Radiowaves to monitor water level

One of the things that I’ve spent some time researching has been using radiowaves (radar) to monitor water level in manholes. Sewers are challenging environments in which to monitor not only because of the caustic gases (like H2S), extremely humid air, possible surcharging, etc., but installation of in-situ depth/flow sensors is, by the very nature of the sensor, involved: requiring entry into the sewer. While there is no substitute (right now) for the data you can obtain when a sensor is physically contacting the water (and other stuff) in the sewer, what about using a non contact technology.

The non-contact technologies on tap are typically ultrasonic (sonar) and radar, with some other ideas like photogrammetry (disclosure: I previously worked on a photomonitoring project which had a patent filing last year), as well as light ranging (laser).

Radar seems a fitting application for sewer monitoring

The cool thing about radar is that it works through many materials. Therefore, you can house your radar and antenna (or antennas) in a box, protected from the elements, and it will happily carry along, sending radio waves down and listening for the return. Additionally, the dielectric permittivity of water is such that it is easily distinguishable from other objects (such as concrete), so it should return a strong characteristic echo. Radar seems a fitting application for sewer monitoring. Radar is not necessarily new to sewers (there are many companies that sell radars for this purpose) but they typically are mounted close to the water (similarly to ultrasonic sensors) OR they must be used in large open channels. We tried using one from many feet away, ranging to a narrow channel.

There are two main radar technologies on the market. Typically, you either will use a pulse radar or a frequency modulated continuous wave (FMCW) radar. FMCWs are in vogue and provide many of the benefits of a pulse-based radar but at a reduced cost (extremely fast circuitry typically used in pulse radar tend to be expensive). There are even many many open-sourcey and low cost projects available such as the following:

We set off to test this technology

Even though there are so many cool open source projects available, we bought one which is not open source, partly because it was ready to go out of the box. It is a pulse based radar that is based on the same technology used for the XeThru radar (which can also see through walls). More info on XeThru here. The radar chip manufacturer was able to get superfast circuitry in this at a reasonable cost! What does that do? It lets us chirp the emission and wait for a response, which is nice for battery power. So, we set off to test this technology. All I have to say is this thing works. It is great. We were able to range to tons of stuff and even test doppler radar. It does work to monitor water level as well, like monitoring changes in river level.

Multipath echoes are a killer

Two attempts were made to measure open channel water level in manholes. The first attempt, at site A, resulted in longer reported distances than we measured using a simple tape measure. However, followup testing in the lab showed distances that were very close to physical tape measurements. The second attempt, at site B, resulted in multiple reported distances, depending on the physical and digital configuration of deployment. The reported distance of the largest return was often farther than the entire depth of the manhole. How could this be? Multipath returns. Multipath echoes are a killer in this type of environment, where there is a smooth and likely damp cylindrical wall competing with the direct reflection from the desired area (the water surface).

Here is a view of site A
Manhole exampleYou can see that the bottom of the manhole is actually very flat and wet but the sides of the manhole are dry.




Here is a view of site B
Example manhole 2You can see that the side walls are slightly more damp here and there is a gradual slope to the bottom of the manhole. Also the water level is a lot higher. Both sides have manhole rungs.




The issue we are running into is that, even when removing clutter, a common approach to prevent things like ladder rungs from being returned as a desired distance, we were getting strong returns that were often as large or sometimes larger than measured. See the following image for an example.

Example Radar Return


What you see in the above image are two large peaks. The largest amplitude peak is actually the correct distance to the water surface, but another very large peak is actually farther than the entire depth of the manhole. This is a multipath return. Sometimes this return is even stronger in magnitude than the return directly from the water. This is perplexing because I would expect it to lose energy as it bounced. Unless the multipath returns are constructive, I am not sure why the magnitude is so high. V

Interestingly, as the water rises and falls, so do the multipath returns. Therefore, I cannot simply remove them as clutter because they continually change.

I want to focus the beam

The good news: radar works to range to water. Note that when I say radar, I am referring to a wide band of frequencies, from <1 GHz to 7 GHz.
The bad news: we need to be able to differentiate between what is water and what is something else.

What can we do? We can try to do some post processing of the return data. This would perhaps allow us to identify the first of the strong returns or identify the first inflection point but this is not robust in that I expect (although I could be wrong) that we will tend to run into slightly different return responses at each manhole. We don’t want to calibrate for each and every deployment.

What is left then is to find a way to focus the beam, creating an aperture that is much smaller than the radiation pattern created by our butterfly antenna. Someone is working on that right now for the chip we are using but you know what is really hard, creating an antenna that can focus radiowaves anywhere between 1 and 7 GHz. If we could focus the beam enough, we could mount the device directly over the water and hopefully diminish the multipath responses.

I want to focus the beam but I think there may be an even more interesting way of doing that, which is called phased array radar (also known as APAR or AESA, or PESA). More info on such radars is available on wikipedia. Greg Charvat’s phased array radars are versions of the PESA as far as I understand. It switches between multiple antennas over time (like a multiplexer) rather than using different waves on all of them simultaneously. Such as system could be used to create a 2D or even 3D view of the manhole and tuned on sight to identify the center of the channel in the radar image. This is beyond my expertise at the moment and when I inquired with XeThru via twitter, they said multiple antenna configurations are on the roadmap but nothing is available currently.

It’s all about the data in the end

If there is a product that is out there or that comes on the market which can do what I am talking about (hopefully at a reasonable cost) then I am all over that. While I love tinkering and development, what I am trying to do is help push the industry to get better at monitoring. If the industry gets better, then I will gladly recommend we use their equipment to grab our golden nuggets of data! It’s all about the data in the end. The tools are a means to an end.






This entry was posted in Environment, Fun_Projects, Tech and tagged . Bookmark the permalink.