Development of statistical models for predicting leachate parameters from simulated landfills.

摘要

Leachate generation and management is recognized as one of the greatest problems associated with environmentally sound operation of landfills, as leachate can cause major pollution problems to surrounding soil, ground water, and surface waters. There are many landfills, especially in developing parts of the world like India, Bangladesh, Africa, and Latin America, where open dump systems are used for final disposal of solid waste rather than engineered landfills. In the near future, regulations in developing countries will likely require installation of liner systems, leachate collection systems, and treatment operations. A major requirement for successful leachate treatment is quantifying its typical composition. Models for predicting leachate parameters would be useful in designing leachate treatment systems for new landfills in developing countries.;Even in the developed countries, it is quite possible that the frequency of monitoring various leachate quality parameters will increase, along with the number of parameters to be measured. In the absence of gas composition data, leachate composition data provides important information about different phases of waste decomposition. However, the analyses of these types of leachate quality parameters are very expensive and time consuming. Models for estimating leachate parameters would be useful in reducing leachate parameter modeling frequency, and thus reducing costs.;Previous studies have shown that waste composition, rainfall and temperature of a landfill significantly influence leachate composition. Most studies have focused on leachate quality data from a single or few regional-specific landfills considering general waste composition, temperature, and moisture content. The few attempts to develop regression models to predict leachate characteristics using statistical techniques have focused on a single or few regional landfills.;The goal of this research was to develop Multivariate Adaptive Regression Splines (MARS) equations for predicting leachate parameters: biochemical oxygen demand (BOD), chemical oxygen demand (COD), alkalinity, pH, conductivity, total dissolved solids (TDS), total suspended solids (TSS), volatile suspended solid (VSS), ammonia-nitrogen (NH3-N), and chloride (Cl -), with basic information on temperature, rainfall, waste composition, and time. A statistical experimental design was developed using incomplete block design to determine leachate quality parameters, where the waste composition served as a blocking variable and combinations of temperature and rainfall were the predictor variables. Leachate characteristics were measured from total 27--16L size lab-scale reactors with varying waste compositions (0-100%); rainfall rates of 2, 6, and 12 mm/day; and temperatures of 70, 85, and 100 °F. Waste components considered for the study were major biodegradable wastes, food, paper, yard, textile, as well as inorganic waste.;Initially many attempts were made on total alkalinity (as CaCO3) to develop a multiple linear regression (MLR) model equation. However, it was concluded that basic MLR method was insufficient to analyze lab-scale leachate data due to nonlinearity between response and predictor variables. Therefore, a more sophisticated modeling approach of regression splines was used for the model development of all leachate parameters. Multivariate Adaptive Regression Splines (MARS) equations were developed using Salford Predictive Modeler Builder, Version 6.6, which incorporated predictor variables (temperature, rainfall, and waste components) in predicting leachate parameters.;Overall, reactors at 70 °F had lower concentrations of almost all leachate parameters. Also, reactors with 100% food waste showed the highest concentrations for all leachate parameters. Time or Rain was the most important variable in the MARS model equations developed for the leachate parameters except NH3-N, where Food variable was given the highest importance. Paper vs. Rain 3D-interaction plots showed decreased concentrations of total alkalinity and TDS with increasing rainfall and paper percentage. Leachate Volume vs. Time 3D-interaction plots showed decreased concentrations for total alkalinity, TDS, and conductivity with increasing time and leachate volume. Furthermore, Temperature vs. Rain, Paper vs. Rain, Food vs. Temperature 3D-interaction plots showed similar trends for TSS and VSS. The total alkalinity model had the highest adjusted R2 value of 0.961; conductivity was second with an adjusted R2 of 0.958. Also, the model equations for COD, TDS and BOD had high adjusted R2 values of 0.950, 0.947, and 0.923, respectively. It was observed that 85 °F was the optimum temperature based on interaction plots for BOD, VSS, and NH3-N.