Are two cyclone chambers required?

[Sthutch]

Steamngn,
So if this were placed in a Top Hat, it would be inverted?  If so, wouldn't the fines be likely to settle in inverted bellmouth area?

How would one use this on the exhaust side of the impeller, in relation to a pleated filter?

I don't have an easily accessible source for the honeycomb that you mentions, but the custodians where I work had some of the plastic tubes that Retired2 used.  So I picked up 6 of them for future incorporation into my system.

This is really intriguing stuff.  Its almost more fun than working with wood!  LOL!

[retired2]

The elliptical bellmouth has been discussed here before.  I don't doubt that it may indeed be the optimum configuration, and for some applications provide a measurable performance improvement.

However, as the outlet port in a top hat separator, my opinion is you would not find any measurable difference between an elliptical bellmouth and the bellmouths that are widely available commercially.  To put this in perspective, look at my test data comparing a straight ended pipe to the bellmouth I purchased from a spiral pipe manufacturer.  The difference is not huge by any means, and a straight pipe is the worst possible end condition for an entry port.

So, if you can fabricate or purchase an elliptical bellmouth for the same money or effort as the readily available ones, I'd say use it, but if not don't fret it, you will never be able to find the difference in performance.

[RobHannon]

Steamngn,
So if this were placed in a Top Hat, it would be inverted?  If so, wouldn't the fines be likely to settle in inverted bellmouth area?

How would one use this on the exhaust side of the impeller, in relation to a pleated filter?

I don't have an easily accessible source for the honeycomb that you mentions, but the custodians where I work had some of the plastic tubes that Retired2 used.  So I picked up 6 of them for future incorporation into my system.

This is really intriguing stuff.  Its almost more fun than working with wood!  LOL!

I don't think there is any reason to put a bellmouth on the exhaust side of the impeller.

[steamngn]

Sthutch,
Correct, this would be inverted inside the separator. As RobHannon mentioned there is no need for a bellmouth on the outlet of the blower.

However, as the outlet port in a top hat separator, my opinion is you would not find any measurable difference between an elliptical bellmouth and the bellmouths that are widely available commercially.  To put this in perspective, look at my test data comparing a straight ended pipe to the bellmouth I purchased from a spiral pipe manufacturer.  The difference is not huge by any means, and a straight pipe is the worst possible end condition for an entry port.

Retired2, your testing and work on these separators has been awesome; since I have access to airflow modeling and testing equipment I am only looking to help refine what you and others have already found... And yes any stock bellmouth would be better that the evil straight pipe! ;D
As for the reverse curve at the end of the bellmouth, we have found some very interesting points:

• When a straight-lipped bellmouth is used the lip produces a sheering effect on the cyclonic action causing more rotational vortexes above the lip. This leads to more individual vortexes (and more turbulence) in the air going into the exhaust stream.
• The reverse lip on the bellmouth is more efficient with the mouth only 1" below the top of the inlet pipe. At this height the air immediately above the bellmouth is nearly static, and the inrush into the bellmouth is very smooth.
• The velocity at the exit (the pipe end going to the blower) of the elliptical bellmouth compared to a straight bellmouth is nearly 3% greater!

That last one is the real difference; as the air comes into the separator chamber it will slow down a bit which of course is the whole point... WHERE it slows down is what we would ideally like to control. If only we could get it to slow down right over the top of the slot in that awesome Thien baffle... separation nirvana! 8)
so more info:

• We know air behaves as a fluid in a closed vessel, so as it enters a larger chamber/pipe it will slow down.
• We cannot directly slow the air over the baffle slot mechanically as this would stop the cyclonic separation effect.
• Because (1) is true we know that if we DON'T allow the air to slow in one place then it will slow somewhere else.
• Because (2) is true maintaining velocity elsewhere helps move the slowing air closer to where we want it.

