Prediction of flyrock throw distance in quarries by variable selection procedures and ANFIS modelling technique

Creative Commons License

Hüdaverdi T.

ENVIRONMENTAL EARTH SCIENCES, vol.81, no.10, 2022 (SCI-Expanded) identifier identifier

  • Publication Type: Article / Article
  • Volume: 81 Issue: 10
  • Publication Date: 2022
  • Doi Number: 10.1007/s12665-022-10408-7
  • Journal Indexes: Science Citation Index Expanded (SCI-EXPANDED), Scopus, IBZ Online, PASCAL, Aerospace Database, Applied Science & Technology Source, Aqualine, Aquatic Science & Fisheries Abstracts (ASFA), BIOSIS, CAB Abstracts, Compendex, Computer & Applied Sciences, Environment Index, Geobase, INSPEC, Pollution Abstracts, Veterinary Science Database, Civil Engineering Abstracts
  • Keywords: Aggregate quarry, Flyrock, Burden, Factor analysis, ANFIS, Accuracy measures, BLAST, FRAGMENTATION, PARAMETERS, SIZE
  • Istanbul Technical University Affiliated: Yes


This study suggests application of variable reduction procedures for flyrock prediction. It was aimed to create robust and non-complex predictive models. Eleven operational blast parameters and rock mass properties were measured in an aggregate quarry. Dominant parameters for flyrock occurrence were determined by multivariate statistical methods. Two parallel ANFIS models were developed for flyrock prediction. The first ANFIS model was constructed based on the results of stepwise regression. Burden-hole diameter ratio, in-situ block size and specific charge are the input parameters of ANFIS 1. The second ANFIS model was created based on the results obtained by factor analysis. Burden-hole diameter ratio, bench height-burden ratio, number of holes and charge weight are used as input parameters for ANFIS 2. The calculated mean absolute percentage errors are lower than eight percent for the ANFIS predictions. The median absolute errors are lower than 5 m. The study also investigates alternative accuracy measures to evaluate forecasting performance. Standardized errors, normalized errors and Nash-Sutcliffe Efficiency were found to be useful for model validation. It is concluded that more than a single model can be created for a specific site. Pre-statistical analysis for variable reduction increases performance of the predictive models. Burden appeared to be a significant parameter for flyrock throw.