Question

Parametric Analysis of Cost and Reliability

Answer 1

Answer: Parametric cost analysis uses equations to map measurable system attributes into cost. The measures of the system attributes are called metrics. The equations are called cost estimating relationships (CER's), and are obtained by the analysis of cost and technical metric data of products analogous to those to be estimated. Examples of system metrics include mass, power, failure_rate, mean_time_to_repair, energy _consumed, payload_to_orbit, pointing_accuracy, manufacturing_complexity, number_of_fasteners, and percent_of_electronics_weight.

The basic assumption is that a measurable relationship exists between system attributes and the cost of the system. If a function exists, the attributes are cost drivers. Candidates for metrics include system requirement metrics and engineering process metrics. Requirements are constraints on the engineering process. From optimization theory we know that any active constraint generates cost by not permitting full optimization of the objective. Thus, requirements are cost drivers. Engineering processes reflect a projection of the requirements onto the corporate culture, engineering technology, and system technology. Engineering processes are an indirect measure of the requirements and, hence, are cost drivers.

Many metrics are obvious. Mass and lines_of_code are measures of unit production effort. Number_of_production_units and number_of_prototypes are measures of program size. Technology is a function of time, so its effects may be measured through changes with time. For expendable launch vehicles the mass of the tankage is proportional to the tank volume. Thus tank _volume, energy_consumed, and fuel_energy _density are functionally related to mass, a measure of production effort. Other metrics are not so obvious.

Parametric analysis normalizes for the effect of metrics xk.

rs

c = ea0 ∏eaixi ∏xjaj

i=1 j=r+1

is a commonly used CER with associated linear form