Mizuno Airflow Calorimetry

Quote from Daniel_G

@THH, I have a question for you from a theme that seems to be popping up now and again after reading your posts. At multiple times you mentioned possible discrepancy about input power measurement of Mizuno not clear if its measured before or after the PSU. Yes, I agree it would be nice if input was measured after the PSU but 100% of all PSU's have a finite efficiency so even if it was measured at the line voltage outlet level, that would over-estimate (not under-estimate) the input power unless you think the PSU efficiency is over 100%. So, yes I agree that these things should be cleared up but in terms of any possible error that would under-estimate input power, this one is a non-starter. Can we agree upon that in principle? As Jed says, there is nothing easier to measure than input power.

In terms of the flow velocity profile, if Paradigmnoia is having problems with mixing, I suggest stuffing a high-porosity sponge filter like what you can buy at K&N, etc. that will remove any vortices, etc. from the fan before the air enters the tube. These would have minimal pressure loss and should solve most of these profile issues being discussed.

Daniel, i would agree with you but for two issues:

I) Mizuno has corrected data based on efficiency etc and not documented that. So we'd need to verify whether this was done with the power input, because it would be easy to do it wrong.

2) the problem is inconsistent measurement. For example measuring calibration input before the PSU could lead to lower calibration power than active power if this is measured after. It sounds silly, should not happen, but the evidence from the 2017 spreadsheet indicates that it might happen.

A lot of my finickyness here is that the methodology is not in this case not entirely trustable, and not clearly reported (ok - perhaps if reported with all data calculations it would be trustable and the things we don't understand would all be understood) - so I want to find as many independent cross-checks as possible.

Tom

Quote from Paradigmnoia

I don’t have any recent spreadsheets to refer to.

You do no need recent spreadsheets. All of the spreadsheets show the same thing. To resolve this issue, you do not need any spreadsheet. You need only three average numbers, which I gave you:

For a 30 W calibration, the averages are: 29.69 W input power, 1.75 K Delta T, heat capacity 1.005 kJ/kg*K

Weight of air kg = Energy kj/s / (temp K * Specific heat kJ/kg*K)

(0.02969 kJ/s / (1.75 K * 1.005 kJ/kg*K) = 0.01688 kg/s of air

Mizuno's instruments showed 0.01679 kg/s

That proves his instruments measured the average weight of air per second correctly. The second anemometer brought by an visitor also proved that. The input power and temperatures were confirmed by many other people, including me.

Unless you can give us a reason why this equation is wrong, or why it does not apply, or why it does not prove this is the correct average flow rate, I will consider this discussion closed.

Quote from THHuxleynew

I) Mizuno has many times corrected data based on efficiency etc.

I think you mean recovery, not efficiency. He shows heat recovered after accounting for losses. However, as I pointed out, you can subtract out this and use only heat recovered in the flow of air. It is still positive, and still so far above input, so there is no chance of a mistake.

You will, of course, ignore this fact, and you will continue confusing efficiency with recovery. Carry on!

Quote from JedRothwell

Weight of air kg = temp K (Energy kJ/s * Specific heat kJ/kg*K)

1.75 K / (0.02969 kJ/s * 1.005 kJ/kg*K) = 0.01688 kg/s of air

The correct equation relating mass flow in a flow calorimeter to temperature rise and energy flow is:

mass (flux) = energy (flux) / (spec heat * temperature rise)

I think you forgot the / and also seem to have it the wrong way around?

Quote from JedRothwell

I think you mean recovery, not efficiency. He shows heat recovered after accounting for losses. However, as I pointed out, you can subtract out this and use only heat recovered in the flow of air. It is still positive, and still so far above input, so there is no chance of a mistake.

You will, of course, ignore this fact, and you will continue confusing efficiency with recovery. Carry on!

I'm happy to use your work, recovery, instead of mine, efficiency. I apologise for such non-standard terminological lapses, although it is quite natural. For certainty we have (as i have used the terms):

heat loss (as fraction of input) = (Pin - Pout) / Pin

heat recovery (or efficiency) = Pout / Pin

heat loss + heat recovery = 1

THH

Quote

the problem is inconsistent measurement. For example measuring calibration input before the PSU could lead to lower calibration power than active power if this is measured after.

