Airflow on/around quadcopter drone

[Daren Zomerman]

Hello,

I'm a journalist interested in using an air-quality sensor on a drone, and I'm wondering how the airflow will affect the readings. This would be assuming that I will use a DJ Phantom 2 quadcopter with a GoPro underneath, the drone specs can be found at this link. For the project, we will compare results using a drone along with a weather balloon — just to be sure that it is actually feasible, but I thought I would come here first, as this community has knowledge that I do not.

So I have a few questions:

What would be the ideal place to position an air-quality sensor?
Could I use a formula to correct for the difference in results?
What factors do I have to consider?
Is there a realistic way to come up with reliable results?

Thank you,
Daren

 

[mfb]

What does the "air-quality" sensor measure (and how)?

 

[Daren Zomerman]

What does the "air-quality" sensor measure (and how)?

It's measuring particulate matter — pm10 and pm2.5. I'm not sure how it does it, though. Here is the information from http://www.tzoa.com/about-tzoa/.

 

[mfb]

I guess testing is the best way.

 

[JBA]

1. If you place the collection point below the rotors' then it will be subjected to turbulence that effect the particulate count. Alternatively, by placing the air collection point above the the rotors' height you may be able to collect a sample from a more linear minimally disturbed air stream. Unfortunately, in either case the air stream due to the air flow required to support the copter you are still going to have a relatively high velocity that can effect the particulates' distribution. Additionally, unlike a floating balloon, unless it is tethered, that can move laterally in concert with the wind, your hovering, or moving copter will have a lateral air flow across it that, again, could effect the particulate distribution of the collected air.

2. I seriously doubt an equation could correct for any error due the amount of unidentifable variations that can be introduced by turbulent air.

3. See no. 1

4. The balloon method already does that.

Regardless of all of the above, your method's accuracy can be determined by your planned balloon comparison test; so, if the resources are available, address all of the issues presented on this thread as best as you can and give it a try.

 

[mfb]

Two ideas to minimize the effect of air flow:
- put the sensor in the middle, preferably on top.
- attach a cable to the drone to have the sensor several meters below the drone

The drone can also follow the wind if necessary.

2. I seriously doubt an equation could correct for any error due the amount of unidentifable variations that can be introduced by turbulent air.

The comparison to the balloon or the sensor far away from the drone could reveal some pattern.

 

[Jupiter5]

What type of sensor are you using?

 

[Daren Zomerman]

I'm using this one:
http://www.tzoa.com/#homepage

However, if you think there is a better one, I'm all for looking at other options as well.

 

[Mech_Engineer]

It seems to me the drone will just mix the air around it; as long as its an electric drone and the sensor is exposed to the free air around the drone I'd be surprised if it affected your readings at all. The main limiting factors will be that the drone will be severely limited in operating altitude and flight time compared to a weather balloon.

 

[berkeman]

I'd be inclined to put the sensor on a vertical boom about a meter above the drone to get in more quiet air, and also put a wind direction and velocity sensor on the boom as well. That way you can follow the wind when you want to take that variable out of the equation...

 

[Daren Zomerman]

I'd be inclined to put the sensor on a vertical boom about a meter above the drone to get in more quiet air, and also put a wind direction and velocity sensor on the boom as well. That way you can follow the wind when you want to take that variable out of the equation...

How much would that cost? And how much weight would it add to the mix?

 

[berkeman]

How much would that cost? And how much weight would it add to the mix?

A quick search looks like you can get something for around $50, and it should be pretty light weight. Heck, as long as you have a visual on the drone, you can just put a red ribbon at the top of the boom with the sensor. Fly the drone to keep the ribbon vertical...

Here is a Google Images search on wind sensors:

https://www.google.com/search?q=win...ChMI8fTI2vGfyAIVCUmICh30ogvp&biw=1164&bih=804

.

 

[rcgldr]

I'd be inclined to put the sensor on a vertical boom about a meter above the drone ...

I'm wondering about the effect on stability of the drone, since the boom would increase angular inertia about the pitch and roll axis, and could also create an adverse torque when moving laterally (due to the relative crosswind).

 

[berkeman]

I'm wondering about the effect on stability of the drone, since the boom would increase angular inertia about the pitch and roll axis.

Excellent point, as usual.

So a good addition would be a symmetric boom on the underside of the drone. Better yet (and more practical for takeoff and landing the drone) would be a 3-element pyramidal structure below the drone that counterweights the top single boom, and forms a triangular landing leg structure. This could be a pretty cool looking drone...

 

[Mech_Engineer]

Is any of that really necessary? The sensor is detecting air particulates nearby, the drone is in the air you're trying to sample. Case closed? The only possibility of a problem is if for whatever reason the prop wash from the drone is somehow incompatible with the sensor?

