Guest post from ‘Lucy Skywalker’.
Last time I described my visit to Roderich Graeff’s seminar. Now we look at the experiments in detail. Experiment trumps theory. Experiment is the final arbiter, as Einstein said. Only after looking at the experiments do I want to consider the theoretical elements.
HISTORY AND CONTEXT
One has to be aware of the depth to which the Second Law has been held most sacrosanct of all the laws of physics. Eddington said famously:
“If someone points out to you that your pet theory of the universe is in disagreement with Maxwell’s equations—then so much the worse for Maxwell’s equations. If it is found to be contradicted by observation—well these experimentalists do bungle things sometimes. But if your theory is found to be against the second law of thermodynamics I can give you no hope; there is nothing for it but to collapse in deepest humiliation.”
But today, Dan Sheehan of the University of San Diego could write in his 2005 book “Challenges to The Second Law of Thermodynamics“:
“The second law has no general theoretical proof. Except perhaps for a few idealized cases like the dilute idal gas, its absolute status rests squarely on empirical evidence. As remarked by Fermi and echoed by others, “support for this law consists mainly in the failure of all efforts that have been made to construct a perpetuum mobile of the second kind”… [yet] One would be hard-pressed to name ANY physics theory, concept, law or principle that has not undergone major revision either in content or interpretation over the last hundred years… The damning question is, why has it taken so long for [the 2LoT's] absolute status to be questioned?”
Dan Sheehan thought it worthwhile to visit Graeff at home and see his experiments. Graeff’s unique challenge is backed by years of engineering-quality experimental work. His apparatus can hardly be thought of as a perpetual motion machine. And his challenge is not to the essence of the Second Law, but a modification needed to one of its commonly supposed consequences, namely, the believed equality of temperature between top and bottom of a column of air in equilibrium. Unbelievably, until Graeff, nobody actually tested this supposition. Graeff’s results, at laboratory scale, are tiny but are highly consistent, significant, and replicable. But perhaps even twenty years ago such results could not have been detected.
Graeff presented an excellent paper at Sheehan’s 2011 conference, that is well worth perusing. He noted wryly that he was the only one at Sheehan’s conference who had actually done experiments. This is from his introduction to his paper:
“It is known that temperature gradients in gases and liquids are stable only up to the adiabatic lapse rate… In order to make greater values possible, the author tried various convection-suppressing designs. It was found that the use of fine powders like glass powder largely eliminated convection currents. This has the added advantage of preventing any heat exchange by radiation within the test setup.
“The basic setup consists of an elongated container in a vertical position, with volume varying from ½ liter up to 2 liters, and height varying from 15 cm up to 100 cm. This container is highly insulated from the environment… The temperature gradient T(Gr) of the inner axis is measured with thermocouples, arranged as a thermopile.”
TWO CLASSIC EXPERIMENTS
This is the setup for the 2007 water experiment B372, from the paper already discussed on this blog, with colouring added for clarity. Pinks are insulating layers; greens are thermal-equalizing layers (metal, water jacket). The core (1-5) is not drawn to scale: it is narrower and taller. Read the details of experiment B372 further down.
TESTING AIR: B74 (from website)
When planning the initial experiments in 1998, Graeff did not know what to expect. If there was a temperature difference, it would be a very small one, as otherwise somebody would have measured it long ago. But was it 0.01 K/meter of height, or 0.001 or only 0.0001K/m?
He settled very quickly on the use of thermocouples to measure the temperature difference. Thermocouples don’t introduce any energy into the experiment. They cannot create temperature differences. Just the opposite in fact.
For measuring the voltage produced by the thermocouple he selected a multimeter with a resolution of 0.1 microvolt, which, dependent on the type of thermocouple selected, would have a resolution of about 0.003 K.
The most difficult task is to insulate the actual experiment from the temperature influences in the space around it. Rooms typically have a temperature gradient of +1K/meter of height, warmer at the top. Temperatures fluctuate from day to night, from winter to summer. How can one expect to measure meaningful temperature differences of a few thousandths of a degree if the outside temperatures show fluctuations more than 100 times as great?
This picture shows the setup. A Dewar insert of a commercial Thermos bottle (1) was mounted within a wide mouth Dewar insert of 1 liter size (2) which was covered by a similar Dewar insert of 1/2 liter (3). The space (4) between the innermost Dewar and the two outside Dewar inserts was filled with fine PET fibers.
