How much volume and pressure do I need?

[Induction Concepts]

Bare with me, I'm very new to this. I am trying to build a 'flow bench'. I need to be able to flow from 0-3000cfm of air through a 4" tube and be able to measure this airflow. Pressure will range from 0-40psi.

I have tried to come up with several different ways to design this. I thought about buying a race supercharger that could produce the airflow, and spin it with an electric motor. But I'm sure the electric motor would be very expensive as it would have to be 100hp. and spin to 8000rpms.

So now, my idea is to get a very large tank to hold compressed air. Like a giant version of what I have in my shop. Then we can just open a valve to vent this compressed air into the tube/pipe running to the flow bench.

So, what I'm trying to figure out is how to determine:
1.) The volume of tank needed.
2.) What kind of pressure do I need to compress the air in this tank to, so the air supply will last long enough for us to take our measurements.
3.) How long will said supply flow air at 3000cfm through a 4" tube?

We will be measuring airflow at 37 different points from 0-3000cfm, not sure what kind of time it will take to run this range and measure the airflow through that scale. This is because I don't know how we are going to adjust the airflow yet.

4.) I would also like some suggestions on what to use for some type of a valve that we can adjust the airflow precisely from 0-3000cfm.

Anyway, any suggestions would be appreciated. I'm trying to learn and at the same time, build this as economically as possible as I don't have deep pockets. I was hoping to possibly buy a used propane tank, but I need to know what size. And if you can help me determine how long the air will last vs what size tank, I can make an educated decision on size.

 

[Cliff_J]

Wow, these are huge numbers!

First, the temperature is going to affect your results and it may quickly cause errors that will render the data nearly useless. This will be the largest factor to control by far.

Second, using a big tank won't work. The flow will cause the pressure vs. time curve to look like a capacitor discharge curve - it will take very little time to drain to very low pressures. I assume you've used a paint gun or air wrench and that can drain even a 60gal tank and that is using a 1/4" feed line. You're talking a pipe 256x bigger and nearly the same pressure a non-HVLP paint gun uses. So we're talking swimming pool sized tanks and lots of psi, sounds more like a disaster than anything.

Your cheapest solution - get an old 350 and rebuild it with some PAW or whatever low-buck parts. Find the nearest dirt-oval track and you can probably get all the parts you need for next to nothing for serious HP if you need it. Should be simple to build a stand and bolt a regular blower pulley to the crank snout and then just mount the roots off to the side of the stand and connect the belt.

If my math is correct a 14-71 would flow around 3450cfm at 6000rpm.

Using Corky Bell's forumla of boost*airflow/229=HP or 3450*40/229 that's 603HP.

Factoring in the 3% belt loss and 40% inefficiency of a roots that's 1035HP needed at the crank to drive the blower.

Ok, better make that a couple very healthy SB Chevys or maybe even a couple of healthy BB Chevys to do your high-psi tests.

Cliff

 

[Cliff_J]

As an interesting sidenote, search on the procedure for starting the SR-71 airplane. It used a couple BB Buick and then later Chevy engines to spin up the engines.

http://www.hill.af.mil/museum/photos/coldwar/ag330.htm
http://www.sr-71.org/blackbird/startercart.htm

That thing got a Hemi?

 

[Gokul43201]

The tank approach, I'm guessing, may not work as weel as you're hoping. Actually, I can't tell because I don't know how long you want to sustain the air flow. If your measurements only last a matter of seconds, then you might find some large heavy duty tank that works.

Here's how you figure it yourself, roughly :

For 1 minute of flow time, you use up 3000 cfm ~ 100,000 liters @ about 1 atm.
So, you could get this with a 1000 liter tank pressurized to 100 atm. (~3000 psi). Such a tank could easily cost several tens of thousands of dollars.

If this is going to be done in a lab in school, I would think of stealing air from the building HVAC, by directly tapping into it. Now you can boost the flow with additional fans, but I don't know, off the top of my head, any resonably sized and priced fans that can handle 3000 cfms - that is a lot of flow. And to drive that kind of flow through a 4" pipe (even if it's a reasonably short length) takes some serious pressure. Still I would think that some kind some biggish fan with a meaty motor might do it.

Edit : I see that Cliff's already sized your motor for you, at a 1000 HP

 

[Gokul43201]

To control airflow, you could feed back the measured flow to a PID controlled actuator (a stepper motor perhaps) that opens/closes a baffle in the supply duct.

 

[brewnog]

Why, oh why, do you need 3000cfm through 4" dia anyway? Sounds like one of those instances where "there's got to be an easier way"...

