Adventures in Measuring Airflow for A Solar Collector

Scott and I and others have been looking at a way to accurately determine the performance of several solar air collector absorber designs. 

One of the things you need to do this is an accurate way to measure airflow through the collector.  Since we are all cheap, it would be nice if the airflow measurement method was inexpensive.


This page describes some attempt over the last week to work out a way to measure collector airflow without spending a fortune. 

If you are trying to measure airflow on a budget, the methods on this page may be of some help (or not). 


If you have any thoughts on this, please let us know.


In addition to what is shown here, Scott is working on a method that uses the voltage generated by a PC type 12 VDC fan that is immersed in the airflow generates -- this looks interesting.

Back to the Solar Air Heating Collector test program home..


This page looks at the results of trying three different techniques to measure airflow:

- Hot Wire Anemometer...


- Fantech IRIS air flow measurement damper...


- A very simple pitot tube...


- The Big Bag method...


- The Contractor Garbage Bag method...




- Collector pressure drop measurements...

All of these tests were done on this "reference" collector, which is meant to be the baseline to compare other air heating collectors against...


For the latest air flow method I'm using see...., and the downspout collector test...



Setup for Flow Tests



A little hard to see all this, but

- Collector is on its side to the right -- we are looking at back of it.

- outlet duct is the one at the bottom (near the wheel)

- Fan is at the end of the 6 inch ducting coming out of the collector exit -- fan is pulling air through the collector.

- inlet duct is at the opposite top far end of collector.

- The inlet duct starts at the sledge hammer head, and goes through these steps:  5 inch duct to the Fantech IRIS damper (with the blue and red bits;  5 inch duct to a 5 to 6 inch duct transition fitting; 6 inch flex duct up to the actual collector inlet.

- The fittings right at the collector are 5 to 6 inch duct transition fittings, so the hole in the back of the collector is 5 inches in diameter, but transitions to 6 inches immediately as it leaves the collector.

There are ports to measure the duct static pressure in the outlet 6 inch galvanized duct section (pic below), and also one in the 6 inch galvanized duct portion of the inlet duct.

The pressure drop across the collector is measured as the difference between these to static ports.

The static port is basically just a hole in the duct wall that you can measure the static pressure right at the duct wall with.


Pressure differences are measured with this Dwyer Magnasense manometer.


It registers down to 0.001 inch of water, but is a bit shaky measuring less than 0.01 inch differences.



Fantech IRIS Damper Flow Measurements

This is a relatively inexpensive gadget sold by Fantech for the purpose of measuring and setting airflow.

Seems like a nice simple solution, but the results so far have been disappointing.

Maybe the you will see something wrong in the way I'm using it?



The blue handle adjusts how far closed the iris is.  You  can see the iris inside the duct in pic above.

The iris damper causes a drop in the duct static pressure which is proportional to the flow rate through the duct.

The pressure drop is measured at the two red static ports on either side of the damper by hooking up a manometer to the taps.

The further the close the iris, the greater the pressure drop.


The flow rate is:


        Flow Rate = K * sqrt(dP)


Where K is the K-Factor -- it just depends on where you set the iris -- you read it off the scale on the damper opposite where you place the blue handle.

dP is the pressure drop you measure across the damper in inches of water.


This is the setup for measuring the flow with the iris damper.

The manometer is showing a drop of 0.068 inches of water across the damper.


I believe that this was the measurement done with the blue handle set to a K-Factor of 100, so the flow rate is:


    Flow Rate = (100)*(sqrt(0.068)) = 26.1 cfm


I tried measuring the flow rate with the iris damper from all the way open (K = 280) down to its most closed position with K = 50.

As the damper goes from fully open, to its most closed position, the pressure drop across the damper increases, and the flow rate decreases due to the flow resistance of the damper.   But, the flow rate as calculated from the formula above stays nearly the same -- odd.


These are the values measured:


Iris Damper  
K factor dP inches Qiris cfm
280 0.007 23.4
150 0.02 21.2
100 0.067 25.9
80 0.108 26.3
50 0.2 22.4


This is really puzzling.

It shows very low flow rates -- only about 25 cfm -- much less than the other methods.

And, the flow rates do not vary much as the iris is closed, while the other methods show a good deal of velocity and flow rate reduction as the iris is closed.


