J. Phil Thien's Projects

Quote from: dbhost on July 20, 2011, 10:57:48 AM
How is it then that Lee Styron can make a profit with his Shark Guards? They are an excellent, small shop built product that is IMHO second to none. And I am relatively certain that he isn't farming his production out to some sampan factory in China.

Yes I know he charges more than $75.00 for his guards, but they at least to me, appear to be far more complex than an impeller for a dust collector...

A piece of rotating equipment, especially a bladed, vaned impeller, is far more complex than it may appear, that's why.  We're talking about fairly sophisticated design, tooling, parts manufacturing, assembling, balancing, and testing processes.

Jim:
Yes, the systems I have worked with are somewhat unique.  What I am experienced with is known as the fluidized bed catalytic cracking process, which is used in oil refineries to make gasoline components.  It uses an air blower at the front end and cyclones at the back end to separate contaminants from the air before it re-enters the atmosphere.  

One part of the process takes filtered air at atmospheric pressure up to about 20 psig using a big axial blower.  Then the air is introduced into a vessel operating at 1400 degrees where it mixes with talcum powder-like catalyst that is contaminated with carbon.  The carbon is burned off from the catalyst in this hot vessel.  Then the carbon-free catalyst is separated from the air using cyclones, which are good to about 20 microns, followed by an electrostatic precipitator, which is good to about 5 microns, and finally followed by a water scrubbing process to remove sulfur oxides and the last of the catalyst.

As I mentioned, I think there is a real benefit for designing a scroll at the suction side of of our shop separator top-hats, and that the pilot plant data used in designing these industrial cyclones can be of potential use.  What will be important is how the data were used in the scale-up.
Bill

QuoteHas this design been compared to a 'traditional', side inlet top-hat design?  
Have any consistent measurements been made?  

I would be very interested in knowing these results.  I am not aware of scroll designs being implemented on the suction side of the system.  They have been in use in blower designs for at least 75 years.  It is a proven way to simultaneously increase both velocity and pressure within the blower housing.  But I have some questions about their effectiveness on the suction side...  Some empirical measurements will be quite enlightening!  

Also, are the data listed in your first image based on the dimensions listed?  That is a HUGE design.  I am not certain how much air would be needed to make a 4' diameter separator work effectively.  I have a 10hp, 4000cfm system and I know it could not drive a 4' diameter Thien separator...  

My largest separator is 24" but all new ones are <16".   I have a 12" unit that separates MDF fines down to <3 microns - sampling shows NO particles 3 microns and larger.  On that unit I paid an extreme amount of detail to edges and corners.  

I will look forward to any measurements that Bill can provide.

Jim:

You are correct.  The cyclone suction drawings and separation results I show in the earlier post are reflective of industrial cyclones operating downstream of blowers in a low positive pressure environment, rather than small cyclones in vacuum service like the separators discussed on this forum.  But I would argue that the separation efficiency is a result of pressure differential and velocity through the device, so absolute pressure in the ranges we're talking about shouldn't be a big factor.  These industrial cyclones are huge, often operate with four or six in parallel, and are driven by large 4000HP motor or turbine driven blowers.

But the industrial designs are based on empirical pilot plant data generated in set-up's not too dissimilar in size from our shop separators.  I will do some digging around and see if I can find the pilot plant data that went into the design of the inlet scroll feature.  It's out there.  I just need to find it.

Bill

I have had good luck with the 30 gallon metal trash can also.  The concern is air leakage into the system rather than dust escaping.  For a lid, I used 3/4" plywood with a diameter about 3" larger than the can opening.  To secure the lid to the can, I cut a groove into the lid about 3/8" deep with an inside diameter just slightly larger than the ID of the can.  This gives a tight friction fit.  To cut the groove I used a 3/8" rabbeting bit and a router circle cutting jig.

To test for leaks, I turned on the system and smoke-tested the can seams and the lid-to-can fit.  No leaks at the seams, but I found a fair amount of leakage between the can and the lid -  right at the two lap joints on the rim where the can is press-fit together.  I mixed up a little bit of Bondo and smoothed over the two lap joints, which solved the problem.

I agree with you, Chuck.  Intricate machinery, if well made, requires a lot of labor to manufacture, despite all the tooling - and even robotics employed today.  You cannot compete as an individual against industry that can mass produce.

