Need some help finding direction for a new project

[sardonicus]

So I operate a poultry farm and we have 18 ton feed bins made of corrugated steel that we have to manually inventory the contents of each week. Typically we do this by banging on the sides of it with a long stick and listening for the hollow sound to identify how far up the bin we still have feed and making an estimation from there. It can be difficult to really hear the difference in sound and I am hoping to make a small robot that can climb the bin and identify the hollow area from the solid area more accurately than our current method.

The actual climbing part of this robot is easy but Im not sure where to even start looking for the other part. Telling the difference between hollow and solid. I would appreciate any thoughts on how to achieve this. Ideally my plan is for there to be a red and green light and the robot will display green for solid and red for hollow. As it travels up the bin taking this measurement, once it gets to the area of the bin with no feed in it - it will register hollow and the light will turn red.

 

[Bluejets]

A ball float similar to measurement of water with a pulley at the top and an indicator on the exterior would be a more practical approach.

 

[Nanren888]

A ball float similar to measurement of water with a pulley at the top and an indicator on the exterior would be a more practical approach.

A simple mechanical solution seems a really good option.
.
An alternative way would be maybe a level sensor, either permanantly mounted on the top, pointing downward, or on a selfie-stick.
There are lots of level-sensors for water & similar. Depending on your feed and geometry, such a sensor might easily do the job. But, you still need the display.
What is "feed"?

 

[sardonicus]

The feed is ground corn mostly. Im sure there are other things in it too but its mostly just ground corn. Has a powdery sand like consistency.

A simple mechanical solution seems a really good option.
.
An alternative way would be maybe a level sensor, either permanantly mounted on the top, pointing downward, or on a selfie-stick.
There are lots of level-sensors for water & similar. Depending on your feed and geometry, such a sensor might easily do the job. But, you still need the display.
What is "feed"?

 

[sardonicus]

I only hesitate to put anything actually in the bin because if something was to break it would get down into the augur at the bottom which is super annoying to work on. Also something like that would not be self resetting in this feed as it is in water and we dont handle feed deliveries ourselves - another company does that and the drivers that deliver the feed could not be counted on to reset it before the filled the bin.

A ball float similar to measurement of water with a pulley at the top and an indicator on the exterior would be a more practical approach.

 

[sardonicus]

The idea of a level sensor is interesting and would be a lot simpler to implement - what kind of sensor did you have in mind? The bins themselves are basically just cylinders with tapered sections at the top and bottom, but primarily the feed levels we are trying to estimate are just in the cylinder section.

An alternative way would be maybe a level sensor, either permanantly mounted on the top, pointing downward, or on a selfie-stick.
There are lots of level-sensors for water & similar. Depending on your feed and geometry, such a sensor might easily do the job. But, you still need the display.

 

[Nanren888]

If the bin is narrow, includes vibration, or otherwise self-levels the top of the feed, then measuring to any point on the surface will give an estimate.
Standard level sensor in air, often for water are ultrasonic and radar.
If your level is distinct, a standard laser distance measurement device as you'd pick up at the hardware store for DIY might be enough (either permanently mounted, or somewhat manually on a stick if you have a port that you can look into, as in top to feed level. Most have a button you need to push to get a reading. Some are bluetooth. Would need to explore what's available versus what you can handle operationally.
.
I've used the acconeer radar units. There are multiple sorts with differing ways to communicate the answers, hence differing extra stuff required to get the answer from it.. Ultrasonic units are also common. For these, radar and ultrasonic, would just have to check that they would detect the feed surface reliably. This does seem likely.

 

[sardonicus]

The bins are approximately 6 feet in diameter if I had to guess, they don't self level so the center of the bin is always lower than the outside since the augur at the bottom is centered. It might be possible to add vibration for the purpose of leveling. What is the range of the typical ultrasonic and radar sensors. Im sure given enough money radar range is crazy far but I'm hoping to not spend a fortune piddling around with this project. The bins are quite tall - I'd guess somewhere around 30 - 40 feet, though I would not have to be able to measure the total distance as the bin is rarely completely empty, and even if it were that low a manual estimation of the feed in it becomes increasingly easier the lower the feed level. So I suppose I would need something that could accommodate a range of about 15 to 25 feet.