Of course it could. But it would be a small percentage of the power consumed by the PSU and thus a very small percentage error in the output/input ratio. And no reflection at all on the output power measurement which is done an entirely different way with a completely different set of instruments. I agree some colossal error is possible. even probable, with Mizuno's experiments but it would be absolutely startling if it were in measuring the input power from the power supply!

Quote from seven_of_twenty

Of course it could. But it would be a small percentage of the power consumed by the PSU and thus a very small percentage error in the output/input ratio. And no reflection at all on the output power measurement which is done an entirely different way with a completely different set of instruments. I agree some colossal error is possible. even probable, with Mizuno's experiments but it would be absolutely startling if it were in measuring the input power from the power supply!

I'm looking at this bit by bit, so not claiming any specific issue is known a problem, but nothing it as a posible problem.

In this case PSU effciency can be low (like, 10%). So with calibration input measured before the CPU the active test output would appear 10X larger, for apparently the same input power, if that is measured after the PSU in the active test.

See what I mean?

mistakes are often startling in their simplicity and obviousness, which is why experimenters adopt very rigorous techniques, documenting all apparatus changes and observations clearly in a logbook. I have not got this level of detail from the Mizuno paper and various misunderstandings have shown that Jed does not have it either.

Quote

See what I mean?

Yes but no. JedRothwell says the power is measured on the input AC side of the power supply and on the output side as well between the PSU and the device's heater. Unless he's lying, that sort of error is essentially impossible. Not to mention that commercially available and properly functioning power supply are more like 80-90% efficient.

I would be more interested in errors that could account for a 50W input resulting in a 250W output, if that is really what Mizuno measured. Anyway, does heat from the PSU go into the reactor? Why would it? If not, what difference does it make if it's inefficient? And the more inefficient, the less power delivered to the load, not more. Did I miss something here? I did enjoy a bit of rum last night.

Quote

In this case PSU effciency can be low (like, 10%). So with calibration input measured before the CPU the active test output would appear 10X larger, for apparently the same input power, if that is measured after the PSU in the active test

Bring that by again please? Which of us had too much rum? Maybe I just need more kawphy.

SOT: lab (variable) PSUs, if linear, have very low efficiency when output voltage is small fraction of max voltage. 10% is very possible.

Even switchers have very variable efficiency with voltage and current, it can be low (though not usually 10%).

mix power measurement before and after a low efficiency PSU, without checking, and you get that factor.

OK so let's suppose I measure 100W into the PSU and it's output is actually 10W and not 80 or 90W. Doesn't that make the reactor appear to have a lower P_out/P_in than it really has? Not to mention that measuring between the PSU and the reactor will always give the right result and JedRothwell said Mizuno always did this, sometimes letting outside people use their own instruments to repeat it.

Like I said, my thinking is a little blurred today and I am trying to get a lot of stuff done. Did I miss something?

Assume measured output of reactor is 250W

Assume power supply consumes 25W and is 10% efficient and is measured between the power cord and the PSU.

Measured power is 25W

But real power to reactor is 25 x 0.1= 2.5W

Output/Input ratio is measured to be 250W/2.5W =100

Now, suppose power supply consumes 25W and is 80% efficient.

Measured power in still 25W

But now real power in to reactor is actually 25 x 0.8=20W

Output/input is now 250W/20W=12.5

So the with 10% efficiency by the power supply, the measured power ratio is higher than it would be with 80% efficiency - not lower. The error would go the wrong direction. No?

I think you are using "input" to mean different things. Jed says Mizono measures input to the heater and you are talking about input to the power supply powering the heater. Your posts would be clearer if you used "heater input" or "power supply input" depending on what you are referring to.

When power is measured to the heater, it will be very precise, whether done with an expensive power meter or just measuring V and I and multiplying. Measuring power to a resistive heater driven by DC is about as easy as it gets.

