Drying Kinetics Prediction of Solid Waste Using Semi-Empirical and Artificial Neural Network Models

Exploring drying kinetics and fate of nutrients in thermal digestion of solid organic waste

Thermal digestion has emerged as a novel technique for the rapid treatment of solid organic waste (SOW). Dehydration mechanism and fate of nutrients during the thermal digestion of the SOW were explored. A series of experiments were carried out in a specially designed laboratory-scale dehydrator to determine its drying kinetics. The statistical analysis revealed that the diffusion model predicted the dehydration profile most accurately than other models. The effective moisture diffusivity coefficient depended on the temperature and varied from 2.81 × 10^-08 m²/s to 8.68 × 10^-08 m²/s at the tested temperature range. The activation energy required for complete dehydration was found to be 26.56 kJ/mol. The artificial neural network (ANN) model was found highly efficient (R² - 0.983) in predicting the total drying time required for attaining equilibrium moisture content. The total N decreased from 2.2% to 1.81% due to evaporation of ammonical nitrogen, while the availability of P and K was increased from 0.38% to 0.43% and 1.47% to 1.75%, respectively when the temperature was increased from 110 °C to 170 °C. The thermal dehydration technique was found effective in digesting the organics and improving the bioavailability of the nutrients, which favours for its re-utilization in agriculture.

New insights into regional differences of the predictions of municipal solid waste generation rates using artificial neural networks

As one of the most popular non-linear models, artificial neural network (ANN) has been successfully applied in the prediction of municipal solid waste (MSW). Despite its high accuracy achieved in a specific city or region, little progress is made on a larger-scale, which would be resulted from the regional difference. In this study, ANN models for MSW prediction in mainland China are developed and optimized. Besides a model aiming for all cities, regional models are developed by grouping these cities into three categories. Impact of regional difference in MSW prediction is analyzed by evaluation of model's dependence on each predictor, and comparisons made between these models. Results show that regional difference has huge impact on MSW prediction. Accuracy of MSW prediction would increase from 0.916 in R² and 59.3 in rooted mean squared error (RMSE) to 0.968/0.946/0.943 in R² and 6.4/9.7/17.6 in RMSE for southern/northern/western region after a three-region division. Models for MSW prediction in southern and northern region of mainland China share much similarity in dependence on predictors, which differs a lot from that for western region. Further cross-prediction process confirmed that models for southern or northern regions might be suitable for the MSW prediction in another, yet not apply to that in western region. Such large-scale based model can be used by cities lacking historical data for prediction of their local MSW generation, the predictive result would be helpful in MSW disposal planning and the analysis of regional difference would be helpful in establishing regional policy, especially for the three regions in mainland China.