Test pressure fluctuations - How to eliminate pressure fluctuations

[pmcenroe]

Hi,

I've just found this site (by accident) and, as usual, wish that I'd found it a long time ago!

Apologies is the post below rambles a bit but I've tried to provide some background information as well. I've listed my questions at the bottom of the post so if you'd like to skip the verbiage....


I've just started to build the second iteration of a flow bench I started working on a few years ago. As usual time is the one resource that's most precious hence the delays in trying to redesign and use the flowbench.

As background, I am interested primarily in checking flow in VW beetle heads and intake manifolds as part of my Formula Vee racing campaign. In Ireland we run 1600 VW beetle engines with a minimum of modifications allowed (in theory none except for accepted refurbishment work), hence the flow figures I'm considering are probably relatively small in comparison to the big-block heads and 4-barrel monster carbs that some people get to play with!

My original bench used a vacuum motor and used plastic drain piping (68 mm) to provide the pipe work for connecting the motor to the test plenum. The plenum I used was a drain pipe "collection" box which I sealed with some MDF. The volume of this box would probably be about 5 litres or so.

I used water manometers and "orifice" plates in measuring both test pressure and for measuring the air flow rate. The orifice plate dimensions were based on figures I'd found on the internet. As I was primarily interested in comparison work I wasn't too worried about absolute levels, hence my orifice plates were not accurately cut. I reckon that 100 cfm at 10" test pressure would be adequate for all my needs.

To control airflow I used a by-pass mechanism to bleed air between the motor and orifice plate.

I used the bench like this for a while, with some success, however the biggest problem I had was pressure flucuations that occurred when running under "steady" conditions. I tested at 10" water and periodically(every couple of seconds or so) the pressure would change slightly, dipping maybe 1/2" before recovering. This pressure drop also affected the flow measurement manometer and I would end up trying to record the flow value as the test pressure passed through 10".

I was able to use the flowbench, though it was frustrating trying to get repeatable figures.


For my second iteration I decided I would make the following changes:

Use a variac to provide speed control for the vacuum motor rather than air bleed

Increase the test plenum volume (now about 25 litres)

Replace the air-flow measuring manometer with an electronic one for ease of reading and some filtering

Check up pipe-flow orifice design practice on the web and try and make use of these practices to design a more theoretically correct flow measuring element.


After doing this work, I have to say that I am still disappointed. Pressure flucations are still apparent and similiar in magnitude to before. The response of the electronic manometer is such that the information from it can no longer be synchronised with the test pressure manometer and overall is probably less useful that the original inclined manometer I was using (a big suprise/disappointment to me).

Finally, my questions.

Why do I see these pressure variations? I haven't measured the supply voltage, however I don't believe this to be an issue (different houses, no other loads, washing machines present etc.)

How can I reduce these pressure flucuations? Should I increase the volume that the vacuum motor uses? (Currently this would be just about 10" of a 68mm pipe)

Is an orifice plate-in-pipe guage the best way to measure flow? (Based on some of the comments I've read on the site, an in-pipe orifice is a poor compromise and an orifice connected to a large free area would seem to be the favoured approach).


I have lots more questions, but I would be interested in peoples thoughts on the above before proceeding much further.

Thanks

Paschal McEnroe.

[Shawn]

Does this happen when you flow with a test oriffice, also or does it just happen when your flowing a cylinder head?
Shawn

[84-1074663779]

There could be several reasons why the pressure could fluctuate, singly or in combination, but two seconds is a very low frequency for a cyclic pressure variation, but it is certainly possible.

The pressure fluctuations I have seen are a sort of flutter or pulsing many times a second.

The first thing to suspect is blower surge. All centrifugal compressors will do this to a greater or lesser extent if you reduce the airflow by throttling the intake or outlet. It is best to vary blower speed or switch the number of blowers rather than just try to cut off the airflow with a restriction in the flow path. Usually the more efficient the blower, the more violently it will surge. Some blowers are particularly bad, with others it is hardly detectable.

Another thing to think about are resonances in the various plenums and pipework. There is a whole science of tuning loudspeaker boxes with tuned ports and vents to resonate at the desired very low audio frequencies. With a bit of imagination a bass reflex loudspeaker enclosure might look like a sort of weird flowbench.

