Descripción
|
|
---|---|
Eight distributions, Weibull, Swebrec, Grady and Gilvarry, and their bi-component varieties, are fitted to 28 sets of screened fragment data from both blasted and crushed rock; such data cover at least three orders of magnitude in size and in some cases more than five. Several fitting criteria (OLS and weighed sum of squares) are used for the non-linear fitting. The merit of each distribution and fitting criterion is discussed in terms of the logarithmic errors in the prediction of sizes. Root mean squared log errors are obtained for different size ranges, arbitrarily defined as coarse, central, fine and very fine, and the behavior of the distribution functions and weighing factors is assessed. Log errors can be less than log 1.25 with most functions in the central range (passing 20-80 %), while only the bi-components keep the error at that level in the fine range (passing < 20 %). Errors in the coarse range (> 80 %) are higher than in the central; bi-components keep the log error around log 1.25, except in the 1/p2-weighed fits, and for the single components it is between log 1.25 and log 1.50, increasing from the OLS to the 1/p2 fits. The very fine range (passing < 2 %) is very difficult to match; only the minimum relative error criterion (1/p2 weights) does acceptably well, though this criterion gives in turn the worst results in the coarse range. | |
Internacional
|
Si |
Nombre congreso
|
9th International Symposium in Rock Fragmentation by Blasting- Fragblast 9 |
Tipo de participación
|
960 |
Lugar del congreso
|
Granada |
Revisores
|
Si |
ISBN o ISSN
|
978-0-415-48296-7 |
DOI
|
|
Fecha inicio congreso
|
13/09/2009 |
Fecha fin congreso
|
17/09/2009 |
Desde la página
|
239 |
Hasta la página
|
248 |
Título de las actas
|
Rock Fragmentation by Blasting: Proceedings of the 9th International Symposiumon Rock Fragmentation by Blasting- Fragblast 9 |