Mizuno Airflow Calorimetry

Quote from JedRothwell

No, it isn't. If you measure 7 point and they give the same answer, you know for sure that every point between those 7 is also the same. Anything between point 1 and 2 will be the same. Or 3 and 4. That's 1/4 of the surface. How likely is it that points to the left and points below are different? I would say zero to none. In any case, the method I just used, working backwards from calibration numbers, shows that the flow rate was 4.1 m/s, just as he said. That could not be a coincidence. It is not likely that it was 4.1 m/s at all 7 points, yet if he had measured 2 more they might have been 5.0 and 3.2 (averaging 4.1). That is implausible.

I have just measured the flow and it varies considerably. It is turbulent. (The 60 cm tube approaches laminar flow) That may mean well-mixed. That does not mean homogenous.

Here they are again: 25 cm tube (top), and 60 cm tube (bottom), same fan as Mizuno, 65 mm ID.

.

Quote from Paradigmnoia

The 2019 CMNS paper claims the blower was run at 6.5 W (page 9 ) and yet the traverse plot (figure 9, page 6) shows a peak power evaluated of 5.5W at 5.0 m/s.

I believe that is a misprint. In any case, all of the data I have for calibrations and excess heat runs shows the blower at 3.8 W. Smack in the middle of the Fig. 4. traverse test. I am sure Mizuno did it that way on purpose.

That could refer to a previous blower. There were two or three defunct ones lying on a shelf when I visited. Those things seem to break often. He said he decided to spring for the best quality one he could find, this time. I guess he was tired or recalibrating every time the blower blew. Needless to say, you have to recalibrate and do the whole traverse song-and-dance when you change out the blower.

He went through an entire calorimeter, plastic box and all. It is sitting off in the corner, being scavenged for spare parts, I assume. Mizuno had shelves, drawers, boxes, and piles of "spare parts" (a.k.a. broken stuff) including 1950s vacuum tubes. I kid you not. As Ed Storms says, you can't do experiments without lots of broken stuff. He said that periodically a new civilian director comes into Los Alamos, and orders people to: "Clean up this mess! Get rid of that junk!" So they toss out the broken machines, and no one can do an experiment again. Experiments are made from broken machines. Along the same lines, pilots describe an airplane as "a collection of spare parts flying in formation."

The very uniform velocity profile shown in the paper just seems weird to me. Obviously it is not what you get after allowing the flow to develop. However close to the blower you would not expect it either, the blower turbulence and asymmetry makes it unlikely - Paradigmnoia's uneven results, for what is notionally the same setup, seem more plausible.

So: a mystery.

Quote from Paradigmnoia

I rotated the probe end CW/CCW maybe 20 degrees each time to get the highest reading (oriented properly), then took the average from that point for each location sampled.

Edit: Averaging the points from the 25 cm long tube, the average velocity was 4.99 m/s, while for the 60 cm long tube it was 4.77 m/s average.

That (I think) shows that with the turbulent flow not fully developed to its final state, and dominated by blower turbulence, we get higher (because more turbulent) measured average speed. Even though the actual average must be the same.You might expect the average from a short tube to be even higher. One thing though - are you sure you should be averaging points? Actually your points look as though they each hold roughly equal cross-sectional area - because the outer ones are closer together than the inner ones. Dis you calculate this?

But I don't understand how the Mizuno blower measured such a flat velocity profile. Maybe the anenometer used was very different from yours?

Quote from Paradigmnoia

It is turbulent. (The 60 cm tube approaches laminar flow)

Paradigmnoia... you may be confused about what laminar flow means.

which traverses do you consider laminar? vertical,diagonalor horizontal?

Quote from THHuxleynew

The very uniform velocity profile shown in the paper just seems weird to me. Obviously it is not what you get after allowing the flow to develop. However close to the blower you would not expect it either, the blower turbulence and asymmetry makes it unlikely - Paradigmnoia's uneven results, for what is notionally the same setup, seem more plausible.

So: a mystery.

That (I think) shows that with the turbulent flow not fully developed to its final state, and dominated by blower turbulence, we get higher (because more turbulent) measured average speed. Even though the actual average must be the same.You might expect the average from a short tube to be even higher. One thing though - are you sure you should be averaging points? Actually your points look as though they each hold roughly equal cross-sectional area - because the outer ones are closer together than the inner ones. Dis you calculate this?

