Thread: Plenum concepts for DE pleated filters..?

Originally Posted by GreggMacPherson

I'm wondering what is the optimal way for the flow to exit the fan and enter the filter(s). I'm thinking about 2 to 3Hp systems, but something may be learned from concepts used on bigger systems... My haphazard scrounging for info hasn't yielded anything much. Sometimes it looks like the air has a short modified duct from the fan outlet to the mouth of the filter. But commonly, boxy shapes are used as a plenum space which then feed the filter(s).
For example..Plenum Boxes • Wynn Environmental

I'm thinking of any volume that connects the fan to the filter as a plenum. Can anyone point to something to read about this..? Does anyone have any ideas or opinions about the physics or design concepts...what the objective is..?

I'm guessing that one wants minimum energy loss before the air exits the filter. So I wonder about..
- Smooth duct shapes with an expansion to the filter diameter that might minimize drag.

Yes ideally smooth shapes help but as long as the pathway inside the plenum has a greater cross sectional area and the smallest dimension is greater than any ducting minimum dimension the plenums are not usually the rate limiting step. At work the smallest filters we used for a single 150 mm ducting was 300 x 300 x 150 and to house these we used 300 x 300 x 300 boxes. For multiple 150 mm ducts and larger filters we upscaled accordingly eg for a 1200 x 600 filter we used a plenum of 1200 x 600 x 150 .At these sizes the shapes of the plenum didn't seem to make any difference

[QUOTE]- The boxy shaped plenums will have large unstable eddies in all the corners, that shed off the main flow and increase drag.[QUOTE]
150mm ducting and most DC air speeds is already unstable and will have eddies - provided the plenums are sized as above the plenum eddies are not that significant. If you want to reduce these the corners can be filled with curved pieces of sheetmetal/plky/plastic and filled with expansion foam. We thought about doing this a few times but never ended up doing it.

[QUOTE]-- Any expansion of volume means a a velocity reduction...so where is the ideal point for this to occur..?
- The energy loss from the filter. Is this the only unavoidable thing..? Is there an optimal pleat density (n pleats/perimeter) and geometry..?

It's pressure loss that filters drop. The best way to minimise this is to use as large a total area of filter as possible, and of course keep filters clean.

-Re that last one, some of the filters being used have so may pleats that one wonders if it impedes the air flow, and cleaning would be obviously compromised.

If the area/volume of the filter drum or plenum is large then having many fine pleats doesn't matter. The finer the pleats the greater the escape area pathway and this usually leads to least pressure losses and better the filtration. Hospital grade type HEPA filters typically have 6-7 folds per inch.

One thing to think about, unless you use a cyclone its better not to have the air stream come direct from a fan and directly impact on any section of a pleater filter because if a large chip or piece of sharp wood or metal does this it can penetrate the filter so either you will need a deflector or offset the filters away from the air path. Another way to protect the filter is by laying a coarse prefilter over the top of the pleated filter but that means cleaning two filters. We did this quite often at work but we had so little dust on the coarse prefilter we were able to put them into a washing machine to clean them.

Thanks Bob. I should have said that a separator of some kind was assumed to be in line before the fan. ...


I can see that some plenum boxes might help protect the filter(s) if a stray chip made it through..maybe the streamlined low drag version less so...I wish I had a feel for what drag is significant in the system, but I don't yet.
Ditto for the form (shape and scale) of the turbulence and instability.

I did buy some cheap junk that may let me experiment if I can find the space and time...an old two bag 2HP Holytek in bits with two 485x600 (wxh) filters. The filters may be an early generation that's been in storage. They have 130 pleats, each pleat with a perimeter of 53mm, so I measure 4.1m^2 per cartridge. They have full length brushes for the paddles and construction seems robust. The geometry is fairly open, looking like it will clean easily.


Gregg.

In commercial setups the filters are usually arranged in a box with the air flowing from the outside to the inside of the filters. You might want to look at the general arrangements they use.

Cartridge Dust Collectors | Donaldson Industrial Dust, Fume & Mist.

Dust Collector Services :: Case Studies

Pete

Yea, I've bumped into material about units like these. I don't know why they like horizontaly oriented cartridges, except that they are easier to swap out. Having made that decision they then go on to redesign the cartridge section from circular to that slightly toroidal shape..

What happened to the Canadian forum Pete? Some text went missing and traffic is down to near zero...

EDIT...Sorry, toroidal is the wrong word. I've forgotten the word describing that rounded triangle shape..

