Plumbing - waterblock test circuit

[RoboTech]

OK, I'm kicking around ideas for how to setup and plumb my WB test bench...



Most of the piping will probably be 1/2" NPT (brass fittings, 3/4" copper tubing, maybe some plastic pieces and a little SS). The chiller I'm considering (Haake DC3-K20) has a 6 L res so I thought adding a larger, insulated external res with internal mixing chamber might give the system better stability (?). The pump will probably be a relatively large Iwaki or Little Giant, 50 micron filter, flow sensor, and then an array of throttle valves for setting flow rate. Primary sensors are T1 (water inlet), P1 (inlet pressure) and P2 (outlet pressure). I'll probably go RTDs for T1 and thermal die simulator and press transducers for pressure and then use a water column or mercury manometer for Delta P. I'm looking at a lot of flow sensors in this range (0.3~7.9 GPM)...

I plan to connect the system to the waterblock under test with two relatively short (12"?) sections of 1/2" ID silicone tubing, attaching at the angels shown to cancel out any preload on the block. I'll put short sections of copper pipe right before the block with pressure taps - another set of these will also serve as 1/2" to 3/8" adapters when needed.

My plan to adjust flow rates is to have an array of five or six 1/4" NPT ball valves and needle valves arranged in parallel branches. Each set of valves (ball and needle) would be rough set to a particular flow rate. The ball valves isolate each branch while the needle valves adj flow. One branch open => min flow setting. Opening each additional branch increments flow upwards. Needle valve in the last branch opened can be used to fine tune flow. This may be overkill, but I won't know until I try it.

I have started to define the specs for particular components and sensors but wanted to get the basic plan out for ideas first. I still have plenty of time while I save up $$$ and study.

OK, thoughts, suggestions... ?

[pHaestus]

I use PVC crosses on the inlet/outlet of the wb like so:

http://phaestus.procooling.com/pvccross.jpg

left is a temperature probe (YSI dual linear thermistor) and right side is the tubing that connects to pressure gauge. By putting them opposite one another I avoid the 90 degree turns you have in your loop for temp readings and I get the temp probe a bit closer to the block inlet and outlet.

I have my setup plumbed as so:

1 gallon res with 3/4"ID tubing connected to pump, rad (no chiller yet), cross, wb, cross, filter, flowmeter (with a length of copper pipe to make sure that it is always being placed according to mfgr specs), T to split flow, needle valve and gate valve on one side and needle valve on the other, both lines going back into res. I run it this way because it's easier to control the flow rate with a few parallel valves than with valves in series. My loop isn't set up at the moment with an eye to rapid and efficient filling and draining; that would be something I'd think about before I soldered anything together

Your chiller isn't large enough to use as the reservoir?

[Since87]

If the 'extension' reservoir is of benefit, I think it would be best to have water flow:

Chiller->test loop->pump->reservoir

and then have water circulating between the reservoir and the chiller. This might require an additional port on the chiller.

I'm not certain what the effect of the added reservoir would be. I would guess, the amplitude of variation in water temperature would be lower, but the time constant of the water loop would get longer. I don't think it's likely, but it's possible the chiller's control loop might actually become less stable. I think you'd just have to try it and see.

I'd have the pump outlet pumping into the chiller/reservoir, so that the heat the pump puts into the water will be dissipated by the chiller before the water reaches the waterblock. The degree to which the pump heats the water will vary depending on how restrictive the test loop is.

[RoboTech]

Thanks for the picture and ideas...

Originally posted by pHaestus
Your chiller isn't large enough to use as the reservoir?

I don't have the chiller yet, but I have a lead on that one used. Maybe it's internal res (6 L) is big enough, I don't know?

What flow meter are you using pHaestus? I've looked at several mag units but they are all way expensive ($2,500~4,000 new). I know Bill warns against turbine types... but I have my eye on one that has an excellent range (0.26 to 7.93 GPM) with a claimed accuracy of +/- 1.0% of reading and repeatability of 0.1% for ~$700. I need to find out more...

Since87,

Yes, I agree... I will need to try the reservoir in various configurations and see what works best! I didn't show it in my rough sketch, but if I do incorporate a secondary res for added volume I plan to put a small (1 gal) "pre-mix" chamber in there so the main test loop is always picking up water coming from the chiller loop unless the main loop flow rate exceeds the chill loop flow rate.

I also didn't show it, but I probably will use a small bypass loop (ball valve and float type flowmeter) to bleed off excess pump flow at the lower test loop flow rates. I want a pump that will have adequate head to overcome losses of the filter and flow meter and still be able to deliver 300~400 GPH max thru the test block, but I also don't want to have the pump essentially dead-headed at lower flow rates.

Thanks guys,

[pHaestus]

I have a paddle wheel type Great Plains unit. Also 1% accuracy.

http://phaestus.procooling.com/gpi.jpg

Magnetic units are too rich for me. You must not be looking at ebay and labx?

[bigben2k]

FWIW (For What It's Worth)


I have to agree with Since87 on the order: put the pump after the block, and before the chiller. To add to that, I'd minimize the restrictions between the chiller outlet, and the block inlet: any of them may add heat, throwing your block inlet temp off, and you may encounter some difficulties stabilizing that temp, as you vary the flow rates. Dunno, might not be significant.

The chiller I've got (ThermoTek 252) only allows me to set the temp in +/- 0.5 deg C increments, even though it'll keep it stable to +/- 0.1 .

I like pHaestus' cross idea. I'll have to remember that one. The thing is, I wouldn't want it to induce any turbulence in the water, just as it enters the block. I can work with it...

How would you incorporate a mercury manometer? That's been puzzling me for some time.

