278 -METHODS OF CALCULATION 



same scale of volts at the terminals corresponding to this reactive 

 delivery, this segment will be equal to the ordinate QN U, taken 

 between the curve and the straight lineylP. In fact, the back ampere- 

 turns will then be equal to AB, and the loss of voltage by dispersion 

 equal to BQ, by reason of similar triangles. Besides, the point Q 

 divides the straight line AP in the ratio of the current Id to the short- 

 circuit current /.. 



Thus arranged, suppose that any reactive current Id be taken 

 experimentally at the corresponding e.m.f. U; the point Q will be 

 determined by these two conditions: its vertical distance from the 

 curve of excitation is equal to U and its radial distance from the 



point A is equal to XAP. 



*cc 



The point is found, therefore, at the intersection of two new 

 curves that are easily drawn: a curve parallel to ONM traced at 

 a vertical distance U below the former, and a curve homothetic 

 to the curve of excitation with respect to the point A, with homo- 

 thetic ratio -=. These two curves are parallel in their rectangular 



*cc 



parts, and separate as much one from the other as the point M is 

 selected further beyond the bend in the characteristic. They would 

 coincide if the point M were below the bend. The experiment should 

 therefore be made with an excitation OA sufficient clearly to pass 

 the bend. 



The test is made by causing the alternator to operate first on 

 short-circuit, and then upon a reactance-coil having an open mag- 

 netic circuit, or upon an under-excited synchronous motor, giving 

 cos^> less 0.20, that is to say, a current almost entirely reactive. 



It is understood that if not only one, but also several, reactive 

 circuits are tried, the straight line PQ will be still better determined 

 thereby, and consequently CDS will be known with correspondingly 

 greater precision. 



Analogies between this Method and that of Potier-Behrend. 

 It is possible also to follow a somewhat different course, by causing 

 the alternator under test to operate upon an inductive circuit with 

 cos0 nearly o (for example, upon a synchronous motor driven by 

 a motor adjusted in a manner to produce just the power consumed 

 at light load), and by varying the excitation of the alternator; in 

 this manner a constant reactive-current curve is obtained, from which 

 may be deduced, by the method of M. Potier, the coefficient of the 



