How to get a good fuel reading (sloshing problem)

[MagnusBomba]

I plan to use a capacitive fuel probe to detect the level of fuel in commercial vehicles. The principal is simple, the probe provides a frequency (or voltage) that is proportionate to the level of fuel. Doing the math based on tank geometry is also OK.

But what I am wondering about is hot to get a proper reading with all the sloshing that is going on in the tank.

What is the best approach to designing an algorithm to do this?

Also, how long does it usually take the fuel tank to settle once the vehicle comes to a stop? (idea would be that rather than taking measurements while vehicle is moving, they could be taken only when standing).

Thanks

 

[MikeHalloran]

If you seriously intend to design fancy instrumentation that only provides a reading when the vehicle is stopped, you might as well save a bunch of money and just provide a dipstick.

If you time- average a decent number of samples over a period exceeding the natural slosh period, you shouldn't have to worry about, or detect, whether the vehicle is in motion.

Doing the measurement quasi- continuously also saves you from embarrassment over truncated and inaccurate readings associated with operators who shut down every system as soon as the vehicle comes to rest.






Mike Halloran
Pembroke Pines, FL, USA

 

[MagnusBomba]

Thanks Mike. Well, the idea to measure while stopping was, well, just an idea.

From what I have read so far, and I am still Googling, of course averaging is the way to go. But what type of average... there are many.

By the way, a related issue is that, the idea is not to provide data logging of continuous level monitoring. That would require way too much storage and anyway I don't see the benefit. My idea is to log regular intervals (say every 1 liter drop) and of course irregularities (like fueling or siphoning).

Any thoughts?

Thanks

 

[patprimmer]

If you put the transducer inside a baffled area just big enough to avoid interference and have quite small holes in the bottom of the baffle, you will seriously reduce the effect of slosh.

Regards
Pat
See FAQ731-376 for tips on use of eng-tips by professional engineers &

http://eng-tips.com/market.cfm

for site rules

 

[MagnusBomba]

Thanks patprimmer but I cannot modify the tank other than to make the hole for the probe to go in.

 

[MikeHalloran]

Modern vehicle body computers clearly do all that you propose, in their spare time, using resistive sensors. It's just scheduled as a task, and executed ... well, I'm guessing not more often than twice a second. And they do detect a fillup without intervention.

My questions:
Are capacitive sensors anywhere near linear with fuel level?
What dynamic range is the raw signal?
If the sensor uses charge/discharge time to measure capacitance, how often will its output change anyway? No point in sampling faster than that.



Mike Halloran
Pembroke Pines, FL, USA

 

[MagnusBomba]

Hi Mike. Capacitive fuel sensors are linear with the height of the tank. It is up to you to do the math to find the volume for a given level of liquid (and I am OK with that).
My concern is as I mentioned the sloshing of fuel in the tank as the vehicle moves and getting a good reading. I know/assume it amounts to averaging, but not sure exactly how.

 

[MikeHalloran]

So, the probe _assembly_ can't include a slightly leaky 'tank' around the probe itself?

No matter; just sample periodically, stuff the data into a ring buffer, and report the level as the AA of the last N samples, where N is the size of the ring buffer. You can work out the time delay/ phase error associated with this sort of reporting, and compare it with any specs you have for step function response, which you didn't mention.



Mike Halloran
Pembroke Pines, FL, USA

 

[MikeHalloran]

You might even program your chip (the cheapest PIC would do) to dynamically shrink the size of the ring buffer until the output started wiggling more than you'd like, then back off a tick. Put a perturbance in the program so it repeats the adjustment once in a while, and you'll have a conditioned sensor that responds to steady changes, e.g. tilt, in a reasonable time period, but damps out the slosh, all by itself.



Mike Halloran
Pembroke Pines, FL, USA

 

[GregLocock]

What happens if you park on a hill?

Why are you reinventing the wheel?

Cheers

Greg Locock

SIGlease see FAQ731-376 for tips on how to make the best use of Eng-Tips.

 

[patprimmer]

Insert slightly leaky blind tube into hole. Insert probe.

Are you an engineer or a student.

Regards
Pat
See FAQ731-376 for tips on use of eng-tips by professional engineers &

http://eng-tips.com/market.cfm

for site rules

 

[Zapster]

For commercial bulk storage tanks, they use stilling wells for this problem. A capacitance probe without a stilling well is a poor choice. Contact one of the many industrial manufacturers of this type of system and they can help you. I have specified both Rosemount and Magnatrol in the past and have found their technical support excellent. BTW, I think due to wetting, you may find any fix besides a stilling well will be marginal. Pat & Greg have given you good advice.

 

[WGJ]

The advice given for use of a 'leaky tube' to surround your capacitive sensor is both the quickest and easiest way to implement the measurement.
What size hole can you have in the tank?

As someone associated with fuel delivery and level sensing for motor vehicles, I worked with many 'clever' electronics and software engineers that spent plenty money and plenty time trying to devise the perfect indicator and never got much further than the performance of a baffled reservoir with a resistive sensor.

Bill

 

[thruthefence]

There's plenty of aircraft out there that plant the capacitance probe right in the tank. I know of none that sit in a baffled enclosure, ( not saying there aren't any, just never seen 'em. ) I can say, that when taxiing, and in flight on say, half tanks, I have never noted excursions ( such as we are trying to design out,here) on the fuel qty indicator.

 

[MikeHalloran]

I do notice excursions with vehicle dynamics in my car.
In fact, I distrust a level indicator that _doesn't_ do it.



Mike Halloran
Pembroke Pines, FL, USA

 

[patprimmer]

Aircraft tend to fly straight and level. Cars don't.

Regards
Pat
See FAQ731-376 for tips on use of eng-tips by professional engineers &

http://eng-tips.com/market.cfm

for site rules

 

[WGJ]

I don't think aircraft tanks are designed for cheap bulk storage either, unlike tankers and automotive applications where the onus is on as bigger vol as possible, but CHEAP!

Linearisation of any sensor is easy if there are no vehicle dynamics to be taken into account. As soon as the tank starts moving, all bets are off until the general relationship between a typical vehicle 'mission' and the reaction of the liquid bulk is studied.

Bill