I once built a prototype replacement steering wheel for a vintage Mopar muscle car made out of solid walnut, with the hope building a lot of them and turning a decent profit.  It was an exact duplicate of the cheap imitation plastic original.  Lots of angled tenoing jigs, circle templates, glue-up jigs, etc. and also many hours of labor.   It generated a lot of activity on ebay - eventually selling for $410.  I made a couple of more, but never got my labor costs above about $2.00/Hr.

In industry we always design the inlets as rectangular.  We also offset the inlet from the body of the cyclone to improve efficiency.  In other words, rather than having the outer wall of the inlet ductwork in the same vertical plane as the cyclone wall, we offset it by a certain amount. This is called a scroll, which helps the dust particles stay closer to the wall and keep them from impinging on, and escaping through the outlet tube. 

The pictures below show what I am trying to say and give an idea of how a scroll will help separation efficiency.  The inlet ductwork starts with an offset, but gradually returns to the same dimension as the cyclone wall through the first 180 deg of sweep.

I built my separator with a top inlet, but am now redoing it with a rectangular side inlet to include the scroll feature.

You would have to knock out quite a number of impellers at $75 a pop just to recover design and tooling costs.  Rotating equipment, especially a well balanced bladed and vaned impeller, is a lot more complex than it appears, in my experience.

You can also use a belt sander to scarf off enough material from ABS pipe to make a good friction fit.  I took about 3/32" off of a 4" ABS SD coupling to make a very tight connection to 5" flex hose. See the image below where I used a coupling as the separator outlet and modified it to fit 5" flex hose.

mj2736:
Very good.  I will do the tests.  Please PM me your email address and I will keep you informed of progress, including testing protocols and results.
Regards, Bill

mj2736:
These look like very interesting and potentially valuable upgrades.  I have a 31 gal set-up, but am using 4" plumbing.  However, I like your ideas, and since my unit is already finished, I can easily run some tests with both of your suggested upgrades.  Directionally, should your upgrades work on 4", then they should work on 5".  I already have the ability to determine the additional pressure loss, if any, that would come from the enhancements, and can also figure out how to determine the efficiency of each of your suggested alternatives.

If you are interested, please post as so, and I will get it done.

Thanks again for your great ideas!!
Bill

Quote from: painterman on April 12, 2011, 12:44:54 AM

Quote from: bill70j on April 11, 2011, 07:30:38 AM
Painterman:
To answer your question about the losses through a separation system.  They can be huge, depending on flow.  At max flow they can be as large as 60% of the unit?s capacity.  And that is what you are feeling with your hand testing.  But under normal conditions, they are probably closer to 10 ? 15% of total capacity.  I agree with others that your issues are likely not with your top hat system.

Here is my experience with separator losses.  I built a trash can separator for my Harbor Freight 2HP unit and measured the actual vacuum at two locations with a gage, just like you did with your hands.  You can see from the images below, that at the separator inlet under no load, the vacuum was 1.75" of W.C, but at the blower inlet under the same conditions, it was 6.0" of W.C.  So at maximum flow conditions, the loss through the can was 4.25" or almost 60% of the total capacity of the unit (which is about 7.3? of W.C.).  I did smoke testing on my unit, so know there are no leaks and the losses are real. 

The reasons for my losses are due to the friction loss through two 90deg elbows, the 45deg wye, plus other friction losses from the air moving from the can back into the blower suction pipe.  Wood Magazine just did an article on calculating losses in a DC system under normal conditions (350 CFM).  If I use their spreadsheet, I come up with a predicted loss for my separator system of 1.05? of W.C.  But that is at normal flow conditions.  At max flow conditions for my unit of about 650 CFM, that figure translates to 3.62? of W.C. which is not too far from the 4.25? I measured.

In Wood Magazine?s 3/2008 article, they show your Grizzley1029 as having  something like 900 ? 1100 CFM capacity at max flow conditions and a max static pressure of 9? of W.C. ? far superior to my HF unit.  So you will see even larger differences of loss between max and normal flow conditions.

