Novel Models for the Prediction of Key-Parameters for Multi-Directional Magnetostriction of Highly Grain Oriented SiFe

Every change of the magnetic state of soft magnetic materials yields a change of shape. Though this phenomenon of magnetostriction (MS) is extremely weak, it causes distinct noise emission of magnetic cores of electric machines. For the first time, applying a novel Rotational Single Sheet Tester (RSST), we investigated MS as resulting from multi-directional magnetization of highly grain oriented SiFe materials. The focus was put on rotational magnetization as being typical for power transformer cores. In an unexpected way, the study revealed extremely high values of MS, a finding which is of direct practical relevance for modern core designs. Magnetostriction represents a very well known phenomenon, and manufacturers of magnetic core materials have succeeded in an effective way to reduce the MS values of their products by more than one order of magnitude in recent time. However, this reduction concerns the special case of magnetization in rolling direction. In the course of increased environmental consciousness, Japanese researchers have investigated also the case of rotational magnetization (RM) which proved to be characterized by a rotation also of the main axes of MS-caused strain. The latter showed moderate extent, however, the investigated material being of conventional non-oriented type. Specifically for the simulation of RM of advanced, highly textured materials, we developed a novel RSST which uses a hexagonal sample for the first time. It offers the advantage that 3-phase excitation can be used in analogy to practical conditions in machines. The time pattern of multi-directional magnetization was controlled by Fuzzy logic algorithms. With nanometer resolution, MS-caused strains were detected by means of three laser interference units. However, in a later state of project, they were replaced by extra-long strain gauges due to more stable behavior. In addition, the RSST yields power losses which are capitalized even higher than noise. Finally, in co-operation with Japanese partners, a laser Kerr effect system was developed for dynamic observation of magnetic domains. The RSST application was focused on the most advanced types of h.g.o. SiFe as used for large power transformers. The results revealed non-rotating main axes of MS-caused strain. The latter showed extremely high orders, exceeding the nominal values (for magnetization in r.d.) by up to a factor 100. This unexpected finding shows direct industrial relevance. Thus the world-wide largest producer of transformers - who offered valuable support to this project - has already started to incorporate the results into his strategy of core design. To facilitate the industrial use of data, we established an artificial neural network which - trained by results of measurement - offers approximate MS predictions also for non-investigated magnetization patterns. Summing up, apart from delays due to crucial tasks, the project ended in a most successful way. The results were published in many papers and found very high international interest. Finally they stimulated the organization of an international scientific meeting on the subject which was held in Bad Gastein (Austria) in September 2000.