Pith. sign in

REVIEW 5 major objections 6 minor 23 references

An end-to-end pipeline turns neural-network predictions for energy systems into LIME/SHAP scores, fidelity and stability checks, then LLM natural-language explanations.

Reviewed by Pith at T0; open to challenge. T0 means a machine referee read the full paper against a public rubric. the ladder, T0–T4 →

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗

T0 review · grok-4.5

2026-07-11 19:38 UTC pith:4JMTMVHQ

load-bearing objection Solid engineering pipeline of LIME/SHAP + fidelity/stability + LLM text on two energy tasks; the human-interpretability claim is untested but the rest is usable. the 5 major comments →

arxiv 2607.04374 v1 pith:4JMTMVHQ submitted 2026-07-05 eess.SY cs.SY

An End-to-End Explainable AI Framework with Automated LLM-Based Natural Language Explanation Generation for Energy Systems

Venkata Sesha Sai Raj Nanduri , Akthar Hussain , Van-Hai Bui This is my paper
classification eess.SY cs.SY
keywords Explainable AILarge language modelLIMESHAPNeural networksPower system fault detectionBuilding energy labelsFidelity and stability
verification ladder T0 review T1 audit T2 compute T3 formal T4 reserved

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

Black-box neural networks can predict power-system faults and building energy use accurately, but their decisions are hard for non-technical people to trust. This paper claims a reusable pipeline solves that: it trains a multilayer perceptron on tabular energy data, generates local and global feature attributions with LIME and SHAP, scores those attributions for fidelity and stability, then feeds the structured numbers into a fixed LLM prompt that returns five short plain-language bullets. On a balanced fault-detection set the network reaches 99 percent accuracy and ROC-AUC 1.0 with near-perfect explanation scores; on building energy labels it reaches R² ≈ 0.67 with high fidelity but lower stability. The authors argue that packaging prediction, quantitative explanation quality, and automatic natural-language translation into one model-agnostic workflow makes high-stakes energy decisions transparent enough for engineers and non-experts alike.

Core claim

A single end-to-end, model-agnostic XAI framework can train neural networks for both classification and regression energy tasks, produce LIME local and SHAP global explanations, quantify those explanations with fidelity and stability, and convert the resulting structured feature scores into concise natural-language summaries via a fixed LLM prompt, thereby rendering black-box decisions understandable to non-technical stakeholders while preserving high predictive performance.

What carries the argument

The integrated pipeline itself: standardized preprocessing → MLP training → LIME/SHAP attribution → fidelity/stability evaluation → fixed five-bullet LLM prompt that turns numerical attributions into plain-language plot overview, key findings, feature importance, model behaviour and potential issues.

Load-bearing premise

That a fixed five-bullet LLM prompt fed only structured feature scores and fidelity/stability numbers (never raw plots or human ratings) produces text that non-technical stakeholders will correctly understand and trust.

What would settle it

A controlled user study in which non-technical readers of the LLM bullets are tested on factual accuracy of the explanations and on measured trust or decision quality; if comprehension or trust scores stay near chance, the natural-language claim fails.

Watch this falsifier — get emailed when new claim-graph text bears on it.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit.

Referee Report

5 major / 6 minor

Summary. The manuscript proposes a reusable end-to-end XAI pipeline for energy systems that (i) trains multilayer perceptrons for classification and regression, (ii) produces local and global post-hoc explanations with LIME and KernelSHAP, (iii) scores those explanations with fidelity and stability metrics, and (iv) feeds structured feature attributions into a fixed LLM prompt to generate five-bullet natural-language explanations for non-technical stakeholders. The pipeline is demonstrated on a power-system fault-detection dataset (≈99% accuracy, ROC-AUC 1.00) and a building energy-labels regression dataset (test R² ≈ 0.67), with a random-forest baseline for comparison. The authors conclude that the framework yields faithful, stable explanations and that the LLM layer improves interpretability of black-box models for everyone.