As for those adventurous souls looking to make a bellmouth out of PVC I say go for it! There has been all sorts of stuff done to heat the PVC (boiling water, boiling oil even!) DON'T. DO. THAT.
Like this:

• Go to your local dollar store and get a cheap stainless steel bowl. It needs to be ~10" on the top and anything less than 6" diameter on the bottom.
• Take that bowl and a $20 bill to your local commercial electrician. Flip him the $20 to wrap his heat blanket around your pipe and heat it up. Get it real hot...
• Put that bowl upside down on the floor, and push that really hot pipe down over it. Keep heating and pushing until the outer lip is ~14" diameter. This will take a while...
• The outer radius needs to be hand-formed. Use a heat gun to warm the outer lip and roll it back. I used a short slug of 1" metal stock as a guide (because I had it laying around)... This is surprisingly easier than it sounds.
• Lastly, trim the edge of the lip so it is even all the way around. I stood my pipe on the bench and set a pencil on a piece of 3/4" pine... and drew a line all the way around the perimeter. Cut it off with a cutoff wheel in a Dremel tool, sand the lip smooth and you're done. Total time was around 2 1/2 hours, cost (including bowl) was $23.90 plus the piece of pipe I already had for this project.

Not bad!
So as I get going on this separator project I will take pictures and post accordingly. Once I have the thing built we can do some real-world testing and see how all this math stuff works out (or not!). My woodworking skills really aren't much (cockeyed birdhouses mostly :o) and I look forward to learning....
Maybe Retired2 can help me build a router table! ;D

[retired2]

steamngn,

I think you have proved my point with your data.  If the difference between the worst possible end condition, i.e. straight pipe, and the most efficient, an elliptical bellmouth, is only 3% then how much difference could there possibly be between the latter and an off the shelf bellmouth.  Maybe you could find some very small measurable difference through modeling, but I submit you would never find it using an anemometer in a real world comparison.

A separator, whether it be a cyclone type or a simple top hat Thien design, imposes a tremendous performance hit on a dust collection system.  And with that being the case, you would think there would be lots of opportunities to cut those losses.  But the basic design has been around for quite a few years and in that time there have only been a few changes.  And those changes have only made small improvements in the losses.

I watch time and time again while people try to re-engineer the separator, and then build an entire piping system with short radius bend ells, and flex hose.  Anyone with an extensive plumbing system could do far more to improve the performance of their system by optimizing the piping system than by tweaking the separator.

Now, with regard to the router table, here is the secret.  Put $1000 in a place where its disposal cannot be monitored by whoever it is that watches what you are spending in your workshop!  Don't we all have someone assigned with that task?

[steamngn]

retired2,
Yes, I am proving your point... mostly! ;D
You are correct that this change will not necessarily be seen with an anemometer, but do not be fooled!
The increase in exit velocity will not change the system velocity (which is what you are measuring with the anemometer); what it WILL change is the efficiency of separation in the separator. By increasing the airflow at the bellmouth inlet we will decrease the airflow elsewhere in the separator:
The modeling showed that the increase in exit velocity slowed the vortex speed along the inner edge of the baffle by quite a bit. I can't say exactly how much the decrease is; The modeling software doesn't provide a speed at every point in the plane due to the complexity of the vortex, just a change of color of the flow line.
As for the financial monitor, well.... i married a bookkeeper! I don't think I could "misplace" $1.00 let alone $1,000! ;D

[DustySanders]

Steamngn - you made a comment in an earlier post in this thread that I think has been unduly ignored -

"Efficiency comes into play big time on the outlet side of things. Blowers (including exhaust fans for ventilation, which is what I am schooled in) are far better "pullers" than "pushers". much is being done to increase the efficiency on the "pulling" side while the "pushing" outlet side not so much.
4" pipe to a separator followed by 6" pipe separator to a blower with an 8" or 10" outlet with no bends... better to put an elbow at the top of the riser from the separator into the blower and have the blower exhaust in a straight line into the filter instead of the elbow on the outlet side.... of course space considerations come into play but you get the idea... I hope!  ;D"

This would actually be an easier set-up for me to build for my shop, and without having done a ton of thinking about it, it might solve some other problems as well.  I'm assuming that the bend from the top of the separator into the blower would be a large radius sweep that would follow the radius recommendations found elsewhere. Would you think that we would still benefit from including a straightening vane between the bellmouth and the blower? If so, would you have a recommendation as to where it would be best located - close to the bellmouth or close to the blower? And would you have any predictions about the amount of efficiency to be gained by having the outlet of the blower going straight into the filter?

