Using a Mass Air Flow Sensor in a Furnace

Discussion in 'Burners and their construction' started by Melterskelter, Jan 9, 2022.

  1. Melterskelter

    Melterskelter Gold Banner Member

    This is the first of what will likely be several posts concerning my attempts to use a MAF Sensor to read air flow rate in my furnace.

    Introduction:
    Measurement of actual air flow rate in a burner setup would be useful in understanding air fuel mixture in one's own setup and would be a useful factor for comparison of burner designs among users. Very often we find ourselves posting comparing our personal burner setups with others but invariably lack objective measurements of comparison as rarely do users have objective measurements of fuel flow rates and never do they have objective measurements of air flow rates. We see folks using very large and powerful centrifugal blowers, leaf blowers of all makes and models, and shop vacs, etc. But no one has any measurements of air flow.

    It is well known that efficient fuel use should result from a stoichiometric ratio of air to fuel of roughly 14.7 with perhaps some modest variation from that ideal for practical reasons. But, so far as I know, no one, including me, knows what their air fuel ratios are. I tune my burner by slowly gained experience, but still don't know if I am close to an ideal mixture.

    Enter the automotive mass air flow sensor. https://en.wikipedia.org/wiki/Mass_flow_sensor

    Having an automotive MAF sensor can provide an objective set point from burn to burn. But to learn actual air flow rates one has to calibrate the sensor against a known air flow rate as the sensor signal outputs are well known to automobile manufacturers, but how those signal values correlate with flow rates is proprietary information so far as I know. (If someone knows of a sensor voltage vs flow rate table for a given sensor PLEASE post that info here.)

    Here follows my plan for attempting to correlate the voltage signal for a MAF I purchased for about 25 dollars to a known air flow rate through that sensor.

    Planned provision of a measured air flow rate through my sensor.
    Available to me is a compressor with a 150PSI maximum pressure 80 gallon tank. It has a pressure gauge. I figure that I can calculate the total volume of air that exits the tank by comparing starting pressure with ending pressure when the motor is turned off as the air is vented. Allowing that air to pass through my sensor will allow calculation of air flow rate and thus correlation with signal voltage. It is true that I will need to hold the flow rate constant despite changes in tank pressure. If I use an air pressure regulator, that should be possible. Much remains to be done to determine how practical this may actually be.

    Here follows some of my reasoning and calculation regarding this process.

    Calculation of tank pressure change needed to calibrate MAF Sensor reading at roughly my furnace burn-rate air consumption using stoichiometric ratio diesel fuel/air approximation.

    Calibration does not require that my burn rate equals the consumption rate but does give me a ball-park range in which I want to measure air flow. This air flow rate is in the same range as the automotive MAF would be expected to operate in a vehicle traveling 70 MPH and consuming 2 to 3 gallons of diesel per hour or 3 x 3.7 (liter per gallon)= 11 liters per hour or .19 liters per minute which is my optimized burn rate in my furnace. That means the sensor will be reading in the middle of the range for which it is optimized.

    Now let's calculate the volume of air needed to burn .19L of fuel

    Stoichiometric ratio of (Diesel fuel)/(air) for total combustion is roughly 14.5
    1 liter of diesel weighs 850gm.
    .19L fuel weighs .19 x 850gm= 161gm. So, I burn 161gm of fuel per min. requiring 161 x 14.5 = 2334gm air per minute
    Air weighs 36gm per cu ft.
    So, I consume about 2334/36 cu ft per min air = 64 cu ft per min

    Now let's calculate the pressure change in my air compressor 80gal tank that equals 64 cu ft as that will allow me to blow off a measured amount of air and that will allow correlation of the signal voltage produced by the sensor at a given flow rate.
    At 150 PSI, the tank contains (150/14.6) x 80 821 gallons of air gallons of air. [Note: 1 ATM = 14.6 PSI]
    1 gallon is .16 cu ft. So we need to blow off 64/.16 = 400 gallons of air.

    The pressure of the tank at 821-400=421 gallons will be 150 x (421/821) = 77 pounds at the end of 1 minute.
    Because, for a given valve setting, air flow rate will be greater at higher pressures, I will need to gradually open the valve as pressure diminishes to maintain a constant air flow rate or use a pressure regulator.