Significance. If the full claim holds, the work would offer a practical, model-agnostic template for deploying explainable neural networks in safety-critical and sustainability-oriented energy applications, where both quantitative explanation quality and stakeholder-facing language matter. Strengths include dual-task validation (classification and regression), explicit fidelity/stability reporting, and a side-by-side neural-network vs. random-forest comparison (Tables 2–3). The distinctive contribution relative to prior energy-XAI work is the automated LLM natural-language layer; that contribution, however, currently rests on an untested assumption about human comprehension and trust, so the paper’s significance is conditional on either adding human evaluation or substantially narrowing the claim.

major comments (5)
  1. The title, abstract, listed contributions, and §2.3 assert that the LLM converts LIME/SHAP outputs into natural-language explanations that non-technical stakeholders can understand and that thereby improve interpretability “to everyone.” Section 2.3 only specifies a fixed five-bullet prompt fed with model name, prediction, ranked feature scores, influence direction, and fidelity/stability numbers; no human study of comprehension, trust, decision usefulness, or factual faithfulness of the generated text is reported, nor is any comparison against raw SHAP/LIME plots. This is load-bearing for the paper’s central claim and must be addressed either by a user study (or at least expert rating of faithfulness/clarity) or by reframing the contribution as automated NL generation without claiming improved stakeholder interpretability.
  2. §2.1 is internally inconsistent on the activation function. The prose states that both the classifier and regressor use ReLU, yet Eq. (2) is written as a tanh activation and the surrounding text then re-describes ReLU. The feed-forward equations, gradient expressions (Eqs. 4–5), and the claimed architecture must be made consistent; as written, a reader cannot reproduce the trained models.
  3. §2.2 / Eq. (9): SHAP fidelity is reported as 1.00 (fault detection) and 0.998 (energy). Exact Shapley values satisfy local accuracy by construction (∑φᵢ + φ₀ = f(x)), so a fidelity of 1.0 is expected for exact SHAP and is only informative for the KernelSHAP approximation residual. The manuscript never states the numerical definition used in Eq. (9) (the equation body is incomplete in the text) nor how many background samples / coalitions were used. Without that definition and without reporting the approximation residual distribution, the “nearly perfect fidelity” claim is difficult to interpret and may be partly tautological.
  4. Abstract and §5 claim the approach produces “stable” explanations, yet §4.3.4 reports SHAP stability of only 0.719 on the energy regression task (vs. 0.998 on fault detection; RF reaches 0.982). The paper correctly notes sensitivity to building-type features, but the unqualified stability language in the abstract and conclusion should be qualified, and the authors should discuss whether explanations with stability ≈0.72 are reliable enough for the intended urban-planning use case.
  5. Reproducibility of the LLM layer is incomplete. §2.3 never names the LLM (model family, version, temperature, or whether a local vs. API model was used), nor does it release the exact structured context strings or example outputs beyond a high-level description. Given that the LLM step is a listed contribution, the model identity and generation settings are free parameters that must be fixed for the pipeline to be reusable.
minor comments (6)
  1. Throughout: numerous grammatical and typographical issues (“block-box”, “aa well as”, “ans 0.8040”, “ProprtyGFATotal”, repeated sentence in §2.4, “Fig 4” referenced without a corresponding figure in the provided text). A careful copy-edit is needed.
  2. §2.1: hidden-layer widths (192 / 448) and iteration caps appear without ablation or selection rationale; a short sensitivity note would help readers judge robustness of the free parameters listed in the architecture.
  3. Figures 5–14 are described but several captions and in-text references are incomplete or inconsistent (e.g., “Fig 4: LLM Integration” appears after the methodology without a visible figure body in the manuscript text). Ensure every figure is present, labeled, and referenced once.
  4. Table 1 and §3: building-type class imbalance (Educational 4%, Institutional 2.3%) is noted but not discussed as a possible driver of the low stability or of SHAP dominance of Residential/Commercial indicators; a brief remark would strengthen the regression analysis.
  5. References: several energy-XAI and LLM-for-XAI works are cited, but TalkToModel and related interactive NL explanation systems are mentioned only briefly; a clearer positioning of what is new relative to Slack et al. (2023) would help.
  6. §4.1.3 waterfall discussion of the misclassified instance is hard to follow (“contributions seemed to point to the fault class the model made an incorrect prediction”); clarify predicted probability vs. true label for both instances.

