In English only
||||
 Topic: Geotechnical design and analysis Subject: Blast Attenuation Confidence LINES and INTERVALS
Follow-up:
 Author: Kristian 2004-01-05 Sorry, Link Shouild be:http://home.netvigator.com/~jksmurf/DowdingLatestBook.htmRgdsk.

Follow-up:
 Author: Kristian 2004-02-26 To close this out I think I now know the answer.Dowding, I believe, is correct (although there is a typo error in the example calc. in his book, for the SEE).The reason is simple. Dowding takes into account that the data is skewed, i.e. not a Symmetric Normal Distibution. In this respect ONLY if the log of the data follows a perfectly symmetric normal distribution will both methods produce the same UCL lines.Apologies to Prof. Dowding for my doubting him!k.

Follow-up:
 Author: Kristian 2004-02-26 Strike that. It's still way off base.When you check the number of data Points above or below the line, it is nowhere near the 1 in 20 predicted for 95%.Darnk.

Follow-up:
 Author: Kristian 2004-02-26 Found it ....Referring to the Regression Details discussed earlier, I believe there is an error (probably typographical) in both the Books "Blast Vibration Monitoring and Control" (1985) and "Construction Vibrations" (1996). The same error in both books.Where it is written (several times), in Chapter 4, that:(a) the 95% line value = the 50% value * 1.645 * 10 ^ StdError, I believe is incorrect and should read:"the 95% line value = the 50% value * 10 ^ (1.645 * StdError)"(b) the 84% line value = the 50% value x 10^StdError is correct but only because because the 84% Std Deviation Multiplier =1.0, however for any other % the general form should probably be:"the XX% line value = the 50% value * 10 ^ (XXMultiplier * StdError)"As a result I believe the examples are incorrect both in the Chapter and in the Example toward the back of the book.Ref. 1 Dowding C.H. "Construction Vibrations" (1996), pp50-55, 526-527Ref. 2 Dowding C.H. "Blast Vibration Monitoring and Control" (1985) approx. pp30-40.Along with the earlier typo on p526-527 of "Construction Vibrations", where the term (1-cc^2) should be on the TOP line. Note that Excel actually has very good function to determine the Standard Error (STEYX).Rgdsk.