Comment on "Wind Power Limitations Associated with Vortices"

Abstract

the costs estimated, although far lower than present-day experience, were about twice what the mature industry correlations predicted. Predictions of costs for SPS components employed factors related to increased use of automation at high production rates. The data base supporting this practice is scantier than the data base for the CER's used to predict development and first unit costs. Nonetheless, the detailed cost analyses that we were able to perform to validate these predictions supported the general methodology. Dr. Hazelrigg argues that the correlations observed in the CER data base are actually caused by market forces and hence are not genuine correlations between product physical parameters and resources required for their production. His argument is false for two reasons: First, the CER data base does include data points for programs which were cancelled. Our own data base includes, for example, the Dyna-Soar and SST programs, both of which proceeded sufficiently far to provide experience in the cost of producing elements of these systems. Second, the argument that high-priced programs are cancelled is not necessarily valid. One can think of several programs that exceeded early cost predictions by large factors and yet were completed. In these cases the market (or need) was able to sustain the resulting price. Program cancellations have occasionally resulted from cost problems, but more often from changes in perceived need. Again, a confusion between cost and price seems to be present; CER's are based on experienced cost and have nothing to do with-the price for which products may be sold. To some degree, cost predictions tend to be self-fulfilling prophecies. This often unstated idea is exploited by the design-to-cost approach, in establishing a reasonable economic price for a product. Design-to-cost proceeds by requiring the design team to develop a system design that can be produced for less cost than the target price. This is not always possible, of course, but is more often successful than might be imagined. The procedure causes system designers to be at least as concerned with the cost effects of their design decisions as with the performance effects. Properly applied, it leads to significantly lower costs than possible for a performance-maximized system design. Design-to-cost has been used in the SPS system studies. As a result, SPS cost predictions have been criticized as mere allocations of targets. They are, however, the results of cost predictions using CER-type models with the best available data base. (The targets we would like to hit are significantly below the present cost prediction results.) Dr. Hazelrigg has, although not in his comment, advocated the use of bottom-up or grass-roots cost estimating for SPS. It is our experience at Boeing that grass-roots cost estimating applied to a new product in an early phase of design study will generally lead to unrealistical ly low figures and that the CER procedure is far more reliable. The success of CER methods, as noted earlier, depends on the validity of analogy between the product being estimated and the products upon which the CER is based. In the case of a solar power satellite system these analogies will range from very good to somewhat suspect. As a result, the cost estimates must be used with due regard for the inherent uncertainties. It is silly to say that costs cannot be forecast although they cannot, of course, be forecast with certainty. Our cost analyses for SPS have convinced us that there is a good chance that SPS's will be an economical energy source. Laboratory investigations to adapt and prove out the critical technologies are a worthwhile investment.