Circularity Check

0 steps flagged

No circularity: standard post-hoc XAI pipeline with independent held-out metrics and no self-definitional reductions.

full rationale

The paper presents an applied engineering framework that trains MLP classifiers/regressors on two tabular energy datasets, applies off-the-shelf LIME and KernelSHAP, evaluates the resulting attributions with the conventional fidelity and stability formulas (local R^{2} of the surrogate and variance under re-perturbation), and feeds the numerical scores into a fixed LLM prompt. All quantitative claims (accuracy 0.99 / ROC-AUC 1.00, R^{2} 0.67, fidelity ≈ 1.0, stability 0.998 / 0.719) are computed on held-out test instances against the already-trained black-box model; none of the reported numbers is obtained by fitting a free parameter to the same quantity that is later called a “prediction.” There are no uniqueness theorems, no self-citations that carry the load of a derivation, and no renaming of a known empirical pattern. The LLM step is pure generation from structured input and is not claimed to derive any numerical result. Consequently the derivation chain contains no circular steps.

Axiom & Free-Parameter Ledger

4 free parameters · 4 axioms · 0 invented entities

The work rests on standard ML and XAI assumptions rather than new physical or mathematical postulates. Free parameters are ordinary hyper-parameters of the MLP and LIME/SHAP sampling; no invented entities appear. The only non-standard modeling choice is the claim that a fixed LLM prompt template is sufficient for stakeholder understanding—an untested domain assumption.

free parameters (4)
  • MLP hidden-layer widths and depths = 192 / 448 units
    Classifier uses two layers of 192 units; regressor uses three layers of 448 units. Chosen by the authors; no ablation or search reported.
  • L2 regularization strengths (alpha) = 1e-4 / 1e-5
    alpha = 0.0001 (classifier) and 0.00001 (regressor); hand-selected to control overfitting.
  • LIME neighborhood size and repetitions = N=5000, K=10
    N = 5000 perturbed samples, K = 10 stability repetitions; defaults that affect fidelity/stability numbers.
  • LLM prompt template and (unspecified) model
    Fixed five-bullet prompt; model name, temperature and decoding parameters never stated, yet the natural-language claim depends on them.
axioms (4)
  • domain assumption Post-hoc local linear surrogates (LIME) and Shapley-value attributions (KernelSHAP) faithfully represent the decision surface of a trained MLP near a given instance.
    Invoked throughout Section 2.2; standard XAI premise but known to fail for highly non-linear or discontinuous regions.
  • domain assumption Fidelity (local R² or additive reconstruction) and stability under small input perturbations are sufficient quantitative proxies for explanation quality.
    Section 2.2 evaluation metrics; no human-grounded validation is supplied.
  • ad hoc to paper A fixed natural-language prompt template fed only structured feature scores produces explanations that non-technical stakeholders correctly understand and trust.
    Section 2.3; never tested with users.
  • standard math Standard feed-forward MLP training with Adam and L2 regularization yields models whose explanations are worth interpreting for safety-critical energy decisions.
    Section 2.1; ordinary supervised-learning assumption.

pith-pipeline@v1.1.0-grok45 · 19472 in / 3131 out tokens · 26279 ms · 2026-07-11T19:38:11.333371+00:00 · methodology

0 comments
read the original abstract

Explainable AI (XAI) is important for deploying machine learning systems in domains where stakes are very high and where transparency, trust and accountability are critical. Although black box models like deep neural networks often perform with high efficiency, interpreting their decisions remains as a difficult task. This paper proposes a reusable end-to-end XAI framework that is the combination of prediction, explanation generation, evaluation and converting these explanations into natural language text of explanation which can be easily understood by the non-technical stakeholders as well. This framework initially trains deep neural network for both classification and regression tasks. Local and global explanations are generated using XAI algorithms, including Local Interpretable Model-agnostic Explanation (LIME) and SHapley Additive exPlanations (SHAP) respectively. To evaluate these explanations, we use fidelity and stability metrics to know how accurately and consistently explanations reflect the model behavior. The generated explanation includes feature importance scores, prediction specific attributes, and then transformed into a structured input to the Large Language Model (LLM), which generates a natural language explanation through which everyone can understand the explanations generated by XAI algorithms. This framework is tested on power system fault dataset detection dataset and building energy labels dataset. For fault detection, the neural network model achieved 99% accuracy with ROC-AUC score of 1.00. For building energy prediction, model achieves R2 score of 0.67. These findings say that the proposed approach produces a stable and faithful explanations while improving the interpretability of black box model to everyone with the help of LLMs.

