To: Tim Smith, Shaheer Muhanna, Don Taylor, Peter Rollo

CC: Wesley McCain, Greg Bareta , David Batchelor

Editing and corrections of SIR 2019

Attached is a copy of the edited 2019 SIR Protocol by me. When NWGLDE and Battelle worked on the new version of the SIR during 2014-2015 we were asked to contribute and email our recommendations and corrections, etc. I emailed Anne Gregg several times with my corrections and suggestions. At the bottom of this email one of those corrections is attached for your viewing. Anne accepted and added most of my suggestions and additions including the one mentioned below in this email. One of my suggestions that didn’t get included was the numbering of the equations. I had recommended the equations to be numbered to make it easy to refer to them.

Unfortunately, at the time I was involved with other projects and the given time frame for all of us contributors to email our comments was short (I guess this was partly because Battelle had limited time to modify all the protocols). Therefore, I could not meet the deadline assigned by Anne and some of my corrections and additions were not emailed on time. I take responsibility for that and attached is a 2019 SIR Protocol I have read and attached my final recommendations and suggestions. Some of the corrections are minor but a few are important and will make a difference in the final analysis of the SIR data analysis or when a vendor wants to read the SIR Protocol and asks questions about the working of it. There are 11 corrections that are included in the attached SIR 2019 Protocol.

Below is one of my recommendations that were accepted by Anne. I recommended taking out from the new SIR a couple of lines. If someone looks at the SIR 2019 version they will see that “P(d) plus P(fa) will always equal exactly 1.0” which were right above the formula for P(d) shown below and was part of the 2015 edited version of SIR are now no longer in the SIR 2019.

“Below is from the 2015 SIR Protocol that we were asked to edit:

5.4.3 Probability Of Detecting A Leak Rate Of R GallonPer Hour, P(d), For Single VersusManifolded Tanks, Separately

The P(d) is the probability that the method will correctly identify a leak of specified size. In general for a leak rate of size R, P(d) is given by:

P(d)=P{t>(C – R-B)/SD}

where C, B, and SD are as before. The probability is calculated from the one-sided t-distribution with (n- 1) df. Note, that if the B is negligible and the threshold is exactly 50 percent of the leak rate, then P(d) plus P(fa) will always equal exactly 1.0.

“

The comment H1 to the right is one of my comments which were added to the new 2019 SIR and the above was deleted from SIR 2019 protocol.

The new edition may be visited on page 20 of SIR 2019 with those two lines deleted.

Locations of the 11 comments:

1. 2 on page “V”

2. 2 on page 5

3. 2 on page 11

4. 1 on page 15

5. 4 on page 17

Regards,

Sam