Here is a plot of temperature data,
all taken with CHT thermocouples.
Rad_in is clamped to the water pipe
just before it splits to go to the two radiators
Rad_out is clamped to the exit water
pipe after the flow from the two rads has been recombined.
Oil_filter is clamped to the oil
line as it exits the filter and enters the engine, and
Oil Cool left end is bonded to the
end away from the entrance/exit on the oil cooler to tests that the oil cooler
is working. As you know the standard Mazda oil cooler sends the oil down to the
end and back. I was getting high oil temperatures on one of my (non
thermocouple ) oil sensors, and wanted to verify that the oil cooler was
working.
Bill Schertz
KIS Cruiser #4045
N343BS
Phase I testing
Sent: Friday, May 14, 2010 5:32 PM
Subject: [FlyRotary] Re: alternative water
pump
Hi Bill,
Had seen your nice data
before, but one thing finally awoke in my old brain when I looked at it this
time that I had not considered before. We know that parallel cores give
slightly better efficiency than a serial core set because the DT decreases for the
second core in the series compared to both parallel cores having the same DT (at least in theory).
However, what jumped out at me this time was the real significance of the
parallel cores in cooling. In this case, I am assuming no thermostat in
the coolant flow.
If I understood your
graphs correctly, it looks like you are getting around 20 gpm flow with a
single core (so presumably you would get a bit less with two cores in series
– but perhaps not significantly), but looks like with parallel cores you
are getting around 32 gpm flow. That is a 20/32 = approx 37 %
more mass coolant flow through the engine. That means (all else
being equal), you should transfer 37% more heat out of the engine per unit time
with the parallel cores compared to the serial cores (assuming cores of same
type and size).
Now the engine is
producing X amount of waste heat at Y HP that it needs to get rid of.
That won’t change for a given power setting Y. So Q (waste heat X)
produced by the engine should be a constant at Y Hp.
So taking Q = M*Dt/Cp and since Q (waste
heat) = constant at power setting Y, then with M (mass flow up 37%)
implies that in this case Dt = (Temp of coolant out of engine – temp of coolant into
engine) should decrease by 37%. When you increase the mass flow and
are removing the same quantity of heat, the DT is of necessity a lower
value.
If that is the case, then
the question is - does this mean the temp of coolant into the engine increases – not necessarily desirable, or does the Temp
of coolant out of the engine decrease? Or a bit of both? I suspect
it’s a bit of both depending on the radiator’s performance.
If your radiators/air flow are the limiting factors, then transferring more
heat per unit time to the radiators is not going to buy you much. The
reason is that if it is not able to get rid of the heat at the faster rate and
the DT between the coolant and air will be less.
But, my guess is that
this theoretical increase in heat removal by using parallel cores could be
useful in some situations – again if you are already limited in the airflow
situation, then this won’t make much difference. It does
suggest that using parallel cores could result in the need for core sizes 37%
smaller. OR did I miss something here?
Like your data in
any case
Ed
From: Rotary
motors in aircraft [mailto:flyrotary@lancaironline.net] On Behalf Of Bill Schertz
Sent: Thursday, May 13, 2010 10:39
AM
To: Rotary motors in aircraft
Subject: [FlyRotary] Re:
alternative water pump
Back in 2002 I measured the flow
from a 13-B pump, attached to the engine but driven with an electric motor. The
curve is attached. I ran the pump at 3 different RPM, established by changing
the pulley size on the motor. At 5594 rpm, the pump produced 19 psi at zero
flow, and 44 gpm at 0 psi. At lower RPM, the pump of course pumps less.
The other test I did was to measure
the flow through one core of the two I was using for my installation. That is
the curve going up to the right with the red dots as the experimental points.
Since I am running my cores in parallel, the right hand rising curve is a
'calculated' flow response for the parallel cores.
Finally, I hooked up the cores to
the system, and pumped water through them. The single large point represents
where the flow and pressure came out, very close to the calculated expected
response.
All flow measurements were done by
the "bucket and stop-watch" technique, with multiple runs to get the
flow.
Bill Schertz
KIS Cruiser #4045
N343BS
Phase I testing
Sent: Wednesday, May 12, 2010 11:54 AM
Subject: [FlyRotary] Re: alternative water
pump
Al,
Are you
sure of the 40 GPM? That seems like a lot. My radiator in/out is
1.25 inches, so the water would be traveling at 628 feet per minute at that
flow rate. That is over 7 miles per hour!
Bill B
When my 20B (with a 13B pump that Atkins referred to as ‘high
flow’) was on the dyno the measured flow was 48 gpm with the standard
pulleys. I expect the dyno cooling loop was fairly low pressure drop
compared to our typical systems, so I’m just guessing 40 gpm is in the
ballpark. 628 fpm (10.5 ft/sec) would not be considered very high - -
above 15 ft/sec I’d consider high.
Al