OCT. 19, 1921 PARSONS AND HARPER: ENGINE RADIATORS 411 



(where H is heat dissipated per unit time per unit frontal area of 

 core, M is mass of air per unit area per unit time through the air 

 tubes of the core, and T is difference between the average tempera- 

 ture of the water in the radiator and the temperature of the air at 

 entrance to the core) was found to hold for all ordinary types of core 

 construction, h and n being constants pertaining to any particular 

 core. The numerical value of n was usually in the neighborhood 

 of 0.8, indicating that the relationship between heat dissipation and 

 air flow through the core of a radiator is not far from the relationship of 

 direct proportionality, although too far to assume such a relation 

 over any considerable range of air flows. 



A second equation for computing the cooling power of a core is 



H = CpMT (1 - e~""') 



where H, M, T have the significance quoted in the preceding para- 

 graph, e is the logarithmic base 2.718, Cp is specific heat of air at 

 constant pressure, Xi is the depth of the radiator core (dimension par- 

 allel to air flow) and « is a constant for a given core and given air 

 flow through it. Since a chosen value for a permits no latitude in 

 air flow, or in changing core construction, the equation in this form 

 is intended primarily for study of the effect upon cooling power of 

 changing the depth of core. Giving due consideration to head re- 

 sistance as a function of frontal area and depth, the formula permits 

 computation of the optimum depth of radiator for given conditions. 

 For very high speed flight, it is advantageous to use cores much deeper 

 than was the common practice in early days of airplane development, 

 8 to 10 or 12 inches depth being desirable for some types of the usual 

 quarter inch to half inch air tube core, in unobstructed positions. 



The formula just quoted is not entirely empirical in its origin, 

 but has a rational basis. The assumptions underlying it are plausible 

 and it is not surprising to find that in applying it to results of measure- 

 ments with almost all common types of core these results are rep- 

 resented very' well. The parameter a is an abbreviation for pq/MCp, 

 where M and Cp have their former significance; p is total perimeter 

 of the air tubes in unit frontal area, provided the cooling surface is all 

 direct surface (and an equivalent value for the cases where part is 

 indirect cooling surface) ; q is the cooling coefficient between the metal 

 and an air stream of velocity corresponding to M, namely, q is heat 

 transferred per unit time per unit area of metal per degree tempera- 