So, don't really know what to make of this -- it appears that the damper is not even close on flow rates.  Not sure if there is something wrong in my setup, or something wrong with the damper, or ???   Any ideas??

I plan to email Fantech about this on Monday and see if they have some advice.

Update Dec 2: I tried a 2nd IRIS damper I had on hand with the same result.
I emailed Fantech support a couple days ago -- no reply yet.

The thing I like about using the iris dampers for flow rate measurement is that its nice and simple and not too expensive.


One thing I had not thought about is that by the time the damper is closed far enough to create a pressure drop that can be measured by a not to expensive manometer, its a fairly large pressure drop, and the fan has to be large enough to supply this drop.  The end result is that you might have to have one fan for testing (with higher pressure capability) and a different fan for regular use.



Hot Wire Anemometer

I have a Test Velocity Stick hot wire anemometer, which I used for this go at flow rate measurement.


For each setting of the iris damper (above), I also measured the duct velocity in the 6 inch section of duct with the hot wire.

The results are in the plot below.

They seem sensible to me.


To get the readings, I drilled a half inch hole in the duct wall, and I insert the hot wire probe into the hole so that it sits about in the center of the duct.

The flow in the pic below is left to right, and there is a several diameters stretch of straight duct to the left of the hot wire reading hole.

And, since the fan is pulling air, and is all the way on the other side of the collector, there is no fan induced swirl that I saw on previous measurement setups.

Hard to see in this pic, but the hot wire probe drops straight down into the duct from just to the left of where my hand is.

I mark the penetration depth with a piece of tape on the hot wire tube.



The problem that I have always had with this hot wire meter is that the readings vary a lot with time.  If you are reading a flow with an average velocity of (say) 400 fpm, you might get readings over 30 seconds that vary from 360 up to 440 fpm.  So, even if you average 10 or so readings, it does not give a lot of confidence that you have really nailed down the velocity closely. 

Not sure if this is a characteristic of the this particular hot wire meter (not enough filtering?), or its a common thing on hot wires?


The hot wire anemometers are also kind of pricey at about $200 for the cheaper ones.


I plan to call Testo and ask them about this high variability in readings thing, but even if they have a good suggestion, it seems like its too expensive a solution.


The other thing is that the hot wire gives a velocity in the center of the stream, and you have to factor that to account for the drop off in velocity as you go out towards the edge of the duct.  Our duct speeds are such that the flow is fully turbulent, so the velocity profile is pretty flat, and the correction is not to large, but its just another thing to contribute error.



Very Simple Pitot Tube

I had sort of dropped the pitot tube as a potential flow measurement method because our duct velocities generate very low Pitot pressures that would difficult to read without very fancy instruments.

But, given that the hot wire and the iris differed by so much, I wanted another method to check against the two.


I used the very very crude Pitot tube that comes with the Dwyer 460 air velocity kit.  Its basically just a 1/4 inch diameter  L shaped tube where the open end of the tube faces into the flow.   You measure the difference between the static port on the duct and the Pitot tube with the manometer, and then calculate the duct velocity from the pressure difference, and then the flow rate from the duct velocity.



The simple Pitot.   The L shaped end goes into the hole in the duct just below pointed into the flow at the center of the duct.
The matching static pressure comes from the static port just in front of the hole in the duct.

This is the same hole that was used for the hot wire measurement.


These Pitot tube measurements were pushing the capability of my manometer, and I'm sure the crude design of the tube did not help either, but the results actually agree pretty well with the hot wire, and are dramatically different from the iris damper.  So, I'd say that this tends to confirm that the iris damper measurements are just not working.



The graph just below shows how the 3 methods compare on their estimates of flow rate through the collector.



Vertical axis is flow in cfm

Horizontal axis is the pressure drop through the collector.

Red are flow rates as estimated with hot wire anemometer.

Blue are flow rates as estimated with the iris camper.

Green are flow rates estimated with the crude Pitot tube.



Big Bag Method

So, none of the methods above are very satisfying.


I got to wondering if the technique where you time how long it takes to fill a garbage bag would work better. 

One problem appears to be that for even a big garbage bag the fill time is quite short, and would be hard to measure accurately.


So, I'm having a go at building a bigger bag from poly sheeting and duct tape.


The bag is a cylinder, a bit over 3 ft in diameter, and about 10 ft long.