A plenum is usually appropriate when you have more than one exit, like examples 2 and 3 from your Wynn link. A plenum adds the entrance loss of the openings to the filters so that cost goes down with the square of the number of filters in parallel.

The entrance loss to the filters depends on the inverse fourth power of the filter's inside diameter so, for a single filter, it's (1 + K)x(fan exit id / filter id)^4 where K ranges from about 0.03 to 0.5 depending on the rounding of the edge. The loss factor for an elbow is as high as.7 for a sharp (.5 r/D) bend. In the US a commonly available sweep is .75 r/D which has a loss factor of 0.33.

For example, assume 150mm fan exit and 200mm filter ID with modest rounding, K = .15, and compare to a 150mm .75r/D elbow. 1.15x(150/200)^4 = .48, only slightly more than the elbow. Going to two filters that drops to .12 which would be very difficult to get with a ducting split.

Note 1 - The above leaves out a lot of detail and is an incomplete description, only looking at the major parts.

Note 2 - "...a separator of some kind was assumed to be in line before the fan..." will have much more loss than the filter connection and is probably more deserving of your attention.

Originally Posted by GreggMacPherson

Yea, I've bumped into material about units like these. I don't know why they like horizontaly oriented cartridges, except that they are easier to swap out. Having made that decision they then go on to redesign the cartridge section from circular to that slightly toroidal shape..

What happened to the Canadian forum Pete? Some text went missing and traffic is down to near zero...

The second link has a little on the change from vertical to horizontal. Having been an aircraft mechanic, anything easier to do gets done more often. I thought that little doghouse roofs over each filter would shed the dust cake from above to the side so it would drop to the hopper below when the pulse jet cleaning cycled rather than fall on the ones under.

The Canadian forum hasn't been fully exorcised of the problems since it upgraded (vBulletan being slow to fix and the forum having a small technical staff). They are a small forum and activity varies, the problem exacerbating participation.

Pete

Later addition to the post. Looks like I wasn't the first with the idea of partially covering the filter to keep dust off the topside. https://www.aqcdust.com/wp-content/u...-collector.pdf

Dave,

"A plenum is usually appropriate when you have more than one exit..."

My curiosity is naturally attracted to the form of whatever volume connects the fan outlet to the filters...the possible configurations for small dust collectors...and of course wondering what is efficient. I read a good piece by you where you incrementally add the elements to the outlet of a 2HP DE fan and plot pressure vs flow rate. I was surprised how bad the seperator dish contribution was. Maybe that's one of the worst ways to configure the filter, and any plenum shape or short expansion duct to the filter diameter is better..

Just to be clear, this pic below is the kind of "plenum" I was intuiting for a single vertical filter stack. Starting rectangular and ending round, with the full filter ID. (drawing shows less diameter than that...I didn't have time to change it). Or a two filter plenum version. I suppose, putting the idea simply, one might try to idealize the velocity and pressure distribution in the volume and eliminate parasitic separated flows.
IMG_20200918_153930_1.jpg

Or is it a waste of time....someone has already measured the pressure increment for clunky rectangular plenum volumes and it is low...?

When you are looking at the "entrance loss to the filters" above, it's ignoring the shape of the plenum for now, right...? Any easily accessible reading online to catch me up on how that calculation arrives, and what K is...?

Gregg.

Let's start with "K". K (usually has a subscript, often upper case "L") is the ratio of heat energy lost per unit volume of the flow to the kinetic energy per unit volume. For fluid flow energy per unit volume is usually expressed in units of pressure. So "velocity pressure" for kinetic energy and "static pressure" for potential energy as well as dissipated energy.

There are lots of tabulated K values for various geometries. Just be careful with changes in velocity from input to output and verify which one is used for the reference. It's usually the smaller diameter, i.e., the higher velocity.

Let's look at an entrance:

BMHRadius.jpg

For example, in Imperial units, 6" duct, 800 CFM is 4000 fpm and 1" velocity pressure. So just to get the air flowing Bernoulli tells us that we need to use 1" of static pressure to get 1" of velocity pressure. In addition, there's thermal loss. From the graphic we we see that loss can be anywhere from 3% or less up to about 80% of the velocity. So the total pressure inside the entrance will range from -1.03 in w.g. to -1.8 in w.g. relative to outside the pipe.

------------------

A plenum, as I was using the term, is an expansion (velocity reduction) in which the relative velocity pressure is low enough to be ignored. So, (almost) all of the kinetic energy is dissipated, just like it was emptied into the room or outdoors. Since velocity pressure in a duct varies with the fourth power of the diameter it really doesn't take much to meet that definition.