Figures

Figures reproduced from arXiv: 2607.04374 by Akthar Hussain, Van-Hai Bui, Venkata Sesha Sai Raj Nanduri.

Figure 1
Figure 1. Figure 1: The flowchart of the Explainable AI framework [PITH_FULL_IMAGE:figures/full_fig_p004_1.png] view at source ↗
Figure 4
Figure 4. Figure 4: LLM Integration 3. Datasets To evaluate the robustness of the proposed XAI framework, so the experiments were conducted on multiple benchmark datasets for both classification and regression tasks. The following datasets which contains mixture of numerical features and categorical features helps us to check the robustness of this framework. These Datasets were selected to evaluate the proposed framework acr… view at source ↗
Figure 5
Figure 5. Figure 5: Fault detection performance metrics 4.1.2 LIME Based local explanations To improve the interpretability of the decisions made by neural network, LIME was used to generate local explanations for two representative test instances: one that was correctly classified and another that was misclassified. These local explanations shows us how individual input features contribute to the model’s prediction and provi… view at source ↗
Figure 6
Figure 6. Figure 6: LIME explanations of 2 instance 4.1.3 SHAP Based Global and Local Explanations Global feature importance was computed using KernelSHAP across the test set. The SHAP bar plot ranks all six features by their mean absolute SHAP value. The current in phase A (current a) is the most important thing that affects the result. It has an impact with mean SHAP value 0.13. The currents in phase B, (current b) and the … view at source ↗
Figure 7
Figure 7. Figure 7: SHAP bar plot The SHAP beeswarm summary plot helps us understand how each feature influence the result. The distribution of the SHAP values in the below plot demonstrates that contribution of each feature varies across different operating conditions, which reflects the nonlinear relationships that neural network learned. When we look at current_a we observe both positive and negative SHAP values depending … view at source ↗
Figure 8
Figure 8. Figure 8: SHAP Summary plot We also looked at SHAP explanations for two specific instances. For the instance that was correctly classified we started with the models baseline expected output of 0.445. The waterfall plot showed that voltage_c contributed -0.13, voltage_b contributed -0.13, current_a contributed -0.12 [PITH_FULL_IMAGE:figures/full_fig_p016_8.png] view at source ↗
Figure 9
Figure 9. Figure 9: SHAP Explanations for 2 instances 4.1.4 Explanation Quality Metrics The proposed framework was evaluated to see how well it explained the decisions it made. It was tested using two measures: Fidelity and stability, the explanations from the framework got a fidelity score of 1.00. This indicates SHAP explanations accurately reproduce the neural network’s predictions without introducing approximation errors.… view at source ↗
Figure 10
Figure 10. Figure 10: Energy regression performance metrics 4.3.2 LIME-Based local explanations To interpret the neural network’s predictions, LIME was applied to two representative test instances: one correctly predicted building and one incorrectly predicted building, as shown in figure below [PITH_FULL_IMAGE:figures/full_fig_p019_10.png] view at source ↗
Figure 12
Figure 12. Figure 12: SHAP Bar plot [PITH_FULL_IMAGE:figures/full_fig_p020_12.png] view at source ↗

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Reference graph

Works this paper leans on

23 extracted references · 3 canonical work pages

  1. [1]

    Explainable Artificial Intelligence (XAI): Concepts, taxonomies, opportunities and challenges toward responsible AI

    Arrieta, A.B.; Díaz -Rodríguez, N.; Del Ser, J.; Bennetot, A.; Tabik, S.; Barbado, A.; Herrera, F. Explainable Artificial Intelligence (XAI): Concepts, taxonomies, opportunities and challenges toward responsible AI. Inf. Fusion 2020, 58, 82–115. https://doi.org/10.1016/j.inffus.2019.12.012

    doi:10.1016/j.inffus.2019.12.012 2020

  2. [2]

    Peeking Inside the Black -Box: A Survey on Explainable Artificial Intelligence (XAI)

