[go: up one dir, main page]
Next Article in Journal
Enhancing Seismic Resilience of Bridge Infrastructure Using Bayesian Belief Network Approach
Previous Article in Journal
Strategic Assessment of E-Learning Platform Selection: A Multi-Criteria Decision Analysis for Students in India
 
 
Search for Articles:
Title / Keyword
Author / Affiliation / Email
Journal
Article Type
 
 
Advanced Search
 
Section
Special Issue
Volume
Issue
Number
Page
 
Journals
Engineering Proceedings
Volume 76
Issue 1
10.3390/engproc2024076019
engproc-logo
Submit to this Journal
► Article Menu

Article Menu

share Share announcement Help format_quote Cite question_answer Discuss in SciProfiles thumb_up
...
Endorse textsms
...
Comment
first_page
settings
Order Article Reprints
Font Type:
Arial Georgia Verdana
Font Size:
Aa Aa Aa
Line Spacing:
Column Width:
Background:
Proceeding Paper

CO2 Emissions Projections of the North American Cement Industry †

by
Ángel Francisco Galaviz Román
,
Seyedmehdi Mirmohammadsadeghi
and
Golam Kabir
*
Industrial Systems Engineering, University of Regina, 3737 Wascana Parkway, Regina, SK S4S 0A2, Canada
*
Author to whom correspondence should be addressed.
†
Presented at the 1st International Conference on Industrial, Manufacturing, and Process Engineering (ICIMP-2024), Regina, Canada, 27–29 June 2024.
Eng. Proc. 2024, 76(1), 19; https://doi.org/10.3390/engproc2024076019
Published: 17 October 2024
Browse Figures
Figure 1
<p>Machine learning fit plot of CO<sub>2</sub> emissions forecasting.</p> "> Figure 2
<p>Integrated GRG model fit plot for CO<sub>2</sub> emissions forecasting.</p> ">
Versions Notes

Abstract

:
Forecasting carbon dioxide (CO2) emissions has become a relevant issue. International organizations have emphasized the necessity of generating a plan to gradually reduce the concentrations of this pollutant to combat climate change. Cement industries represent one of the key sectors expected to solve this problematic. The objective of this study is to predict CO2 emissions for North American cement industries. To achieve this, a multi-objective mathematical model is developed, integrating various machine learning algorithms. The results demonstrate a considerable improvement in accuracy metrics, with a 48.13% reduction in Mean Absolute Error achieved using the Generalized Reduced Gradient method (GRG). The forecasts reveal an increment in emissions from about 0.58 MtCO2 every year between 2020 and 2050. The proposed framework can help decision makers and policy makers focus on the technical and logistics requirements to meet net-zero emissions targets.

1. Introduction

Cement has experienced increasing demand over the last couple of decades. This trend is expected to continue with an increasing in production of around 40% in the upcoming years, as cement remains indispensable in the daily lives of civilizations [1]. Consequently, emissions related to the production process will increase accordingly. Different literature studies estimate a total contribution equivalent to 5% to 8% of anthropogenic CO2 emissions associated with cement industries worldwide [2,3]. It is estimated that the total emission is around 564 kg of CO2 per ton of cement produced [4].
To achieve the 2 °C scenario outlined in the Paris Agreement, a common goal of reducing emissions through a global effort has been established, with a first objective of a 25% reduction by 2030 [5]. The United States, Canada, and Mexico represent key players in this scenario. The first two are among the top 10 global emitters, with a total of 6001 MtCO2 and 736 MtCO2, respectively, while the third one hosts the headquarters of the fourth-largest cement industry worldwide [6].
The United States stands out as the main contributor in the cement-manufacturing sector among the mentioned countries. The U.S. was the world’s fifth-largest contributor in 2021, generating 41.3 million metric tons of carbon dioxide [7]. Canada’s cement industry reports a lower contribution to global emissions. In 2020, cement manufacturing accounted for just 9.7 MtCO2 or 1.4% of Canada’s total emissions. However, it is pertinent to mention that the Cement Association of Canada anticipates a constant growth of 1% per year in concrete production to support the investment in new infrastructure [8].
The proposed study aims to provide valuable information through the generation of reliable predictions for the North American cement industry, which is one of the crucial manufacturing sectors of modern society. The overall goals of this study are the following:
  • To generate an accurate CO2 emissions forecast for cement industries located in Mexico, the United States, and Canada based on historical data collected between 1990 and 2019;
  • To enhance the performance of machine learning algorithms through the employment of an integrated mathematical model capable of improving accuracy metrics obtained from individual computations.
The results of the proposed study may be considered crucial information for decision makers and policy makers to develop adequate strategies to achieve net-zero emission targets by starting from a more reliable prediction. Some of the contributions of this research are as follows:
  • A novel multi-objective mathematical model is developed based on a Generalized Reduced Gradient (GRG) algorithm integrated by machine learning models capable of achieving accurate results with a small dataset;
  • To the best of the authors’ knowledge, this study is one of the first research efforts focused on developing a reliable forecasting model to predict CO2 emissions from cement industries in North America.
The remainder of this paper is organized as follows. Section 2 presents a review of the available literature focused on the development of advanced methods to predict data behaviors such as CO2 emissions from cement industries. The methodology followed in this study is presented in Section 3. Finally, in Section 4, conclusions, limitations, and future research directions are highlighted.