The innermost Dewar (1) of 1/2 litre was filled with a fine powder in order to eliminate convection currents and radiation between the inner wall surfaces. A thermocouple (5) was arranged vertically in the middle axis with a distance between junctions of 170 mm. A second thermocouple (6) was taped to the outside of the Dewar insert (1) with a vertical distance of 180 mm. A third thermocouple (7) was taped to the outside of Dewar insert 2 with a vertical distance of 230 mm.
The Dewar inserts were held in place by fine PET fibers within a box (8) fabricated out of 40 mm thick polystyrene foam panels. This box was surrounded by 50 mm thick panels (9) consisting of copper wires pressed together into rectangular blocks. The whole setup was insulated from the air in the room by 100 mm thick polystyrene foam panels (10).
The diameter of the thermocouple wire is kept as small as possible in order to reduce any heat conduction through these wires. The measuring instrument measures the voltage of the thermocouple with a resolution of 0.001K.
TEST RESULTS OF B74:
The temperature differences shown by the three thermocouples were recorded every 4 minutes over a period of half a year. In this graph, each point represents the average of 60 measurements over a 4 hour time span.
Any point above the zero line means that the upper junction of the thermocouple is warmer than the lower junction.
It can easily be seen that the temperature difference on the outside of Dewar insert 2 is quite constant and always warmer at the upper junction than at the lower one by about .045K. This is the result of the temperature gradient in the surrounding room air, which is warmer at the top and cooler at the bottom, typically by 1K/m of height. The insulation around the Dewar inserts, and especially the thermal equalizing effect of the thick copper plates, are why this gradient for the Dewar inserts is so low.
An opposite effect can be seen at the inner axis, where most of the time the temperature of the upper junction is lower than the lower one.
This result is more clearly demonstrated in the following graph. Each point represents the average of all values from the beginning of the test up to that point [date].This cumulative calculating procedure shows a clear average steadily emerging from noisy data:
The middle axis shows, on average, a negative temperature gradient, cooler at the top and warmer at the bottom, that cannot be explained by applying the normal formulas for the conduction of heat. Under these laws, heat can travel only from an area with a higher temperature to an area with a lower temperature. The value of the inner axis of -0.0705 is surprisingly close to the theoretical calculated value of -0.07. One would expect a measured value somewhat closer to that of the surrounding gradients, and certainly not one in opposition to them.
Both calculation and explanation require consideration of the effect of gravity on the molecules in the innermost space of Dewar insert 1, translating kinetic energy acquired into heat transported from top to bottom. Only one of the degrees of freedom is affected by gravity, and therefore the normal specific heat must be divided by the number of degrees of freedom involved, to calculate the theoretical temperature gradient. Normally this gradient is lessened by convection in the atmosphere, and seems to be completely neutralized by convection in the oceans – thankfully, otherwise the planet would be uninhabitable like Venus.
TESTING WATER: B372 (from Graeff’s paper). See setup diagram higher up.
The temperatures inside the test setup are measured by thermocouples and by thermistors. These are mounted at the tops and at the bottoms of the inner axes of the two glass tubes. Additional sensors are mounted on the outside of these glass tubes and on the outside of the two PVC tubes. The temperatures of the double wall aluminum housing are measured 3 cm below the top and above the bottom.
The test setup was installed in May 2006. All sensors were connected to DMM Multimeter Keithley model 2700 and the data fed into a computer. Measurement results are reported from December 2006 through March 2007, a time period long after the setup, so that it can be expected that equilibrium conditions had been reached.
WATER TEST RESULTS
Caveat:: I’m waiting for Graeff to say whether the numbers 9-14 below are correctly allocated.
Each point of the curve represents a 10 value average (of a ten times repeated reading of the same object) measured every hour, using the scale on the left side of the graph. The smooth lines represent the thermistor measurements, each value measured hourly in centigrade, using the scale on the right side.
This graph yields many secrets. First, see how far the convection-inhibited water (1) depresses the negative temperature gradient. (1) is firmly negative. (6) and (7) are firmly close to zero. (8) is firmly positive. By the standard interpretation of the Second Law there should be nothing at all in the blue section. The “free” water (2) also shows a negative gradient, but far closer to zero, and to its PVC casing. Has it simply picked up the gradient of the convection-inhibited water? Plus that of the surrounding convection-inhibited air in the lagging? Now see how steady the results are holding There are no wild flickers as with the air experiment (B74). This is because thermal capacity is far higher so the experiment is more robust to external fluctuations. All the dots connect coherently into sets of lines that run parallel, so we can trust that something is being measured reliably. And what a fine scale – this is because they are 10x more sensitive to gradient than to absolute temperature. Yet the thermistors also give very steady curves, running parallel showing they too are measuring coherent values. See how the gradients all together blip up shortly after the temperature stops falling and starts rising? Conversely, all the gradients dip down shortly after the temperature stops rising and starts falling. Why? Evidence suggests that the top of the core is affected by external temperature changes before the bottom of the core is affected. The thin lines in mid-January indicate a nonfunctioning datalogger for that period – but results are not affected.