    Adadi, A.; Berrada, M. Peeking Inside the Black -Box: A Survey on Explainable Artificial Intelligence (XAI). IEEE Access 2018, 6, 52138–52160. https://doi.org/10.1109/ACCESS.2018.2870052

    doi:10.1109/access.2018.2870052 2018

  3. [3]

    and Aha, D.W

    Gunning, D. and Aha, D.W. (2019), DARPA's Explainable Artificial Intelligence Program. AI Magazine, 40: 44-

    2019

  4. [4]

    https://doi.org/10.1609/aimag.v40i2.2850

    doi:10.1609/aimag.v40i2.2850

  5. [5]

    Stop explaining black box machine learning models for high stakes decisions and use interpretable models instead

    Rudin, C. Stop explaining black box machine learning models for high stakes decisions and use interpretable models instead. Nat Mach Intell 1, 206–215 (2019). https://doi.org/10.1038/s42256-019-0048-x

    doi:10.1038/s42256-019-0048-x 2019

  6. [6]

    Explaining Explanations: An Overview of Interpretability of Machine Learning,

    L. H. Gilpin, D. Bau, B. Z. Yuan, A. Bajwa, M. Specter and L. Kagal, "Explaining Explanations: An Overview of Interpretability of Machine Learning," 2018 IEEE 5th International Conference on Data Science and Advanced Analytics (DSAA), Turin, Italy, 2018, pp. 80-89, doi: 10.1109/DSAA.2018.00018

    doi:10.1109/dsaa.2018.00018 2018

  7. [7]

    Towards a Rigorous Science of Interpretable Machine Learning

    Doshi -Velez, F.; Kim, B. Towards a Rigorous Science of Interpretable Machine Learning. arXiv 2017, arXiv:1702.08608. https://doi.org/10.48550/arXiv.1702.08608

    Pith/arXiv arXiv doi:10.48550/arxiv.1702.08608 2017

  8. [8]

    Interpretable Machine Learning: A Guide for Making Black Box Models Explainable, 2nd ed.; Leanpub: Victoria, BC, Canada, 2022

    Molnar, C. Interpretable Machine Learning: A Guide for Making Black Box Models Explainable, 2nd ed.; Leanpub: Victoria, BC, Canada, 2022. Available online: https://christophm.github.io/interpretable-ml-book

    2022

  9. [9]

    Why Should I Trust You?

    Ribeiro, M.T.; Singh, S.; Guestrin, C. “Why Should I Trust You?”: Explaining the Predictions of Any Classifier. In Proceedings of the 22nd ACM SIGKDD International Conference on Knowledge Discovery and Data Mining, San Francisco, CA, USA, 13–17 August 2016; pp. 1135–1144. https://doi.org/10.1145/2939672.2939778

    doi:10.1145/2939672.2939778 2016

  10. [10]

    A Unified Approach to Interpreting Model Predictions

    Lundberg, S.M.; Lee, S.-I. A Unified Approach to Interpreting Model Predictions. In Advances in Neural Information Processing Systems; Guyon, I., Luxburg, U.V., Bengio, S., Wallach, H., Fergus, R., Vishwanathan, S., Garnett, R., Eds.; Curran Associates: Red Hook, NY, USA, 2017; Volume 30. https://proceedings.neurips.cc/paper/2017/hash/8a20a8621978632d76c4...

    2017

  11. [11]

    From Local Explanations to Global Understanding with Explainable AI for Trees

    Lundberg, S.M.; Erion, G.; Chen, H.; DeGrave, A.; Prutkin, J.M.; Nair, B.; Katz, R.; Himmelfarb, J.; Bansal, N.; Lee, S.-I. From Local Explanations to Global Understanding with Explainable AI for Trees. Nat. Mach. Intell. 2020, 2, 56–67. https://doi.org/10.1038/s42256-019-0138-9

    doi:10.1038/s42256-019-0138-9 2020

  12. [12]

    Anchors: High -Precision Model-Agnostic Explanations

    Ribeiro, M.T.; Singh, S.; Guestrin, C. Anchors: High -Precision Model-Agnostic Explanations. In Proceedings of the AAAI Conference on Artificial Intelligence, New Orleans, LA, USA, 2–7 February 2018; Volume 32

    2018

  13. [13]

    & sayres, R

    Kim, B., Wattenberg, M., Gilmer, J., Cai, C., Wexler, J., Viegas, F. &amp; sayres, R.. (2018). Interpretability Beyond Feature Attribution: Quantitative Testing with Concept Activation Vectors (TCAV). <i>Proceedings of the 35th International Conference on Machine Learning</i>, in <i>Proceedings of Machine Learning Research</i> 80:2668-2677 Available from ...

