(Georgia Institute of Technology, 2005-06)
Ristow, Alan; Begović, Miroslav; Rohatgi, Ajeet
The cost of energy produced by a photovoltaic system is dependent upon the amount of energy produced by the system and the amortized cost of the system's components. Existing simulation tools either use crude estimators of system reliability or ignore the effects of system downtime on energy production altogether. Furthermore, the costs associated with system components are often not known precisely during system planning. However, it is difficult to reflect this uncertainty in energy cost calculations using conventional deterministic
techniques. This work attempts to address these deficiencies by applying a stochastic model of system reliability to the prediction of energy production over a system's life. Similarly, it uses a stochastic model that encompasses the
uncertainties associated with system component prices to estimate the uncertainty in the total installed system cost. Finally, using these two results, it computes the uncertainty in the cost of energy produced by the system. Preliminary testing of this approach, using failure data obtained from an actual system, produces cost estimates of
$0.300–0.400/kWh, with a mean of $0.349/kWh, consistent with contemporary residential system cost analyses. The link between engineering and economics suggests that the proposed method may be useful as an optimization tool if an appropriate database can be developed from which to draw realistic input distributions.
(Georgia Institute of Technology, 2004-06)
Ristow, Alan; Begović, Miroslav; Rohatgi, Ajeet
Forecasting of the manufacturing cost of PV modules is governed by a large number of uncertain factors. Cost estimates are frequently based upon imperfect information and, as a result, may not be perfectly accurate. Existing studies of these uncertainties focus on the sensitivity of the manufacturing cost to individual cost inputs, examining the effects of each input in isolation. Such methods of analysis neglect statistical correlations between inputs, provide no measure of the uncertainty in the projected manufacturing cost, and do not permit the assignment of
probability distributions to the inputs in the case that one range of values is thought to be more likely than another. This work describes the development of a stochastic modeling framework that addresses these deficiencies. Furthermore, it demonstrates how sensitivity to particular inputs may be ranked in order to help determine the most effective path to cost reduction. The result is a method with great potential for exploring the link between engineering design, PV module cost, and the manufacturing process.
(Georgia Institute of Technology, 2001-10)
Pregelj, Aleksandar; Begović, Miroslav; Rohatgi, Ajeet; Ristow, Alan
In evaluating the payback time and energy price per kWh generated by a renewable energy system, the system is usually assumed to work without interruptions over its entire life. Renewable energy systems are fairly reliable, but like any complex system, they may fail, and the effects of failures should be analysed and accounted for. This paper presents a procedure for filtering-out failures associated with the "infant-mortality" failure mode, and enables the estimation of distribution parameters of the constant failure rate mode that governs the behavior of the system throughout its useful service life. The procedure is best suited for cases when a large number of data are available, such as when a large number of similar systems are monitored.