I'm trying to follow the example you state here,
This is the DeltaT of coldplate and chiller water? Higher DeltaT is what you are saying in this case for the lower C/W. This is coinciding with higher in chiller flow rate?

Bear with me, not trying to contradict this just making sure I understand your example.

Interesting (to me) observation:

Extrapolating Bill's data - the MCW5002 has a lower C/W than White Water with flowrates a bit lower than 1 lpm.

I predict a thermal resistance of ~.255 C/W at 0.1 lpm for an MCW5002 and ~.285 C/W for a White Water at 0.1 lpm. (Inasmuch as these numbers have any relevance, it is WRT Bill's die simulator.)

I don't really want to go into a detailed explanation of how I came up with these numbers. In brief, I used my 'hydropower' spreadsheet to make fairly straight lines on a log graph out of the C/W vs 'hydropower' curves, and then extrapolated those lines to the 0.1 lpm point.

Warning: Mathematical nightmare approaching.

Some points for consideration:

A 226 Watt TEC is actually a two dimensional array of many smaller 'tecs'. Each of these small 'tecs' can be operating at a different point of the dT vs Q tradeoff. Thinking of a 226 Watt TEC as a single monolithic block works reasonably well when 'high' flowrates are used, because of the small dT between inlet and outlet coolant. As the flowrate is lowered, it becomes increasingly important to consider the operating dT and Q of the smaller 'tecs' within the TEC.

Edit: I'm just beginning to think about a grossly simplified model that attempts to consider the points above, and my head already hurts. There's a lot to be said for finding things out experimentally.

I believe there are 127 in the array.
http://www.arrakis.es/~cidete/iti.htm

lol, the livin' proof here

your extrapolations seem correct, the slope of the MCW5002's C/W curve is less at its low end,
though at 0.3gpm the WW is 0.002°C/W lower, rather difficult for anyone to measure

which is the specific advantage - for all TEC apps - of a wb with opposite corner connections

Something like this:-
http://www.jr001b4751.pwp.blueyonder.co.uk/Chiller3.jpg

Cannot vouch for accuracy of data points(inpection of gifs).

EDIT: Revised/updated graph to include experimental data points. Thanks Bill

Yep, except a bit more convoluted on my part due to using an existing spreadsheet as a shortcut.

Same here.

Ah pimpage. LOL

Actually if there is a benefit to the dT/Q gradient that develops at low flowrates, the benefit likely becomes substantially greater if the the chilled water flows past the two TEC's in series. Taken to it's logical conclusion 'diagonality' becomes irrelevant.

(I'm still not certain that there is an explanation for your experimental results in the dT/Q gradient alone. I have no other hypotheses though.)

Somewhat coincidentally, several at O/C forums seem to get 'good' results from chillers containing ten '50 Watt' TEC's. I haven't paid enough attention to know anything about the construction of those chillers though.

I suspect there should also be some benefit to running the hotside water counterflow to the chilled water. Have you tested this Bill?

shakin' ma booty, pimpage rules

"two TEC's in series", na - one chamber with a TEC on either side
no, diagonally is relevant - consider the areas of stagnation

I have a liquid/liquid xchr which is counter flow, only way to get good results
but here there is a TEC plus 2 rather hefty bps between the 2 flows, no significance I suspect

we have some 360W TECs, perhaps all 6 sides ?
lets see, 6x24A = 144A @ 24V = 3456 Watts
would that get it done ?

will send you and Les the datapoints

You've seen examples of my flow visualization abilities. Why would you think that if I "consider the areas of stagnation" I will get better results? Got pictures? drawings?

I presume there is some small amount of stagnation in the 'non-barbed' corners. On the other hand, an open chamber type with barbs centered in two opposite sides, will have worse stagnation in all four corners. Reasonably accurate?

exactamundo
and center inlet, corner exit is worse yet