But I don't understand how the Mizuno blower measured such a flat velocity profile. Maybe the anenometer used was very different from yours?

Display More

The readings bounce around a fair bit at each location, maybe +/- 0.2 m/s but generally a bit less. I didn't want to take the high number from each point, I just rotated the tip to ensure that it was reading as high as it would go (which looked like it was aimed straight into the flow), and let the probe sort it out after that. It is a lot harder to do a good traverse than it might seem. I have scratched measurement increments onto the probe end paint that continue those on the extension shaft (which zero at the hot wire center). Notably my first average of the many-point traverses for the 25 cm long tube was 5.12 m/s which compares favorably with 4.99 m/s the second time. I was aiming for equal areas but it won't be exact unless I build a rig to hold the anemometer in place for long periods.

Quote from RobertBryant

Paradigmnoia... you may be confused about what laminar flow means.

which traverses do you consider laminar? vertical,diagonalor horizontal?

I was considering the shape of the profile as a whole. Laminar flow in a full tube should have low(er) velocity along the edges and the peak flow rate in the center, with sort of a smooth parabola of flow rates with the peak in the center, no matter which traverse cross section angle is used. Perhaps that is wrong.

As you can see, even at a length 9.2x the tube diameter the velocity profile has not settled down. At 3.8 x D, the velocity profile is very messy, and clearly dependent on the fan projection characteristics of the fan outlet.

Just a little confirmation of physics and pardon me if I missed something here but the specific heat of air calculation you are using assumes dry air which I am positive is not the case. It would improve accuracy of the airflow calorimeter if RH was measured on the inlet and outlet, to get a better specific heat calculation. Has this been done? Even if both temp and RH were measured at one point together, one could calculate the actual specific heat with a psychrometric chart.

Quote from Paradigmnoia

Laminar flow in a full tube should have low(er) velocity along the edges and the peak flow rate in the center, with sort of a smooth parabola of flow rates

This is true,, but the traverses you have don't seem parabolic.

If you have hairy carpet on the walls you can laminarise the flow ..but you only have rough cardboard,

I think you mean well-mixed..rather than laminar .. perhaps there is better mixing at 60 cm.

http://article.sciencepublishi….j.ijmea.20150304.13.html

Quote from RobertBryant

This is true,, but the traverses you have don't seem parabolic.

If you have hairy carpet on the walls you can laminarise the flow ..but you only have rough cardboard,

I think you mean well-mixed..rather than laminar .. perhaps there is better mixing at 60 cm.

http://article.sciencepublishi….j.ijmea.20150304.13.html

Display More

I merely noted that the 60 cm tube is approaching laminar flow, in comparison to the 25 cm tube which is very asymmetrical.

Definitely even 60 cm is not enough to smooth things out using cardboard. I could stuff some nails or thick wire through the tube near the fan end and probably break up the asymetricity without drastically affecting the overall flow.

.

Quote from Daniel_G

Just a little confirmation of physics and pardon me if I missed something here but the specific heat of air calculation you are using assumes dry air which I am positive is not the case.

The difference is negligible. The Ohio U. data does not even show that. It does show variations with temperature. I ignored them, but Mizuno has them in the spreadsheet.

https://www.ohio.edu/mechanica…tables/air/air_Cp_Cv.html

Quote from Daniel_G

It would improve accuracy of the airflow calorimeter if RH was measured on the inlet and outlet, to get a better specific heat calculation.

I didn't bother to include that either, but Mizuno usually does.

I also did not bother to include heat losses from the walls. I assumed that 100% of the input power was recovered in the air flow. That is impossible, but for a first approximation at low power it works okay. How do we know it works okay? Look at the numbers I computed. I said these are STP, from the O.U. site. You can see that my result is remarkably close to Mizuno's.

My purpose was to show that the flow rate is uniform across time and across the orifice. That is what I did. I was not looking to compute the flow rate to 6 significant digits.

Quote from Paradigmnoia

I have just measured the flow and it varies considerably. It is turbulent.

You have that backwards. When the flow varies considerably, it is laminar. When it is turbulent the flow rate is uniform.

I get those two mixed up all the time!