    The other consideration is that I use a siphon nozzle. So, I will have to account for both combustion air provided by my leaf blower, but also compressed atomization air. I have a plan for that.

    Tomorrow I will take some pics of the sensor I have. It is a commonly used sensor having been installed in many Izusu and GM products. It has only 3 wires---sensors commonly have 3, 4, or 5 wires. Frustratingly, just discovering which wire is 12v power, ground and signal required at least an hour of searching. But I think I have it. When I hook it up tomorrow, if I am wrong, POOF, the sensor is toast.

    Denis
     
    HT1 likes this.
  2. I'd thought about measuring airflow too, maybe a larger diameter plastic sock made from a roll of builder's film so a 2M circumference and maybe 20 metres long (63.662 cubic metres) and videoing the filling of it to determine the fill time to inflate it from the frame rate. You could then compare with the airflow sensor result to get a calibration. The film is supplied folded in half so you'd have to tape the two edges together as you unroll it to form a tube.


    https://www.bunnings.com.au/grunt-2-x-20m-clear-100um-multi-use-plastic-film_p0810292
     
    Last edited: Jan 10, 2022
  3. GTS225

    GTS225 Silver

    Maybe I'm off my rocker, here, but wouldn't you also need an O2 sensor in the exhaust stream to develop any truely useful information?

    Roger
     
  4. rocco

    rocco Silver

    This is quite a rabbit hole we're going down here, intuitively it seems to make sense that we should be running our furnaces at the ideal stoichiometric ratio but is that actually the case? What is our metric for performance? Do we want power or economy? A gasoline engine for example makes the most horsepower running a little on the rich side, around 12.5:1 but the gasoline engine powered car tends to gets it's best fuel economy (mpg) when the engine is running on the lean side 18:1 or so. In the context of a foundry furnace, the question might be, do we tune our burner to minimize melt time or to minimize the amount of fuel required to bring our metal up to pouring temperature? Are these necessarily conflicting requirements and how do they relate to the ideal stoichiometric ratio for the fuel we're using? I don't have any answers and I'm not sure I'll be able to contribute anything useful to this thread but I'll be watching with interest.
     
    Last edited: Jan 10, 2022
  5. Al2O3

    Al2O3 Administrator Staff Member Banner Member

    You'll have to measure temperature and correct for that too. Some of the multi wire MAFs have the temp sensors built in. Humidity can materially alter the mass flow measurement as well. There are/were a few aftermarket manufacturers that supply full calibration table with purchase, but they are few and far between because, everyone cracked the high production (very low cost) OE units. If you join one of the ECU manufacturers forums you can probably search and find the pin outs and tables for popular MAFs.

    Since you already have a MAF, I'd think a sharp edge orifice plate would be a sufficient standard. I think you can measure pressure accurately enough with a water column. The automotive manifold absolute pressure sensors are very good, but again you have to process the signal. You can also use your tank and regulated pressure but you'd need a high flow regulator. The expansion will cool the air, so again, a good temp sensor is also required. I don't know how accurate you want to be, but depending upon your altitude and scheme, you may need absolute pressure for your standard.

    It's been a long time Denis, but back in the AA days, and being an engine and motorsport guy, I was actually going to use one of my fully instrumented aftermarket ECUs to make a burner. It would have had a diesel pump and injector, MAF, Air Temp sensor, and wide band O2. I decided it would have been more of a science project than a burner build so I didn't do it.

    Honestly, I think you'd be ahead of the game to just use a lamba/wideband O2 sensor to measure the flu gas stoichiometry. If you buy a heated one, all you need to do is build a little ejector to withdraw a small sample. Just because you can precisely measure the amount of air and fuel going into your furnace doesn't mean your burning it all.......and that's what you really want to know, and of course, as you already know, where it's being burnt in your furnace also matters as far as melt temp/time.