The third possibility is turbulence at various points in the air path. Vortex shedding can cause cyclic variations in pressure. It is the same mechanism that causes flags to wave in moving air.

I suspect that there might be a bench resonance which is causing surges in pressure. Some sort of flow instability either in the blower or pipework might set this off. All you can really do is experiment with different pipe lengths and see if the nature of the pulsing changes. The other thing you might try is heavy acoustic damping in your plenum. Try lining it with very thick foam rubber. It might prevent the surging from building up to a resonant peak.

For more information on this effect, look up "Helmholtz resonators" or resonance, on google. A closed box and pipe stub can be tuned to VERY low frequencies, and I think that is what you may be seeing.

One other thing. These flow instabilities can drive a digital manometer totally nuts. The electronics samples the output of the pressure transducer, and if it is varying, the readout will be all over the place. What you need is a big acoustic muffler in the measurement pipe to the digital manometer.

Use a very small bore pipe into a large averaging tank, then another small pipe to the manometer. The volume of the tank will slow down the response and average out low frequency fluctuations. Fill the tank entirely with sound absorbent material. Cotton wool or something similar works fine.

A water manometer is not so susceptible because the water in the tube is very heavy and small fast cyclic pressure fluctuations just cannot move it fast enough to show up.

Oh the joys of flowbenching are endless..............

[pmcenroe]

Hi,

Thanks very much for the prompt replies. Lots of food for thought. To discuss the points raised:

Shawn:
I don't have a test orifice as I haven't tried to calibrate my flowbench to any great degree yet (and haven't gotten round to it!). The closest I have is an adaptor I set up to simulate the flow resistance of the carb when testing manifolds. (That was after the wire holding the throttle open slipped and everywhere got sprayed with red water!).

With this plate, on the old bench, I was seeing similar fluctuations as well.

On the question of test orifices, has anyone tried using a CD centre hole (or multiples of) as a possible test orifice? The reason I ask is that old CDs are readily available and I would assume that the standards applied for the hole dimenstions are quite tight and universal.

Tony:
Lots of interesting questions here.

I had given some thought to turbulence but as you mention 2 secs is a very low frequency and I had thought that the timescales for turbulence in the flow bench would be much less. One experiment I tried in the past was to place some oven-hood filter material over the pressure take off point in the hope that this would dissapate any turbulence affecting the pressure point area.

This mod had no impact on the fluctuations.

The question of blower surge may be more relevant. I use a single blower running quite close to it's limit. My original bench used air bleeds/bypasses after the orifice to control flow through the orifice. My current bench uses speed control however this does not seem to have made any noticable difference.

When I get the box built to attach the blower motor to I will then have the option of mounting an additional blower motor. In the meantime I will also try some experiments at lower test pressures to see if the fluctuations are less.

I hadn't thought about possible resonances, to be honest, as I thought flow would be "steady" during testing. I will try some of the mechanisms to damp resonances if some of experiments above do not have the desired effect.

There are some good ideas on conditioning the signals and again I will try these. I'll persevere with the digital manometer for a while longer but I have a feeling I may move back to a water based one.

On your last point: do flow bench projects take on a life of their own? I'm sure I should be doing lots of other work on my car but I'm just so annoyed that what reads so simply on the web should prove to be frustrating in reality.

Thanks again

Paschal

[Terry_Zakis]

These vacuum motors operate at very high RPM, and must have at least 5-6 vanes on the impeller. Therefore, I don't think that a blade pass frequency would be causing this problem.

As far as surge from the motor, I'm not sure if that could be the cause. Bruce was kind enough to post a few of the Excel files that I developed which show performance curves for nearly all of the Ametek vacuum motors. The plots of flow vs. vacuum are relatively steep, which typically would not lead to a an operating line instability.

If you look at some flow vs. head curves for centrifugal pumps, then you will indeed see the curve flattening off at low flow levels. In those cases, the operating line can become unstable. But I don't think this could be happening with the single vacuum motor.