There is no doubt your flow is laminar and Mizuno's is turbulent. I am sure you are both making the measurements correctly. Because after all, it isn't that hard to make them. Modern electronic instruments are hard to fool. Let us take these readings at face value. Your physical setup and equipment different for some reason. So what? Who cares? Make it the same and move on.

You cannot do calorimetry with the configuration you now have. If this thread is about "Airflow Calorimetry" then make it right and do air flow calorimetry. Don't fret about a trivial problem that you can fix easily, probably by smooshing the cardboard tube down to make it into a Venturi. After you see a uniform flow rate, carry on!

Quote from JedRothwell

You have that backwards. When the flow varies considerably, it is laminar. When it is turbulent the flow rate is uniform.

I get those two mixed up all the time!

There is no doubt your flow is laminar and Mizuno's is turbulent. I am sure you are both making the measurements correctly. Because after all, it isn't that hard to make them. Modern electronic instruments are hard to fool. Let us take these readings at face value. Your physical setup and equipment different for some reason. So what? Who cares? Make it the same and move on.

You cannot do calorimetry with the configuration you now have. If this thread is about "Airflow Calorimetry" then make it right and do air flow calorimetry. Don't fret about a trivial problem that you can fix easily, probably by smooshing the cardboard tube down to make it into a Venturi. After you see a uniform flow rate, carry on!

Display More

It is possible that Mizuno only showed the 7 points in the diagram because plotting many more traverse points would just make an ugly blob on the graph, but at this point I don't believe that the 7 points, (if those are the only ones), are sufficient to characterize the velocity profile in a tube that close to the fan, unless something was done intentionally to mix the uneven air flow profile. Sorry, but my data and the experience of many HVAC companies (which my data seems to agree with) seem to disagree with what is reported. It does not mean that the velocity numbers used by Mizuno are wrong. It just seems like some important detail is missing that would help it all make more sense.

Does Mizuno squash his tube into a venturi? Does he install some rings of some sort inside to keep the paper tube from getting too squashed?

I am certain that the temperature of the air in the outlet tube is well-mixed. The air basically goes through a blender before being ejected from the fan.

I can fix the velocity profile, but that is not the point at this moment. I will get the calorimeter working. However I am off for another trip for another 10 days or so, so it won't happen soon.

Quote from Paradigmnoia

I merely noted that the 60 cm tube is approaching laminar flow,

It isn't... it may be slightly more uniform than at 25 cm,, make sure the cardboard is smooth and flexible

vibrations in the walls might affect the profile,, who knows.. perhaps a Venturi a la Rothwell might get it uniform,,

plaudits for your persistence,

at least you have established that there is no zero velocity annulus which was postulated to cause 36% error in the anemometry..

Quote from JedRothwell

You have that backwards. When the flow varies considerably, it is laminar. When it is turbulent the flow rate is uniform.

To be a little more specific, I expect you have a laminar flow with moving streamlines. The fluid is not mixed, so layers of it move faster than other layers. Then they jump around. This happens with badly designed water flow calorimeters. You move the temperature sensor a little, or tap on it, and the answer changes. That's not an instrument problem. The temperature really is different in one part of the pipe compared to another. tapping disrupts the streamlines. Leave it alone and after a while the temperature will change abruptly as the streamlines decide to move. To fix that problem with a water flow calorimeter, you install an inline mixer or Venturi. Since you have the same problem with your air flow fluid, fix it. Stir it.

After you fix it, take 7 measurements in the same pattern as Mizuno's. You will see they are the same. Take 7 more, or 7 times 7. They will all be the same, I guarantee. As I said, with your present configuration and the data you already have, when you look at the 7 positions Mizuno used, you can see that your flow rates are not uniform. You have confirmed that your instrument is different from his. It ain't working. You don't need to measure all the other points. Fix it, make those 7 the same, and the others will fall in line.

Quote from Paradigmnoia

It is possible that Mizuno only showed the 7 points in the diagram because plotting many more traverse points would just make an ugly blob on the graph, but at this point I don't believe that the 7 points, (if those are the only ones), are sufficient to characterize the velocity profile in a tube that close to the fan

You have no reason to doubt that. Those same 7 points in your graph are indisputable proof that your flow rates are not uniform. If you measured only those 7, no one would argue that it was actually uniform and you happened to pick the only points in the whole circle that are off. That would be ridiculous. So, if you can prove that with only 7 points, why do you doubt that he proves the opposite? It makes no sense.