Since87 has a point about the time it would take, for the temp to stabilize. I'd stick with the chiller, straight. In my case, the unit only gives out max 0.5 gpm, but with a max pressure of 8 psi (!). Useless to me, so I'm looking into swapping out the pump: ThermoTek is being unusually uncooperative (maybe they're tired of answering questions from eBay buyers?!? ).

I think you mean "flow meter", and not "flow sensor": the latter detects flow and reports a yes/no condition, according to the minimum flow rate its designed to detect. I also need a good flow meter. As a general guideline, you ought to keep 10d of straight run before and after the meter (that's 10 times the diameter). This spec may vary depending on the meter you end up with. I've seen 10d "before", and 5 "after". Just be on the look out for it.

Otherwise, I'll be using 3/4" tubing, except for the block connections (1/2", or 3/8" if that's the limit). The reason for 3/4", is because I'd like to be able to test a higher flow rate ( >= 2.5 gpm), without having to deal with the tubing restrictions.

What made you pick 50 microns, for the filter?

I like the idea of the valve array.

[pHaestus]

There is a good reason from a NPSHR standpoint to have a reservoir at the inlet of your pump with large ID tubing connecting the two. If you put the pump after the blocks without a res then it's possible cavitation will occur.

[RoboTech]

Thanks Ben,

Yes, there are many possible piping arrangements, especially the order of components. I definately will not be placing the pump after the block. I'm a firm believer in only throttling a centrifugal pumps discharge - always try to minimize any restrictions on the suction side (I agree with pH). I guess I could place the block upstream in the flow path (pump > block > filter > flow meter > throttle valves > res) to minimize the affects of heat generated by restrictions. I will just have to try it different ways to see if there is any measurable difference.

The chiller I'm looking at has a digital controller and can be programmed with a temp offset, which may be useful it setting actual block inlet temp.

Yes, the recirculating pumps built-in to most chillers do not have the head or flow needed for the main test loop.

Manometer - to measure the Delta P across the block you would just connect each side of the manometer U-tube to the inlet and outlet press ports of the block.

I believe flow "sensor" is a generic term that includes all types of flow measuring and detection devices. A flow "meter" specifically measures flow rate while a flow "indicator" generally gives a yes/no (on/off) indication of flow.

The 10 D rule is applicable to certain types of invasive flow meters (turbines, paddle wheels, etc). Some units have flow straighteners built-in and some other types don't care.

50 micron was just an initial guesstimate - more of a strainer than a filter, fine enough to keep the chunks out (protect flow meter) but not so fine as to require frequent maintenace and create a substantial press drop.

[nikhsub1]

Originally posted by pHaestus
There is a good reason from a NPSHR standpoint to have a reservoir at the inlet of your pump with large ID tubing connecting the two. If you put the pump after the blocks without a res then it's possible cavitation will occur.

Exactly. And, it really makes no difference AS LONG AS the res/chiller water temp is consistant. It should not make a bit of difference from block to block as the setup/water temps will remain constant.

[Since87]

The dT between pump inlet and outlet will vary from block to block and with different settings of flowrate control valves. Minimizing the heat that enters/leaves the water between chiller outlet and block inlet will optimize the stability of the system, and therefore reduce time spent waiting on temperatures to reach a new stable state.

Providing NPSH at the pump inlet, and preventing the pump from heating the water as it leaves the chiller, are different issues. It is possible to have water flowing pump->chiller and still provide adequate NPSH.

Edit: I guess I'm assuming the chiller reservoir is sealed aside from inlet and outlet tubes. Is that the case?

[RoboTech]

Agreed...

Originally posted by Since87
Edit: I guess I'm assuming the chiller reservoir is sealed aside from inlet and outlet tubes. Is that the case?

No, most of the recirculating lab chillers I have seen all have an open reservor. I suspect this is why BillA uses a liquid-liquid heat exchanger between his chiller and test loop (?). I'm not clear on his actual layout so don't know for sure.

[bigben2k]

I actually have this option, of using my chiller "as is" and using a heat exchanger of some kind, "liquid-to-liquid", if I find that I can't swap the pump without affecting the accuracy.

Maybe the logic is similar?

[BillA]

test loop is as follows:
3qt sealed res ->
pump (1" copper) ->
cross-flow liquid/liquid heat exchanger (recirculation through one side of xchr, submerged in chiller bath as well) ->
hanging swivel with petcock for venting (to change out components) ->
3/8" bronze cross (to minimize volume change) ->
wb, rad, etc. ->
cross ->
swivel with vent ->
mag flow meter ->
multiple parallel needle valves ->
3/4" return line ->
vented 3qt res and filter with valves to switch in and out ->

any other configuration works worse,
I know because I've tried each one

- if the pump is placed in the bath a xchr is not needed, BUT the total pump heat (which can be VERY considerable when throttled back) will have to be dealt with by the chiller

[myv65]

I'd like to get some clarification on a point Bill mentions. To the best of my understanding, centrifugal pump energy always increases with increasing flow. As the motor essentially runs at a constant RPM (not quite, but close), the motor efficiency should be comparatively constant.

If submerged, all pump energy gets into the water either directly as pressure*flow, non-effectual flow (turbulence), or thermal energy off the motor casing.

I believe the real answer is that energy drops with dropping flow, but energy does not drop as quickly as flow. Translation, less energy gets put in with decreasing flow, but the resulting delta-T due to the pump increases. ie, lower energy but more energy per unit volume of fluid pumped.

Thoughts?

[Since87]

Originally posted by myv65

I believe the real answer is that energy drops with dropping flow, but energy does not drop as quickly as flow. Translation, less energy gets put in with decreasing flow, but the resulting delta-T due to the pump increases. ie, lower energy but more energy put unit volume of fluid pumped.

Thoughts?

That's consistent with measurements of pump power consumption that I've done.