Using Velostat (or other method) to accurately sense flow rate or water level

Hi everyone,

I'm working on a project where I need to deliver specific amounts of water from a reservoir tank through a boiler. The rate at which the boiler delivers this water is variable, but the end total must always be the same (within +/- 2%). Currently, I'm using a flow meter in line with this flow. It's essentially a hall-effect sensor reading a rotating magnet on a paddle wheel. It works well, however it's relatively expensive and (more importantly) restricts the flow significantly, increasing total delivery time.

I've been hunting for alternative solutions and was inspired by some projects utilizing Velostat, a material which changes its resistance based on pressure. I plan on mounting the material underneath the water reservoir and using it in a voltage divider to generate an analog value. My issue is that my pressure are low; the difference in force between a full and empty reservoir is only around 15N as the water column is only about 5" tall. If I want to be able to resolve the volume of water moved to some reasonable accuracy, say 2mL (a change of 0.02N in force), I don't know if my voltage divider will give me that kind of resolution.

So, I guess I have a multipronged question:

  • Does anyone have experience with Velostat and think that my application would/would not be fitting?
  • Are there any other indirect methods of continuous water volume measurement that are LOW COST? (i.e. ultrasonic is out for cost, float switches are out for lack of continuity)


I can't visualize the setup you are trying to describe. It sounds like you have a reservoir with a relatively small amount of water (a few litres?) and you want to move water from that into a boiler and measure accurately how much is moving. It also sounds like the reservoir is not being refilled while this is happening.

You mention measuring amounts as small as 2 milli-litres (I think).

I'm curious to know how you are measuring that right now, because some time ago I was interested in measuring small flow rates (500 - 2000 ml/hour) and couldn't find a solution.

If you are just drawing water from a tank can you not mount it on an electronic scales / loadcell that can output data to a PC or Arduino?


You are correct in your assumption about the setup. At max I have 1.5L and at minimum it will be nearly empty. A user fills the reservoir and I monitor the amount that has been transferred to the boiler (which then boils it and creates the flow).

The total volumes I am delivering range from 330 mL to 1400 mL (1.4L), thus I'm trying to keep accuracy to +/- 3mL (1% of smallest volume).

My flow meter looks similar to this:

There's a hall effect sensor monitoring the position of a paddle wheel inside the unit. As the paddle wheel spins, two oppositely-charged magnets pass by the sensor, opening and closing the switch. Digital read on that will give you a total count of half-spins (or full spins depending on how you read it) which you can then translate to volumes through simple calibration. It's a simple product, but it's relatively expensive ($10 USD, likely in the $1.50-$2 range in China) and requires that the flow be choked down to create a jet to turn the paddle wheel. This one ( is rated for 0.15L/min up to 1.5L/min which is just outside of your bounds. That's 9-90 L/hr where you're looking for .5-2 L/hr. In my experience with these, when flow rates drop below a critical level, your calibrations go to hell as water will bypass the paddle wheel without spinning it.

Unfortunately, all the "elegant" solutions like scales are just too pricy for this application. Thus the Velostat option which may end up being pennies (if it works).

The only flow meters I've used have worked over much higher ranges of volume flow.

If you want to measure really small flow rates, maybe you could tackle that indirectly by measuring how much water is needed to replenish the reservoir? You could do that using a calibrated injector with a regulated pressure supply. I don't know what sort of absolute pressure your system will be at, but this seems like the sort of thing that could potentially be inserted into the middle of a supply line if you wanted.

(The flow regulator on a medical drip works on a similar principle.)

Hello jr4284

I don't have any experience with Velostat but I think that the current flow meter that your using now will work the best for the small quantities that your using. You mentioned that the water bypass the paddle if the flow is very low. A possible work around to this would be if you could install your current flow meter in a vertical position so that the fluid travels upwards vertically. This way the water will have to flow over the paddle and the paddle will turn. Just a suggestion.

Thanks for the suggestion, Dirka. I have tried it with the flow meter in a vertical configuration directly beneath the reservoir (flow goes down) and the results were... Well, the plot shows it:

... compared to horizontal:

Hello Jr4284

You should not install it in vertical downward position as that won't guarantee that the pipe will remain full. Example see the installation guide for electromagnetic flow meters, the principle however remains the same regardless of what technology you use.

Edit: You should always have a full pipeline when measuring flow through a flow meter and the best way to be able to get this is to install it either in a vertical position so that liquid flows upwards or alternatively in a siphon position.

If you install it in a vertical position with the flow going downwards you might have liquid in the middle of the pipe but not at the sides etc.

The straight runs that are shown in the picture should also be observed. If you have a 90 Deg bent for example then you should have at least 3 diameter of straight pipeline before your meter and two after to be able to have good flow characteristics of the fluid going through your metering point.

Flowing fluid should show lower side pressure than static fluid, maybe you can make some kind of pitot tube and watch the level of water in it?

But 2 cc accuracy out of a 1.5 liter container?

I would wonder if a version of capacitive sensing would work, see how long it takes to charge a container of water (using a small current flow) with different amounts of water in it? That might need isolating though.

You can use a pressure sensor for your project
try here... NXP® Semiconductors Official Site | Home
Good luck