measured speed related to system efficiency?

[PerKr]

I'm trying to measure the change in efficiency for a duct system. As it is, the fan is blowing air straight into a wall and the air then travels down a channel towards a series of outlets (imagine a fan situated in a short circular tube connected perpendicularly to a flat and wide rectangular duct). I am looking at the design of the inlet as well as the geometry around the fan. All I have to check whether or not I have found an improvement is an airspeed indicator. Now, if I place this a bit down into the channel I can see how the airspeed changes as I use different inlet geometries and different geometries around the fan. Am I right in assuming that an increase in the measured airspeed indicates fewer aerodynamic losses?

 

[rb1957]

you've got multiple outlets and only one probe ? i assume you're proposing to measure only one outlet ??

you're assuming that your changes in duct geometry aren't affecting the distribution of the airflow. an increase in airflow in one channel could be due to two causes ... reduced losses and redistribution of the airflow (from one outlet to another).

i'd measure each outlet with each setup.

could you put a force balance on the plate where the fan impacts ? this'd measure the lost air pressure.

 

[MikeHalloran]

You may see an improvement by putting a very shallow cone or wedge splitter on the planar surface where the fan discharge impacts.

That's a standard improvement to gas turbine exit plenums, which are it seems always designed first in the worst possible way 'for packaging reasons', even when the causative packaging limitations are not present.


You might do well to study the illustrated tables of duct component losses in the ASHRAE Handbook.




Mike Halloran
Pembroke Pines, FL, USA

 

[rb1957]

"the fan is blowing air straight into a wall and the air then travels down a channel"

this thing sounds like a "piccolo tube".

is the fan blowing into a closed chamber (plenum) with only one outlet ? and you're improving the outlet orifice geometry ??

 

[PerKr]

maybe I need to explain a bit better (and if this doesn't do it, I should probably provide you guys with some sketches to show you what I mean. I'm not very good at explaining stuff anyway). Are you guys familiar with frost-free freezers? that's what this is. The system is divided into two major components: a box at the top of the compartment containing the evaporator and fan and a duct along the back going from the box down to the bottom surface.

The duct in this case is wide and flat. The width is about twice the fan diameter. This duct geometry extends to the top of the evaporator housing.
The fan (axial fan, 100mm diameter I think) is situated at the very rear of the evaporator housing, about 40mm from the rear wall. It's perpendicular to the wall and the motor is behind the fan (that is, it's in the duct)

The duct has several small slits along its length where the air is meant to escape and cool whatever is in the storage compartment.

The things I can work with are the inlet and adding components in the duct. So far, my only idea has been to add styrofoam components to make the top end of the duct conical to stop air from moving upwards and to the sides but rather directing the air towards the base of the cone (inspired by this design:

http://www.tradingdepot.co.uk/external/commerce/1/gfx/hires/MAN40400.jpg).

For testing, I only have the evaporator housing with fan and inlet and half a duct. My goal here is to compare the effectiveness of various designs for inlets and geometries close to the fan outlet, so I might as well (no, I "should") tape over the slits in the duct for now.
Airspeed is measured at the end of my short duct using an instrument we got from some other R&D department which was closed down years ago(I'm not sure what the type of sensor is called. it doesn't use a propeller. I think it's based on a comparison between current and reference temperatures of a heated element).
Obviously, we have very limited resources when it comes to measuring pressure, flow, acceleration...

My thinking was that if the inlet is restrictive, this would show as a lower airspeed at the open end of the short duct (seems to hold true judging by initial test results). I then figured that when I vary the geometry around the fan outlet, an increased airspeed should indicate an improvement there.

Once I have a couple of solutions to reduce losses related to the inlet and geometry behind the evaporator carrier I will investigate what happens at the slits/outlets for the complete duct and perform energy consumption tests (those take at least 2 weeks though, so if the results of my simple test doesn't indicate anything I'll just keep being a frustrated engineer crying for the fancy software we're not allowed to have)

MikeHalloran: There is a recess in the back wall behind the fan motor for packaging reasons, but I could try to make some sort of geometry around this recess ressembling what you suggest. Thanks for the suggestion.

 

[MikeHalloran]

I think the sensor you describe is a hot-wire anemometer.

But I think maybe you're optimizing a small part of a big system, and optimizing the duct in isolation may/will compromise the performance of the system.

What I'd like to see for something like that is a 3D array of fixed temperature sensors, located throughout both boxes, to give you an idea of how well the fan is working. I.e., measure the gradients, and use that as a baseline; you don't want to make them worse.

In addition, I'd like to see a couple of hot-wire anemometers, mounted by gum or tape or something, so you can move them around and 'see' the airflow that you are trying to modify.

I.e., I don't think that optimizing the duct in isolation is going to improve the product; you need to sell the idea of working on the product as a whole. On the bright side, multichannel temperature recorders/ transducers are not hugely expensive, especially since your environmental requirements are relatively benign.




Mike Halloran
Pembroke Pines, FL, USA

 

[berkshire]

PerKr
Your housing looks halfway to being a centrifugal fan.
Why not put a centrifugal in there instead of the axial flow fan.
You will get better airflow if you do it right and the air will be directed down the duct. Now it is being accelerated by the fan, stopping at the back wall and using static pressure regain to get the turn.
B.E.

 

[PerKr]

berkshire: it's all about cost. That and our lack of knowledge in aerodynamics and fan blade design (someone tried designing a radial fan blade for use with this motor some time back and the results were pretty poor).

