where the nj are random variables with mean zero and standard deviation a. An exact solution for the mean and variance (although not the entire distribution) of the levelized costs, regarded as random variables, can be obtained. Another problem in such comparisons is to take correlations into account. If, for example, the cost of electric equipment for one system is high it will probably be high for an alternative. This is dealt with by formulating the difference between levelized costs as the random variable. The result is to indicate a wider difference between two technologies than if correlations were ignored.* Technical Uncertainty The major uncertainty regarding the capital cost of SPS is with the nature and performance of the ultimate configuration. The solar cells and the space transportation and construction are the main uncertain elements. There is little basis in experience for these costs other than the observation that early estimates of such costs are consistenly low. This suggests the use of a skewed probability distribution of SPS costs. An estimate of the most probable cost of SPS installations in production has been provided by M. Samsa of Argonne National Laboratory; namely $3,650 per kilowatt (1978 $), together with a lower limit of $3,150. In lieu of an upper limit, which is much more conjectural, I have used a parameter f, which is the probability that the SPS capital cost lies below the most probable value. In accordance with the above the value of f is assumed to be less than 0.5. The shape of the distribution is only of significance to these calculations below the most probable value, and I have assumed a straight line. To illustrate the kind of comparison that is possible the input numbers for an example are assumed to be: * Another problem, with correlations, whose correct treatment will generally have the opposite influence, arises if the capital cost of a single system is treated as the sum of independently costed subsystems. This effect is not studied here. A* This estimate of today's cost is based on the assumed solution of all technical problems leading to full production. All initial numbers were furnished by M. Samsa of Argonne National Laboratory.
RkJQdWJsaXNoZXIy MTU5NjU0Mg==