Wind Tunnel Construction

[bigben2k]

I took the blower apart last night and this morning. Good news, it's reversible (the housing can be mounted on either side). In plain english, I can screw up on the rotation (CW or CCW) for the replacement motor, and make up for it.

The not-so-good news is that the blower wheel isn't balanced. There's been an apparent effort made to balance it (i.e. small disks drilled out on the perimeter) but not good enough; the wheel freely rotates to the same position. The outside hub appears to be perfectly centered, but the inside one is off.

I'll look at this issue again, when I have the replacement motor (no spare shaft to play with, right now).

Otherwise I may have found a (free) source (co-worker) for a plastic tube in 24" diameter, but the length is in question.

Now focusing on the temporary rig for the temperature controller test.

[bigben2k]

Am getting parts for a test rig.

Gonna have to be patient on the replacement motor; they're more scarce than I anticipated. In the mean time, I'm reselling unused components.


Does anyone have opinions on a plastic nozzle? (I may be able to get them fabricated).

[BillA]

those nozzles are critical, absolutely to spec - or cal the tunnel, ~$4000
lets get that spec first

re the balance, dynamic is what is relevant
wait 'till you can spin it up

[bigben2k]

Got a replacement motor:
http://cgi.ebay.com/ws/eBayISAPI.dll...tem=7592920115

Leeson 1/3 HP, 1 Phase, 1/2" Shaft, TEFC (Totally Enclosed Fan Cooled), 3450 RPM, G48 Frame Motor, 115/208-230 Volt, New in Box
More info: www.leeson.com/

(the item had the wrong picture posted, here's a correct one: http://cgi.ebay.com/ws/eBayISAPI.dll...tem=7595637153)

Been in touch with Jon; he's out, not equipped for spinning (a reminder that turning and spinning are not the same!). Progressing on the heat die though.

I hear you on the balance, but right now I expect it to vibrate. Grommet time. will see, shortly.

Gonna put up some items on eBay to get rid of unused things (including that 220 v motor), and make costs back.

edit: motor listing: http://cgi.ebay.com/ws/eBayISAPI.dll...tem=7595805284
shameless plug: My old ISA GPIB card: http://cgi.ebay.com/ws/eBayISAPI.dll...tem=7595813019

[bigben2k]

Picked up an inclined manometer:
http://cgi.ebay.com/ws/eBayISAPI.dll...tem=7594051088
Dwyer Mark II
0-3"
(just for a visual indicator)

Got a couple of watchers on the motor.

[bigben2k]

Gerry produced more than one paper:
http://www.me.pdx.edu/~gerry/class/ME449/weekly/pdf/flowRateSlides.pdf
(PDF, 27 pages, 210 kb)

More general.

Edit: his main page:
http://www.me.pdx.edu/~gerry/

Now teaching an Applied CFD class.

[bigben2k]

A few more hits of interest:
http://www.eere.energy.gov/buildings...ler_app8_6.pdf


The AMCA documentation on 210-99 standard is available, for a price ($90):
http://cart.amca.org/publications/product.asp?PN=210/N

NASA's KSC library has a copy of 210-99, but not available for loan.

Rice shot down my online request for a loan. I might have to pay them a visit.

A Chinese library (Wanfang) seems to have a copy but there's no way to it. $3.00 (!)


The AMCA certification program documentation:
http://cart.amca.org/publications/pr...asp?PN=211F/ND
and
http://cart.amca.org/publications/pr...asp?PN=611F/ND
(I haven't read them all yet)

[bigben2k]

I received the replacement blower motor last week. Friday, I installed it, and was able to take it through a quick test run. It works well. It's quieter than I expected, similar to a household vacuum cleaner; very tolerable.

I picked up another manometer:
Meriam Instruments 40GD10
link
($0.99 + $17.81 shipping & handling)
I hope to get more precise readings from it, until I go digital.

