MTBF II

[Eamon Searson]

[Originally Posted: 3/13/00-- Transferred by ReliaSoft Moderator]

Hello, I have a technical question regarding reliability growth analysis. The Duane plot is created on log/log paper (log to the base 10), from which I can easily calculate the slope. In determining the slope(Beta) the SEMI spec I have calls for Beta =1- (r-1)/(sigma(lnT/ti) (note natural log). In order to help clarify this, I have pasted in an example, and from the data given I calculate a slope (Beta) of 0.8 using the above ln formula, while I also calculate slope of 0.523 by taking the log base 10 approach. The sample data is: eleven fails occur at the following times: 18, 20, 35, 41, 67, 180, 252, 287, 390, 410 and 511 hours. What is the improvement slope?

Can you possibly help clarify this apparent inconsistency for me?

Thanks in advance,

Eamon Searson

[Adamantios Mettas]

[Originally Posted: 4/3/00-- Transferred by ReliaSoft Moderator]

The reason why these two values are different is because two different parameter estimation methods were used. The formula that you used for estimating beta is actually the MLE equation (I believe that the value returned by this equation is 0.4 and not 0.8). The second approach is the Graphical method. In this method the result is greatly dependent on how the line was drawn (the disadvantage of manual plotting). To avoid this drawback of the Graphical approach you can use the Least Squares method