The paper on B372 has two more graphs which are worth revisiting: it is vital to appreciate the precision of work here, and how it has been possible to extract further precision through cunning application of statistics. The question still needs to be asked: why is this measured gradient a tiny bit larger than the calculated gradient? But even this question cannot take away the fact of a negative gradient surrounded by a positive-gradient environment.
We will elaborate on the theory next time. For now, it is important to establish the validity and importance of the experiments. Until Climate Science with Santer, Trenberth, and others, the primary importance of significant “rogue” data, either from experiment or from observations, has always been the case with Science. Almost always, theories have followed data discoveries; the existence of theory-less data is not an indication of bad science, in fact quite the opposite. This should not need to be said, but I have seen too many scientists claiming in effect that without a theory, the data does not constitute Science.
Meanwhile, for those already interested, there are calculations, with theory explained more or less, here and here. Graeff’s book contains two chapters I regard as important, which explain nicely his own journey to understanding, and how his building of theory happened.
Graeff had a drum made from thick aluminium cylinders, with inner and outer containers capable of revolving separately, very slowly. First he rotated the outer drum, to provide an external environment with a much more constant temperature – only to prove that this did not materially affect his essential core results anyway. He then rotated the inner drum with the tests, and showed that about a day after inversion, the negative temperature gradient would reestablish itself. Here it is difficult to explain these results by anything other than gravity.
Graeff has tested metals and other materials like oil. Xenon testing is hoped but difficult. Prof Chuanping Liao has done interesting tests including using a centrifuge. But as Graeff says:
“These results are only meaningful as long as the environment shows a positive gradient, opposite to the negative gradient measured for the inner axis. This question Prof. Chuanping Liao does not discuss in his paper”.
Some mention is made of Graeff’s range of experiments in his 2011 paper. The experiments are ongoing and more has been tested since then. Graeff will now say that whereas he would have had doubts earlier that his experimental results could be flukes or have some non-gravity explanation, he cannot really assert this any longer, in the wake of his hundreds of experiments containing data that is both statistically significant and well-tested for experimental bias.
I am planning another post to look at Graeff’s theoretical work, and to discuss my plans for replication. I thought I could do both these in this post, but I see we will overload, and important insights and directions for discussion will be neglected. IMHO Graeff has worked out all the theory we need, in a straightforward way and with a delightfully human touch.
I am hoping to return to Graeff in the Black Forest presently with a small group, to apprise ourselves of exactly what we need for replication, to discuss what we should focus on, and how to get the word out where it needs to go ie universities. This will also be an excellent opportunity to learn to work together as a team.
My mind is in some ways quite simple and my scientific virtuosity is sometimes limited, and I really don’t want to get any more complicated than we need. I am focussed above all on restoring integrity to Climate Science, and by extension, to Science in general again – and to open up the field to participation with amateurs, as was always originally the case. I have tried to tackle one issue at a time, with enough thoroughness and politeness so that like Steve McIntyre we can work towards stopping all back exits to the current levels of slovenliness, bias, corruption and stupidity.
To me, it is plain that this experimental challenge of Graeff’s is highly important, both to Science in general, and to Climate Science in particular. I personally am also concerned that this important work has so far not been accorded the status of reposting at blogs like WUWT. I’m aware of course that problems arose over Willis’ perception of Joel Shore’s treatment here, and over widespread dismissal of Nikolov and Zeller that was largely because of their alleged flouting of the Second Law. So after explaining Graeff’s theory, I propose to write an article on Science By Blog, effectively Experimentum Summas Iudex part 3, with reference also to some familiar issues of “Freedom of Speech.”
For now, I ask commenters to stick with the overall message of this post as far as you can please! I think it is important that we take one step at a time, and validate the experiments first, before proceeding to theory and thence to the questions of getting a fair hearing, both on the climate skeptics’ blogs and in the universities’ research departments.