I also got a set of four SSR (Solid State Relays), model D1D20:
($30.01 + $8.49 S&H)
Specifications
(PDF, 2 pages, 48 kb)
I'll be using these along with my Harrison PSU (0-36Vdc, 0-5A) to heat up some nichrome wires, in that test run with the Fuji temperature controller.

Still working on the tunnel tubing. A pair of 55 gallon drums is starting to look attractive.

Money is becoming tight, and I'm going to have to cool off the acquisitions for a while.


Saturday, I was able to test some of my other testbench instruments: the three RTD temperature transmitters, the industrial RTD probe, the whole lot of small PSUs (5v, 12v, 15v, and 24v, DC).
(see list here)

Picked up a pair of lithium batteries for my Sponsler flowmeter ($17 with shipping). The meter won't power up without them.

I will be picking up a set of thermocouples, and two Fluke 2190A from Bill (graciously calibrated!).

Am going to have to pickup:
.826 specific gravity, red oil for both manometers (can probably get locally).

250 ohm precision resistors for the RTD transmitters (the 4-20 mA output is through the power connection, (!) ).

[bobo5195]

Gerry looks like the dude there are a couple of things on his website that are quite useful for me.

he uses the same cfd program (star cd) that we use. It is an awful awful program but it was developed in the early 90's by one of the profs (he also did alot of stuff on engine intake modeling as well for all the car nerds out there and that is where the program comes from i believe)

[bigben2k]

Sounds in line with his work; a wind tunnel for Peterbuilt freight trucks.

I'd like to see more about the TIM joint tester that a group of his students put together: http://www.me.pdx.edu/~gerry/research/TIMtester/
Most puzzling.

[bigben2k]

Help, I need someone's input!

I need a solution to mount the nichrome wire, something cheap, that won't burst into flames

I'm going to pickup an assortment of nichrome wire with gauges 30, 33, 34, 36, and 38. The temp controller is going to drive the SSR DC, and my Harrision PSU will power (0-36 v, 0-5 amps) the nichrome wires.

I've got a couple of heavy cardboard tubes, 80 mm diameter, and ~ 6 inches long. I'll slap a spare 80 mm fan at one end, mount the nichrome wire assembly in the middle, then the thermocouple at the end.

What I don't have, is a cheap way to mount the nichrome wires, probably in some kind of ziz-zag arrangement. Does anyone have any ideas?

All I can think of right now, is some kind of flimsy wood frame with steel screws. either that, or sacrifice a hair dryer or the spare toaster. McMaster has ceramic sheets that I might be able to cut, somehow...

[Brians256]

Cheap? Use coat hangars or wire and then attach by putting loops around standard insulators that you find for electric fences.

[bigben2k]

I like it, keep 'em coming.

[bigben2k]

Came across an NIST paper about calibration of the nozzles:
http://www.cstl.nist.gov/div836/836..../NCSL_4e03.pdf

[bigben2k]

Did I post these yet?
http://www.ceesi.com/pubs_venturis.aspx
http://www.ccdi1.com/efa-flow/064/064-desc/064-desc.htm
http://in3.dem.ist.utl.pt/lxlaser200...paper_22_5.pdf
http://www.flowmaxx.com/
(that last one's mostly useless)

[bigben2k]

I thought I'd share a few thoughts I had about temperature control. (I'm having to re-write this; my wife's laptop crashed when I was almost done ).

First a recap: I have a temperature controller with an input for a thermocouple (among others), and a PWM (Pulse Width Modulation) SSR (Solid State Relay) drive output. The output is varied on the duty cycle in what I'm assuming is 1% increments (I'll be verifying that experimentally). The frequency can be set to a maximum of 1 Hertz (not great, but I'll live).

Going over the details of the test setup that I'm preparing (an 80 mm fan blowing into a heavy cardboard tube, with a nichrome heating wire arrangement in the middle, and the thermocouple at the end) I came to realize what difficulties Bill told me that he encountered trying to maintain a steady air temperature prior to acquiring the environmental chamber.