2. Literature Review

Studies have focused on generating forecasting models to predict CO2 emissions using algorithms capable of obtaining reliable results with small datasets. Recently, researchers have shown interest in predicting total carbon emissions from various countries by employing different methodologies that integrate and enhance the results obtained from individual machine learning algorithms. An example of this is the improvement in the accuracy of results from Fast Learning Network and Extreme Learning Machine algorithms through the implementation of Chicken Swarm Optimization. Diverse factors, such as energy-related carbon emissions, technology carbon emissions from cement production, and forest carbon sinks, were calculated based on data from 1995 to 2019. The conclusions demonstrate that the CSO-FLN model obtained better results compared to the other algorithms tested, achieving lower values on three error indicators (MAE, MAPE, and RMSE) [9].
The literature review conducted has revealed several relevant gaps, which are summarized as follows:
  • Available studies have not evaluated the impact of implementing technologies developed by the scientific community in recent years. This information is crucial for stakeholders, as such projects will play a key role in meeting emission targets in the upcoming years;
  • Almost every literature review has focused on analyzing the current status in Eastern countries, thereby overlooking the high impact that the American continent represents in achieving the goals set by the Intergovernmental Panel on Climate Change (IPCC).
The proposed research aims to address these limitations by developing an enhanced forecasting model that elucidates the impact of pathways recommended by international organizations and the research community for reducing emissions. Additionally, this study specifically concentrates on the contribution of cement industries in Western countries, namely the United States, Canada, and Mexico.

3. Framework Development

The proposed framework integrates various machine learning algorithms and employs the Generalized Reduced Gradient Optimization algorithm to predict CO2 emissions from cement industries in North America. Firstly, emissions data from cement industries in the United States, Canada, and Mexico between 1990 and 2019 were collected. Subsequently, using the gathered data, SARIMA, AR, and Regression ARMA were individually employed and tested. Four different error indicators (MAE, RMSE, NRMSE, and MAPE) were utilized for evaluation. Following this, the forecasts obtained from each algorithm in every year were used as inputs for the development of the Generalized Reduced Gradient (GRG) model. This MOMM determined optimal coefficient values for each algorithm considered in its constitution.

3.1. Data Collection and Data Pretreatment

Initially, total CO2 emissions from each country spanning the years 1990 to 2019 were compiled from statistics published by the World Resources Institute [6]. Subsequently, it was necessary to calculate the corresponding emissions from the cement-manufacturing sector. According to the International Energy Agency [10], this sector contributes 3% to the total anthropogenic CO2 emissions. To ascertain the specific contribution from the cement industry in North America during the same period, it was crucial to compute 3% of the global CO2 emissions for these countries.

3.2. Machine Learning Algorithms Employed

This research employs various methods such as AR, SARIMA, and Regression ARMA, known for their ability to deliver a robust performance with small datasets. To train this set of algorithms, 80% of the available data (covering the period from 1990 to 2013) was utilized, while the remaining 20% (spanning 2014 to 2019) was reserved for testing purposes.

3.2.1. Seasonal Autoregressive Integrated Moving Average Algorithm (SARIMA)

SARIMA is an extension of the ARMA model; however, it is employed when the data analyzed is not stationary across the years and presents different mean and other statistics for a given season or period of time. This algorithm is developed by following three major phases: identification, estimation, and diagnostic check [11]. The results obtained demonstrated a lower Bayesian Information Criterion, meaning that including seasonality in the forecasts improved the accuracy of the model developed. Equation (1) describes the formula employed in this case study.
1 − ∅ 1 L − ⋯ − ∅ 9 L 9 ( 1 − ∅ 2 L 2 ) ( 1 − L ) y t = C + ( 1 + θ 1 L ) ( 1 + θ 2 L 2 )   ε t

