Development of performance prediction model for block-cutting (S/T) machines with circular saws used in natural stone factories

Natural stones have been used worldwide as construction and decorative materials. It is one of the oldest material used in buildings for thousand of years and it holds an outstanding position among all building materials. In recent years, the demand for natural stones has highly increased because of steadily developing technologies and rapidly spreading innovative design ideas from all over the world. Natural stones are more and more preferred because of their durability and natural beauty offering a wide variety. Turkey is one of the biggest exporters of natural stones worldwide. From global perspective the country holds almost 33% of the overall natural stone reserves and in 2015, almost 50% of the country's exports of mines and stones comprised natural stones (in terms of prices). This clearly indicates that natural stones have a significant place in the economy of Turkey. For such an important industry, in-depth studies are required in order to meet the need of the growing market demands by supplying high quality products. Therefore, the machines rather used for the extraction of natural stones from quarries or for further stone processing in factories must be examined in detail. Especially, block-cutting machines, which are used in stone processing plants, have the most important position in natural stone factories. It is used for the production of stone slabs in different sizes. Block-cutting machines offer the most effective way to produce slabs with smooth surfaces which do not require any extra work for shaping. The machine can be easily operated only with one person. All of these features make the machine very cost effective compared to other machines commonly used in natural stone processing plants. For this reason, the present study chooses block-cutting machines with circular saws as the central object of research. Here the main objective is to generate different models to predict areal slab production rates of block-cutting machines. Several researchers have already published some studies about the machine performance using specific methods with some particular physical and mechanical properties of limited types of natural stones. Hence, it is aimed at selecting a wide range of natural stone types for the present study: Burdur beige, Söğüt beige, Korkuteli beige, Denizli pink marble, Kavaklıdere white marble, Uşak white marble, Bucak white travertine, Kaklık travertine and Denizli yellow travertine. Actual areal slab production rates were measured on the blocks of the selected natural stones in the processing plants. Three different branded block-cutting machines of same four footed type were used for the cutting phases. By measuring the area that the saws penetrate and recording the time consumed for this work, areal slab production rates (m2/h) were obtained for every natural stone sample. Some laboratory studies were also carried out to determine the mechanical and physical properties of the natural stones that were selected for the present work. For the laboratory tests, samples were taken from natural stone blocks to do relevant tests for measuring density, uniaxial compressive strength, Brazilian tensile strength, Cerchar abrasivity index, Schmidt hammer hardness value, Shore scleroscope hardness value. These tests were performed on procedures described by ASTM and ISRM. Thin section petrographic analyses were also performed in order to describe the stone textures. Images of the analyses are given in the thesis. The results of the laboratory tests indicate that density values range from 2.14 to 2.73 g/cm3; uniaxial compressive strength values range from 11.0 to 120.3 MPa; Brazilian tensile strength values range from 4.0 to 9.4 MPa; Cherchar abrasivity index values range from 0.4 to 3.4; Schmidt hammer hardness values range from 39.0 to 65.6; Shore scleroscope hardness values range from 26.6 to 62.3. The actual areal slab production rates measured in the processing plants range between 6.1 – 20.0 m2/h. By using the independent variables of the database created by laboratory and field tests, the present study tries to develope a model for predicting areal slab production rate using the natural stones' mechanical and/or phsyical properties. For this purpose, statistical analyses were applied for every property of natural stones one by one. As first step, simple regression analysis was performed to find the best relationship between one phsyical or mechanical property and areal slab production rate. In statistical modeling, regression analysis is a statistical process for estimating the relationships among variables. It includes many techniques for modeling and analyzing several variables, when the focus is on the relationship between a dependent variable and one or more independent variables. The strongest relationship was determined between uniaxial compressive strength and areal slab production rate as a result of analyses carried out by Excel. It was selected as the first model and estimation of areal slab production rate from uniaxial compressive strength is given in the study (Equation 7.1). After selecting the first model based on simple regression analysis, the multiple regression analysis was performed by the stepwise method using SPSS statistical software. Multiple regression is an extension of simple linear regression. It is used for predicting the value of a variable based on the value of two or more other variables. In this case, areal slab production rate is called the dependent variable. Stepwise regression procedure eliminated all the other parameters and left with only uniaxial compressive strength (UCS) because SPSS was not able to find any other entry that gives significant relationships when it put next to uniaxial compressive strength. With the generated results, analyses continued to be performed by Excel, one by one for every possible combination of variables. This resulted by two new models with multiple parameters used. The best fitting model was obtained using density and Schmidt hammer hardness value as independent variables. The analysis of variance table is cautiously investigated for testing significance of regression in multiple regressions. Significance F-value for the second model is smaller than 0.05 which indicates the 95% confidence level. F-ratio is also compared with the F-tabulated value obtained from the F distribution table. This also confirmed the significance of the regression within the 95% confidence. The t-test ratio is also used to determine the patentioal value of each variable. Tabulated t value in 95% confidence is obtained from the t distribution table and compared with the t-ratio values. Tests of the analysis indicated that the second model (Equation 7.2) is statistically meaningful with very high confidence level. The second model may be suggested for the use of predicting areal slab production rate of block-cutting machines. A nomogram based on this model (Equation 7.2) is also suggested so that it may be used for easily predicting areal slab production rate by density and Schmidt hammer hardness values. The last model developed by multiple regression analysis is obtained using the independent variables: density, Schmidt hammer hardness value and Cerchar abrasivity index. However, this model is able to succeed the tests with 90% confidence level. Actual performance data obtained from past studies was used in order to test the reliability of all three models. This database contains data from Karahallı white marble, Karahallı grey marble, Mustafa Kemal Pasa white marble and Sivaslı purple marble. Both mechanical and physical test values needed for the models were entered in and results obtained. Predicted and actual production values were placed on figures to be seen and compared in one to one scale. This test also verified the models with high coefficient of determinations for each models. The purpose of the present work is to develope significant and useful models to estimate the areal slab production rate of block-cutting machines. Generated models are satisfying enough for the sample size used in this study. This study should be improved by collecting more data from different types of natural stones and parameters to create more reliable statistical relationships and scientific generalization. It must be considered that there are several different parameters affecting the performance of block-cutting machines. Operator's experience on the machine, machine type and properties, sawing mode etc. may be considered as some of these major factors.