 

[Induction Concepts]

Wow, these are huge numbers!

First, the temperature is going to affect your results and it may quickly cause errors that will render the data nearly useless. This will be the largest factor to control by far.

Second, using a big tank won't work. The flow will cause the pressure vs. time curve to look like a capacitor discharge curve - it will take very little time to drain to very low pressures. I assume you've used a paint gun or air wrench and that can drain even a 60gal tank and that is using a 1/4" feed line. You're talking a pipe 256x bigger and nearly the same pressure a non-HVLP paint gun uses. So we're talking swimming pool sized tanks and lots of psi, sounds more like a disaster than anything.

Your cheapest solution - get an old 350 and rebuild it with some PAW or whatever low-buck parts. Find the nearest dirt-oval track and you can probably get all the parts you need for next to nothing for serious HP if you need it. Should be simple to build a stand and bolt a regular blower pulley to the crank snout and then just mount the roots off to the side of the stand and connect the belt.

If my math is correct a 14-71 would flow around 3450cfm at 6000rpm.

Using Corky Bell's forumla of boost*airflow/229=HP or 3450*40/229 that's 603HP.

Factoring in the 3% belt loss and 40% inefficiency of a roots that's 1035HP needed at the crank to drive the blower.

Ok, better make that a couple very healthy SB Chevys or maybe even a couple of healthy BB Chevys to do your high-psi tests.

Cliff

This was my original idea, use a 14-71, only spin it with an electric motor and that's where I went off track because that will end up being more expensive than building an engine to run it. I can build a 1000+ crank hp engine, won't be cheap because of the parts needed to make it live any time at all. How would you couple two engines to run the blower and keep them sync'd?

Next question, how will you control and vary the airflow precisely? What I plan to do is have a 'known' accurate meter upstream. I guess I can log the two readings of these meters against time and be able to determine where the meter I'm testing needs to be calibrated?

 

[Induction Concepts]

Why, oh why, do you need 3000cfm through 4" dia anyway? Sounds like one of those instances where "there's got to be an easier way"...

Because I need to be able to calibrate a 4" meter that will see this kind of airflow.

I'm trying to find the easiest way, that's why I'm here. I'm completely open to suggestions.

 

[Cliff_J]

If I wanted to couple two large HP V8s I'd see what those pulling tractor guys are doing - they typically do things as simple as possible which should be cheap and dependable as well.

Accuracy will be a big question here. Starting with NTP of 68F and 14.7psia if we want 40psi above that the pressure ratio is 54.7/14.7 or 3.72:1. By raising to the .28 power and multiplying by absolute temp (460 + temp) we get the temp rise, 234 degrees. But this needs the thermal efficiency of the blower factored in and I used a generous 60% above (some roots are closer to 35-40% efficient) and now we're talking 391 degrees of gain.

At 459F that's going to take something like an EGT sensor to read properly and the air is only 57% as dense (final absolute temp/starting absolute temp) as when it started due to all that heating. As the air cools the density ratio will change - the same lbs/min of air will be flowing but the characteristics like temperature and pressure will change too.

In addition, we have serious problems extending the models out this far because even the efficiency of the roots style blower is going to change relative to the level of boost (charts from Eaton and others show this quite well) and the power needed will increase and the temperature gain will as well. This is fighting the wrong side of an exponential curve and the non-linearities could quickly become massive.

What I'm saying in short is that this is a very ambitious project to build this setup. And if the goal is to calibrate an instrument the variables are too large to overcome for accuracy. I'd assumed you had an instrument already and wanted to test tubing variations or heat exchanger/intercooler performance.

Perhaps you can find a different goal for your project. All an internal combustion engine cares about is getting enough lbs of air for the lbs of fuel ingested. With wideband O2 sensors so cheap nowadays it would seem cheaper to just start gathering empirical data from testing and especially if you have electronic fuel delivery means that can adjust in closed loop.

Or if you truly want to measure things, I'd have to imagine that an aerospace or heavy industry diesel application (imagine one of those giant ship diesels) would have flow/pressure meters well beyond your needs.

Please PM me with the application, these numbers are just plain massive!

Cliff

 

[Artman]

3000 SCFM will leave a 4" pipe at 34377 feet per minute. The pressure drop is at least 3.11 inches per foot of duct if it's smooth.

How long does the tube have to be? This will impact your HP requirements.

 

[russ_watters]

Have you considered a nitrogen cylinder? At 2200 psi and 60 cubic feet, that'll give you 55 seconds at 3000cfm.

