Prediction of Municipal Solid Waste Generation and its corresponding Lower Heating Value in the EU : A Case Study of Romania

Sammanfattning

In order to apply sufficient waste management systems and processes that can utilize waste for potential energy recovery and material extraction it is important to be able to predict future waste generation. By knowing how much and what types of waste that will be generated in the future countries are better prepared for collecting waste and using is for their benefit. Collected waste can be incinerated and in the process release heat which can be used for heating or electricity production. It therefore also becomes valuable to know how much energy that can be utilized from waste. Thus having a forecasting model that can determine the amount of future waste and the energy that can be utilized from it is wanted for countries. The study thus consists of the development of such a model with the means of being applied for an EU member country. To exemplify this the developed model was applied in a case study for Romania. Predicting the countries waste generation and corresponding potential energy recovery for the years 2025 and 2035. In this research an artificial neural network (ANN) model was used to forecast generated quantities of different municipal solid waste (MSW) categories. The neural network was compared with a developed multivariate linear regression (MLR) model. Five performance assessment metrics wereused for this comparison; Pearson’s correlation coefficient (R), Coefficient of determination (R2), Root Mean Square Error (RMSE), Mean Absolute Percentage Error (MAP E) and Median Absolute Percentage Error (M dAP E). The predicted MSW composition combined with two linear models were used to forecast the lower heating value (LHV) of the generated MSW. The developed ANN model (R = 0.8297, R2 = 0.6332, RMSE = 1.7223, MAP E = 60.3269, M dAP E = 38.2395) outperformed the MLR model (R = 0.8009, R2 = 0.5375, RMSE = 1.9340,MAP E = 97.9781, M dAP E = 51.4378) in predicting waste generation by composition and in using the predicted MSW composition to estimate the amount of energy that can be recovered from the waste. During the case study of Romania it was estimated that the country would generate 14.729M tMSW in 2025 and 11.3222M t in 2035. Furthermore the models predicted that the corresponding LHV was around 14030 - 14525 kJ/kg for 2025 and around 15191 - 15465 kJ/kg for 2035. This corresponds to 58593 - 60660 T J and 65249 - 66426 T J for 2025 and 2035 respectively. This increase in LHV, despite the decrease in total MSW generation, is mainly due to an increase in paper and cardboard waste generation which has a larger positive impact on LHV than other waste categories.