    Best,
    Kelly
     
  6. Melterskelter

    Melterskelter Gold Banner Member

    Rabbit hole? Might be for those that can set up their furnaces right on the edge of ideal performance by eye. That’s not me though. I’ll tell you what, since I put an accurate and precise meter on my fuel flow rate, my burn efficiency in terms of both time and fuel quantity have improved considerably. That is the goal here—-To turn on the burner, nail the fuel flow rate AND air flow from the start and maintain it throughout knowing what I am doing throughout.. A secondary goal is to suggest a method that will allow us to compare setups objectively whether a person is located in Washington State, New York, The UK, Australia or South Africa. (To me it has been frustrating to have no way to quantitatively know how much air someone having trouble with their furnace is blowing) and distant third goal is to know how close to some ideal ratio fuel is actually burned which likely is not the stoichiometric ratio—-it is not in automobiles or gasoline aircraft, nor, I am told in commercial boilers.

    I have found it challenging to judge flame color, length, and sound consistently across days of bright summer sunshine, evening or night burns, winter days close to freezing and light drizzle. Unless one can accurately measure both air and fuel flows, one is left to guessing. Sure, it is possible to set a needle valve to a rough setting and a shutter on the air too. But, I have learned that very very small changes in either make a significant difference in furnace performance.

    For those not needing to know this info, there is no need to follow me down this rabbit hole. It may not end up working out for a variety of reasons. But, yup, that was me that just popped out of sight. :)

    Denis
     
  7. Melterskelter

    Melterskelter Gold Banner Member

    Kelley, it is my belief that the MAFs are temperature compensated internally. I’ve looked at mine and it is more than a heated wire. What I have read indicates compensation.

    An O2 sensor on the exhaust side would be nice. I have bought for the purpose. But, the exhaust temp they are supposed operate at is far exceeded by the furnace as you know. I had a scheme dreamed up to “sniff” the exhaust. But implementation wasdaunting enough that I did not quite do it, yet. http://forums.thehomefoundry.org/index.php?threads/measuring-air-fuel-mix-in-an-oil-burner.243/

    D
    enis
     
  8. Al2O3

    Al2O3 Administrator Staff Member Banner Member

    The exhaust temp is no problem. Just use a small diameter metallic tube that just penetrates your flue stack/silencer wall and coil a length of it. It will conduct heat away from the attach point to the stack so that part of the sense tube will live. The initial tube could probably be a steal brake line. If you have an old K-Type TC steel the Inconel sheath off of it, then use a fitting for copper tube. If you keep the sample rate low enough and the tube diameter small, I think you'll find it loses heat so rapidly you wont need to fuss with a coil and the temperature will be basically ambient by the time it hits the sensor, thus the need for a heated sensor. The sniffer is no problem either. The pressure drop in your flue is probably enough to propel the sample to the sensor. You may need to put a small length of hose on the output to prevent backward diffusion of air to the sensor. If not, just run the output hose to the low pressure inlet side of you blower and that will do it. -Simple, and very reliable.

    Best,
    Kelly
     
  9. Al2O3

    Al2O3 Administrator Staff Member Banner Member

    Assuming the ultimate goal is achieving higher furnace temps, there is a related topic that has been touched on before, and that is optimum fuel/air flow rate for a given furnace. Meaning, there is a limit to how much energy in the form of fuel/air/burn that can be achieved in a given size furnace. I think I understand the limiting factors and could maybe come up with some rules of thumb, but the variables are such that it is probably far better determined experimentally.

    The primary variables are the free furnace volume (furnace volume less crucible and plinth volume, let’s call it FFV), flowrate of air/fuel (FR), and the flame propagation speed (FPS) of the fuel.

    For liquid fuels, the flame propagation speed can be dramatically affected by atomization. For gaseous fuels like natural gas and propane, the FPS is well established.

    As an estimation, if you divide FFV/FR you’ll get the average time a unit of fuel air can spend in the furnace. Now you have to use your imagination a bit about the path it travels but if you estimate a couple trips around the perimeter you would now have the distance traveled by a unit of fuel air. If you divide that buy the time, this figure needs to be less than the FPS, and will be the upper bound for how much fuel/air can be put through given furnace.

    Now keep in mind, the distance traveled will be a nebulous and ever changing thing, and getting the burn to occur in the correct location to transfer heat to the crucible will still be a significant factor in melt time, but it is a plausible explanation for those that have observed the limits of any given furnace.

    If you really want high temps, add oxygen....