@SOT has reiterated my argument exactly. We both agree the postulated argument goes the wrong way but @THH is saying that if there is an adjustment based on the PSU inefficiency that then that could show up as excess power. I think Jed has cleared the this issue already as the power in is measured between the PSU AND also checked at the source as a “reality check” so no possible source for error here. QED.

Quote from seven_of_twenty

OK so let's suppose I measure 100W into the PSU and it's output is actually 10W and not 80 or 90W. Doesn't that make the reactor appear to have a lower P_out/P_in than it really has? Not to mention that measuring between the PSU and the reactor will always give the right result and JedRothwell said Mizuno always did this, sometimes letting outside people use their own instruments to repeat it.

So the with 10% efficiency by the power supply, the measured power ratio is higher than it would be with 80% efficiency - not lower. The error would go the wrong direction. No?

Correct. I've noted that. However it means a possible mistake on relative (calibration vs active) results. There would also need to be a corresponding error or mistake in absolute results.

Quote from seven_of_twenty

I agree some colossal error is possible. even probable,

Why probable? Is there some technical aspects of this that makes you think a colossal error? Or is that just your intuition? Your intuition is unreliable in this case, because you failed to understand how insulation works, and you still do not understand that input power is not noise, and input power can be measured with ~100 times better precision than output heat. Your understanding of calorimetry is weak, and you cannot judge results based on your own understanding.

Quote from Robert Horst

Jed says Mizono measures input to the heater and you are talking about input to the power supply powering the heater.

Yes, and that is obvious from the data because it often shows zero input power, such as in Table 1 for 5/18. If this were the power going into the power supply, it would never be zero, as long as the power supply is turned on.

Quote from Robert Horst

Measuring power to a resistive heater driven by DC is about as easy as it gets.

Yes. THH says it is "complex" but he is wrong.

Quote from Daniel_G

I think Jed has cleared the this issue already as the power in is measured between the PSU AND also checked at the source as a “reality check” so no possible source for error here.

A reality check done at my suggestion. The plug in power meter costs about $70. U.S. versions are cheaper. It does not have a data port, so it cannot be included in the spreadsheet. It just sits there displaying watts and kilowatt hours, which accumulate until you reset it. These things are widely used to monitor the cost of appliances. Here is one that cost $18, with 0.2% accuracy. In other words, for $18, anyone can reduce the error margin for Mizuno's largest input power to ~400 mW, which is less than he can even detect on the output heat side. This is something Seven_of_twenty fails to understand.

https://express.google.com/u/0…7873416101415994_10918132

Quote from Curbina

I can only understand THHuxleynew obsession with power input to the heater from an assumption of fraud. If someone was determined to commit a fraud in this, that person could use some convoluted circuitry and tricks to have a concealed input masked within an exotic waveform. The Italian that I shall not name was accused of this kind of meter fooling shenanigans and AFAIK it was always suspected and accused of doing so but never conclusively proven other than this being the only “reasonable explanation” before conceding excess heat, but I will not address that any more as it is a well pass dead horse beating.

So, I think one cannot be faulted to denounce that this obsession with the heater power is a de facto tacit accusation of fraud, as no one would obtain a false reading without purposefully altering the system to do so.

Curbina What obsession? And what fraud? I have repeatedly stated that I do not look for fraud here. It seems unlikely. However mistake is infinitely likely. Maybe you would jump to fraud from blatant mistakes - if so i belive you are both uncharitable and wrong.

Quote from Robert Horst

When power is measured to the heater, it will be very precise, whether done with an expensive power meter or just measuring V and I and multiplying. Measuring power to a resistive heater driven by DC is about as easy as it gets.

I agree Robert. However I refer you to Jed who says that in some cases this measurement was done with a power meter on the input side of the heater PSU.

Quote from Curbina

You are the one obsessed to find a fault in the simple most undebatable of the reported data: the heater power consumption. There is no way that data is wrong as it was measured, and if you suspect that, you are tacitly accusing Mizuno of fraud.

Ok - so we must then disagree. And you must explain those 2017 data that concern ascoli.