It has a volume in the area 60 cubic feet, so it should take almost a minute to fill.


If this bag approach works, then it might be easier just to use a length of this "layflat" poly ducting:

to build the bag -- 20 ft of 24 inch would do it.



The big bag -- its a 10 ft wide length of poly where the edges are duct taped together to make a cylinder.

Ends are closed off by clamping between two pieces of wood and duct tape.

The fill fitting is sandwiched together between two pieces of plywood.


I used the Gorilla brand duct tape -- have to say its pretty impressive stuff :)


First Air

Did the first inflation -- that was fun!


Measured a few things and did a couple rough timings.



It inflates nicely.  Takes about 48 seconds to full.

The poly and duct tape will take the full fan stagnation pressure of 1.2 inches of water with no apparent strain.

No leaks detectable with the hand test on any part of the bag.


Fully inflated -- left it this way for about 5 minutes -- no problems.



Static pressure goes up to 1.23 inches of water with full inflation.


The Pictures below show the shape of inflated bag:


Need to figure out how to calculate the volume.


Volume for now:

Diameter is (119 inches)/Pi = 38 inches.


Inflated length is 120 inches for wood strip to wood strip.

Assume that end shape effectively loses 10 inches of full diameter volume, so effective full diameter length = 120 -10*2 = 100 inches


Rough volume = (38^2)*(0.785)*(100)/1728 = 65.6 cf  very rough




Flow Rate

Using the rough volume estimate above, this says that with a 48 sec inflation, the airflow = 65.6 cf/ 48 sec = 1.37 cf per sec, or 82 cfm.

This agrees quite well with the hot wire calculated fill rate with the iris full open (the iris was out for the bag test).


Back Pressure:

One concern is that the filling balloon might cause some back pressure that would slow the flow rate down.


I hooked the manometer up to the static port near in the 6 inch duct just before the inflation fitting.

Inflation duct (collector exit duct) free in room                                               0.09 inches water


Duct connected, bag filling, bag loose, wood strips on floor                      0.09 inches water


Bag lifting wood strips off the floor and balloon nearing final shape         0.14 inches water


Bag full flow stagnated                                                                                     1.23 inches water


So, for the first half plus of the inflation, the bag causes no back pressure.


As the wood strips that terminate the end of the bag are being lifted off the floor, and bag is nearing its final shape, a back pressure of 0.05 inches of water ( 0.14 - 0.09) is present.  Based on the tests above, this might slow the inflation rate for the last part by as much as 10 cfm.


I later hung the bag from the ceiling to try to reduce the backpressure that occurs as the wood strips are lifted off the floor.


This works very nicely, and the ramp up in back pressure as the bag gets near full does not start until about 3 seconds from the bag being full.


Here is an avi  video of the inflation to download and play on your PC ...    (4mb)

In this breathtaking video, the Big Bag is hooked up to the exit duct of the collector.  It takes about 50 seconds to blow up, and you can see how easy/hard it is to judge the full inflation point -- seems pretty easy to me.

Yes -- I know my shop needs a good cleaning !



Extra Large Contractor Garbage Bags

This approach uses one or two off the shelf 55 gallon garbage bags instead of making an extra large bag.


I found a couple of extra large "contractor" garbage bags purchased some time back for some forgotten reason.  They are a nominal 55 gallons (7.35 cf).  I decided to just try one of these.


To put something together quickly, I just tied the bag onto the end of the duct, and then switched on the fan to start the timing.  This is not ideal in that the blower takes a little bit to ramp up to full speed, but it gives a pretty good idea.



55 gallon bag attached and ready to start inflation.




The time was about 9.3 seconds.

If the flow was 82 cfm, the time should have been (7.3 cf/65 cfm)/60 = 5.3 seconds.  The longer actual time was very likely due to the blower ramp up time.

So, some kind of fixture that could be attached to the bag, and then very quickly placed over the end of the duct with the fan already running would be needed -- something like what I have on the Big Bag.


Judging when the bag is full is easier than it would seem, since the bag sort snaps into the fully full position.

I would guess that one could get to better than half a second -- maybe even better.


So, this is less accurate than the Big Bag, but does seem feasible, and probably better than any of the instrument approaches.


If we could find some even larger bags -- maybe around a hundred gallons, I think this would work fine.