 

[rcgldr]

So a good addition would be a symmetric boom on the underside of the drone.

That would further increase the angular inertia about the roll and pitch axis. As long as the center of mass is below the plane of the rotors, that should be good enough for stability, and if the boom is light, then any added mass would not have to be that large or located that far below the the plane of the rotors.

The main issue would be potential adverse torque due to aerodynamic drag during lateral movement (relative crosswind). Having a boom extend below would compensate, but the increased angular inertia could be an issue. All but the lightest of cameras are usually gimbal mounted, effectively acting as a bearing (when holding the camera's orientation fixed), reducing the effective angular inertia somewhat.

An alternative would be to attach a horizontal boom (carbon fiber tube) to the under body of the drone, with both ends extending outwards just enough to be past the induced downwash, with a sensor on one end and a dummy load on the other end. The boom would have to be mounted so that it doesn't interfere with the camcorder. This should involve less of an increase in angular inertia, and also keep the center of mass below the plane of the rotors.

The other issue is if the drone can be set to remain level and just drift with the wind to avoid cross wind issues. Trying to use ribbons and have the pilot compensate could be difficult.

 

[Mech_Engineer]

Again, it seems to me there shouldn't be any problem in directly sampling the particulates in the turbulent prop wash air?

 

[JBA]

Because of the CG shjift issue, as an alternative to placing the collection point above the copter, I was among those who suggested it, what about extending (two diametrically opposed for balance) horzontal tubes from the copter with cup ends joining under the copter at a common collection chamber with the sensor enclosed and a bottom opening so the downwash from the copter blades will draw the undisturbed air sampling thru the tubes and past the sensor.

 

[Mech_Engineer]

Why bother? Why not just sample directly in the prop's down wash??

 

[JBA]

Because the downwash area is turbulent. In the extended arms arrangement the turbulence will be downstream of the sensor and away from the collectors sampling regions

 

[Mech_Engineer]

What's wrong with sampling in turbulent flow? I don't see anything obvious from the sensor's description that makes me think it can't sample in turbulent flow? Just sitting on your belt on a windy day would be turbulent flow...

 

[rcgldr]

I'm wondering if particle density (mass per unit volume), is different than the air, then the acceleration of the induced wash through the props could affect the ppm readings, but I'm also wondering how those particles end up suspended in the air if their density is significantly different.

 

[JBA]

Mech-Engr: You would have to be walking in an extremely strong wind to even begin to simulate the flow velocity below a flying quadcopter, particularly a large one. For example, have you ever observed or experienced the downdraft from a helicopter when it is idling on the ground.

rcgldr: The best example of the basis of my concern is that air turbulence can tend to cause the concentration of particulates in some regions of flow similar to the functioning of a vortex separator.
Another, probably much better example, is the smoke patterns observed in windtunnel testing. In this case the high velocity laminar streamlines tend to concentrate the smoke particulates by creating a low presssure boundry that draws them into and along the flow stream. I just wish I had thought of this obvious corollary sooner. I should have, because I have performed extensive low pressure noncritical, ie subsonic flow testing for low pressure relief valve body internal flow patterns and efficiency.

 

[Nidum]

See if there is any guidance in the pictures on this site :

http://www3.epa.gov/ttnamti1/files/2014conference/wedngambaxter.pdf

 

[JBA]

Absolutely, note the location of the sensor box in the overhead pictures.

 

[Mech_Engineer]

Mech-Engr: You would have to be walking in an extremely strong wind to even begin to simulate the flow velocity below a flying quadcopter, particularly a large one. For example, have you ever observed or experienced the downdraft from a helicopter when it is idling on the ground.

My point isn't that the wind speed would be the same as a quad copter, but instead that wind blowing past you on the ground probably transitions to turbulent flow at relatively low speeds. With this being the case, even wearing the sensor on your belt in the wind is still sampling in "turbulent flow", which you claim will effect the measured result. So I will ask again, what specifically is the problem with using this sensor to sample suspended particulates in turbulent wind conditions?

Estimate for Reynold's number for 1 m/s wind: 16,000 (transition is 10,000).

 

[JBA]

As for the specific issue of the sensor, when viewing its website, I saw no indication that its results are certified to be accurate. As a simple tool for indivduals to get a general idea of the pollutants around them then that is not an issue; and, if this quadcopter application is not intended for serious science, then this sensor and any other vagarities in its application may effect their results are fine.
My purpose in posting is simply to present the issues that I believe can cause possible errors in the accuracy and repeatability of the sampling. Beyond that it is up to the actual project team developing the unit as whether they decide that what I offer is of value based upon the evidence I present.