I like Tony's suspcion that there may be a resonance issue from the plenum. I haven't looked at the mathematics of the helmholtz resonator in a while. This could be a cause.

Then there's always the possibility that there's a flow separation in the test piece that could be causing this.

Best of luck,

Terry

[pmcenroe]

Hi Terry,

Thanks for the reply.

I bought the motor I'm using about 4 years ago and it was a new, generic, no-name replacement motor bought from an appliance repair shop. Your mention of the name Ametek reminded me of the research I'd done at the time and having had another look, the following URL



seems to be very close to the motor I have. The height is approx 170mm and the diameter 145mm. I think it was rated at about 1400W.

Is this type of motor suitable for a flow bench or are it's characteristics perhaps too unstable for accurate flow measurement?

I did a little more experimenting tonight. I finished the large box on which I could mount the motor. I was disappointed to find that it did not appear to change the pressure fluctuations I am seeing. I measured the pressure in the motor "box" and could see similiar fluctuations.

Measuring the flow pressure drop using the digital manometer I have showed that it too was changing. As an earlier poster mentioned it's readings seemed to be particular upset as the variation was quite large.

I tried checking the input voltage using a digital meter I have. I'm not sure what the response time for the meter is, however the voltage seemed steady at 185-186 Volts. (Domestic supply here is a nominal 230V 50Hz. I'm using a variac (I think it's called) for speed control). Changes were slow And I don't think the variations would account for the pressure changes I was seeing (perhaps 1/4" - 1/2").

That's it for now. Thanks again for the pointers.

Regards

Paschal

[Terry_Zakis]

Hello Paschal,

I did notice that this motor is a radial discharge style, as opposed to the ones I've purchase, which are flow-through. I would think that this kind of radial discharge motor would be more likely to cause pressure fluctuations, since the impeller vanes pass right by discrete openings in the casing.

In contrast, the flow-through Ametek style motor routes the air flow between the armature and field windings. Flow-through may pulse less. Hard to say.

They didn't have any flow vs. vacuum data, so I can't plot up a flow characteristic curve.

Since you are only using one motor, you may want to try flow-through type for comparison.

Best Regards,

Terry

[pmcenroe]

Hi,

I've had a chance to do some more experimentation and, rather build up the story like a TV detective story or Discovery Channel revelation, I now think that supply voltage fluctuations are the cause of the test pressure fluctuations I have been seeing.

Just shows how dangerous assumptions can be.

After Terrys advice I decided to try using the house vacuum cleaner as a source. I didn't disassemble anything but I reckoned the motor design would be sufficently different to eliminate some of the effects of the motor design.

Although the overall flow was at a much lower level I could still see fluctuations.

I then remembered the old analogue volt meter that hadn't seen the light of day in 5 years. I hooked this up to the mains in parallel with my digital voltmeter, without running any motors.

The analogue meter could be seen to be changing slightly (response was slow) and monitoring the digital volmeter at the same time it looked like the voltage varied between 230 and 233 Volts.

Whilst not sounding like much, assuming that motor speed is proportional to voltage and air flow is proportional to motor speed, this indicates possible fluctuations of 1.28%.

Given the square root relationship for pressure difference over an orifice, this would give rise to potential variations of 2.55% or 1/4" at 10". This would match the levels of variation I have been seeing.

In hindsight this is something I should have checked first and I apologise if I've wasted anyones time.

I am encouraged by the potential accuracies inherent in the system. I am also encouraged by the fact that the configuration of the flow bench does not appear to have a large effect on the ability of the flowbench to measure flow and detect changes.

I hope to get my hands on a surplus invertor/motor controller that I'm told should allow the motor speed to be controlled very precisely. Having had a quick look on the internet, Voltage stabalisation can be an expensive business so I may need to look at ideas to work around the problem.

I'll wait and see how this works out (and I'll report back).

However one conclusion I would make is that the principles of orifices and water manometers provide a sound basis for building a flow bench. If you want to spend money, spend it on providing yourself with precise control of the operating parameters (flow etc.) rather than expensive measurement devices.

