Model predviđanja potresa od miniranja na površinskim kopovima primenom simulacione metode Monte Karlo i fazi logike

Abstract

Modeliranje predviđanja potresa tla od miniranja primenom fazi logike i metode Monte Karlo zasnovano je na podacima dobijenim terenskim merenjima potresa uzrokovanih miniranjem. Promenljive modela definisane su analizom parametara bušenja i miniranja minskih serija, kao i velosigrama izmerenih potresa od miniranja. Iskorišćene su prednosti fazi logike u implementaciji mnogo većeg broja ulaznih i izlaznih veličina za modeliranje u odnosu na postojeće empirijske modele. Na osnovu identičnih podataka, dobijen je regresioni model predviđanja potresa, da bi se izvršilo testiranje novog modela. U cilju povećanja pouzdanosti modela predviđanja, izvršena je analiza izmerenih karakteristika potresa prema postojećim standardima, izračunati su spektri odziva na osnovu velosigrama, izvršeno je merenje i analiza pomeraja pukotina usled potresa od miniranja i dnevnih promena temperature i relativne vlažnosti vazduha na jednom od stambenih objekata. Tokom istraživanja, registrovan je zemljotres 03. novembra 2010. god. sa epicentrom u Kraljevu, što je omogućilo analizu velosigrama potresa uzrokovanih zemljotresom. Testiranjem modela utvrđena je visoka pouzdanost predviđanja u odnosu na postojeći regresioni model. Uzimajući u obzir kompleksnost problema predviđanja potresa od miniranja, kao i određen stepen neizvesnosti, povezan sa karakteristikama stenskog materijala, primenjen je Monte Karlo model za određivanje učestalosti pojavljivanja kritičnih vrednosti pokazatelja potresa, za konkretne slučajeve miniranja. Metodologija modeliranja može se primeniti pri predviđanju potresa uzrokovanih miniranjem na svim površinskim kopovima.

Modeling the prediction of blast-induced ground vibrations using fuzzy logic and Monte Carlo method is based on data obtained by field measurements of ground vibrations induced by blasting. The variables of the model are defined by the analysis of drilling and blasting parameters of blasting rounds as well as vibration time histories of measured vibrations. The advantages of fuzzy logic in the implementation of a much larger number of input and output variables for modeling compared to the existing empirical models are apllied. The regression model for the prediction of vibrations is obtained on the basis of identical data for testing the new model. In order to increase the reliability of prediction model, the analysis of measured vibration characteristics according to existing standards, the response spectra on the basis of vibration time histories as well as the measurement and analysis of crack displacement due to vibration from blasting and daily changes in temperature and relative humidity in one of the residential buildings have been carried out. During the research, the earthquake was registered on November 3rd 2010. with its epicenter in Kraljevo, which enabled the analysis of vibration time histories induced by earthquake. Testing the model revealed the high reliability of prediction in relation to existing regression model. Taking into account the complexity of prediction of blast-induced vibrations as well as a certain degree of uncertainty associated with the characteristics of rock mass, the Monte Carlo model has been applied to determine the frequency of occurrence of ground vibration critical values for the specific blasting cases. The methodology of modeling can be applied for the prediction of vibration from blasting in all open pit mines.