Data Driven

Thursday, March 10, 2011 by Jim Otte

My previous blog discussed a “Should Cost” methodology used by PRICE Systems to complete an analysis. In the article I included a chart depicting calibration results for manufacturing complexities for each weapon system (X-Axis). Manufacturing complexities are a major cost driver within the model. This parameter can be derived from model knowledge tables, generators or from calibration. Many times the calibrated results are simply averaged and used for predicting cost for the new system. This assumes that the new system is very similar in technology and performance as the systems used for calibration. In general this is not the case. Below I discuss how data driven cost estimating relationships (CER) can be developed between the calibration results and a selected independent variable. The CER will then be used to derive the manufacturing complexity, which then matches much closer to the new technology.

graph one
Figure 1. Calibration Results for each Weapon System

Since the data was already collected and calibrated within the model, the next step was to identify an independent variable. For manufacturing complexity for structure, thrust was selected. However, other variables could have been selected.  Maximum take-off weight, range, ceiling, empty weight, thrust per pound, or speed were all candidates for an independent variable. For manufacturing complexity for Electronics, frequency speed was selected as an independent variable. Again there are many potential candidates for an independent variable for electronics. 

The next step was to graph and complete regression analysis. Figure 2, depicts the results for manufacturing complexity for structure, and figure 3, depicts the results for manufacturing complexity for electronics.  I normally select a power function when regressing data. One could select any function except linear. Linear regression does not always work very well and as a practical matter technology advancements are rarely linear. 


graph two
Figure 2.  Manufacturing Complexity for Structure


graph three
Figure 3. Manufacturing Complexity for Electronics

Using the above equations now require information be collected on the aircraft thrust requirement for structure, and frequency speed for electronics. These requirements are then used to calculate the appropriate manufacturing complexity. This approach insures that the cost estimate now match up with the requirements. 

The next time you need to complete a cost analysis, talk to one of our expert PRICE Solutions consultants for help.

Comments for Data Driven

Wednesday, March 16, 2011 by Kevin:
So the 'above equation' can be used to calculate a Complexity given an aircraft thrust?
Thursday, March 17, 2011 by Jim Otte:
Yes a manufacturing complexity for structure can be derived based on thrust. This assumes that you calibrated the model to solve for this complexity, along with completing regression analysis to derive the equation. The above example also excludes the electronics manufacturing complexity.
Friday, March 25, 2011 by Kevin:
Ok, well that is why it does not make any sense. If your x-axis is system number (as stated above), the above equation does not predict Complexity based on thrust.
Tuesday, March 29, 2011 by Jim Otte:
The x axis refects thrust. i changed the values around to maintain data control, since this information is being broadcast outside the US. The regression curve reflects what can be done with calibration and relating the calibration results to an independent variable.

Leave a comment





Captcha