Paschal

[84-1074663779]

Blower speed stability is an interesting issue. Unfortunately the ac/dc brush type motors as used in vacuum cleaners are very sensitive to voltage changes. This makes them ideal candidates for a very simple speed adjustment system, but it has the disadvantage that unwanted mains voltage variations may be troublesome.

Ac induction motors on the other hand, particularly three phase induction motors are particularly insensitive to voltage changes. It is mains frequency, and the speed of the internal rotating magnetic field that establishes motor RPM.

The ultimate system would be a three phase induction motor running from an electronic variable frequency drive, a fairly expensive solution. But If you ever have the opportunity to pick up a cheap secondhand electronic VSD, the three phase motors themselves are worth almost nothing. These systems will run from single phase up to typically about three horsepower. If you are lucky enough to have three phase power available, it is an ideal solution for a large flowbench. I run a 10 BHP variable speed drive in my bench and it is excellent..

Blower surge is an interesting phenomena. A centrifugal blower works by accelerating the air to high velocity within a rotor, and then converting this velocity to static pressure in some sort of diffuser, scroll, or volute. The higher the velocity (rotor tip speed) the higher the differential pressure the blower is capable of.

If you totally block all flow to a centrifugal compressor, the rotor will try to accelerate the air until some peak pressure is reached. At that point the flow will drop to zero, (because the flow is blocked).

The velocity mechanism that generated the pressure differential in the first place has now ceased to work, and so air just flows backwards through the still spinning rotor. When the differential pressure has fallen sufficiently, the air then starts to flow back in the right direction again for a short while. Flow then decreases to zero again.

This causes complete flow reversal through the rotor, and the pressure to go up and down in a cyclic manner, and is called blower surge. The flow does not have to be completely blocked, but if you reduce the flow by throttling, below some undefined threshold, your blower may begin to surge.

How fast it surges depends mostly on pipe volumes. If the blower is filling or emptying a large sealed plenum, the surge frequency may be very low indeed. With a flowbench it might not be terribly difficult to find yourself in a situation like that, and flow does not need to drop right to zero to start seeing some cyclic pressure fluctuations.

Unfortunately the more efficient the blower, the more prone it will be to surge. Something really crappy like a leaf blower is not going to have any detectable surge. Something much more efficient like a turbocharger compressor, or a centrifugal supercharger, might surge quite badly under severely throttled conditions. Vacuum cleaner motors are highly variable, but they are mostly not very efficient, and not terribly prone to surge.

If your flowbench has some serious internal resonance problem, it might provoke an almost stable blower into surge. Difficult to separate cause and effect. It could be the blower, the bench, or an unfortunate combination of both.

There are two ways of fixing it, one is to simply bleed some air around the blower at low flows. The blower still gets some flow even if the bench is then sealed right off. When doing high CFM testing, the blower bypass can be shut. Adjustable blower speed also helps, but may not be a complete solution if surge is really violent.



.

[Terry_Zakis]

Beautiful rundown on blower surge Tony! I have to say that I didn't even think of that one on the flow bench. Shame on me, because I've seen this before and heard it one of our gas turbines operating at very low power output. That was actually caused by the liquid fuel pump.

Anyway, if you look at the Ametek vacuum motor files that Bruce was so kind for post for me, you can examine the plots of vacuum vs. Flow in CFM, to get an idea of when the surge will/can start. Granted, the characteristic flow vs. pressure curve for a blower, or flow vs. vacuum for a vacuum motor, are specific to each blower/vacuum motor model.

For the Ametek 056 series, the 115923, and 116604-00 models, you can see that the plot of flow vs. vacuum has a nice slope to it, from 20" H20 through 100" of H20. However, as Tony mentioned, if you start to close off the flow, and push the vacuum motor higher up the operating line, it will approach a point where the curve starts to lay over. This is the point on the operating curve when surge or flow instability will start to occur.

I don't recall if the test pressure was mentioned in the earlier post, but it would appear from these Ametek plots, that operation above 20" H20 should not get into a surge problem.

It may be a good idea to try and search the manufacturers website, or call their technical support to get some data you can look at, and plot up, to see where the motor starts to surge.

Again, great call Tony.

Best Regards,

Terry_Zakis