So are you saying that everything is normal and that my test was really an unreliable test?  You say that I will see an even larger differnce  of loss between max and normal flow conditions.  Is that bad?
Thanks,
John

John:
I am saying that if your system is hooked up, then the test you did would give a different result - you would feel much less difference between the suction at the separator inlet and the suction at the blower inlet.  That's because with the system hooked up, the flow is much less due to the friction loss through all the parts of the system, starting from the machine you're sucking on back through the flex hose, the ells, the wyes, the blast gates, the reducer, and the ductwork.  So if you do a test with all of that hooked up, your flow will be reduced from your max blower capacity of 1100CFM to much less that that, say 550CFM.  The pressure drop through your separator at 550CFM vs the loss at 1100CFM would be four times less.  And I am saying that the bigger your blower, the more discrepency there would be between your original hand test and the real result with the system hooked up - not a bad thing, just a different result.

Painterman:
To answer your question about the losses through a separation system.  They can be huge, depending on flow.  At max flow they can be as large as 60% of the unit?s capacity.  And that is what you are feeling with your hand testing.  But under normal conditions, they are probably closer to 10 ? 15% of total capacity.  I agree with others that your issues are likely not with your top hat system.

Here is my experience with separator losses.  I built a trash can separator for my Harbor Freight 2HP unit and measured the actual vacuum at two locations with a gage, just like you did with your hands.  You can see from the images below, that at the separator inlet under no load, the vacuum was 1.75" of W.C, but at the blower inlet under the same conditions, it was 6.0" of W.C.  So at maximum flow conditions, the loss through the can was 4.25" or almost 60% of the total capacity of the unit (which is about 7.3? of W.C.).  I did smoke testing on my unit, so know there are no leaks and the losses are real. 

The reasons for my losses are due to the friction loss through two 90deg elbows, the 45deg wye, plus other friction losses from the air moving from the can back into the blower suction pipe.  Wood Magazine just did an article on calculating losses in a DC system under normal conditions (350 CFM).  If I use their spreadsheet, I come up with a predicted loss for my separator system of 1.05? of W.C.  But that is at normal flow conditions.  At max flow conditions for my unit of about 650 CFM, that figure translates to 3.62? of W.C. which is not too far from the 4.25? I measured.

In Wood Magazine?s 3/2008 article, they show your Grizzley1029 as having  something like 900 ? 1100 CFM capacity at max flow conditions and a max static pressure of 9? of W.C. ? far superior to my HF unit.  So you will see even larger differences of loss between max and normal flow conditions.

Thanks again for your help, Dave.  Everything seems to work.  I haven't decided what to do with the manometer.  Maybe leave it hooked up to see if it can help determine when to backflush the Wynn cartridge.  But that may be a loooong time from now judging by how well the Thien separator is working.  I have had virtually nothing carry over into the HF poly bag after dumping several trashcanfulls of stuff.

I completed pressure testing and installing a vacuum relief device to prevent my separator can from collapsing under dead-head conditions.  Thanks to Vodkaman (3/27/11 post) for your ideas on the design.

I began by installing a pressure gage on my mobile rig as shown in the very last image.  Then while watching the gage, began closing the blast gate on the inlet wye until I saw the can start to deflect.  That happened with the blast gate about 90% closed at a reading of about 7.2" of W.C.

I converted the 7.2" of W.C. to psi.  (7.2/12X14.7/33.91) = 0.26psi.  Then I converted that to pounds force on a 3.5"D disc.  (0.26X1.75X1.75X3.14159) = 2.5 pounds force.  I took out my box of springs and, using the kitchen scale, found one that compressed well to the 2.5 pound figure (assuming for this design that pounds mass and pounds force are the same).

Then I got materials I had laying around to build the device.  I had on hand the 1/2" MDF, 3/4" dowel stock, Lexan, 5/16" hex bolt, and 2" ABS SD pipe - so needed only to buy the 3.5" O-ring.

I built the device, installed it, adjusted the spring to the 2.5 pound mark and fired up the blower.  I began closing the blast gate.  And as it approached the fully closed position, I could see the spring begin to compress, opening up the disc.  That happened at about 7.3" of W.C.  So it looks like my can is safe under all conditions.

Hope this can be helpful to others.

Dave:

Thanks for the input.  Appreciate your test.  And great idea to start out with too many holes.  I had not thought of that. 

So I am going to go out and start some testing on my setup.  Using a blast gate to vary pressure, I'll hook up a manometer of some sort and also rig up a dial indicator to measure can deflection.  At varying pressure levels, I'll decrease the number of holes in the membrane until it fails.

Also, I think I'll consider your rupture disc design vs. the relief valve for a permanent solution.  Sounds a lot more simple.  Have you given any thought to size?  Maybe I can get some design info from my test data.

Bill