Elsevier

Waste Management

Volume 78, August 2018, Pages 955-968
Waste Management

An artificial neural network approach for the estimation of the primary production of energy from municipal solid waste and its application to the Balkan countries

https://doi.org/10.1016/j.wasman.2018.07.012Get rights and content

Highlights

  • The model for primary production of energy from municipal solid waste was developed.

  • General regression neural network architecture was used for the model developing.

  • The model performances were improved using correlation and sensitivity analysis.

  • The obtained model has demonstrated excellent level of accuracy.

  • The model was successfully applied on the Balkan countries.

Abstract

Although the use of municipal solid waste to generate energy can decrease dependency on fossil fuels and consequently reduces greenhouse gases emissions and areas that waste occupies, in many countries municipal solid waste is not recognized as a valuable resource and possible alternative fuel. The aim of this study is to develop a model for the prediction of primary energy production from municipal solid waste in the European countries and then to apply it to the Balkan countries in order to assess their potentials in that field. For this purpose, general regression neural network architecture was applied, and correlation and sensitivity analyses were used for optimisation of the model. The data for 16 countries from the European Union and Norway for the period 2006–2015 was used for the development of the model. The model with the best performance (coefficient of determination R2 = 0.995 and the mean absolute percentage error MAPE = 7.757%) was applied to the data for the Balkan countries from 2006 to 2015. The obtained results indicate that there is a significant potential for utilization of municipal solid waste for energy production, which should lead to substantial savings of fossil fuels, primarily lignite which is the most common fossil fuel in the Balkans.

Introduction

High rate of population growth and even higher rates of urbanisation and industrialisation led to increased energy consumption worldwide and at the same time significantly increased waste generation (Ouda et al., 2016). One of the possible ways to cover some of this expanding energy demand and deal with growing waste issue is to produce energy from municipal solid waste (MSW) (Abila, 2014), using processes like anaerobic digestion (Li et al., 2018, Yano and Sakai, 2016), landfills (Lee et al., 2017, Mboowa et al., 2017), pyrolysis (Chen et al., 2015), ethanol fermentation (Wu et al., 2015) and incineration (Li et al., 2016, Lu et al., 2017).

In 2015, fossil fuels accounted for 86.02% of the global primary energy consumption (thereof oil 33.12%, coal 28.88% and natural gas 24.01%) (BP, 2017). The common use of fossil fuels to meet energy demand is nowadays associated with growing greenhouse gasses (GHG) emissions and the resulting climate change caused by global warming (Bilgen, 2014).

At the United Nations Framework Convention on Climate Changes which was held in Paris, France, in December 2015 on the 21st Conference of the Parties, 196 countries agreed to limit global warming to less than 2 °C compared to pre-industrial levels. This aim can be achieved only by reduction of the GHG emissions achieved through the use of low carbon fuels (Murphy and McDonnell, 2017).

Utilization of renewable energy sources, accompanied with implementation of energy efficiency measures, is essential in supply of low-carbon energy (International Energy Agency, Eurostat, 2004). According to International Energy Agency’s classification of renewables and waste (International Energy Agency, Eurostat, 2004) MSW is a combustible fuel, which can be renewable or non-renewable. Renewable waste is the portion of MSW which is biological (biodegradable) material and non-renewable waste is the portion of MSW which is non-biological (non-biodegradable) material (European Commission – Eurostat, 2015).

MSW can be a significant source of water, soil and air pollution and consequently represents a huge risk for public health, but also, if properly managed, MSW can become a useful raw material and a significant energy source. Using municipal solid waste to generate energy is beneficial for multiple reasons, such as decreased dependency on fossil fuels and reduction of greenhouse gases emissions from fuel combustion.

It is well known that incineration of municipal solid waste with aim to generate heat or/and electricity (waste-to-energy process), releases lower emissions of GHG, sulphur oxide, nitrogen oxide and particular matter compared to combustion of methane from landfills (Chandel et al., 2012, Kaplan et al., 2009) or coal combustion (Kaplan et al., 2009). In addition, to reduce GHG and other pollutants emissions, utilization of MSW for energy production also reduces the volume of MSW that ends up in landfills. However, in many countries, especially in developing countries, such as the majority of Balkan countries, not even existing potentials for the use of municipal solid waste in the primary energy production is known, which is crucial if higher usage of municipal solid waste as fuel is to be achieved.

Energy statistics on supply, transformation and consumption usually are the basis for any energy, economic or environmental policy-related decisions. Decision makers need to have reliable information about production, trade, stocks, and consumption of energy sources, but basic energy data are not always available and reliable due to diminishing statistical expertise caused by budget cuts and simultaneous increase of data requests, liberalisation of energy markets which led to difficult gathering of basic information and other factors (International Energy Agency, Eurostat, 2004). One of solutions to stop current erosion in data quality, coverage and timeliness is development of proper prediction models of supply (Gils et al., 2017, Sveinbjörnsson et al., 2017), transformation (Volkart et al., 2017) and consumption (Aydin, 2014) for any particular energy source.

