use electricity to power the pump motor [Stanley et al., 1980]. A
sample simulation program is presented in Appendix C.

 Economic and Technical Assumptions

 Utilizing data contained in Harrison [1978], a hypothetical per-
manent overhead system irrigating 40 acres of citrus was modeled. Each
irrigation application was defined as two acre inches (gross of water
applied in a period of six days with the system operating 18 hours per
day. A total of six applications per year was assumed (12 acre-inches
per year) and held constant throughout the program. Assuming a total
dynamic head of 250 feed [Stanley et al., 1980], the motor size necessary
to power the system was estimated to be 32 HP. When in operation, the
motor required a continuous electrical load of approximately 24 KWH/H.
 Based on the photovoltaic system design rules discussed previously,
the array size necessary to power the irrigation system was calculated
to be approximately 766 m2 (61KWP). Under average solar insolation con-
ditions this system is estimated to produce about 89,865 KWH of electri-
city annually. The estimated annual amount of electricity needed for'
irrigation is approximately 15,510 KWH, yielding a net generation of
electricity for resale to the utility of 74,355 KWH.
 Three basic economic scenarios were simulated for the irrigation
system described above (Table 2). The scenarios (base, optimistic, and
pessimistic) differ primarily on the basis of expectations regarding the
rate at which the cost of photovoltaic systems declines over time and
the real rate of increase in electricity KWH rates over time. The
scenarios also differ in the degree to which economic and technical
assumptions are amenable to establishing the economic feasibility of
photovoltaic powered irrigation systems. For each of these scenarios
the buy-back ratio linking the price of resale electricity to utility
rates is varied from 0.25 to 1.5 in 0.25 increments.

 Simulation Results

 For each of the three economic scenarios presented in Table 2, a
 total of six simulations were run. This resulted in a total of 18
 different model scenarios. In all but two cases, the discounted cost