INVESTIGATION of PHYSICAL RISK FACTORS in NATURAL STONE QUARRIES AND ESTIMATION OF PM CONCENTRATIONS WITH M5P DECISION TREE

In this study, particulate matter (PM) concentrations (TSP, PM10, PM2.5, and PM1), as well as noise and ground vibrations, were measured and evaluated during monolamella and diamond wire cutting operations in two travertine quarries in Sivas province. PM concentrations released during the monolamella process exhibited a wider distribution compared to the diamond wire cutting method, posing a higher health risk to workers. PM measurement data did not exhibit a normal distribution, supported by the significant difference between the arithmetic mean and standard deviation. Due to the wide distribution, calculated Leq(PM) values were close to the upper limit of the data range. Therefore, the use of histogram analyses in addition to arithmetic means could yield more realistic and representative results. When PM measurement data were evaluated according to the arithmetic mean, the measured values were generally below the legislative limit values. Conventional regression methods were inadequate in estimating PM concentrations; instead, the M5P decision tree algorithm was successfully implemented, with correlation coefficients exceeding 0.88 in all tests. Ground vibration measurements conducted according to the TS ISO 2631-1 standard were below the limit values, while noise levels measured according to the TS EN ISO 9612 standard were mostly acceptable.