It should be taken into account that municipal solid waste as a nonhomogeneous fuel is greatly different from conventional fossil fuels and therefore the calculation of the net calorific value (lower heating value) is very complex and can lead to substantial errors. Consequently, development of a model for the prediction of energy produced from MSW is more complex than for conventional sources of energy.

Relatively accurate predictions of energy consumption (Antanasijević et al., 2015, Ekonomou, 2010, Szoplik, 2015), as well as the predictions of various aspects of the application of MSW for energy recovery (Dong et al., 2003, Nabavi-Pelesaraei et al., 2017, Ogwueleka and Ogwueleka, 2010, Shu et al., 2006) were obtained using artificial neural networks (ANNs). In their studies, Dong et al., 2003, Shu et al., 2006, Ogwueleka and Ogwueleka, 2010 applied ANNs to predict of the lower heating value of MSW, whilst Nabavi-Pelesaraei et al. (2017) used ANN and Life Cycle Assessment (LCA) to analyse the energy consumption and environmental impact of MSW recycling system in Teheran, Iran.

ANNs are a non-linear computing system based on the concept of learning by example as the human brain does (Freeman and Skapura, 1991) and it can use various broadly available socio-economic, industrial, demographic and environmental indicators as its inputs.

This paper presents the development of a model for the prediction of annual primary production of energy from MSW (PPEMSW) at the national level for 16 European Union (EU) countries and Norway, using general regression neural network (GRNN), an improved technique in neural networks based on the nonparametric regression. GRNN relies on one-pass learning algorithm, which does not need iterative training, and therefore, unlike the other artificial neural networks architectures, GRNN has no problems related to network overtraining. One of the main advantages of GRNN in comparison with other ANN techniques is the speed of training (Kisi, 2006, Thwin and Quah, 2005). Also, it is considerably simpler than other ANN techniques and it has ability to make accurate predictions even with small or incomplete data series (Gheyas and Smith, 2011).

The main aim was to obtain accurate, but low complex model, which can be easily applied on developing countries, e.g. eleven Balkan countries, that lack or have highly uncertain data related to the amount and heating value of municipal solid waste.

Unlike previous studies that dealt with the prediction of net calorific values of MSW (Dong et al., 2003, Ogwueleka and Ogwueleka, 2010, Shu et al., 2006) and energy consumption related to the local MSW management (Nabavi-Pelesaraei et al., 2017), the topic of this research was modelling of the annual primary production of energy from MSW at the national level and application of the obtained model to the countries that mostly do not utilise MSW for the production of energy.

Section snippets

Primary production of energy from MSW (PPEMSW)

The first step of this study is the development of a model for the prediction of PPEMSW in EU countries and Norway.

Primary production of energy from all products refers to the sum of the first energy form of all products, which includes: energy contents (in case of coal, crude oil, natural gas, biomass and waste), heat (in case of nuclear, geothermal and solar thermal) and electricity (in case of solar photovoltaic, wind, hydro, tide, wave and ocean) (European Commission – Eurostat, 2015).

Selection of inputs

In order to identify mutually correlated input variables, Pearson’s correlation analysis was carried out and the results are presented in Table 4.

It is considered that variables are highly correlated when their mutual correlation coefficient is higher than 0.8 (Hamilton, 1991). For the analysed dataset, the following highly correlated variables (Table 3) were detected:

  • -

    FEC (final energy consumption) with GDP (gross domestic product), ET (energy taxes) and with NMER (number of main electricity

Conclusion

An artificial neural network (ANN) model for the prediction of annual primary production of energy from municipal solid waste (PPEMSW), based on general regression neural network (GRNN) architecture, was developed using data from European countries. During the development process, from broadly available socio-economic, industrial, demographic, waste management and energy indicators, thirteen variables had been selected as initial model inputs and then further refined using correlation and

Acknowledgements

The authors are grateful to the Ministry of Education, Science and Technological Development of the Republic of Serbia, Project No. 172007 for financial support.

References (69)

  • I.A. Gheyas et al.

    A novel neural network ensemble architecture for time series forecasting

    Neurocomputing

    (2011)
  • H.C. Gils et al.

    Integrated modelling of variable renewable energy-based power supply in Europe

    Energy

    (2017)
  • S. Kathiravale et al.

    Modeling the heating value of municipal solid waste

    Fuel

    (2003)
  • U. Lee et al.

    Evaluation of landfill gas emissions from municipal solid waste landfills for the life-cycle analysis of waste-to-energy pathways

    J. Cleaner Prod.

    (2017)
  • W. Li et al.

    Two-stage anaerobic digestion of food waste and horticultural waste in high-solid system

    Appl. Energy

    (2018)
  • X. Li et al.

    The status of municipal solid waste incineration (MSWI) in China and its clean development

    Energy Procedia

    (2016)
  • X. Lin et al.

    A simple method for predicting the lower heating value of municipal solid waste in China based on wet physical composition

    Waste Manage. (Oxford)

    (2015)
  • J.-W. Lu et al.