Mike: you're probably right. and I'm sure my upcoming measurements will show that. So once I have improved the flow inside the duct, I will need to figure out how to get the air out of the duct, without messing with the surfaces visible to the customer.

 

[rb1957]

thx, the 2nd explanation helps lots.

how much more efficient do you think you can be ?

how to the duct areas compare ? the area of the fan exhaust ? the area of the rectangular duct ?

i was thinking one failed experiment doesn't doom the concept (using a centrifugual fan instead of axial. my 2nd thought was you really need axial flow along the rectangular section, which you won't get (well) with the centrifugual fan.

i think you're thinking that the fan exhaust hitting the flat side of the duct is causing a lot of the momentum losses that you think you have.

i wonder if strakes along the rectangular duct will help (to align the flow).

"The duct has several small slits along its length where the air is meant to escape and cool whatever is in the storage compartment." ... this is the useful work done by the fan. how are the slits arranged ? there'd be more dynamic pressure (= airflow) at the beginning of the duct and less at the far end (where most of the airflow has been exhausted into the cooler). so you might get better results (more uniform airflow into the cooler) if the slits have a varible distribution (fewer at the beginning, more at the rear) ?

how's the duct sized around the fan exhaust ? if the duct is a little bigger (and open to the atmosphere) you'll entrain air from the surroundings and have more flow in your duct. careful too much bigger and air will escape instead.

 

[PerKr]

the slits are rectangular cutouts in the face of the duct with the short side of the rectangle perpendicular to the flow. There is no geometry to help guide the air towards the outlets.

According to an analysis we ordered from a company late last year, the highest pressure is just behind the fan (imagine that) and at the bottom of the duct. Looking closer at the lower part of the duct this becomes rather obvious as the duct is almost completely closed off by a wall. The idea behind that wall was to avoid having all of the cold air going into the bottom drawer. The same results should have been possible to reach by redirecting the air towards the outlets, but without the time and knowledge to work that out the guys who designed the duct took the easy way out. Can't blame them.

I intend to test a couple of ideas for directing the air towards the outlets:
*the first idea was to have angled and slightly radiused fins to guide the air
*the second idea was to simply put a shelf directly under the outlet to build up pressure locally
*the third idea, which might be a bit far fetched, was to put a shelf directly above the outlet to get the air to sort of tumble out the outlet (doesn't make a lot of sense, I know, but I might as well test it since it crept up on me).
either way I will have increased turbulence, but hopefully it won't be too much of a problem as it should be concentrated towards the duct wall where heat transfer shouldn't be a problem (as opposed to the back surface where this would mean increased losses to the outside of the product)

 

[MikeHalloran]

Given that you lack CFD, the next best thing is to take the test rig outside and blow smoke into the fan inlet, and then into small holes near the fan discharge, and take lots of photographs, to try and map the airflow.



Mike Halloran
Pembroke Pines, FL, USA

 

[rb1957]

i understand your 1st idea ... 2 and 3 leave me wondering about what this "shelf" is and why you'd want to build up pressure ?? i'd've thought you would be trying to direct (encourage) the airflow into the rectangular duct.

is the system performance acceptable ? and are you "just" trying to improve it ?

is the system unacceptable and you're trying to fix it ??

if the 2nd, then i suggest you try more radical things ... align the fan with the duct, move the fan so you have more space to adapt the fan outlet to the duct.

 

[PerKr]

The system is acceptable, but still needs improvement as the requirements keep going up for energy consumption and cost reduction.

But yes, more radical solutions should be tested as well in case the opportunity of a re-design presents itself. Because sooner or later (when the customers start looking for A+++ products in stores), the current system will be unacceptable

you're right, #1 does make more sense than the others.

 

[rb1957]

the current dsign is compromised by the design decisions taken. this isn't to say they were bad decisions ... clearly not if the system is acceptable.

i'd suggest an intelligent way forward is to study the current design ... how much usefull airflow (energy) is being provided ? measure the airflow (and temp?) at each slot. could you release the structural attachments and measure the force needed to keep the duct in place (ie how much force is the airflow applying to the duct face it's hitting) ?

an alternative (less data driven) approach is to try stuff you think'll work ... reposition the fan so that it's in-line with the duct ... should be easy enough to rig up.

 

[PerKr]

Ok, so the idea of shaping the channel to a cone around the fan outlet worked. Gave me a 10% increase in airflow out of the channel outlets and 5% decrease in energy consumption.

I'm now concentrating on the frond end of the system, the inlet. The original inlet had several small ribs and we know from a CFD report that this creates a wake at the bottom of the module.
A new design used fewer and larger ribs. This should have caused a larger wake at the module bottom. Measurements also show that it reduces the airflow. This inlet increased the energy consumption by 2-3%.
An idea I tested showed an increase in airspeed both at the inlet (major increase) and the outlets, but the energy consumption was only marginally better compared to the large-ribbed design. For this design, which was based hugely on the first design, the inlet was stripped of all ribs except the bottom one and the wake area was filled with a piece of polystyrene to get higher airspeeds. I'm thinking this also meant that the flow became mostly laminar and that's why no real improvement was seen.

I'm now testing this third inlet with the addition of a grid (basically a steel plate with lots and lots of holes) to cause turbulence. The question is though: how do I estimate whether or not the air will still be turbulent as it reaches the evaporator?