Hackaday Prize Entry: Watching Out For Forest Fires

Hackaday Prize entrant [Danie Copnradie] lives in South Africa where wildfires are a major problem. Every year, humans and animals are killed, crops are destroyed, and property is lost. The FireBreakNet project aims to deploy wireless environmental sensors that alert farmers, park rangers, and emergency personnel when fires break out.

According to [Danie], firefighting services are underfunded in South Africa, with farmers and their employees having to do a lot of the work involved in firefighting with their own equipment. Having access to a network of early warning sensors would allow for faster response times, saving money and lives.

The goals of the project include a low price, easy deployment, low power consumption, physical ruggedness, and scalability. Currently, [Danie] is testing Adafruit Feather as well as Texas Instruments LaunchPad for the brains of each node, taking readings from CO2 and temperature sensors, optical air quality sensors as well as optical flame sensors.

27 thoughts on “Hackaday Prize Entry: Watching Out For Forest Fires

    1. That’s way too many sensors. :P

      Some giant solar plane could do a pre-planned set of loops until a malfunction forces it to land for repair, then they just launch another to continue watching for heat.
      But, the price could be prohibitive.

      1. How much more (if at all) would a 100m radio-style tower with a high resolution FLIR camera and fancy zoom optics on top of it cost?
        This could easily cover a circle with a radius of at least 35km (on flat ground), more if on a hill.
        Yes, it’s expensive (so is a solar plane), but you need just one for ~3850km^2, only frequent maintenance is cleaning the lens assembly, the tower itself is needs a paintjob and guy-wires checked every few years…

        Power would have to be from local renewable sources and it would need a high-speed internet connection.

        1. Doesn’t need high-speed Internet. You just set the sensors to look for any temperature over 100C, to pick an arbitrary number. That means “FIRE!” so you send off a signal, essentially 1-bit. If you send it as an SMS you may as well throw in a bit more information.

  1. A simple radio circuit powered by a bimetallic strip may be all you need for many nodes. The radio is buried and only the thermoelectric generator and antenna are above ground. As a fire goes over the radio powers up and sends a coded message that gives it’s ID which has it’s location logged in a GIS database. The back end computer can the give you a real-time map of the fire-front and can eliminate false positives too. When the fire goes over the heat charges up a capacitor that then powers the radio for as long as there is enough heat, this duration of transmission is also meaningful. Frequency allocation and management is the hardest part of the design but how much bandwidth does it take to send a 20 bit number and checksum anyway? It may be possible to combine both above ground parts into one unit that is flexible enough to survive been stepped on or run over, it just bends and then springs back up again.

    I can’t think of anything more economical, robust and reusable, particularly if you then get them made in their thousands in China, just test them before you deploy them and make sure you don’t get cheated on the quality of materials as the units need a long life.

    1. Surely there’s some PID, standard americium ionization, or other detector you can just spin the sensitivity up on. Stick ’em on a tree or flagpole, one per ## square miles or so oughta provide a decent early detection in areas at risk.
      If there’s no farming around spacing can increase since there’s not as big an impact. Then other slower / lower res options should suffice.

      1. Nope, smoke goes upward due to convection, detectors only work in buildings because they are located on the ceiling where the hot air and smoke is trapped. IR is short range and directional, and requires the heat source to move across the “view” of two detectors at a high enough rate. You can’t rely on trees or poles as they are not always where you need them, and they may burn down. You need something that can survive a grass fire, and that will not get destroyed by curious animals, birds in particular. Where I am the parrots have been know to destroy anything that is not in an armoured box, just because they can, we call them pirates of the sky for good reason :-). You can’t assume that the sensor density is linked to habitation as the need to detect a fire early means doing so before it gets near people or farms. They move as fast as the wind in some cases due to embers spotting new fires ahead of the main front, so you need a significant safety buffer to have enough time to react effectively to a fire threat.

        1. When you say IR are you talking about domestic smoke detectors? I think Leithoa was talking about something like FLIR, bolometers, detecting heat from a distance. There’s surely some way of making them spot over distances. Even if you have to, say, put several “telescopes” in front of one, each pointing in a different direction. So you’d have “zones”, a small part of each area, covered by the telescopes. If that area burns, you assume surrounding ones are too. The telescopes give the greater sensitivity you might need.

          I’m pretty sure long-wave IR optics already exist to make the telescopes.

          As for local fauna, stick it on a metal pole, with some of those downward-facing spikes and barbed wire. Must be enough to keep most perpetrators away. Maybe stick a shocking voltage on the metal housing. Someone I know has a small horse paddock, all of the horses are separated just by a shock fence, that works from a small (6″ or so) solar panel. The “fence” is just plastic tape with metal thread sewn through it. Of course it only takes a couple of shocks before the horses learn to stay where they are, but still it’s a low-power operation.

          1. I dismissed the possibility of the FLIR type sensors and metal pole options because that simple does not scale well as the cost blow out very rapidly.

          2. Nope, I was talking about the photoelectric smoke detectors found in most commercial buildings and increasingly in homes. They work by the smoke from smoldering fires drifting across a beam path reflecting the light onto a sensor. So probably the IR ones ???? ???? dismissed.
            Dialing up the gain on these may give you false positives from dust storms but perhaps there’s a happy medium. Or build a redundency into the system so you need a couple hits to trigger a response.
            The other option is PIDs commonly used in gas sensors for high end chemistry instruments but also found in ruggedized first responder and construction tool kits.They use a UV lamp to ionize gas and across a high voltage gap that can be calibrated for a variety of gasses based on the current conducted.Detection levels can be ppm or ppb for some things. They run as low as 500USD but if your livelihood is on the line 10k in sensors is a good investment.
            Given the state of Fire science, analytical chemistry, and land management there’s got to be a table of the most common gasses for wood and grass fires. Terpenes probably wouldn’t be a bad place to start.

          3. Actually you could set up a perimeter type fire fence using directional FLIR sensors for a reduced cost, I just didn’t consider the linear option and was thinking in terms of 2D maps.

    2. Add a master node for each zone that has more sensors and solar power and have it use an *Pi with a USB digital TV dongle and gnu-radio for a SDR to monitor the dumb sensors and pass the data back to the base station with time stamps etc.. It could even send out a timing pulse so that you can coordinate the spread of the dumb sensor transmission timing to avoid collisions. They wake up on heat, listen for the timing signal then wait an offset for their group and transmit on the unique frequency for their position within that sub-group.

  2. @John Baichtal, I’ve never seen the surname Copnradie before – I think it should be Conradie (without the “p”). How would you like it if someone spelled your name John Bapichtal?

Leave a Reply

Please be kind and respectful to help make the comments section excellent. (Comment Policy)

This site uses Akismet to reduce spam. Learn how your comment data is processed.