222 THE BELL SYSTEM TECHNICAL JOURNAL, JAXUARY 1954 



In carrying out an actual study, a point of reference is needed first. 

 This involves (a) fixing a standard against which all costs will be com- 

 pared, and (b) expressing the value of all features in a common lan- 

 guage. Once this has been done, every factor may be evaluated as an 

 incremental "cost penalt}^" over and above the reference standard. 

 Finally, in making comparisons, the reference standard will alwaj^s 

 subtract out, so that one needs only to add all incremental cost penalties 

 to get the total cost of the design changes in mind. 



The evaluation of all variables on a common basis is covered in Part I, 

 which considers various manufacturing costs, power costs, and the cost 

 of functioning time. 



The remaining parts of this paper then de\'elop relations for maximum 

 economy in the switching system, for some important cases that arise 

 in practice: 



Coil design for maximum power economy. 



Economical number of occasionally unused extra parts. 



Economical adjustment for speed relaJ^ 



Coil design for maximum combined power and speed economy. 



So far as possible, results are given in general form, permitting one to 

 work from charts when considering specific application problems. Of 

 further benefit are charts which permit the designer to decide the eco- 

 nomic disad\'antage of a design which may depart from optimum. 



Design for systems economy through relay design includes many other 

 important topics not considered here, as for example: how to make best 

 use of molding, welding and other manufacturing processes; or how to 

 design for long life, reliability, and other maintenance concerns. This 

 article covers only some typical problems for which optimizing methods 

 are readily applied. 



Part I — Expressing Systems Performance and Manufacturing 

 Variables in a Common Language 



In the over-all switching system, one must e^'aluate various effects 

 whose costs seem ciuite different, and a decision must be made as to 

 what design course to follow, no matter how varied the conditions may 

 be. Many of the most important such cases can be handled quite easily 

 by a process of converting actual cost in the telephone plant back to an 

 equivalent cost in terms of the factory production cost of each indi- 

 vidual relay. Once all such costs are so stated, it is possible to examine 

 the incremental effect of each change, and confidently draw conclusions. 

 For example, if it can be stated that a change in a relaj^ coil will save an 

 amount of power that is worth 50 cents per relay equiA'alent first cost, 