 

[Cliff_J]

Russ - I figure that to be a cylinder about 3.5' in diameter and over 6' tall to get 60 cubic feet of container volume. Otherwise I figure its 60 cubic feet of air crammed into a much smaller volume of about say .4 cubic feet which would be about right for 6 inch inside diameter and 2ft tall cynlinder. Am I missing something simple?

Artman - I come up with the same 34377 fpm (391 MPH) for the airspeed but my fluid dynamics is rusty - are you saying 3.11 psi drop or 3.11 in/Hg drop per linear foot of pipe?

I must say WOW again at the size of these numbers.

Cliff

 

[Artman]

Artman - I come up with the same 34377 fpm (391 MPH) for the airspeed but my fluid dynamics is rusty - are you saying 3.11 psi drop or 3.11 in/Hg drop per linear foot of pipe?

I must say WOW again at the size of these numbers.

Cliff

Just inches of water. I was surprised the losses were so low, but I may be using the wrong calculation for such a high velocity.

I say wow at the size of the numbers as well.

I also say use extreme caution with this experiment. That velocity of air can cut a body part off. No kidding.

 

[russ_watters]

Russ - I figure that to be a cylinder about 3.5' in diameter and over 6' tall to get 60 cubic feet of container volume. Otherwise I figure its 60 cubic feet of air crammed into a much smaller volume of about say .4 cubic feet which would be about right for 6 inch inside diameter and 2ft tall cynlinder. Am I missing something simple?

Boy, that was a dumb mistake - no, you're right. 60 cubic feet at atmospheric pressure or 3.4 cubic feet at 3000, which gives 1.2 seconds at 3000 cfm.

 

[brewnog]

But as far as nitrogen tanks are concerned, a few metres tall and a metre or so wide doesn't sound far off the mark does it?

 

[Cliff_J]

Russ - with numbers this big its too easy to make mistakes!

Artman - what you think air velocities and temps similar to a small jet engine output are dangerous? Oh, and after conversion that's about .11psi drop per linear foot of straight smooth pipe according to my math? With 40psi to begin with, that's only .3% (per foot) so that easily would be masked by density differences due to temperature change.

I'm picturing some turbocharged drag racing Peterbuilt or some other diesel application, a gas/alcohol engine would be making power large enough its hard to consider what it would be used for besides...well 6 sec drag strip passes, offshore boat racing, land speed car...

This thread needs pics - before the experiment!

Cliff

 

[Artman]

Russ - with numbers this big its too easy to make mistakes!

Artman - what you think air velocities and temps similar to a small jet engine output are dangerous? Oh, and after conversion that's about .11psi drop per linear foot of straight smooth pipe according to my math? With 40psi to begin with, that's only .3% (per foot) so that easily would be masked by density differences due to temperature change.

I'm picturing some turbocharged drag racing Peterbuilt or some other diesel application, a gas/alcohol engine would be making power large enough its hard to consider what it would be used for besides...well 6 sec drag strip passes, offshore boat racing, land speed car...

This thread needs pics - before the experiment!

Cliff

And of the carnage just after. I hope I'm wrong, but it is dangerous. Just holding the pipe still is going to be a challange. 40 psi and 3000 CFM of air popping out at nearly 35,000 fpm, wow.

 

[russ_watters]

Russ - with numbers this big its too easy to make mistakes!

Having just learned to scuba dive, that's not a good enough excuse...

But as far as nitrogen tanks are concerned, a few metres tall and a metre or so wide doesn't sound far off the mark does it?

Maybe not - I used a supersonic wind tunnel that was a bank of large, permanently mounted nitrogen tanks charged from a cryogenic tank. That's where I got the idea. How about a battery of scuba tanks and a compressor?

 

[Cliff_J]

Yes, I would imagine in scuba diving this is an important thing to know.

Regarding the initial purpose of the thread, what method of detection could be used here anyway to find the lbs/min of air traveling through the pipe. Would an air temperature sensor and some sort of obstruction with a strain sensor be adequate? In approaching STP pressure/temp and low flow this seems horribly inaccurate.

I know the typical MAF for automotive application uses a heated wire and measures the heat loss to calculate flow, but if we're compressing the air the temperature would need to be greatly reduced before its introduced into the pipe. Or if from a bank of tanks then maybe just correct for the temp.