A laser would probably detect the surface of the feed just fine but I might be concerned with the dust. Dust may actually be a problem for anything I choose - Im not sure. That will probably depend on how I am able to mount it. The underside of the bins lid seems like the most obvious choice but that will take some investigating.

 

[sardonicus]

I was just looking into those acconeer radar units, if im reading the specs correctly most of them will read distances between .5 meter and 10 meter which would be more than adequate for these bins I think.

 

[Nanren888]

Yes, acconeer will likely do it. They also offer lenses, which will change field of view & range: as field of view narrows, range increases, as more power into smaller beam.
They need programming and the result is as data of course, you still need to get the answer out somehow, eg a controller. The cost does add up somewhat if you need the unit & the demo board & something to connect to &/or some programming to set it up & retrieve the answer. Depends on your capabilities.

If you bins are metal sided, the radar will likely bounce nicely off the sides. Range will likely not be a problem. If the corrugations are horizontal, they might give interference patterns, so positioning, maybe to point away from the sides may be worth some experimentation.
Not sure what your feed dust is like, but while the acconeer units are low power, I've not seen anywhere saying the acconeer units are certified intrinsically safe, as in for use in potentially flammable atmospheres.

I've seen reference to radar being preferred over ultrasonic for cases where foam might exist on top of a liquid level. It might offer similar improvements for dust, but have no evidence to say that.
.
I guess if you are currently tapping the side of the bins, you are sensing level only at the sides with some sort of mental compensation for actual surface profile.
.
If you search, you will likely also find other radar & ultrasonic units off the shelf, potentially with included display options.

 

[sardonicus]

I appreciate all your input on this. The feed dust is officially flammable, though only when its airborne. the bin itself is not dusty all the time - only when its recently been loaded so I dont think it'll be a fire hazard. It is good to know the beam can focused with a lense. Im not worried about the range but you made an interesting point about the potential problems that might be caused by the corrugations.

 

[Harald Kapp]

I was just looking into those acconeer radar units,

Radar would have to be mounted on the inside on top looking down into the bin as radar will not pass from the outside through the steel walls.

With an ultrasonic sensor like e.g. this one you could sense from the outside of the bin.

Another project uses an Arduino and LoRaWan to remotely sense water level - applicable to food grains too, I think. But needs to be mounted inside the bin on top, too. The advantage here is that you can collect data on fill level remotely in your office.
Here's another project that detects fill level of a trash can - a "toy version" of your foodstock bins?

The disadvantage of any sensor mounted inside the bin is here:

the bin itself is not dusty all the time - only when its recently been loaded

This dust will settle on any sensor mounted within the bin and clog it. The sensor will have to be cleaned regularly, Here's where the outside sensor can play all its charm as it can be sealed and will not be in contact with the dust.

A completely different idea:
The empty part of the bin should act as a resonating body. When stimulated it will emit sound (effectively what you currently do by tapping the walls). The lower the fill level, the higher the air volume the lower the sound frequency. So the resonating frequency is an indirect indicator of the fill level. One could pick up the resonating sound from the bin with a microphone, feed it to a computer, analyze the spectrum and look for the resonance frequency. Making a few measurements with known fill levels should result in tables that allow to determine the fill level from the spectrum. Possibly each bin will require its own table, or not - depending on the accuracy you need and the different resonance curves of the different bins.
These measurements can be done with a standard laptop and microphone, a small hammer to trigger the resonance and free software (e.g. audacity) for analysis.

 

[Nanren888]

The empty part of the bin should act as a resonating body.

While either might work, you might find, as we did, that reverberation time is an even better indicator than resonant frequency if you try this method.

 

[Harald Kapp]

you might find, as we did, that reverberation time is an even better indicator than resonant frequency

Can be measured by the same method.

 

[sardonicus]

The idea of being able to do this from outside the bin is interesting though the one you linked I dont think would have the range I need. What makes that one different from other ultrasonic transducers that it would work outside and they would not?

 

[Harald Kapp]

I think this is an inspection tool that registers the difference between a blank piece of steel and a piece of steel with food grain behind by the different absorption of the ultrasonic sound. The instrument I linked will not sense the height of the grain in the bin from above - this is the method the other projects would use. The instrument mimics what you currently do manually: You need to trace the length of the bin from bottom to top (or vice versa) and you'll get an indication when there is a change from "bin and grain" to "bin without grain". So there is considerable manual effort still required to check each bin manually, only the operator's ears are replaced by the instrument.

