frequency) on edges is p mnt only but cylindrical-like waves also and this is of great significance. To establish the success criteria of UTD at low frequency the fdlom'ng needs to be defined: - The type of UTD derived resuîts to be assessed (Le. radiation patterns, field values, surface currents on scatterers, etc.) - The formulation of UTD to be usd specially in wmputer programs. - The independent rnethod (able to sohre similar problems as UTD) and its cornputer formulation to be used to test the UTD results. Obviously this method should be totally UTD independent from the theoretical formulation point of view. It should be also a -Il-proven rnethod already u d with succesS. The above mentioned subjects will be awered in the folldng dons.
2.2 CHOlCE B€IWEEN UT D RADIATION PATTERNS & SURFACE CURRENTS At the beginning of this investigation. radiation pattems wre considerd to be more pradical as a UTD end result. Sine the interest however was to explore the low frequency limit of UTD, the scatterers to tm useâ must k eIectrïcally small (dimension below 1 a). Several amputer FORTRAN codes (some atready existent and sorne created specifically for this task) were used to plot radiation pattems of line sources above dectrically small plates and cylinders. The rewL compareci well betwwn themselves. The patterns hanever were almt isotropic. This was something to expecî for a far field pattern when the only scatterers dishirbing the incident fields had dimensions of 0.2 a or below. Figure 2.2 shows the radiation pattern of a rnagnetic line source above an infinite strip (plate) and also above an infinite rectangular cylinder. Both the plate and the redangular cylinder have a Mdth of 0.2~ . The magnetic line source is 0.05A above the plate or cylinâer surface. The code used is based on the moment method. Figure 2.3 shows the radiation pattern of a mgnetic line source above an infinite plate with the same geomby as for Figure 2.2 except that the code used is based on UTD. It is quite obvious that both methods ernphasize the fact that at such a low frequency (electricaly smll scatterers) the radiation patterns are almorrt uniforni and omni diredional. Som plots also wenr done for the su- wrrents and they uuere almost idenücal non wiai~tanding the rnethod useâ to produœ them. It is interesting to note houmver that changes made to moment method codes to use a smaller number of sagmnts rawlteâ in naticsable differecc#, in the surface cumnt plots but almost none in the rediation pattern ones. ARar this observation it was decided to use surface cunent values and plots as the criteria