Fix the problem. Then measure those 7 points, or any other 7, or 77 if you like. They will all be the same.

IN SHORT, you need to replicate his results. You need to show those 7 points being the same. THEN maybe you can demonstrate that 7 points are not sufficient. I don't think you will succeed, but you cannot even start now, and you have no basis to judge. Make those points the same first, replicate his results with those 7 points first, and then find out if it is sufficient, by measuring other points.

Quote from JedRothwell

You have no reason to doubt that. Those same 7 points in your graph are indisputable proof that your flow rates are not uniform. If you measured only those 7, no one would argue that it was actually uniform and you happened to pick the only points in the whole circle that are off. That would be ridiculous. So, if you can prove that with only 7 points, why do you doubt that he proves the opposite? It makes no sense.

Fix the problem. Then measure those 7 points, or any other 7, or 77 if you like. They will all be the same.

IN SHORT, you need to replicate his results. You need to show those 7 points being the same. THEN maybe you can demonstrate that 7 points are not sufficient. I don't think you will succeed, but you cannot even start now, and you have no basis to judge. Make those points the same first, replicate his results with those 7 points first, and then find out if it is sufficient, by measuring other points.

I think we will be better served if someone else attaches a 65 of 66 mm ID 25 cm long (or 20 cm long or whatever it is) tube to the same type of fan as Mizuno's, does a bunch of traverses with an anemometer, and either agrees with my results or agrees with Mizuno's.

Then we will see where to go looking for answers.

Quote from Paradigmnoia

I think we will be better served if someone else attaches a 65 of 66 mm ID 25 cm long (or 20 cm long or whatever it is) tube to the same type of fan as Mizuno's, does a bunch of traverses with an anemometer, and either agrees with my results or agrees with Mizuno's.

That would not prove anything! Your flow is laminar; his is turbulent. If the new results agree with Mizuno's, that person has a turbulent flow. If the results agree with yours, the flow is laminar. These are two physically different systems. That's all there is to it.

You have asserted that 7 measurements are not enough. That can be tested. First, make sure the flow is turbulent, by measuring several points. These 7, or some other 7, or 40, or whatever. After you confirm the flow is turbulent and the flow rate is the same everywhere, then go back and measure the 7 points Mizuno measured. Put them in a spreadsheet and look for significant differences, and do statistical tests to see if this sample is sufficient to characterize the entire orifice surface. If it is, Mizuno was right, and 7 points are enough. If it is not, and if more points show better R-squared values (or some other test), then you are right. But not by much, I guarantee. I proved that above, by running the equations the other way, and deriving the flow rate from the input power and temperatures. It cannot be a coincidence that my answer agrees with Mizuno's so closely. If there were any significant differences in the flow rate from one place to another, and if those 7 points were not sufficient, he would get the wrong answer. His answer would not agree with mine. That is not the case, so you will not find a significant problem, no matter how hard you look.

Quote from Paradigmnoia

I am certain that the temperature of the air in the outlet tube is well-mixed. The air basically goes through a blender before being ejected from the fan.

It can't be! If it were well-mixed the flow rate would be the same everywhere you measure it. Some of the streamlines of air are moving faster than others. They cannot be mixed together if they are moving at different rates.

Also, when you measure differences by location, those difference persist. Right? Don't they? That's the usual situation. That means a streamline is forming and sticking around for a while. Not mixing.

If the numbers changed from second to second, and yet there were large differences in the air flow rate, that would be mind-boggling. That would take Maxwell's demons or something.

Quote from JedRothwell

It can't be! If it were well-mixed the flow rate would be the same everywhere you measure it. Some of the streamlines of air are moving faster than others. They cannot be mixed together if they are moving at different rates.

The temperature can be mixed perfectly while the air flow profile could be very asymmetrical.

What do you think is happening when ambient air goes through the fan? The air temperature separates into cold and hot layers preferentially with velocity?

The streamlines going into the fan intake, which could have uneven heating, get scrambled severely in the squirrel cage, then exits in a velocity profile according to the fan design characteristics.

(I don't dispute that keeping the air mixing helps ensure good air temperature mixing)