Gill fuel flow sensor failure modes

[DaveW]

There is an old adage amongst "measurement" people - never measure anything twice, it will only cause problems. It seems to very true in this case.....

Most transducers don't measure directly the properties of interest. In the present case, the measurement attempted appears to be an average velocity of fuel consumed, and not the average mass flow, as required by the regulation. Many opportunities for foot shooting there, as pointed out by others.

What puzzles me is why the regulation is there at all. Fuel consumption requirements are automatically take care of by initial fuel state for a race.

[Jef Patat]

It must have been mentioned a thousand times by now: it's about total fuel consumption over the total race distance AND about peak consumption which cannot go over 100kg/h, even not for a fraction of a split second. The second one cannot be checked by looking at the available/remaining fuel.

[DaveW]

It must have been mentioned a thousand times by now: it's about total fuel consumption over the total race distance AND about peak consumption which cannot go over 100kg/h, even not for a fraction of a split second. The second one cannot be checked by looking at the available/remaining fuel.

Neither, apparently, can the peak fuel consumption be estimated unequivocally by Gill transducers.

[Jef Patat]

It must have been mentioned a thousand times by now: it's about total fuel consumption over the total race distance AND about peak consumption which cannot go over 100kg/h, even not for a fraction of a split second. The second one cannot be checked by looking at the available/remaining fuel.

Neither, apparently, can the peak fuel consumption be estimated unequivocally by Gill transducers.

You didn't mention anything about that aspect. The peak consumption will be better estimated by a (bad) sensor than by looking at the total consumption after the race. Whether the Gill sensor is doing a good job is very hard for us to tell. It's word against word. At the moment it's a matter of who you believe most: FIA, RB, forum members, others. Personally I assume the sensor works as specified. It happens seldom that this is not the case. That's why my personally idea is that RB were trying to outsmart an drew the shortest straw this time.

[langwadt]

It must have been mentioned a thousand times by now: it's about total fuel consumption over the total race distance AND about peak consumption which cannot go over 100kg/h, even not for a fraction of a split second. The second one cannot be checked by looking at the available/remaining fuel.

exactly, it would be the equivalent of saying you are free to do 100mph in the city as long as you go slow now and then to
keep your average from A to B under 35

[ppj13]

The Gill ultrasonic fuel flow sensor did experience an acknowledged calibration shift over the course of RB's practice runs according to the FIA explanation of Ricciardo's disqualification. Given this...

The FIA acknowledged RBR SAID there was a calibration shift in the sensor. FIA never acknowledged it was true.

Even trusting his sensor better than RBR word, and for the sake of extra verification and clarity, FIA allowed RBR to replace the sensor. But that second (brand new) unit didn't work at all at time practice! (Conveniently for RBR). That alone is extremely odd.

So FIA asked RBR to revert to the only working sensor tested so far, and commanded an offset IN THE ENGINE MAP. Not in the sensor. And RBR didn't do it.

I guess they said "let's see how it goes" in the start, to back up during the race if in fact the sensor again read more than 100. But afterwards, in the heat of the battle and in the prospect of 2nd place, they decided to play strong against FIA.

As I posted in the engine thread, controlling instantaneous fuel flow with sensors is utopia, given the accuracy the teams and everybody will require as "fair". Only solution to me at this moment is duplicate or quadruplicate the sensor and accept the random variation as other random parts of the game. An extremely disappointing solution for everyone, but the only realistic one I think at the moment.

[DaveW]

You didn't mention anything about that aspect..

You might like to read my post again. Fluid velocity is not Fluid mass rate. If mass rate is required, then measure it directly (if you can), or think of another way of limiting mass rate, or free it up altogether.

[Jef Patat]

Sorry, I only meant to respond to the second part of your post.

Mass rate can be measured directly but AFAIK it always involves some kind of interference with the flow. This might be a serious disadvantage in some cases but sometimes it might be a better way of measureing. Ultrasonic has the advantage of not interfering with the flow. Even if mass rate is required, in a lot of cases volume flow is used. If you know the density of the fluid a conversion is straight forward. The sensor is also temperature compensated (ultrasonic wave speed depends on temperature) and an ultrasonic sensor compensates for flow friction (speed of flow is higher in the center than it is where the fluid touches the tube). I don't think the fuels of the several teams will diverse that much in density to be able to cause a problem there neither. IMHO flow rate to mass rate is simple maths in this case.