 

[hevans1944]

An alternative, and usual, method to determine the amount of silo content is to weigh the grain silo. Ideally, this would be accomplished with load cells inserted under the supporting legs during silo construction, but clearly this accommodation was not made before the silos were installed. All is not lost however. Visit this link for an "aftermarket" solution to measuring silo weight using "bolt on" strain-gauge instrumentation. Or visit this page for more generalized information.

The links don't say how much the "L"-shaped transducers cost, or how many are needed for accurate grain weighing, but the principle and methods used are legitimate and have long been used for this purpose. This might even be a DIY project for someone who knows their way around small-signal analog electronics signal conditioning and takes the time and effort to learn how to purchase and attach strain gauges to the "L"-shaped metal transducers. I would imagine the "L" shape compensates for temperature variations that will cause the strain in the support legs of the silo to vary with ambient temperature as well as with the weight of the silo contents.

It may be possible to "get by" with attaching strain gauges to just one support leg of the silo, and not bothering with temperature compensation of the strain gauges, if a simple go/no-go indication of when it is time to re-fill the silo is adequate. One could also use a "look-up table" (experimentally determined) to compensate for ambient temperature variations. A simple thermistor temperature sensor, mounted near the strain gauges, would be necessary to measure the ambient temperature.

 

[sardonicus]

We have looked into a load cell type system called bintrac - its about 5000 dollars per bin to implement and we were hoping to not spend that much. A readymade system like this would be great and do a lot more than just give us feed inventory but its a huge investment that we can't make. Just spent 50k on a compact track loader so the coffers are pretty sparse at the moment.

 

[hevans1944]

Hmmm. I wonder what $5000 per bin buys these days? That sort of cost reminds me of the 1960s. This is a DIY site you are posting on, so why not try a few experiments to see if you can produce acceptable results and get the cost down to, say, $50 per bin?

Strain gauges are inexpensive (in many applications they are disposable) but they are also fragile until cemented in place, usually with a cyanoacrylate adhesive (super glue) and then coated with a protective rubber sealant. Wire connections to individual strain gauges require a small intermediate terminal strip, similar in size to the strain gauges, that serves as a strain relief when the data cable is attached. This terminal strip is mounted close to the strain gauges using cyanoacrylate adhesive and covered with a protective rubber sealant after the data cables are soldered to the terminals and thin wire connections are soldered to the strain gauges.

All this messing around with strain gauges is a lot of work, so it may be desirable to make your own "L"-shaped transducers for attachment to silo support legs. Or, you can purchase an inexpensive load cell for about ten bux, bolt it firmly to one leg of your silo, and then see how much its output changes as you empty a grain silo. Or wait until a silo is empty, and see how much the load cell output changes as the silo is filled.

This is where experimentation comes in because using strap-on (or bolt-on) load cells will not be nearly as sensitive as placing load cells on the ground directly under the silo supports, or attaching strain gauges directly to the silo support legs.

Some signal-conditioning electronics could be installed at the strain gauge sites, with shielded cables carrying amplified strain gauge signals to an Arduino with its 8-bit analog-to-digital converter interfaced to a laptop computer. Other microprocessors, as well as more bits of A-to-D conversion, are also possible. The important thing is to get the raw analog strain-gauge signals digitized in a format that is presentable to the laptop computer, which will do all the "heavy lifting" of displaying the results and selecting which silo weight to display.

Many other configurations are possible of course, but assuming a laptop is available the cost for readout instrumentation should be less than $100 for as many silos as you wish, since you only need to measure the weight of one silo at a time. Um... be prepared to write some software, but keep it simple. No need for fancy colored bar-graph displays. Also no need for a Wi-Fi network to get data from the silos... cable is inexpensive and much less sensitive to RF signal loss and interference.

If all this seems to be more than you care to get involved with, your climbing robot with a solenoid-activated striker and a sensitive microphone to "listen" for the response could be your best bet. How would that climbing puppy work? Magnetic treads to grip the side of the silo? Might need some fancy on-board audio processing software to make sense of the "echo" after banging on the side of the silo, but it appears to be a "do-able" solution.