    Best,
    Kelly
     
  10. Smoking Shoe

    Smoking Shoe Silver

    Those values are known, for at least some units. Give me some time and I'll dig out my data from an EFI project I was working on. I even plugged the numbers into Excel to generate a formula so I didn't have to use a lookup table in the software.

    I've though about this problem, even before I jointed this list, and I was thinking of attacking the project from the other direction and more directly by using an automotive oxygen sensor.
    The problem with that is that the cheap old style Lambda probe is practical only around stoichiometric balance. We would probably need to use a wide band O2 sensor and the electronic drive for them is quite a bit more complicated. They would also likely require a bit of an add on to the furnace to keep them from sniffing the atmosphere and be in a reasonable range as far as temperature of the exhaust.
    With the wide band O2 and an Arduino we could dial in any ratio within reason and let the software run the fan/fuel valve as needed.
     
  11. Melterskelter

    Melterskelter Gold Banner Member

    All of the above seems true. That is why getting a handle on what is going into the furnace will be important. It is true that furnace builds vary some, but I have suspicion that fuel/air inputs vary wildly/widely. By the end of the week I should have a pretty good idea whether this was all a time waste or will provide some useful data. Hope springs eternal…

    Denis
     
  12. Melterskelter

    Melterskelter Gold Banner Member

    I was really hoping that one or more of you automotive guys would have access to some data. It is out there somewhere, but I tried pretty hard to find it with zip to show for it. I did learn that the signal generated tends to be near 1v for low flow rates and is somewhat linear (sags in the middle) as flow increases to around 8 or 9 v. I think the units may be outputting PWM signals rather than a smooth line voltage. My oscilloscope will soon answer that question.

    Here is the sensor I bought. It has a long list of vehicles in which it is used.
    7EF98405-4181-4BF5-9F7B-24D33C2FABD1.jpeg

    So, if someone has the data, that would be a huge plus. Even for a different sensor, data would be great as other folks, if interested, might well buy a sensor and have all the correlated data at hand without the need for doing quantitative flow tests and without having to guess as to which pin is 12v, ground, and signal.

    Denis
     
  13. Mach

    Mach Silver

  14. Melterskelter

    Melterskelter Gold Banner Member

  15. Melterskelter

    Melterskelter Gold Banner Member

    Mach,

    I have done Arduino work in the past. I will have to look into that Arduino Sketch more closely. It looks great. I am not surprised that there is very little lag time in response of the MAF as rapid response would be very important for engine regulation.

    I had found the very image you linked regarding the pin out of the MAF.

    Thank You!!

    Denis
     
  16. Smoking Shoe

    Smoking Shoe Silver

    Here is the data on some Bosch units. A lot of other sensors use the same basic hot film sensor but the numbers can be a bit different depending on the housing they get stuffed into.
    Rather than build a data table the numbers can be put into a graph in Excel and then a trendline fit to get a formula for the curve. That eliminated having to resort to interpolation and if you have a good fit (R^2) then accuracy goes up too.

    I can't find my old data right now but I seem to remember a 3rd or 4th order polynomial giving a very good fit.
     

    Attached Files:

  17. One of the old school ways of measuring unburnt fuel on the exhaust side was some platinum mesh which would glow and ignite unburnt fuel in a rich exhaust. It might be possible to "borrow" :rolleyes: someone's catalytic converter core to do the same thing.
     
  18. Melterskelter

    Melterskelter Gold Banner Member

    I have used a piece of steel on a long handle to test for soot deposition as a indicator of rich/lean. And that gives a gross "yes/no" type of answer. I am looking for a "how much" type of answer with respect to optimized mixture which may not necessarily be a stoichiometric ratio. But, once I get things dialed in, I should be able to find that sweet spot quickly and reliably.

    Denis
     
  19. HT1

    HT1 Gold Banner Member

    any fact on where this "best" Heat transfer location is???

    And I'm watching this thread very close, my new furnace is still giving me heart ache, but I will take that to my own thread

    V/r HT1
     
  20. Al2O3

    Al2O3 Administrator Staff Member Banner Member

    I posted questions in that thread.

    Best,
    Kelly
     
    HT1 likes this.

Share This Page