    Status and perspectives of municipal solid waste incineration in China: a comparison with developed regions

    Waste Manage. (Oxford)

    (2017)
  • D. Mboowa et al.

    Qualitative determination of energy potential and methane generation from municipal solid waste (MSW) in Dhanbad (India)

    Energy

    (2017)
  • F. Murphy et al.

    Investigation of the potential impact of the Paris Agreement on national mitigation policies and the risk of carbon leakage; an analysis of the Irish bioenergy industry

    Energy Policy

    (2017)
  • A. Nabavi-Pelesaraei et al.

    Modeling of energy consumption and environmental life cycle assessment for incineration and landfill systems of municipal solid waste management – A case study in Tehran Metropolis of Iran

    J. Clean. Product.

    (2017)
  • O.K.M. Ouda et al.

    Waste to energy potential: a case study of Saudi Arabia

    Renew. Sustain. Energy Rev.

    (2016)
  • D. Sveinbjörnsson et al.

    Energy supply modelling of a low-CO2 emitting energy system: case study of a Danish municipality

    Appl. Energy

    (2017)
  • J. Szoplik

    Forecasting of natural gas consumption with artificial neural networks

    Energy

    (2015)
  • M.M.T. Thwin et al.

    Application of neural networks for software quality prediction using object-oriented metrics

    J. Syst. Software

    (2005)
  • K. Volkart et al.

    Multi-criteria decision analysis of energy system transformation pathways: a case study for Switzerland

    Energy Policy

    (2017)
  • S. Walczak et al.

    Heuristic principles for the design of artificial neural networks

    Inf. Softw. Technol.

    (1999)
  • C. Wu et al.

    Effect of ethanol pre-fermentation and inoculum-to-substrate ratio on methane yield from food waste and distillers’ grains

    Appl. Energy

    (2015)
  • H. You et al.

    Comparison of ANN (MLP), ANFIS, SVM, and RF models for the online classification of heating value of burning municipal solid waste in circulating fluidized bed incinerators

    Waste Manage. (Oxford)

    (2017)
  • BP, 2017. BP Statistical Review of World Energy 2017. London, UK....
  • M.K. Chandel et al.

    The potential of waste-to-energy in reducing GHG emissions

    Carbon Manage.

    (2012)
  • C. Chen et al.

    A new accuracy measure based on bounded relative error for time series forecasting

    PLoS ONE

    (2017)
  • M. Dehghani et al.

    Uncertainty analysis of streamflow drought forecast using artificial neural networks and Monte-Carlo simulation

    Int. J. Climatol.

    (2014)
  • P. Douglas et al.

    Estimating particulate exposure from modern municipal waste incinerators in Great Britain

    Environ. Sci. Technol.

    (2017)
  • Cited by (25)

    • Comprehensive evaluation of municipal solid waste power generation and carbon emission potential in Tianjin based on Grey Relation Analysis and Long Short Term Memory

      2022, Process Safety and Environmental Protection
      Citation Excerpt :

      In many countries municipal solid waste (MSW) plays an important role as an energy source, burning waste with a very high conversion efficiency. There is a great potential to produce energy from municipal solid waste, which will save a lot of fossil fuels (Adamović et al., 2018), but municipal solid waste consists mainly of degradable materials, which can lead to a significant increase in greenhouse gas emissions and pollute the air (Yaman et al., 2020). Kumar and Madlener (2016) showed that fossil-fueled power plants produce significant CO2 emissions and that power generation through waste-to-energy not only mitigates energy problems but also achieves GHG emission reduction targets.

    • Türkiye's energy projection for 2050

      2022, Renewable Energy Focus
      Citation Excerpt :

      In another study, an artificial neural network model was used to estimate the energy required for hot asphalt production [124]. In another study where energy production from domestic solid wastes was estimated, artificial neural networks were preferred [125]. In another study that can be an example of application areas, artificial neural networks were used for load estimation [126].

    • Analysis of input set characteristics and variances on k-fold cross validation for a Recurrent Neural Network model on waste disposal rate estimation

      2022, Journal of Environmental Management
      Citation Excerpt :

      Input data screening and pre-processing also contributed to improved predictive model performance (Younes et al., 2015; Kannangara et al., 2018). Collinearity among input parameters may confuse ANN model training algorithms and should be avoided (Adamovic et al., 2018). Vu et al. (2019b) applied lagged inputs for ANN time series analysis and obtained better yard waste estimates.

    • Machine learning models for prediction the Higher Heating Value (HHV) of Municipal Solid Waste (MSW) for waste-to-energy evaluation

      2022, Case Studies in Thermal Engineering
      Citation Excerpt :

      The researchers presented three types of principle models to predict the HV for MSW (Fig. 1). Artificial Neural Networks (ANNs) are one of the modern and computational methods that are used for machine learning, display knowledge, and ultimately apply the knowledge gained to predict output responses from complex systems [28]. ANNs are used as a powerful tool for modeling where the relationships between inputs and outputs are complex and nonlinear [29].

    View all citing articles on Scopus
    View full text