DIY Super Accurate Flowbench

[FlowSpecialist]

Inspired by DV's articles on DIY flowbenches I thought I'd write an article on how I designed and made my own 20 years ago and how you can do the same for pocket change. All the pipe work is standard domestic plastic soil pipe available from your local DIY shed. Off the shelf bends and connectors to join different pipe sizes make it a doddle to assemble pretty much any configuration you want.

The design follows the requirements of British Standard 1042 on the measurement of fluid flow in closed conduits. Similar standards will exist in other countries no doubt and can be found at a main library.

The bench is a two manometer design with one manometer measuring the pressure drop across the test piece and the other across the calibration piece - an accurately made sharp edged orifice plate. The basic elements are as follows.

1) The Downpipe.

This in my case is a 2 foot length of 4" soil pipe mounted in a wooden cabinet, an old kitchen unit I think it was, so the top of the pipe is flush with the top of the cabinet. The bore adaptor and cylinder head then go on top of that and I seal round each time with plasticene. You could easily make up something more permanent.

At the bottom of the downpipe is a standard 90 degree bend and adapator to mate to the 2.5" orifice plate pipe which runs horizontally out of the cabinet.

2) The Orifice Plate Pipe.

This is 2.5" soil pipe (standard drainpipe) and you need 20 diameters of pipe length upstream of where the orifice plate is going to go so that the flow has time to settle down after the bend and become laminar. About 4 feet in my case.

3) The Orifice Plate

This is a sharp edged orifice plate which serves as the reference by which the pressure drop across the head is evaluated. Ideally it needs to be sized so that it flows about 50% of whatever the highest flowing test item will flow at the same pressure drop. That will ensure it isn't restrictive but has a high enough pressure drop across it at low valve lift to give accurate results. An orifice plate will have a discharge coefficient of about 0.60 so it will flow 60% of what a venturi tube of the same area will flow which is 137.86 CFM per square inch at 25" pressure drop.

I only envisaged flowing 2V heads up to a maximum of about 200 CFM so I plumped for a 32mm plate with an area of 1.25 sq in and a target flow of about 1.25 x 0.6 x 138 = 103 CFM. For flowing bigger heads like Chevies and 4V items you could up the size accordingly. For target flows up to 350 CFM a 42mm plate would be about right.

Downstream of the orifice plate is more 2.5" pipe, the length of which doesn't really matter but gives you something to which the vacuum will be connected.

4) The Vacuum

Just a standard 1kW wet and dry shop vac which spends most of its time hoovering crap off the lathe and mill and gets connected to the flowbench as needed. Happily one of the bits of pipe it came with was a perfect fit into the 2.5" drainpipe so no further mods were needed.

5) The Manometers

Clear plastic 1/4" tube filled with water to half height and pinned to a bit of 6 foot board. For measurement I bought two 6 foot steel tape measures and cut the tapes out of them and pinned them to the board between the tubes.

6) The Pressure Tappings

One at the top of the downpipe (it must be at the very top) goes to manometer 1 and two situated 1" either side of the orifice plate go to manometer 2. I made little brass fittings and glued them to the plastic tube but you could screw something into place just as well. I found that the flow, and the manometer readings, pulsed quite a lot so I also made dampers to fit into each manometer tube which were 2" lengths of 1/4" brass with a 1mm hole drilled through them.

Measuring Flow

This is the complex bit if you want to follow the full flow equations including Reynolds number and all the other stuff. I wrote a computer program to take the two manometer readings and convert into CFM but there is a shortcut which gives reasonable accuracy. With M1 being the reading in mm of the pressure drop across manometer 1 and M2 similarly the flow will be square root (M2/M1) x 103 CFM (the target flow of the orifice plate).

Accuracy

To check and calibrate the bench I had a friend make me 4 test orifice plates in various sizes out of thin steel plate. The orifices must be sharp edged and the way to do that is clamp the plates between other bits of scrap metal and bore through the lot.

BS1042 tells you that compressible flow (gases) is inherently chaotic and the best you should aim to achieve is 1% margin of error even if you follow every recommendation in the standard. I measured the 4 orifice plates to 0.01mm accuracy, calculated their target flow at 25" using the full flow equations and tested the bench with each plate mounted on a 91mm bore adaptor (Ford Pinto size) to simulate how a cylinder head would be mounted. The CFM results I got were as follows.

Plate.....Target....Actual
1...........33.9........34.2
2...........78.5........78.6
3..........124.6......123.9
4..........165.1......165.0
Total.....402.1......401.7

Average error less than 0.1%

That's not just accurate it's f***ing extraordinary. I've used my calibration plates to test several 'professional' flow benches over the years and found errors up to 15% or even worse. That's partly why I take no notice of other people's flow figures because they can be so far out it's untrue.

Total cost less than 20 quid. I needed a shop vac anyway. It might be a bit more if you don't have a mate who can make orifice plates for you. Accuracy better than anything you can buy.

Dave

[MilesTugoh]

Thanks for a very useful post, I will build one of these as soon as I figure out what 20 quid is in US $. No currency converter help on the quid ..