 

[mfb]

That would further increase the angular inertia about the roll and pitch axis. As long as the center of mass is below the plane of the rotors, that should be good enough for stability, and if the boom is light, then any added mass would not have to be that large or located that far below the the plane of the rotors.

I'm not sure about that. Without active control, thrust is always following the orientation. It should be metastable with every geometry in every orientation in the same way rockets are. We had a great thread about that topic a while ago but I don't find it any more.

 

[rcgldr]

As long as the center of mass is below the plane of the rotors, that should be good enough for stability ...

I'm not sure about that. Without active control, thrust is always following the orientation.

Updated my prior post, the main issue would be aerodynamic drag and related torque effects during lateral motion (relative crosswind).

The other suggestion made by others and myself in this thread would be to mount a horizontal boom below the main body of the drone, which would probably be simpler and also stable.

 

[Daren Zomerman]

True, but with the center of mass below the plane of the rotors, stability would be greater than with the center of mass above the plane of the rotors.

The other suggestion made by me and at least one other in this thread would be to mount a horizontal boom below the main body of the drone, which would probably be simpler and also stable.

We'd also be using this underneath
http://eye-mirror.com/cart/index.php?route=product/product&path=60&product_id=70

If mounted correctly, could this be used as a boom?

As for the specific issue of the sensor, when viewing its website, I saw no indication that its results are certified to be accurate. As a simple tool for individuals to get a general idea of the pollutants around them then that is not an issue; and, if this quadcopter application is not intended for serious science, then this sensor and any other vagarities in its application may affect their results are fine.
My purpose in posting is simply to present the issues that I believe can cause possible errors in the accuracy and repeatability of the sampling. Beyond that it is up to the actual project team developing the unit as whether they decide that what I offer is of value based upon the evidence I present.

You're absolutely right. I'm in contact with the developer, and I going to get as much information as possible from them before deciding to go with this instead of another, perhaps more expensive, proven device.

See if there is any guidance in the pictures on this site :

http://www3.epa.gov/ttnamti1/files/2014conference/wedngambaxter.pdf

This is fantastic -- would that box around the top give it enough air for the readings?

 

[mfb]

True, but with the center of mass below the plane of the rotors, stability would be greater than with the center of mass above the plane of the rotors.

I doubt that. It is not true for rockets, and quadrocopters are not that different from rockets in terms of passive stability.

 

[rcgldr]

True, but with the center of mass below the plane of the rotors, stability would be greater than with the center of mass above the plane of the rotors.

I doubt that. It is not true for rockets, and quadrocopters are not that different from rockets in terms of passive stability.

OK, my comment seems like the inverted pendulum fallacy for rockets. There are algorithms for quad rotors to balance inverted pendulums, ones that are bearing mounted. These rely on lateral movements to keep the inverted pendulum balanced. The same approach applies to a non-inverted pendulum, but the situation is more stable, since the pendulum would just swing back and forth even with no lateral correction by the quad rotor, and friction in bearing would dampen the oscillation.

With the pendulum directly connnected to the quad rotor, angular inertia is increased, and adverse torque could occur during lateral motion due to aerodynamic drag (due to the relative crosswind).

 

[mfb]

Torque needed to restore a vertical position is identical in both cases. The quadrocopter needs significantly more torque than usual, but again, that is common to both cases (above and below).
If you mount it as a pendulum which is free to rotate, things are different, but that's not the case here.

 

[rcgldr]

Torque needed to restore a vertical position is identical in both cases. The quadrocopter needs significantly more torque than usual, but again, that is common to both cases (above and below).

and more torque needed in order to lean for lateral movement. I updated my prior posts.

I was thinking more of the aerodynamic aspect and lateral stability in the case of a gusting crosswind or when trying to achieve and maintain lateral motion. In the case of a gusting crosswind, the inverted pendulum could cause the quad rotor to lean the "wrong" way (lean downwind), increasing the recovery time, while a non-inverted pendulum would cause the quad rotor to lean the "right" way (lean upwind, into the wind). In the case of lateral motion with a relative crosswind, the inverted pendulum could oppose the lean, while the non-inverted pendulum would increase the lean, some of which might be needed to overcome the lateral drag from the non-inverted pendulum.

I'm also wondering about any arrangement that would signficantly increase the angular momentum of a drone. Usually cameras are gimbal mounted, with separate gyro and servor drivers. When holding a camera orientation, the gimbal mount effectively acts as a bearing, reducing the impact of the camera and housing on the angular inertia perceived by the main part of the drone.

Getting back to the original post, and a later post with an article about showing a top mounted sensor on a quad rotor, this particular quad rotor uses relatively large rotors closely enough spaced that there may not be room to locate a sensor on top of the quad rotor.