The Impact of Renewable Electricity Output on Sustainability in the Context of Circular Economy: A Global Perspective
Round 1
Reviewer 1 Report
The manuscript entitled "The Impact of Renewable Electricity Output on
Sustainability in the Context of Circular Economy. A Global Perspective,"
requires more work to substantiate the conclusion of the manuscript. Below
are my further comments:
1. The literature review has low correlation with the goal, lacking of relevant
and latest literature; likewise, the direction and content of the problem research is not clear enough
2. The manuscript needs further English proof to the native English speaker
3. The research method of the manuscript is quite simple and the
manuscript lacks innovation
4. Under "section 5: machine learning and predictions" include the "convolutional neural networks"
5. Provide more convincing results such as graphical and table of comparison/table of differences in terms of convergence speed, implementation issues, algorithm complexities, drawbacks & dependencies, error budget analysis, among others
6. Summarize all the variables that you used in the manuscript and
cite the sources
Author Response
- The literature review has low correlation with the goal, lacking of relevant and latest literature; likewise, the direction and content of the problem research is not clear enough
We have increase the references of cited articles in a more systematic way.
The manuscript needs further English proof to the native English speaker
We have revised all the text.
The research method of the manuscript is quite simple and the manuscript lacks innovation
We have adjunct Convolutional Neural Network analysis.
Under "section 5: machine learning and predictions" include the "convolutional neural networks"
We have increased the analysis with a Convolutional Neural Network-CNN.
Provide more convincing results such as graphical and table of comparison/table of differences in terms of convergence speed, implementation issues, algorithm complexities, drawbacks & dependencies, error budget analysis, among others
We have inserted the following results: statistical results in terms of R^2, MAE, MSE, RMSE. Furthermore, we have adjunct the ranking of algorithms and 2 figures that represent the prediction with the best predictive algorithm.
Summarize all the variables that you used in the manuscript and cite the sources
We have summarized all the variables and cited all the sources.
Reviewer 2 Report
This article focuses on investigating the correlation between Renewable generation output and various selected factors based on the data from World Bank ESG. Coefficients and their signs obtained confirm the relationship of the various factors to the renewable electricity output. This serves as a confirmation towards relationship that has also been shown in the literature.
Four clusters are identified with K-means clustering that best justifies and corroborates the observed correlations. Finally, various ML algorithms are used to verify the renewable generation output for thee 193 countries for the year 2020.
The work deals with a subject that is very timely. Article is written well. However, the novelty and contributions of the article are not clear.
Please find my major concerns below.
1. The authors should consider including Scope-2 and Scope-3 emissions in this analysis. Including Scope-1 direct emissions only can result to incorrect correlations. For example, another factor that reflects a nation's economic growth is its infrastructural growth. This requires usage of construction materials that can increase CO2 emissions (through indirect Scope-2 and Scope-3 emissions). Thus for a nation with high infrastructural growth, the renewable electricity output increase by itself can not account for emissions alone. These couplings need to be analyzed in the article.
2. Similar to above, how do the authors reconcile the fact that one nation's policy decisions on using carbon-based generation can impact the environmental conditions, climate and other factors. This is not captured by the k-means clusters. Moreover, there are shared socio-economic pathways that couple the decisions on carbon-use among nations. This coupling has been assumed to not exist in this study. Authors should explicitly mention this and give a justification for this assumption.
3. Authors should justify their choice for a linear model while formulating the econometric model as well as selection of the 13 variables. There are many other factors that should influence the model such as geographical location of a nation (impacting wind speed, duration, solar irradiation) etc.
4. Authors claim that the renewable electricity output is negatively correlated with cooling degree days, but is it not true that during those cooling degree days, there is an increase in demand for heating systems, thereby causing an increase in renewable electricity output. This seems to be the case (see report US Energy Information Administration - https://www.eia.gov/todayinenergy/detail.php?id=29112)
5. In the ML and predictions section, the motivation for comparing the various algorithms is not clear. It would be more helpful to predict a future forecast of time-series of what portion of a nation's electrical infrastructure should be renewable energy system, given a desired environmental condition, economic growth rate target and policy parameters for likelihood to adopt renewable technologies, etc.
Some minor concerns that can help improve the quality of the manuscript:
1. Some of the abbreviations (such as OLS, WLS) as well as terminologies like Panel Data etc. should be defined in the manuscript.
2. Figure names and captions are missing. Please include those. What are the axes for Figure 2?
Author Response
- The authors should consider including Scope-2 and Scope-3 emissions in this analysis. Including Scope-1 direct emissions only can result to incorrect correlations. For example, another factor that reflects a nation's economic growth is its infrastructural growth. This requires usage of construction materials that can increase CO2 emissions (through indirect Scope-2 and Scope-3 emissions). Thus for a nation with high infrastructural growth, the renewable electricity output increase by itself can not account for emissions alone. These couplings need to be analyzed in the article.
We have added the following analysis in the paragraph of cluster analysis.