Thanks, Miles

Edit : Australian dollar = .95616 USD x 20 = $19.12

[Warpspeed]

Great post Dave.

For anyone really serious about building a home built flow bench, check out the best resource anywhere on the internet at the PTS Flowbench and Dyno Forum:

Flowbench & Dyno Discussion

[FlowSpecialist]

Originally Posted by MilesTugoh

Thanks for a very useful post, I will build one of these as soon as I figure out what 20 quid is in US $. No currency converter help on the quid ..

Thanks, Miles

Edit : Australian dollar = .95616 USD x 20 = $19.12

Are you trying to call me an Aussie you cheeky f**ker?

Dave

[Rick360]

Dave,
Interesting post, as are some of your other posts.

When you refer to a "sharp edged" orifcice are you really referring to a "square edged" orifice? To be a sharp edged orifice shouldn't the inside edge of the orifice be machined at an angle down to a "sharp" point?

Do your calculations for "flow" convert to ACFM or SCFM?

Are we really concerned with how much air "flows" thru a head on a flowbench? It will be different depending on the atmospheric conditions from day to day, room temperature, baro, etc etc. Isn't what we really want to know how easily air flows thru a head? I don't believe that a SF (or any other typical flowbench) can accurately measure airflow. They do however measure the resistance to flow very well and will give a repeatable number in a broad range of conditions. A SF style (internal orifice) bench will compare the resistance to flow to the internal range orifice.

When you measure your accuracy and indicate they are less than 1% you are using a testing method that also has a measurement error of 1%. It is my understanding that you can never be better than 1% when using an orifice. After they are calibrated to some much better standard (such as LFE or nozzle in a lab environment) and placed into service, you are never certain it is better than +/- 1%.

Your comments about pressure recovery inside a bench (in another thread) was interesting and I appreciate the enlightenment, it was something I hadn't thought of.

I built a orifice style bench many years ago and used a head that had been flowed on a SF600 to "calibrate" it. How much the "actual" flow is doesn't matter. It will give similar results to most SF benches I've compared to, which is really only for comparing to others, which always allows for a wide margin for error. I use a "CFM" number to talk to others about airflow, even though I know it may not actually be that many CFM, it is just used for comparison of resistance to flow. It has always repeated very well.

Rick

[rookie]

Rick,
Check out DV's article on simple benches.
Porting School #4 - Budget Bench Electronics

[FlowSpecialist]

Originally Posted by Rick360

Dave,
Interesting post, as are some of your other posts.

When you refer to a "sharp edged" orifcice are you really referring to a "square edged" orifice? To be a sharp edged orifice shouldn't the inside edge of the orifice be machined at an angle down to a "sharp" point?

Yes I'm referrring to a square edged orifice plate that has been correctly made so there is absolutely no rounding of the edges. With mine you can drag them across the back of a fingernail and slice material away.

Do your calculations for "flow" convert to ACFM or SCFM?

Not sure what you mean here but I'll guess the A and S are intended to mean Actual and Standard CFM. My computer program works out all of them. Actual CFM at actual test conditions, Actual CFM at standard (15c, 1013.25mb) conditions and nominal flow at standard pressure drop of say 10", 25" or anything else I enter. I say 'nominal' because the generally accepted square root law to convert from one pressure drop to another doesn't actually work but that's a topic for another time.

When you measure your accuracy and indicate they are less than 1% you are using a testing method that also has a measurement error of 1%. It is my understanding that you can never be better than 1% when using an orifice. After they are calibrated to some much better standard (such as LFE or nozzle in a lab environment) and placed into service, you are never certain it is better than +/- 1%.

Yes, as I said the margin of uncertainty in compressible flow tests is 1%. It just happens that my bench comes out with exactly the numbers that the flow equations predict for each test plate.

Dave

[Randy W.]

Hello flow specialist:
I find your posts very helpful. Was wondering if u had an e- mail address where we could converse on a one to one basis.
I find your posts and replies very helpful.
Interested in hearing from you.
Regards
Randy

[Rick360]

Originally Posted by rookie

Rick,
Check out DV's article on simple benches.
Porting School #4 - Budget Bench Electronics

I read it, was there something in particular in the article you were referring to, or just thought I'd be interested in more flowbench articles?

Can't say I agree with with the cal method in article 3, but everyone has their own opinion. Article 4 has some good stuff with the FlowQuik and I have seen and helped setup some of Audie's flowbench stuff. His automated testing setup works pretty slick.

I certainly am not a proponent of testing in this manner described in the article (although it would be better than nothing) and the only testing I would ever do below .200" lift is to see what makes it flow the least. I do like to test at higher depressions, but at higher lifts to more closely simulate the actual air velocity in the engine. I have seen some heads that look good at 10" that take a dive when tested at 28" and some that looked good at 28" take a dive when tested at 40".

Rick

[Stan Weiss]

Isn't all flow bench testing subjective? In a running engine, you start around zero pressure differential (when the intake valve opens how well is the exhaust doing its job) and go up to maybe 100" or more. At 6500 RPM one crank rev (360 degrees) takes 9.231 Milliseconds.
What is being done is the same think as step testing on an engine dyno verses true acceleration testing.