A Comparison Among Some Non-linear Prediction Tools on Indirect Determination of Uniaxial Compressive Strength and Modulus of Elasticity of Basalt


GÜLTEKİN N., GÖKÇEOĞLU C.

JOURNAL OF NONDESTRUCTIVE EVALUATION, vol.41, no.1, 2022 (SCI-Expanded) identifier identifier

  • Publication Type: Article / Article
  • Volume: 41 Issue: 1
  • Publication Date: 2022
  • Doi Number: 10.1007/s10921-021-00841-2
  • Journal Name: JOURNAL OF NONDESTRUCTIVE EVALUATION
  • Journal Indexes: Science Citation Index Expanded (SCI-EXPANDED), Scopus, Academic Search Premier, IBZ Online, Aerospace Database, Aquatic Science & Fisheries Abstracts (ASFA), Communication Abstracts, Compendex, INSPEC, Metadex, Civil Engineering Abstracts
  • Keywords: Basalt, Uniaxial compressive strength, Modulus of elasticity, ANFIS, ANN, Non-linear multiple regression, POINT LOAD STRENGTH, DEFORMATION MODULUS, NEURAL-NETWORKS, ENGINEERING PROPERTIES, FUZZY INFERENCE, ROCK MASS, MECHANICAL-PROPERTIES, GRANITIC-ROCKS, SEDIMENTARY, TURKEY
  • Hacettepe University Affiliated: Yes

Abstract

Basalt is among the most used rocks as aggregate, ballast, ornamental stone and for other construction purposes. Therefore, the uniaxial compressive strength (UCS) and elasticity modulus (E-i) of intact rock are required to be known for several purposes. For this reason, the purpose of the present study is to develop various non-linear prediction Model s for UCS and E-i by employing simple and non-destructive test results. Here, a dataset including 137 cases was analyzed. Each case includes unit weight, porosity, sonic velocity, E-i and UCS. The non-linear multiple regression (NLMR), adaptive-neuro fuzzy inference system (ANFIS) and artificial neural networks (ANN) were utilized as non-linear prediction algorithms. The performances of the developed Model s were assessed using various metrics such as coefficient of correlation (R-2), values account for (VAF), root mean squared error (RMSE) and a20-index. To obtain these metrics, a ranking approach was employed. When the metrics were compared, the performance of ANFIS was found slightly higher for the Model s that predict UCS. The ANN was the most successful prediction tool for the Model s predicting E-i. Also, a series of Taylor diagrams were constructed to analyze the Model performances. According to the results, the Model s using porosity and sonic velocity as input parameters for predicting UCS exhibit the highest correlation with the observed data. Regarding the E-i prediction, the Model s with three inputs have the highest performance. The results show that the investigated algorithms reveal comparable performances and the Model s developed here can be used in feasibility assessment stages.