To resolve the question of the relationship between the various types of CO2, the relationship between the demand for CO2 in the industrial sector in OECD countries and the percentage of renewable energy out of the total energy produced was analysed. The CO2 demand of the industrial sector approximates the definition of Scope-2. The available dataset relating to Scope 2 emissions has data up to 2015. Specifically, first of all a clustering was carried out with an optimized k-Means algorithm with the Silhouette coefficient based on the value of Scope-2 emissions. Subsequently, for each cluster a comparative analysis was carried out with the level of Renewable Energy Output to verify the presence of a general positive or negative trend through a graphical representation method. In particular, the variable used to approximate the Scope 2 value is "CO2 emissions embodied in domestic final demand, by source country and industry". The variable is calculated in millions of tons.
A clustering with k-Means algorithm is therefore carried out by applying the Elbow method. The analysis shows the presence of 4 clusters.
The clusters are composed as follows:
- Cluster 1: Taipei, Thailand, Netherlands, Argentina, Malaysia, Kazakhstan, Belgium, Greece, Vietnam, Czech Republic, Malta, Iceland, Philippines, Brunei Darussalam, Switzerland, Cambodia, Cyprus, Luxembourg, Hong Kong, Latvia, Costa Rica, Austria, Israel, Romania, Estonia, Lithuania, Slovenia, Colombia, Chile, Croatia, Tunisia, Sweden, Denmark, Slovakia, Peru, Bulgaria, Portugal, Singapore, New Zealand, Ireland, Finland, Hungary, Morocco, Norway,
- Cluster 2: Japan, Russia, India, Germany;
- Cluster 3: Poland, South Africa, Turkey, Spain, United Kingdom, Indonesia, Saudi Arabia, Brazil, Italy, South Korea, Canada, Australia, France, Mexico;
- Cluster 4: China, USA.
The value in millions of tons, specifically it is possible to identify the following ordering of the clusters, namely: C4=6,886.2>C2=1,264.25>C3=460.2>C1=61.6. It therefore follows that from the point of view of the emissions produced in the industrial sector, China and the USA have very high values of CO2 pollution from industry, i.e. a value equal to about 40% among the countries considered in the OECD dataset.
As can be seen from the analysis, it appears that the relationship between the percentage of renewable energy production and CO2 emissions in the industrial sector, with reference to 2015, is negative for the first and second clusters, tends to be zero for the third cluster and is positive for the fourth cluster. Probably the countries that have a greater balance between the level of CO2 emissions from the industrial sector and the level of renewable energy output are the countries of cluster 3. It follows that between renewable energy output and CO2 emissions from industrialization there is no it is necessarily a negative relationship. There are some countries, such as for example the countries of cluster 3, which manage to invest in the production of renewable energy despite the presence of CO2 emissions from the industrial sector.
- Similar to above, how do the authors reconcile the fact that one nation's policy decisions on using carbon-based generation can impact the environmental conditions, climate and other factors. This is not captured by the k-means clusters. Moreover, there are shared socio-economic pathways that couple the decisions on carbon-use among nations. This coupling has been assumed to not exist in this study. Authors should explicitly mention this and give a justification for this assumption.
In the final part of the cluster analysis we have added the following considerations:
Anyway, the ability of countries to invest in renewable energy output can be considered as a function of a series of socio-economic factors that comprehend also demographic and cultural elements. However, overall, the investment in renewable energy output at global level is the consequence of the development of a green conscious and the implementation of green political economies. In this sense it is not sufficient consider the economic consequences and incentives to the implementation of renewable energy source. In effect, the promotion of political movement that are centred on the fight against climate change, has a positive impact on the improvement of renewable energy output at global level. The idea of environmental sustainability and circular economy together with the application of Cost-Benefit Analysis and worst-case scenario oriented to visualize and compute the negative and adverse consequences of climate change, are powerful and useful tools to promote renewable energy outputs even in high developed and pollutant countries.
- Authors should justify their choice for a linear model while formulating the econometric model as well as selection of the 13 variables. There are many other factors that should influence the model such as geographical location of a nation (impacting wind speed, duration, solar irradiation) etc.
In the paragraph in which we propose the econometric model, we add the following sentences:
We used the 13-variable regression model with the aim of investigating the variables connected to the environmental economics contained in the World Bank database entitled ESG-Environment Sustainability Governance. The thirteen variables that were used were chosen on the basis of statistical significance within the dataset analysed. Certainly, the variables included in the ESG World Bank dataset do not exhaust the complex of environmental and geographical variables and factors that can have an impact on the production of renewable energy. An analysis that goes beyond the dimension of nations and which instead considers the geographical dimension from an environmental point of view would probably also be necessary. In fact, the parts of the territory that are included in the geographical definition of a nation can be very heterogeneous in terms of endowment of natural resources useful for renewables, while on the contrary, parts of the territory belonging to different countries can show high levels of homogeneity. However, the possibility of considering these geographical factors at the level would require a shift of the analysis from the national dimension to the regional dimension.