The temperature controller is a pretty clever device. Beyond the obvious, it can be programmed to respond proportionally to a temperature i.e. it can be programmed to apply the right amount of heat to make up for a specific difference in temperature. Not only that, it also has a self-tuning feature so that it can program itself (saves me a lot of work ).

A bit about the output.

The output of the controller drives an SSR (Solid State Relay). The SSRs that I ordered are of the DC drive for DC output (a bit unusual, since most SSRs drive an AC output).They can handle an output maximum of 100 volts and 20 amps. Maximum turn-on time is 0.1 msec and max turn off time is 1.0 msec, well within the maximum frequency of the controller's output: 1 Hertz. (Unfortunately, it'll be at least two weeks before I receive them GR). While they are usually mounted to heatsinks, right now I don't expect to run too much power through them.

I got an assortment of nichrome wires. 10 feet of each gauge (30, 33, 34, 36, and 38 AWG, of different metals, more on that later). I calculated the applied power at several voltages and lengths, and everything looks in line for me to use my Harrison PSU (0-36 volts, 0-5 amps).

The controller output voltage is within the SSRs input range. I've calculated the amperage draw, and I expect that it will fall within the controller's max output rating (but I'll have to check it experimentally).

What became clear to me, came when I looked at the details of the above, in terms of PWM. There are three variables of interest here: the frequency, amplitude, and the duty cycle.

In my test setup (where I'll raise the air temp to 25 deg C), I am working with one assumption; that there is an infinite supply of air at 20 deg C. Below are the reasons why that is a problem.

The PWM frequency ought to be fairly high, so that any temperature change can quickly be corrected. Also, the duty cycle cannot be 0 or 100%. If the duty cycle tops out, then the temperature has fallen below the SV (Setpoint Value) and the temperature is no longer being controlled. Conversely, if the duty cycle hits 0% (or otherwise fluctuates between 0% and 1%), temp control is (essentially) lost.

*** In short, the heater must always be ON, for the temperature to be controlled. ***

While it might be tempting to state that a 50% duty cycle would be ideal to keep a safe margin "either way", this isn't actually the case: the temperature is "in control" as long as the duty cycle is 1%, 99% or anything in between. But there's more.

Let me use an example where I'm running a relatively high amount of power to the heaters, and I end up with a very low duty cycle. To correct it, I lower the applied power, which increases the duty cycle.

(In actuality, it isn't quite that simple: since the applied power has changed, the heater response has also changed, so the controller has to be re-programmed or re-tuned).

Enter the refrigeration unit.

While visiting Bill, he explained to me the function of his environmental chamber (BTW, the Haake 82 water chiller works the same way). In short, it's made up of a refrigeration unit (the usual compressor, evaporator, condenser arrangement) that runs "flat out", as he put it. The recirculated air is cooled, then heated back up by an arrangement of nichrome wire coils, to the desired temperature.

Without the refrigeration unit, keeping in mind that the heater must always be ON, it should become clear that that arrangement cannot function, where the fluid (air or water) is recirculated.

I've determined that the refrigeration unit serves many purposes.

While I'm sure that the temperature controller and heater (alone) arrangement can work well, I believe that there are a number of issues that arise, when the inlet and outlet temperatures are very close to each other.

1st, as we all know, the transfer of heat within a very small temperature difference is difficult because it is a very slow process. Picture two stacked blocks of copper where one is at say 23 deg C and the other is at 24 deg C, and try to imagine how long it would take for both blocks to be at the same temperature. Now imagine the same blocks, 100 deg C apart, and try to imagine how long it would take for both blocks to be within 99 degrees from each other. Obviously it's quicker in the latter (it's obvious, right? ).

I figure that the nichrome wire heater works best when it operates at a significant temperature difference. Further, given some relatively minor inlet temperature variation, the same heater should easily (read quickly and just as effectively) be able to heat the fluid back up. What I'm getting at, is that the combination of the temperature controller and heater arrangement, may not be able to deal with a significant change in the heater effectiveness, as happens when the heater is called to apply heat levels that vary over a wide range, or more specifically, near and far from a very low range.