    2018

  14. [14]

    On the Robustness of Interpretability Methods

    Alvarez-Melis, D.; Jaakkola, T.S. On the Robustness of Interpretability Methods. arXiv 2018, arXiv:1806.08049. https://doi.org/10.48550/arXiv.1806.08049

    Pith/arXiv arXiv doi:10.48550/arxiv.1806.08049 2018

  15. [15]

    On the (In)fidelity and Sensitivity of Explanations

    Yeh, C.-K.; Hsieh, C.-Y.; Suggala, A.; Inouye, D.I.; Ravikumar, P. On the (In)fidelity and Sensitivity of Explanations. In Advances in Neural Information Processing Systems; Wallach, H., Larochelle, H., Beygelzimer, A., d’Alché-Buc, F., Fox, E., Garnet t, R., Eds.; Curran Associates: Red Hook, NY, USA, 2019; Volume 32. https://proceedings.neurips.cc/paper...

    2019

  16. [16]

    Look at the Variance! Efficient Black-box Explanations with Sobol -based Sensitivity Analysis

    Fel, T.; Cadene, R.; Chalvidal, M.; Cord, M.; Vigouroux, D.; Serre, T. Look at the Variance! Efficient Black-box Explanations with Sobol -based Sensitivity Analysis. In Advances in Neural Information Processing Systems; Ranzato, M., Beygelzimer, A., Dau phin, Y., Liang, P.S., Vaughan, J.W., Eds.; Curran Associates: Red Hook, NY, USA, 2021; Volume 34, pp. ...

    2021

  17. [17]

    Explaining Machine Learning Models with Interactive Natural Language Conversations Using TalkToModel

    Slack, D.; Krishna, S.; Lakkaraju, H.; Singh, S. Explaining Machine Learning Models with Interactive Natural Language Conversations Using TalkToModel. Nat. Mach. Intell. 2023, 5, 873–883. https://doi.org/10.1038/s42256- 023-00692-8

    doi:10.1038/s42256- 2023

  18. [18]

    & Zhou, D

    Wei, J., Wang, X., Schuurmans, D., Bosma, M., Xia, F., Chi, E., ... & Zhou, D. (2022). Chain -of-thought prompting elicits reasoning in large language models. Advances in neural information processing systems, 35, 24824- 24837

    2022

  19. [19]

    Training language models to follow instructions with human feedback

    Ouyang, Long, et al. "Training language models to follow instructions with human feedback." Advances in neural information processing systems 35 (2022): 27730-27744

    2022

  20. [20]

    Fault Detection and Classification in Power Systems Using Machine Learning Algorithms

    Anand, V.; Yammani, C. Fault Detection and Classification in Power Systems Using Machine Learning Algorithms. Int. J. Electr. Power Energy Syst. 2021, 131, 106975. https://doi.org/10.1016/j.ijepes.2021.106975

    work page doi:10.1016/j.ijepes.2021.106975 2021

  21. [21]

    Chen, Z., Xiao, F., Guo, F., & Yan, J. (2023). Interpretable machine learning for building energy management: A state-of-the-art review. Advances in Applied Energy, 9, 100123. https://doi.org/10.1016/j.adapen.2023.100123

    doi:10.1016/j.adapen.2023.100123 2023

  22. [22]

    Explainable approaches for forecasting building electricity consumption

    Sakkas, Nikos, et al. "Explainable approaches for forecasting building electricity consumption." Energies 16.20 (2023): 7210. https://doi.org/10.3390/en16207210

    work page doi:10.3390/en16207210 2023

  23. [23]

    K., Wu, M., Chen, J., & Zhang, L

    Zhang, Y., Teoh, B. K., Wu, M., Chen, J., & Zhang, L. (2023). Data -driven estimation of building energy consumption and GHG emissions using explainable artificial intelligence. Energy, 262, 125468. https://doi.org/10.1016/j.energy.2022.125468

    work page doi:10.1016/j.energy.2022.125468 2023