So let's say your plenum has about twice the effective diameter of the fan outlet. Then any additional losses due to plenum geometry are effectively reduced by a factor of sixteen. That is, being able to ignore those losses is essentially my definition of a plenum!

-----------------

Your tapered sweep 90 probably has a bit less loss than a non-swept 90 but my guess is that it wouldn't be worth any added effort. My eyeball measures your sweep to be about 2.5 r/D which has a K of about .12 so not much room for improvement.

----------------

Bottom line, for a single filter or a stack with a single entrance your fabricated long sweep elbow should have less loss than a plenum. For two filters in parallel a plenum would be my choice, and shape doesn't really matter as long as the flow path is sufficiently large.

Thanks for your patience Dave... a very useful nudge.. I'll try and do a little reading.

Gregg.

While the curved corner principles for port entrances and exits are correct the diagrams shown by Dave above assume a large open volume on the LHS but your situation is not like that as it is enclosed..

This means you may still get turbulence in the corners like this.
Screen Shot 2020-09-30 at 6.01.15 am.png

In that case I would just go with a fully tapered junction - something like this which is really just an extreme example of curved corners.
In the first instance I would go with the solid lines.
The dotted below line may be better but that requires modding the impeller output and this could disturb the impellers performance.

GMacsImpeller.jpg

Unfortunately making a 3D curved round transition is not easy.

If you make it large enough you can make it rectangular in cross section and tapered in length
This one is not tapered but would be easy enough to make it tapered.
Mens Shed Dust Collection

Hey Bob,
your sketch with the solid red is close to one version of my idea. The expansion (transition) ending with close to the filter ID. Dave promted me to think about how low the velocites are going with the expansion. I assume drag is ~ V^2, so I get the ^4 relationship to radius. The separated vortices that may exist in corners won't matter much if the velocities are low enough.


Anyway, looking at possible forms for transitions with streamlined flow from fan exit to filter entry, the velocity (area) ratios are big. This drawing isn't quite finished. Imagine it with the transition exit diameter = the filter ID. The elbow sweep could change to almost anything. Vanes could be used if useful.

filter plenum 344mm exit.jpg


Transition geometry...
Inlet wxh = 150x125=17671mm^2.
Dh, effective diam = 136.4m^2.
Area from Dh, call it A1= 14612mm^2


Outlet diameter = 430mm. (the pleated filter ID)
Area of outlet, call it A2 = 145220mm^2


A2/A1 = 9.9
And (A2/A1)^2 = 98.....!


And the surface area of element slices at the entry and exit have ratio 433/136.4 = 3.2.
So, drag from the surfaces of the "plenum"...outlet/inlet ~ 98/3.2 = 31


Nonsense..? Bob n Dave will probably correct me.
Re the difficulty of making these shapes...fairly easy and cheap to make using methods commonly used for composite prototyping...

EDIT..sorry, fixed a couple errors...
EDIT...fixed units

Originally Posted by GreggMacPherson

Anyway, looking at possible forms for transitions with streamlined flow from fan exit to filter entry,

Hummm . . . . . it definitely won't be streamlined coming out of such an impeller. Nevertheless one does try to keep things as smooth and as large radius as possible so a large bed 3D printer would be handy at this stage.

Bob, do you have some insights into what the air is doing as it exits the fan? It would seem a little weird if we have arrived at a point in history where the sims (CFD) boys have a better idea than the theorists or experimental guys. I'm keen to know, is the air in semi striated layers that are sort of homogenized, and if so, what is the scale of the rotational flow that is going on in each layer....Are there big wads of air that are thwacked out from each paddle in a rythmical pulse... Please only entertain this if it is really interesting.

[QUOTE=GreggMacPherson;2211066]Bob, do you have some insights into what the air is doing as it exits the fan? It would seem a little weird if we have arrived at a point in history where the sims (CFD) boys have a better idea than the theorists or experimental guys. [QUOTE]

I don't think it's that weird because measuring this is a bit of a nightmare.
Some really interesting CFD pics and vids here.
Computational Fluid Dynamics for fans and plants

I'm keen to know, is the air in semi striated layers that are sort of homogenized, and if so, what is the scale of the rotational flow that is going on in each layer....Are there big wads of air that are thwacked out from each paddle in a rythmical pulse... Please only entertain this if it is really interesting.

As each paddle passes the exit rim this will cause some pulsing.
See vids in links above.
I reckon that the way the filter bags on big DCs (high flow) vibrate might be a result of pulsing - this could be useful if it knocks dust off filters?