttle acceleration at low speeds, slower exit speeds from corners, increased drag when plowing through surface water etc. etc...

[Aren]

Greg, we have "limited" access to the track.

The race uses the GP layout of Circuit de Catalunya (the same used by F1), but that layout is only accessible to the 24 Hours race participants a couple of times two weeks prior to the race so, in the mean time, we have to test in what is called "the national" layout (you can check it out in this image - in front of grandstand E you have to turn right and that leads you to corner 7).

It is also quite possible that we test in other places, not just Circuit de Catalunya.

I think that getting the fuel consumption curve shouldn't be a problem, but just need to check that with Uri, the team manager and our technical sponsor.

If we're able to get that information (fuel consumption curve and track layout), what would you suggest us to do?

Yelnats, that's a very well known problem for us . Not having data transmission between car and pit creates this issue and that's why we want to have an "accurate" model for "normal" racing conditions. Having that, I can always model the software to make it "play safe".

Also, the racing conditions during the tests are going to be quite different than those during the race itself (the entry list for the race will include more than 70 cars) and as we aren't what can be called a "frontrunner" we've got to be ready to be lapped quite a few times (it's still quite early but I think we're going to be 10-12 seconds slower than the leading cars in a +2 minutes lap -fastest lap of the race last year was 2:01:xx if I'm not wrong).

As you can see, it's a quite difficult problem, but I hope we can find a way to sort it out

Cheers!

[RDV]

Check your PM, have several variations of spreadsheets in Excel...what is size of mailbox? IE capacity as for a 24 hr can be quite big.. regards RDV

[DEVO]

How about taking a completely non empirical data approach to this problem and use the rules to your advantage. Let the laughter begin.

Can you list the rules for pitting?

For example:

Can you have both a driver change and fuel within the same pit stop? You mention they are in 2 different areas (pit box, and fuel area)... but can a change of driver be located in the fuel area?

What is the maximum speed of a safety car during a full course yellow? How many laps are required to run under yellow (min of 3?)?

Can you pit during a full course yellow? If you are in the rear of the pack then pit towards the end of the yellow window.

How long does it take to refuel from pit entrance to pit exit?

How long does it take to pit for tires?

How long does it take to pit for driver change?

What are the lap speed with an 1 hour worth of fuel?

The reason I ask these question is that couldn't you come up with a fuel strategy or pit strategy that doesn't require a database of fuel consumption or depend on fuel consumption... rather base it on what the rules allow.

So it's a procedural approach to the problem rather then relying solely on empirical data. The strategy can change depending on a dummy light for your tank. For example a light goes off when half the tank is left... another light goes off when 2 laps of fuel left (or something like that).

So if a driver change is required (2 hours have passed and half tank light is on) do both a driver change and a fuel stop...

If yellow comes out (assuming fueling is allowed) then always top off when 1/2 tank light is on. That way you can change your stops dynamically rather then when the predictive programs tells you to refuel.

I'm sure if you think about this you can come up with a whole list of procedures to follow given the conditions at hand without ever having to know exactly how much fuel was consumed per lap.

Ok... so it's a far fetched approach to the problem but I can't stand it when a top F1 team goofs on pit stop strategy, when a typical fan (non technical) gets it right.

[Aren]

DEVO, I think both Uri and I gave you the idea that we want to completely rely on that piece of software to decide when to refuel and that's far from true, so I think I will go back to my English books and check once again the chapters related to "Speaking in public: how to make sure people gets the whole idea of what you're trying to explain"

The main aim of the software is to provide us some (let's hope) reliable information that can be used to decide what to do in "different" circunstamces (like the ones you've talked about).

From our first estimations it seems the car will be able to run for an hour and forty minutes with a full tank (of course we need to test more to get a more accurate figure) so, in normal race conditions, the driver will just drive and whenever he gets the instructions from the pits, he will go first to refuelling area and then to our box to do a driver change (and tyre change, system checks and so on), as it's not allowed to do it at the refueling area.

But if there's a safety car period (both refueling and driver change are permitted), we want to have some info so we can say: "There are X liters of fuel in the tank, the driver has been driving for Y minutes, and the safety has been on track for Z laps. Will he be able to reach his two hours time limit or should we do the refuel/driver change now?"

What you describe in your post is what we've been doing for the last three years and what we will continue doing but this time we want to have an additional tool that might help us.

Hope it's more clear now

[DEVO]

sounds good. I got the wrong impression. It was my mistake.

[Cappy]

One thing to consider is whether you want to consider fuel consumed as a function of time or fuel consumed as a function of distance.

Fuel consumption per distance will be more constant, because during safety laps you will consume fuel at a slower rate (per time) but it will take more time to cover equal distance. These tend to cancel eachother, but obviously the fuel consumption per distance will decrease.

Fuel consumption per time may prove more useful for calculating when to change drivers though.