Stochastic study of fugitive dust concentration from stone-processing plants at Nah Pra Laan using monte-carlo simulation of ISCST3 model

Abstract

In this study, Monte-Carlo technique and the ISCST3 Guassian air dispersion model are employed to simulate the 24-hour average concentration of ambient PM10 released from 48 stone-processing plants in Na Pra Laan and vicinity in Saraburi Province. The annual trend of the 24-hour average concentration and their statistical values at 5 receptors are investigated under the assumption of an 80% across-the-board reduction of the plant emission rates after the introduction of dust control systems. Two types of inputs with uncertainty are investigated, meteorological data and source emission. For uncertain meteorological inputs, past meteorological data are first statistically analyzed to find out the proper distribution functions and weighting parameters (α). Next random values of stochastic variables representing future meteorological data are generated. It is found that the gamma distribution is the best-fit distribution of all meteorological inputs (wind speed, wind direction, ambient temperature, mixing height, and cloudiness) and a proper value of α is 0.5. For uncertain emission inputs, values of normal random emission factor based on U.S. EPA recommended values are generated using specified mean and standard deviation of 0.05275 kg/ton and 0.005275 kg/ton, respectively. According to the 50 Monte-Carlo simulation results, it is found that the predominant southeastern wind direction has the most influence on the predicted concentration of ambient PM10 in the study area compared with other wind directions. As a result, the high probability of the 24-hr average PM10 value exceeding the ambient standard value (120 μg/m3) cannot be neglected particularly at the receptors located in the northwestern region. Besides, it is found that the magnitude of uncertainty in the 24-hour PM10 in the case of random meteorological inputs is larger than that in the case of random emission rate inputs. Therefore it may be concluded that uncertainty in meteorological inputs has more significant effect on PM10 uncertainty than in the emission rate inputs. Furthermore, a test of the effect of autocorrelation in the wind speed and direction on the transient behavior of the PM10 by varying weighting parameters (α) shows that the degree of autocorrelation in the wind direction has more effect on the behavior of the PM10 concentration than that of the wind speed because of the predominance of the calm wind condition in the study area (wind speed < 2.0 m/s).