3.2.2. Autoregressive Algorithm (AR)

The AR model consists of generating a forecast based on previous values from the same time series, where the order of autoregression is the number of previous periods considered to predict the value at the present time [12]. Considering input data, the optimal number of lags that allowed the obtaining of the lowest Bayesian Information Criterion was six; this means that in order to the predict present year emissions, it is necessary to consider values from the preceding six years. Equation (2) represents the formula used in this study; this is a six-order autoregression model.
1 − ∅ 1 L − ⋯ − ∅ 6 L 6 y t = C + ε t

3.2.3. Autoregressive Moving Average Model (ARMA)

The ARMA model shows a good performance on stationary time series and presents a good prediction capability; the constitution of this model considers not only the residuals obtained but also the parameters used and the number of available observations [13]. Furthermore, the moving average means that the estimation for the current year will also have a smooth impact on the results expected for further years; considering the input data, it was perceived to offer the best fitness by including the regressive emissions from the previous eight years and a moving average constituted by the preceding two years. Equation (3) depicts the formula constructed in this case.
1 − ∅ 1 L − ⋯ − ∅ 8 L 8 μ t = 1 + θ 1 L + θ 2 L 2 ε t

3.3. Multi-Objective Mathematical Model Development

A Generalized Reduced Gradient nonlinear algorithm was implemented with the aim of integrating results obtained by the multiple machine learning algorithms previously mentioned and reducing the error indicator values individually obtained. The developed MOMM is constituted as follows:
Objective   1 :   Mean   Absolute   Error   ( MAE ) = 1 N ∑ i = 1 N | T i − T |
Objective   2 :   Minimize   Root   Mean   Square   Error   ( RMSE ) = ∑ i = 1 N ( T i − T ) 2 N
Objective   3 :   Minimize   Normalized   Root   Mean   Square   Error   ( NRMSE ) = ∑ i = 1 N ( T i − T ) 2 N   T m a x − T m i n
s . t .   T i = ∑ X = 1 X ρ x T x i + ρ o ,   0 ≤ ρ x ≤ 1
where Equations (4)–(6) represent the objective functions for this case study, which are minimizing the results from the error indicators used to measure the individual performance from the developed machine learning algorithms. Equation (7) depicts the calculation followed to determine the value for T i in every year of the testing scenario; ρ x is the optimal coefficient determined for the X t h machine learning algorithm; and T x is the forecast obtained for year i from the algorithm X. Finally, ρ o represents the intercept value.
Once optimum values are obtained for the testing years, Equation (8) can be applied for predicting emissions in the K t h year, where T k is the forecasted value for the K t h point and T x K is the prediction obtained of the K t h term by the X t h machine learning algorithm.
T k = ∑ X = 1 j = X ρ x T x K + ρ o

4. Results

The machine learning algorithms were tested using MATLAB, while Python was employed to solve the multi-objective mathematical model through the development of the GRG algorithm. The results of the forecasts and accuracy comparisons between models are presented in this section.

4.1. Prediction Results with Individual Machine Learning Algorithms

Table 1 depicts the error values obtained for the testing data from each of the implemented algorithms. SARIMA demonstrates the highest predicting accuracy for CO2 emissions in the North American cement industry, followed by AR and ARMA, which outline the next level of reliability in their estimations. Figure 1 displays a graphical representation of the forecasts obtained.

4.2. Prediction Results with Multi-Objective Mathematical Model

Forecasts obtained from the SARIMA, ARMA, and AR algorithms between the years 2014 and 2050 (including testing data) were used as inputs for the development of the GRG model. Considering the developed objective functions and following the methodology presented in Section 3.3, the Generalized Reduced Gradient (GRG) model returned optimal coefficients for each of the employed algorithms (Figure 2 and Table 2), generating an integrated model that allowed the enhancement of the individually obtained predictions.
Error indicator values obtained from the generated model are depicted in Table 3. A reduction of 48.13% in Mean Absolute Error and 33.99% in Root Mean Square Error can be observed in comparison to SARIMA, which is the best individual machine learning algorithm obtained.
Table A1 outline the forecasts obtained between the years 2020 and 2050. The expected emissions will be based on forecasts obtained through the utilization of the GRG model. Thus, an increment of around 7.6044% in 2050 is foreseen compared to the year 2020. This is equivalent to a constant yearly rise of 0.5866 MtCO2, resulting in total emissions equal to 251.0848 MtCO2 for 2050.

