1

**Performance Based Engineering (PBE) / Obtaining +1 std costs and downtime directly from DL summary files**

« **on:**May 30, 2022, 10:26:13 PM »

I am posting a discussion I had in email with Vaafoulay K., with their permission, to preserve it for others to see.

Vaafoulay: Can I obtain the +1 standard deviation costs and downtime directly from the DL summary files?

Adam: If you are interested in the mean+1 standard deviation, I suggest looking at another file, the DL_summary_stats.csv. That file provides statistics on the main outputs, including repair cost and repair time. If you take the value in the mean row and add the value in the std row, you'll get the mean + 1 standard deviation result.

Now, you might want to look at some statistics that are not included in that file. For example, it is quite common to assume that the distribution of the results is lognormal and we are interested in the median + 1 log standard deviation. Such results are not provided in the stats file, so you'd have to calculate them using the sample in the DL_summary.csv. So, you'd open the DL summary file, take the column of interest (e.g., reconstruction/cost) and use an app of your choice (e.g., Excel, Matlab, Python scripts) to calculate the desired statistic based on the available realizations.

Vaafoulay: Can I obtain the +1 standard deviation costs and downtime directly from the DL summary files?

Adam: If you are interested in the mean+1 standard deviation, I suggest looking at another file, the DL_summary_stats.csv. That file provides statistics on the main outputs, including repair cost and repair time. If you take the value in the mean row and add the value in the std row, you'll get the mean + 1 standard deviation result.

Now, you might want to look at some statistics that are not included in that file. For example, it is quite common to assume that the distribution of the results is lognormal and we are interested in the median + 1 log standard deviation. Such results are not provided in the stats file, so you'd have to calculate them using the sample in the DL_summary.csv. So, you'd open the DL summary file, take the column of interest (e.g., reconstruction/cost) and use an app of your choice (e.g., Excel, Matlab, Python scripts) to calculate the desired statistic based on the available realizations.