Again, there is the problem of arriving at an exact (or fairly exact) actual volume for the bags.


Here is an .avi video of the 55 gallon bag inflating to download and play on your PC...  (1 mb)

In this thrilling video, the bag is already hooked up to the collector outlet duct.  I plug in the fan to start the inflation and say "go".

You can get an idea of the timing and how easy the full inflation point is to see.


One advantage of the garbage bag approach is that it is reported that the actual volume of a 55 gallon bag is pretty close to 55 gallons, so for rough flow measurements, you could just use the nominal volume. 

Another approach that would be workable with the 55 gallon bags is to fill a bag with packing peanuts, and then pore the peanuts out into a box to measure the volume.  One could start by putting boxes or tubes into the bag first, and then use peanuts for the last part.


I think that with a little thought one could also do a pretty accurate volume calculation from dimensions -- maybe treat it as a series of ellipses, and then manually integrate over the ellipses?  It would be fairly easy to mark (say) 5 inch sections along the long axis of the bag, and measure the major and minor axes and perimeter of each section.



Dec 2 -- tried two of these 55 gallon bags together

Since one 55 gallon bag fills a little fast for accurate timing, I tried using a 4 inch plastic Tee fitting to hook two of the 55 gallon bags to.

This gives 110 gallons or 14.7 cf.

Fill time is about 16 seconds, which allows for pretty accurate timing.


The bags are attached to the straight through Tee fittings with nylon wire ties.

The worm gear type metal clamps would probably be better than wire ties, but I did not have any of that size.


So, this is a pretty easy solution for anyone to put together with easy to fine parts.



The two 55 gal bags attached to 4 inch plastic Tee with nylon wire ties.
The open leg of the Tee can be quickly placed over the open end of the 4 inch plastic pipe/duct.

Fully inflated.


Closeup on the Tee to bag attachment.


Jumbo Storage Bag

This is a large bag that our local moving and storage place had.

It is 60 inches by 108 inches.

It inflates to a cylinder with a diameter of (2*60)/ Pi = 38.2 inches = 3.18 ft


The inflated effective length is about (roughly) 84 inches ( 7 ft).


Volume = (3.18^2)*0.785*7 = 55.6 ft -- almost as large as the "Big Bag"\



Its made by Chateau Products, inc


Cost was $3.25 per bag with minimum order of 2.



This is the 60 by 108 inch bag tied off to a 4 inch plastic pipe coupling.


The coupling is quickly placed over the end of the 4 inch exit duct to start the fill process.


Back pressure during filling was minimal (0.01 to 0.02 inches of water) right up until the end.

The onset of the pressure rise at the end of the fill was very rapid, and one could use this to stop the timing if one had a sensitive manometer.  But, timing the end by just looking at the bag was also quite precise.


The full bag.


The 2 mil plastic is fairly lightweight, and this bag has to be treated with some care to avoid damage.


Collector Pressure Drop

I also measured the pressure drop through the collector for all the same iris settings where flow rates were measured.


The pressure drop was obtained by putting the tube from static duct pressure port on the inlet side of the collector to one port of the manometer and hooking the tube from the static port on the outlet end of the collector to the other manometer port.  Since both ports are located in 6 inch diameter sections of the duct, the velocity head should be the same, so the change in static pressure should be the actual pressure drop?


Here are all the data measurements with the collector pressure drop being the left column:


    Iris Damper   Hot Wire Anemometer Pitot Tube  
dPo   K factor dP inches Qiris cfm Veloc fpm Qhotwire cfm dP inches (4) Veloc fpm Qpitot cfm (5)
0.815 1 280 0.007 23.4 480 84.8 0.015 451.5 96.1
0.800 2 150 0.02 21.2 470 83.0      
0.788 3 100 0.067 25.9   0.0 0.012 404 86
0.762 4 80 0.108 26.3 370 65.4      
0.710 5 50 0.2 22.4 300 53.0 0.011 386.9 82.4



Note that the pressure drops are pretty high.  A lot of fans would have trouble producing the about 0.7 inches of water pressure drop that goes with getting the flow up to about 2 cfm per sqft of collector.


It makes me wonder if the 1 inch flow channel height is not a bit to shallow, or if the baffle arrangement is not to agressive?


Gary  November 28, 2010, Dec 2, 2010