Predicting the effect of seasonal variation on the physical composition of municipal solid waste: A case study of the city of johannesburg

Abstract

Several factors influence the physical, chemical, and thermal properties of waste at different sources. One of the major indexes to variation in the morphological composition of municipal solid waste is the season. A significant discrepancy in the composition of municipal solid waste at different seasons has been reported in the literature. However, this study explores the adaptive neuro-fuzzy inference system (ANFIS) with a fuzzy c-means (FCM) clustering technique to predict the physical content of waste in South Africa based on the varying weather parameters at different seasons. Four different models (I–IV) were developed to forecast the percentage fraction of organics, plastics, paper, and textile, respectively. The choice of these streams was because a closer look at the historical data reveals a significant variation in the percentage of these waste fractions at different seasons with little or no difference in other waste streams. The percentage composition of samples of waste collected and characterized at Marie Louise Landfill, Johannesburg, in summer 2015 and winter 2016 was used as the output variable. Weather parameters for the same period were extracted from South Africa Weather Service data and used as the input variables. M-file script was written and computed on a workstation with configurations of 64 bits, 4 GB ram Intel(R) core(TM) i3. The performance of the ANFIS models I–IV was evaluated using mean absolute deviation (MAD), root mean square error (RMSE), and mean absolute percentage error (MAPE).