Another factor that can apply here, is that the DUT (Device Under Test) may throw in its own share of heat or cooling, but the heater should still be applying a significant amount of heat. The DUT effect is a factor in the design of the recirculator. If the DUT is expected to induce heat, then the whole assembly might be preset with a lower duty cycle.

To go back to the (false) ideal of a 50% duty cycle...

50% duty cycle is actually ideal, to minimize the temperature fluctuations, but isn't necessarily relevant.

See the graphs below, where green is the average temp, and blue/purple the actual temp. Sorry for the rough graphs, I hope you all get the point. Note the peak high and low.

graph 1 50% duty


graph 2 low duty cycle.


While 50% might seem ideal, it may not be relevant if the DUT (by design) is unable to be affected by the amplitude and frequency of the temperature fluctuation, but if any fluctuations exist, one should be aware of them.

As previously stated, while 50% is ideal, the design may have to accomodate a DUT that adds a heat load, which means that it would be configured with a higher than 50% duty cycle.

I've also determined that the response time of the temperature probe plays a critical role. As I've previously found out, RTD type probes usually have a slower response time than other types, probably because they are encased and/or shielded, where thermocouples can be found bare. In order for the temperature probe to report a new temperature, its mass must be brought to the new temperature. What can help in getting a better response time, is the fluid's flow rate; the higher it is, the quicker the response will be (note that any flow rate also induces heat). I believe that the temp probe response time is by far the most limiting factor in maintaining a steady temperature.

The test setup.

My test setup effort will mostly consist of varying the applied power, then re-configuring the controller each time, until I hit a duty cycle around 50%, then measuring and noting the fluctuations.

I will first start by trying to calculate the necessary power, based on the the energy required to heat a certain flow rate of air by 5 deg C.

I'll be using a type T thermocouple, 40 AWG (0.0031 inch diameter) that I recently ordered from Omega.com (a $12 item!). I'm hoping that the type and small size of this probe will make for a responsive setup.

With all of the above, I believe that I am well on my way to maintain a steady temperature, once I incorporate a refrigeration unit.


A note about analog.

An analog controller would (essentially) connect the temp probe directly to the heater, where more power is applied when the temperature drops. However, I believe that it would have to be finely tuned, so that the heater and the temperature probe response is matched, otherwise the system would over/under compensate for itself, which would cause fluctuations. A similar effect is known in flight, where the plane dips forward and back. Otherwise, a proportionally controlled heater like that would probably be ideal for steady temperatures, if it is possible, because a digital (PWM drive) output is a sure way to fluctuations (although they would be easier to identify).

Ok that's all for now.

[bigben2k]

Argh! Someone stop me!

I "accidentally" picked up another Fluke 2190A:
http://cgi.ebay.com/ws/eBayISAPI.dll...tem=7601329663
with a 2300A this time. I just made a minimal bid, I didn't expect to win!

I also picked up a differential pressure transmitter:
http://cgi.ebay.com/ws/eBayISAPI.dll...tem=7600553910
(Dwyer 647, Wet/Wet Differential Pressure Transmitter)
unfortunately, I only noticed afterwards that the accuracy is very poor. Oh well.

[bigben2k]

Still waiting on the relays.


I calculated the needed power as follows:
The knowns:
-typical cfm of an 80 mm fan: 35
-density of air: 1.2 kg / m^3
-heat capacity of air: 1004 Joules / kg
-1 Watt/hour = 3600 Joules

I then calculate the needed power to be 20 Watts. 19.9 actually, but this is a rough estimate, as it doesn't account for barometric pressure, heat transfer resistance, secondary losses, or actual airflow rate.

As I mentionned before, the target is 50% duty cycle, so I'll setup the nichrome wire to apply 40 Watts at full load.

Using one foot of 30 gauge nichrome wire, and applying 2.5 amps would result in a voltage of 16.3 V. This would result in a wire temperature of ~760 deg C.

Using one foot of 33 gauge nichrome wire, and applying 1.8 amps would result in a voltage of 23 V. This would result in a wire temperature of ~870 deg C.