5. Conclusions

In addition to energy consumption and transportation, cement production stands out as one of the main sources of carbon dioxide emissions worldwide. This industry, along with iron and steel, plays a crucial role in meeting net-zero emission targets, as the demand for these materials will increase in tandem with populations and urbanization. This research develops a data-based forecast to predict CO2 emissions from the North American cement industry by formulating an optimized model capable of enhancing predictions. As one of the conclusions of this study, a clear improvement in forecasts was observed through the utilization of a multi-objective mathematical model. In fact, an improvement of 33.99% in RMSE was computed in comparison with SARIMA, the machine learning algorithm that demonstrated the best individual performance. The proposed model estimates that CO2 emissions will reach 251.08 Mt by 2050, representing an increase in emissions equivalent to 0.58 MtCO2 every year starting from 2020.
The findings and information presented in this study could be used as relevant input for precise effort and resources management while developing strategies addressing CO2 emissions issues. It must be noticed that this research has some limitations that should be assessed. First, some parameters considered for implementing machine learning algorithms were chosen manually; for example, the degree order of every model was obtained based on optimum results from the Bayesian Information Criterion. Secondly, the available literature has explored different multi-objective mathematical models such as PSO, CSO, or WOA for the integration of individual algorithms; these models may outperform the results obtained by the GRG model.
For further research, the methodology followed in this study can be employed for predicting CO2 emissions and evaluating the future situation at specific site locations, considering factors such as total cement production and energy requirements for a particular cement production facility. Furthermore, additional machine learning algorithms can be analyzed and integrated into the MOMM, as they can improve the predictions obtained and provide more reliable forecasts.

Author Contributions

Á.F.G.R.: Conceptualization, Methodology, Software, Formal analysis, Investigation, Writing—original draft. G.K.: Conceptualization, Methodology, Software, Investigation, Supervision, Writing—review and editing. S.M.: Methodology, Writing—review and editing. All authors have read and agreed to the published version of the manuscript.

Funding

The third author acknowledges the financial support through President’s Research Seed Grant of the University of Regina.

Institutional Review Board Statement

Not applicable.

Informed Consent Statement

Not applicable.

Data Availability Statement

Data supporting the reported results can be found at https://www.wri.org/insights/4-charts-explain-greenhouse-gas-emissions-countries-and-sectors (accessed on 7 August 2023).

Acknowledgments

The authors acknowledge the financial support through the Mitacs Fellowship program as well as University of Regina and Faculty of Graduate Studies Research funding.

Conflicts of Interest

The authors declare no conflicts of interest.

Appendix A

Table A1. CO2 emissions forecasting for North America cement industries.
Table A1. CO2 emissions forecasting for North America cement industries.
YearARSARIMAARMAGRG- MOMMYearARSARIMAARMAGRG- MOMM
2020227.03246.83266.79238.492035239.12255.83277.67246.30
2021228.56243.58272.86240.822036238.05252.83269.09243.31
2022231.02247.21280.07243.852037236.92262.49281.04246.98
2023232.29251.59274.37242.622038235.79251.09287.97248.56
2024234.30259.61276.37244.142039234.76253.08288.53248.46
2025236.43259.34285.83247.702040233.84254.55292.63249.53
2026237.79256.25272.30243.852041233.08257.07301.40252.13
2027239.29271.67274.36245.422042232.51253.66306.55253.37
2028240.52262.87277.86246.682043232.16253.84299.01250.93
2029241.15262.55269.44244.262044232.00266.82311.13255.11
2030241.63269.40267.09244.012045232.06256.47311.39254.82
2031241.74264.75268.27244.342046232.30262.92301.28251.92
2032241.41263.21271.35245.092047232.70271.11308.73254.61
2033240.90258.81261.87241.992048233.22267.45305.98253.70
2034240.12266.56269.64244.372049233.83271.24300.55252.27