[DaveW]

... IMHO flow rate to mass rate is simple maths in this case.

I don't wish to disagree with you (in different circumstances I would argue your case). However, in this case it would appear that RB had the option of following the FIA's directive and (perhaps) unfairly squander performance, or have sufficient confidence in their own model to ignore the FIA's directive. In the event it cost Ricciardo valuable championship points. That should not depend on "simple maths", when the reality is far from simple.

[Jef Patat]

... IMHO flow rate to mass rate is simple maths in this case.

I don't wish to disagree with you (in different circumstances I would argue your case). However, in this case it would appear that RB had the option of following the FIA's directive and (perhaps) unfairly squander performance, or have sufficient confidence in their own model to ignore the FIA's directive. In the event it cost Ricciardo valuable championship points. That should not depend on "simple maths", when the reality is far from simple.

It's not always the case, far from indeed. That's why I mentioned 'in this case'.

[Kiril Varbanov]

Other teams were doubtful as well - http://www.autosport.com/news/report.php/id/112973

Note the use of "slightly" had to adjust, as opposed to the consistently in Dan R's case. Note: Vettel has not had the issue at all. To be honest, I don't know where the truth is, but I hope to get meaningful technical details once the case is reviewed.

[erikejw]

It seems FIA is wrong here if it is true they had to downgrade the flowrate to 96kg/hour.
According to the specifications the fuel meter have to live up to.

The device shall measure cumulative total flow to an accuracy of ‐ 1.0/+0% (<0 mean that the sensor reads lower than reality)

96kg means 4%.
I guess since Australia is a high fuel circuit FIA did not want cars to run out of fuel or coast too much.
The Mercedes powered cars had to fuel save alot even though there was a safety car and 2 formation laps.
So FIA deliberately forced calibration far off what is allowed. RB will probably argue this and then none of the fuel sensors during the race was up to spec so either disqualify everyone or noone.

If the fuel sensor was up to specification then RB has no case since the fuel sensor is the judge, as long as it runs inside the specs. RB would not take a chance on that. The Renault is probably good on fuel, the contrary to the high powered Mercs.

The temperature sensors are superslow. 4 seconds before it has to give a correct reading. By alternating the temperature in the flow they can fool the device in every acceleration phase. Quite neat. Ferrari should build a 10M$ fuel sensor research center instead of building the new wind tunnel. That will give them more bang/buck.

[dragosmp]

It seems to me there could be a more basic way of limiting mass flow by setting a restriction on the:
*max diameter tubing
*tank pressure
*fuel temperature

With the fixed tubing, primary pressure and fuel viscosity characteristics (dependent on temperature) one can compute pretty accurately the amount of fuel that could pass. I think temperature and pressure sensors are based on quite mature technology, as opposed to this sensor.

[Tim.Wright]

It must have been mentioned a thousand times by now: it's about total fuel consumption over the total race distance AND about peak consumption which cannot go over 100kg/h, even not for a fraction of a split second. The second one cannot be checked by looking at the available/remaining fuel.

Yes but if you integrate the FFM sensor and get a total fuel use which is significantly more than the amount of fuel which went in the car in the first place, then it follows that it is not holding its +0/-1% accuracy spec on flow rate.

If Red Bull show that the sensor implies that the motor took more fuel then they can physically fit in the tank, they've got a pretty good case.

[olefud]

Sorry, I only meant to respond to the second part of your post.

Mass rate can be measured directly but AFAIK it always involves some kind of interference with the flow. This might be a serious disadvantage in some cases but sometimes it might be a better way of measureing. Ultrasonic has the advantage of not interfering with the flow. Even if mass rate is required, in a lot of cases volume flow is used. If you know the density of the fluid a conversion is straight forward. The sensor is also temperature compensated (ultrasonic wave speed depends on temperature) and an ultrasonic sensor compensates for flow friction (speed of flow is higher in the center than it is where the fluid touches the tube). I don't think the fuels of the several teams will diverse that much in density to be able to cause a problem there neither. IMHO flow rate to mass rate is simple maths in this case.

Mass flow can be measured directly without impeding flow by Coriolis force flow meters. But G forces would probably compromise results. It would probably be simpler to limit fuel pressure through a standard orifice.