Both temps are within the maximum specs for these wires.

The numbers don't feel right, so I'll double check them again.

I have yet to figure out a mounting solution for this wire, but I've determined that I can use screw terminals to connect it, as long as the screw terminal isn't cheap plastic.

[bigben2k]

...and many months later!

I'm about to complete construction of the simulator, to flex the temp controller. (Otherwise the temp controller will be re-dedicated to a water heater).

I need some suggestions, on cutting these things:


They are:
Alumina (Al2O3) ceramic
1" long by 0.19 inches in diameter
Hole diameter is 0.13 inches

I need to cut them in 1/4" length or less. The best success I've had so far, is with a grinding wheel on a Dremel, but still very, very slow.

Anyone?

[billbartuska]

Originally Posted by bigben2k

I need to cut them in 1/4" length or less. The best success I've had so far, is with a grinding wheel on a Dremel, but still very, very slow.
Anyone?

Put a variable speed, reversible drill in a vice and then cut with the Dremel. May not be faster, but you'll get nice square cuts. A little water for cooling may help.

[bigben2k]

I stopped by Lowes today, and came across a jigsaw blade by Bosch, special for ceramics.
http://www.boschtools.com/accessories/

I'll mount the ceramic tube in a drill to secure it, and see if I can chop it that way.

I'll report results Sunday.

Thanks,
Ben

[bigben2k]

Sometimes, the simplest solution works best:

The $8 jigsaw blade was useless but the drill chuck happily crushed these tubes to the length I wanted.

[bigben2k]

A quick update:

I'm still building the simulator. I picked up a couple of power terminals, and I'll be using pieces of a hacksaw blade to route the power in and out of the tube. Nuts and bolts will secure the hot wire inside.

I lined the inside with a steel wire, and used JB weld to secure it. I'll be able to use JB weld again to secure the ceramic tubes, and mount the wire. Then I'll add some aluminium tape to line part of the inside (just because I have it, and it's a cardboard tube). Then I can fire it up, and run some tests.

Pics later.

[bigben2k]

Assembly complete, and I had my first test run last weekend.

I was only able to maintain a steady temp within +/- 0.4 deg C and could only set the temp within a very narrow range (i.e. 2 or 3 degrees above ambient, no more, no less). As soon as the house AC kicked in, it dropped the room temp by 3 deg C, and the system couldn't keep up anymore.

It was interesting to see the hotwire flex, on those moments when the controller was running it at ~50% duty cycle.

What I observed is that the TC I used would return temps that would jump by 0.2 deg C for no apparent reason. I was going to try my miniature RTDs, but they turned out to be thermistors, and won't link up to the Fuji temp controller.

I'll have a few more runs at it, as is, but I plan to replace the temp probe with an RTD. I might add an RTD to measure the inlet temp. I'll also review the temp controller features; it's a bit more powerful than I expected.

[bigben2k]

Another update.

I removed the eyelet slug at the end of my thermocouple, to see if I could get a quicker response. It didn't seem to have any impact.

I've got a set of RTD probes on the way, but I'm still going to try my original idea of using my 40 gauge thermocouple (normally I wouldn't use them here, because they're very small, and a small thermocouple like this can oxidize quickly in relatively low temperatures ( > 100 deg C, which ruins the TC probe), but I'm running with "near room" temps here, so should be fine.

Otherwise I'm still seeing 0.2 degree C temperature reading jumps, and I haven't identified why yet. I'll plug my scope on it, and see if it'll help me figure out what's going on.

I might also try replacing the 1 foot long nichrome wire with recently acquired nichrome coil, to apply more power. If successfull, I might be able to apply enough heat to get a 40 deg C temperature rise, which would allow me to have the temperature controller run a self-programming routine.

Going over this thread, I realized that I've already spent over $100, just to create a test rig.


On related topics, I've begun tooling up for phase change, and I'll be building the cooling unit myself, for the environmental chamber that I'm putting together. Details in the forums: http://xtremesystems.org/