- Authors claim that the renewable electricity output is negatively correlated with cooling degree days, but is it not true that during those cooling degree days, there is an increase in demand for heating systems, thereby causing an increase in renewable electricity output. This seems to be the case (see report US Energy Information Administration - https://www.eia.gov/todayinenergy/detail.php?id=29112
We have added the following considerations in page 7:
Energy consumption is positively associated with both heating and cooling temperatures. In particular, the energy cost of cooling buildings should not be underestimated. Above all because the places in the world that are most populated are also the places in the world where temperatures tend to be higher. For this reason, if the number of "Cooling Degree Days" increases, it is possible to save on energy costs. The situation results in a reduction of renewable energy consumption.
- In the ML and predictions section, the motivation for comparing the various algorithms is not clear. It would be more helpful to predict a future forecast of time-series of what portion of a nation's electrical infrastructure should be renewable energy system, given a desired environmental condition, economic growth rate target and policy parameters for likelihood to adopt renewable technologies, etc.
We used a set of different machine learning algorithms with the aim of identifying which algorithm is more efficient in prediction. In fact, it is not possible to know, without carrying out the necessary analyses, which algorithms are efficient from a predictive point of view without first making a comparison and an analysis from a statistical point of view. The abundance of algorithmic tools used for prediction requires a comparative approach. The comparative approach, which is based on the analysis of the predictive performance, is necessary, on the one hand in order not to underestimate the contribution of the algorithms to the prediction, and on the other hand to create rankings of the most efficient algorithms for the analyzed dataset. Specifically, the trend of the data has essentially a linear structure, due to the fact that the percentage variations in the production of renewable energy at the country level tend to grow slowly, or in any case tend to be devoid of maximums and minimums, whether absolute or relative. Certainly there are limitations in the dataset used which do not allow to connect the production of renewable energy to the energy infrastructure at country level, and which do not even offer the possibility to check for the presence of economic policies which encourage the installation of energy production plants renewable. In fact, it would be necessary to consider the cost, including the institutional one, associated with the growth of the production of renewable energy, due to the need to design appropriate incentives, both for the plants and for the networks. Furthermore, it would also be necessary to calculate the capacity of renewable energy to guide the country towards the achievement of environmental sustainability objectives compatible with international agreements.
Some minor concerns that can help improve the quality of the manuscript:
- Some of the abbreviations (such as OLS, WLS) as well as terminologies like Panel Data etc. should be defined in the manuscript.
We have specified abbreviation Ordinary Least Squares-OLS and Weighted Least Squares-WLS.
We have added the following sentences:
The use of data in panel format was necessary due to the structure of the data used. In fact, the data analyzed do not consist only of a historical series, nor are they observations relating to countries in a single reference period. On the contrary, it is a dataset that combines on the one hand the historical series, consisting of 10 years, and on the other hand the number of observations from the countries, equal to 193 units. The result is a matrix that measures and which therefore needs to be analyzed through the use of panel models which are precisely suitable for data structures that have both a historical series depth and a heterogeneity of individual observations.
- Figure names and captions are missing. Please include those. What are the axes for Figure 2?
We have added captions to all the figures. With respect to Figure 2 we added the following caption:
Figure 2. Clusters of countries for the value of Renewable Energy Output based on k-Means algorithm. Here there is a representation with Multidimensional Scaling-MDS. The x axis represents the input proximities and the y-axis represents either the distances either the input proximities.
Author Response File: 
Author Response.pdf
Reviewer 3 Report
The manuscript is interesting, and chapters are structured according to a template. It is a very good summary of an excellent research work. On the other hand, the scientific level of this manuscript is average.
As a result of the well-composed, logical structure of the abstract, the reader can easily understand the purpose of the research. The aim of the research is clear, the methodology is well-detailed.
The number and the name of the figures are missing. In fugures, the font size is very small, can be resized? Can be sharpened the picture, or the colours should be brighter. Can you improve the quality?
In my opinion, this manuscript similar a review article, but the references are very little.
Overall, the study is of acceptable quality, supporting the claims of the author. The length of the article is also adequate.
Author Response
Q1. The number and the name of the figures are missing. In fugures, the font size is very small, can be resized? Can be sharpened the picture, or the colours should be brighter. Can you improve the quality?
A1. We have added description for each figure and added other figures with better quality.
Q.2 In my opinion, this manuscript similar a review article, but the references are very little.
A.2 We have added a more structured literature review.
Round 2
Reviewer 1 Report
Based on the response of the authors in the manuscript "The Impact
of Renewable Electricity Output on Sustainability in the Context
of Circular Economy. A Global Perspective," I, as a peer-reviewer
is satisfied with the answers to reviewer's comments, and the
revisions made in the manuscript
Reviewer 2 Report
I thank the authors for considering my comments and incorporating them in the article. Authors have now addressed all my concerns in the revised manuscript.
