The following is a paper by H. Aspden published by the Institution of Electrical Engineers as Monograph No. 164M (January 1956) and later in Proc. I.E.E. vol. 103C at pp. 279-285 (1956).
THE EDDY-CURRENT ANOMALY IN ELECTRICAL SHEET STEEL
Abstract: A theory which accounts for the well-known discrepancy between eddy-current losses in electrical sheet steels and the experimentally observed values is presented. The anomaly is shown to be due partly to the magnetic inhomogeneity arising from ferromagnetic domain structure and partly to a time-lag effect caused by the finite speed of domain boundary movements. A new experimental approach to the study of the eddy-current anomaly is described. This involves the use of a method of measuring the anomaly factor as it applies instantaneously at a point in the magnetization cycle.
Commentary: The advance reported in this paper was the step of measuring the eddy-current loss anomaly factor, not as an averaged effect taken over a full cycle of magnetization, but rather as an incremental effect over selected portions of the B-H loop. It was found that the main anomaly effect was occurring over the low flux density range. Indeed, over such a range it could be far in excess of the mean value normally attributed to the phenomenon.
In an extreme case one can show that magnetic domain inhomogeneities could account for a loss anomaly factor as high as 3 in thin sheet steel, but the research reported here revealed loss factors appreciably higher than 3. It followed that a time-lag effect enhancing the hysteresis loss could not be ruled out.
In retrospect, these notes being written some 43 years on from the date of the subject paper, the author now admits that the research suffered from a rather grave omission in that no account had been taken of the effect of the heat generated by eddy-current and hysteresis loss. This heat would flow from the magnetized core in the plane of the laminations and so in a direction at right angles to the magnetic polarization within the magnetic domains.
By the Nernst Effect this is a recipe for the induction of mutually orthogonal electric fields powered by tapping that heat and it is now realized that this would, in fact, add to the induced EMF driving currents around the eddy-current loop, enhancing eddy-current flow and so escalating into an anomalous loss effect.
The only excuse which the author can offer for this oversight is that there seemed at the time no reason to even think about thermoelectric effects affecting electric current flow in a single metal, steel, and even had that been contemplated the limited temperature range involved would hardly have suggested scope for efficient conversion of heat into electricity on the scale needed to explain the anomaly. Now, also seeing all this in retrospect, there is good reason for questioning the validity of what one had been taught concerning the second law of thermodynamics, or at least the applicability of this law to specific conditions prevalent inside a metal conductor where absolute temperature has no significance and only the temperature differential contributes to the effect considered.
This aspect of the eddy-current anomaly can be far more important than the direct implications of the actual power loss. Indeed, one can begin to see scope for a new method of electrical power generation drawing on the ambient heat of our environment. This prospect gains strength from the discovery that eddy-current anomaly factors much higher than 3 were observed, particularly over the low flux density range where the maximum action attributable to the Nernst Effect will occur. The reason for this is that the domain magnetization is then equally shared by the two polarization directions, so that the augmenting EMFs induced will see flow paths of least resistance.
See also the related papers: [1956a] and [1957a].