The nice thing about the scuba gear is its already regulated. Granted its at lower flow rates but with multiple in parallel... But the thing that pops up in my mind about a bank of tanks is the cost. While searching around for large nitrogen tanks I ran across a website dealing with industrial carts to hold multiple tanks. They mentioned that just the waste in purging the tanks was $300-500 for a cart designed for 3300 CFH. I guess no one said science is cheap!

Cliff

 

[Q_Goest]

The program I have for compressor power shows just under 350 hp required for compressing air from 0 to 40 psig at 3000 SCFM (SCFM = Standard Cubic Feet per Minute).

The program I have for pressure drop shows:
- roughly 1 psi drop in 10 feet if the pressure is at 40 psig (154 ft/sec)
- and a 4 psi drop in 10 feet if the pressure is at 0 psig. (573 ft/sec)
(looks like a 4" sched 40 plastic pipe should suffice)

Because you want to hold the pressure constant, the easiest thing would be to put a pressure regulator on the outlet of your air supply and have it control to 40 psig (or whatever pressure you need). But this doesn't determine flow. For that you might make an orifice plate, since you need restriction on the outlet anyway. Put a pressure gage on either side to get rough estimates of flow rate. The location of the taps can be provided if you'd like. I can also give you dimensions for the orifice plate if you want.

Either before or after the plate, install a valve to provide a restriction. By closing off or opening up this valve, you will change flow rate through the test section.

Air cylinders used by industrial gas companies come in various sizes, the most common being an "A" size which contains roughly 250 SCF of air at 2200 psig. You can get them manifolded together in a 12 pack or other sizes. Problem is I doubt they'll give you the flow you need because the interconnecting tubing is too small.

"Water volume" is a common term used to give the actual volume of a cylinder, or how much volume water would take up in a given cylinder. To determine how long your air supply will last, you'll need the water volume of your air supply, the maximum pressure you can get it up to, and the minimum pressure you can drop it to. Then calculate as follows:

t = (((Pmx - Pmn) / 14.7) * V ) / Q
Where: t = time (minutes)
Pmx = Max Pressure in Cylinder (psig)
Pmn = Minimum Pressure in Cylinder (psig)
V = Water volume (ft3)
Q = Flow rate (SCFM)
EX: For a 100 Cu Ft supply from 100 psig to 50 psig at 3000 CFM, you should get:
t = (((100 - 50) / 14.7) * 100 ) / 3000
t = .11 minutes = 6.8 seconds

You'll need a regulator with a Cv of at least 150 per my calculations, I'd suggest looking for one with a Cv of about 300 though. If you need help finding one, let me know. You may find a donator on Eng-Tips forum, but sizing this is just a bit tricky. Many regs have significant "droop" and don't lend themselves well to this type of service. There will be a slight drop in temperature across the regulator which is calculable as an isenthalpic process. But even given 100 psig inlet and 40 psig outlet, the temperature drop is only 2 degrees F, so that shouldn't be an issue.

For the air tank, you might consider using large bore plastic pipe. Charging it may prove difficult though. Perhaps it could be charged using a liquid cylinder of nitrogen if N2 is acceptable for the experiment. Alternatively, you'll need to put an air compressor on it for a considerable time.

 

[FredGarvin]

I have built a couple of systems for short duration runs. I can tell you that the flow rate range you are looking for and the notion of low cost are not going to go hand in hand. I think there are two things you seriously need to sit down and figure out before you start throwing together your system:

1.) You have to have a budget. Not simply phrases like "as cheap as possible." Put real numbers to it that someone can stick to and use. COST WILL BE the number one factor in determining what your system will be.

2.) You must decide what kind of inaccuracy you are will ing to live with over the range of flows for this particular valve you are going to be calibrating. If the valve operates over a relatively narrow band during 80% of it's operating, I would suggest concetrating your efforts around that area and opening up the acccuracy requirements at the extremes of the flow range. That will help with the costs. Once you establish this aspect, you can determine what methods of flow measurement you are going to use, the necessities for items such as sonic nozzles and other items which will then tie right back to item #1.

You did a pretty good job for establishing your general operating regime, now get the specifics nailed down.

I will also add this to think about...Many people are shocked by the cost of producing compressed air. The numbers you are talking are not a small task. 3000 cfm is A LOTof air. You would have to have a huge bank of tanks to make keep enough air to make even a short run worth while. I routinely rent oil free compressors that output 1500 scfm for high flow testing of components and engines. I have had tests with as many as 22 compressors manifolded together to provide the required flow. Each of these compressors averages about $7500 per month to rent.

Get some more details hammered out and we can discuss this further.