References

  1. Xu, C.; Gong, Y.; Yan, G. Research on Cement Demand Forecast and Low Carbon Development Strategy in Shandong Province. Atmosphere 2023, 14, 267. [Google Scholar] [CrossRef]
  2. Amran, M.; Makul, N.; Fediuk, R.; Lee, Y.H.; Vatin, N.I.; Lee, Y.Y.; Mohammed, K. Global carbon recoverability experiences from the cement industry. Case Stud. Constr. Mater. 2022, 17, e01439. [Google Scholar] [CrossRef]
  3. Wei, L.; Shubin, G. Prospective on energy related carbon emissions peak integrating optimized intelligent algorithm with dry process technique application for China’s cement industry. Energy 2018, 165, 33–54. [Google Scholar] [CrossRef]
  4. Costa, F.; Ribeiro, D. Reduction in CO2 emissions during production of cement, with partial replacement of traditional raw materials by civil construction waste. J. Clean. Prod. 2020, 276, 123302. [Google Scholar] [CrossRef]
  5. Global Cement and Concrete Association. Concrete Future. 2021. Available online: https://gccassociation.org/concretefuture/wp-content/uploads/2021/10/GCCA-Concrete-Future-Roadmap-Document-AW.pdf (accessed on 15 October 2023).
  6. Ge, M.; Friedrich, J.; Vigna, L. 4 Charts Explain Greenhouse Gas Emissions by Countries and Sectors. World Resources Institute. 2020. Available online: https://www.wri.org/insights/4-charts-explain-greenhouse-gas-emissions-countries-and-sectors (accessed on 7 October 2023).
  7. Tiseo, I. Greenhouse Gas Emissions from Cement Production in the United States from 1990 to 2021. Statista. 2023. Available online: https://www.statista.com/statistics/451804/green-house-gasemissions-in-united-states-from-cementproduction/#:~:text=Cement%20produced%20in%20the%20United,percent%20compared%20with%202005%20levels (accessed on 14 September 2023).
  8. Cement Association of Canada. Canada’s Cement and Concrete Industry Action Plan to Net-Zero, Concrete Zero. 2023. Available online: https://cement.ca/wp-content/uploads/2023/05/ConcreteZero-Report-FINAL-reduced.pdf (accessed on 14 August 2023).
  9. Feng, R.; Dinghong, L. Carbon emission forecasting and scenario analysis in Guangdong Province based on optimized Fast Learning Network. J. Clean. Prod. 2021, 317, 128408. [Google Scholar] [CrossRef]
  10. IEA. CO2 Emissions in 2022. IEA, License: CC BY 4.0. 2023. Available online: https://www.iea.org/reports/co2-emissions-in-2022 (accessed on 14 September 2023).
  11. Kotu, V.; Deshpande, B. Time Series Forecasting. Data Science; Elsevier: Amsterdam; The Netherlands, 2019; pp. 395–445. [Google Scholar] [CrossRef]
  12. Eid, M.; Kutscher, T. Statistical models for analyzing stability and change in happiness. In Stability of Happiness; Academic Press: Cambridge, MA, USA, 2014; pp. 261–297. [Google Scholar] [CrossRef]
  13. Gilli, M.; Maringer, D.; Schumann, E. A gentle introduction to financial simulation. In Numerical Methods and Optimization in Finance; Academic Press: Cambridge, MA, USA, 2019; pp. 153–188. ISBN 9780128150658. [Google Scholar] [CrossRef]
Engproc 76 00019 g001
Figure 1. Machine learning fit plot of CO2 emissions forecasting.
Figure 1. Machine learning fit plot of CO2 emissions forecasting.
Engproc 76 00019 g001
Engproc 76 00019 g002
Figure 2. Integrated GRG model fit plot for CO2 emissions forecasting.
Figure 2. Integrated GRG model fit plot for CO2 emissions forecasting.
Engproc 76 00019 g002
Table 1. Machine learning algorithm metrics.
Table 1. Machine learning algorithm metrics.
MAEMAPERMSENRMSE
ARMA18.99740.08158.62120.3160
AR7.11960.03043.07450.2430
SARIMA4.05940.01741.87880.1485
Table 2. Optimal coefficients to integrate GRG MOMMs.
Table 2. Optimal coefficients to integrate GRG MOMMs.
ARMASARIMAARIntercept
GRG0.31280.02720.316474.1726
Table 3. GRG algorithm metrics.
Table 3. GRG algorithm metrics.
MAERMSENRMSE
GRG2.1054491.2400930.165179
Disclaimer/Publisher’s Note: The statements, opinions and data contained in all publications are solely those of the individual author(s) and contributor(s) and not of MDPI and/or the editor(s). MDPI and/or the editor(s) disclaim responsibility for any injury to people or property resulting from any ideas, methods, instructions or products referred to in the content.

Share and Cite

MDPI and ACS Style

Román, Á.F.G.; Mirmohammadsadeghi, S.; Kabir, G. CO2 Emissions Projections of the North American Cement Industry. Eng. Proc. 2024, 76, 19. https://doi.org/10.3390/engproc2024076019

AMA Style

Román ÁFG, Mirmohammadsadeghi S, Kabir G. CO2 Emissions Projections of the North American Cement Industry. Engineering Proceedings. 2024; 76(1):19. https://doi.org/10.3390/engproc2024076019

Chicago/Turabian Style

Román, Ángel Francisco Galaviz, Seyedmehdi Mirmohammadsadeghi, and Golam Kabir. 2024. "CO2 Emissions Projections of the North American Cement Industry" Engineering Proceedings 76, no. 1: 19. https://doi.org/10.3390/engproc2024076019

Article Metrics

Article metric data becomes available approximately 24 hours after publication online.
Back to TopTop