Thanks in advance for all of your help.

[steamngn]

Dusty,
Sorry for the delay in responding, work and all...  ;)
Read on for some interesting updates:
We spent two hours trying to bend a smooth radius offset 90 in a piece of 6" pipe... only to find out we could not get the bend tight enough to fit in the space we had! It was at that point one of the guys said this:

It isn't necessarily how long the sweep is, only that it is a smooth radius.... right?

Well... RIGHT! So we took a conventional 90 degree elbow and examined the thickness of the plastic along the inner throat area; you know, where it makes the "sharp" bend. What we found was more than enough material to get in there with a port grinder (you could use a Dremel or sanding drum) and contour it into a smooth radius. Remember the bell-mouth discussion earlier in this thread? That outer 3/4" radius? Turns out that 3/4" radius was the magic number for the inside of the 90, and there was JUST enough material to get that in there (we blew a hole in the first elbow, be careful)! We flow-tested the connection in the attached image and it is almost EXACTLY the same as the model with a longer-radius 90. Lesson learned! As for the air straightener, you definitely want them regardless of whether your blower is close-coupled to your cyclone or not.. Straightening the airflow not only increases efficiency it also changes the characteristics of the airflow in the cyclone. In the modeling we noticed that straightening the airflow quickly in the outlet actually increased the velocity of the cyclonic air (the air going round-n-round in the separator). I haven't had time to finish the build yet (Yes, I promise to post pictures!) but the simulator indicates the best position is just above the separator outlet within the first 12"- 18' or so. We are working with 5 vertical vanes (don't have finish dimensions yet but the modeling indicates ~ 1-1/2" wide) that are 10" tall around the perimeter of the outlet pipe. As for cumulative efficiency, even without finished testing I can assure you it will be better. Have little or no resistance at the outlet prevents back pressure; back pressure will cause open-vane blowers like these to sheer and unload - basically the vanes "cut" through the air instead of "pushing" it... not good!

[retired2]

If you only have one tight radius bend in your system, you may not be able to measure a difference, but there are many published engineering tables that show there is indeed a big difference between loses in tight radius bend vs long radius.  So, it is incorrect to say, "it is not how long the sweep, only that it is smooth".  Also, there is published data advising against 90degree ells close to the blower inlet. That does not mean your system will not function, it just means it will not function as efficiently as it could.

Certainly any bend needs to be smooth, that is why flex hose has three times the losses of an equivalent length of smooth pipe.

[steamngn]

Ah Retired2,
I did not explain myself very well... :o
First let me explain what i mean when i say "modeling" vs "testing":

• When i say "modeling" we are running whatever part(s) through a computer simulation. This is awesome, but is not always able to take every odd thing that could change the outcome into account. For instance I continuously wonder what larger debris in the airstream will do.
  
• When I say "testing" (or sometimes "flow-testing") we are actually pushing air through the assembly, monitoring loss. Sometimes we also add colored smoke to the stream to see the effect. This airstream is typically generated from the flow bench and NOT from the cyclone blower, and sometimes changing the two changes the result. With the bench we can test CFM, Velocity, temperature change (friction)... we do not test amperage draw; Retired2 can attest to the fickle nature of what amp draw results tell us so we leave them out.

You are 100% correct that there is plenty of documentation in loss differences between tight and long radius bends; it is the reason we went to great lengths bending one in the first place. What we found when testing the two side-by-side after we modified the pre-made elbow was very surprising and interesting indeed.
As long as the airflow turbulence was minimal coming in to the elbows there was very little difference. Look, I'm certainly not a guru or all-knowing anything, I'm just reporting our test results.... even when they are sometimes not at ALL what we expect.
As for the elbow right at the blower- Right On! if you look at the image you will notice a short straight length of pipe; this pipe may or may not be long enough to prevent whatever turbulence the elbow introduces from fouling the blower. Haven't had the chance to test it yet, and that is why the blower is screwed to a chunk of 2x6 and not mounted on bushings yet. According to the modeling the stub that is on there should be fine if the airflow is good but when we introduce a small amount of turbulence then that stub needs to be a bit longer... or we need to address the turbulence with additional straighteners. Or something.

Keep in mind we are going at this a one big experiment; testing and challenging along the way. All this testing and fussing is why it is taking forever and a month of Sundays to get the job finished! like everyone else doing these things we're just looking for the best possible outcome for the application and restrictions at hand.
And hopefully shed some insight or helpful ideas along the way! ;D

[DustySanders]

steamngn, thanks for your reply. And no worries about the delay, I'n not in a huge rush.

I'll be able to use a large radius bend, and even have a section of straight pipe between the bend and the blower inlet, and that is where I intend to put the straightener, with the blower output going through a short straight pipe into one of the big cartridge filters. A whitepaper (Attached) that I probably found in one of R2's great threads is a very through study of air straighteners, hopefully you will find it interesting.

Thanks again!

[retired2]

steamngn, I do not dispute any of your measurements.  One tight radius ell in a fully plumbed system is not likely to make a measurable difference because it is such a small percentage of the total system losses.  However, your comments may lead others to think they can pipe their entire system with 1D bends from a big box store and think that it will not have any impact - it will!  And the impact is greater with smaller diameter pipe than large.

[steamngn]

Retired2,
You are correct, I am not being nearly as clear as I should be. Also important to note that the box store bends HAVE BEEN MODIFIED TO REACH THIS RESULT. I may not have made that clear enough either.
Also important to note that while the pic is of the short-radius bend THIS MAY NOT BE THE FINAL ASSEMBLY. This is during testing! That's why I haven't been posting pics yet...
Everyone in the shop was very surprised at the resultant airflow difference after what seemed like such a simple change. And we did test the short radius, long radius and the custom sweep to compare each. One thing that we have not finished testing yet is the effect of the short-radius vs long radius being so close to the blower. We want to know what differences (if any) there are on blower sheer compared to the distance the elbow is from the blower itself. It appears at this point that it is much more important to keep the air stream smooth coming into the bend initially; whether or not this solves that issue is still undetermined.

I would very much appreciate some feedback from you regarding real-world performance of your cyclone with the bellmouth installed. We did an initial flow test with one of the custom heat-blanket-formed bellmouths, and apparently the pipe had a crack in the inlet bend which failed. Since we didn't have another one made we decided to cut off the bellmouth such that the inlet was now more of a simple "flare" of approximately 45 degrees. When we tested it on the bench we found some minimal increase vena contracta as expected..... BUT.... BUT.... the cyclonic air action was much more circular and stayed closer to the perimeter of the cyclone.
This doesn't show up as any sort of volume/cfm/load change, but it could indicate a more (or maybe less!) efficient cyclonic separation, especially for fine light dust and the such.
So, putting all test numbers aside, did you happen to take notice of how the material moved around in your cyclone with and without the bellmouth? The design we are working with will allow quick-change of the inlet pipe, so we can real-world test different designs if there seems to be a benefit.
... after proofreading this post I have come to the realization that I need to get a life... :P

[retired2]

i cannot say any of my tweaks produced readily visible differences.  I don't have much faith in subjective appraisals about performance, too easily influenced by the placebo effect, and often the improvements are too small for observation.  My reason for every tweak was the measured data to verify it made an improvement. 

My system is quieter and moves more air with the tweaks I've made.  Can I see the difference that 50 or even a few more CFM's make?  Nope, but it has to help in some small way.  Possibly the pick-up is a bit better on my further-most tool bed.

My opinion of the Thien separator is it adds a big SP loss to a DC system, but probably no more than a cyclone.  In spite of that it works well, and more importantly is very forgiving of poor construction methods and material choices.  Conversely, no one has come up with any tweaks that make a truly significant difference in performance.  Bellmouths, air straighteners, rectangular inlets, 2X high chambers all just add small incremental improvements to the basic design.

People keep coming up with crazy new ideas, but I think they would be better served by just buying a bigger DC.

[DustySanders]

I'm sure at the time there were people who thought bellmouths, rectangular inlets, and double height chambers were crazy ideas, too. The great thing about this forum is the spirit of collaborative experimentation, imho.