Academic Fraud Investigation Report: 'A low cost strategy on energy harvesting of flapping foil with time-warping optimization' (Energy, 2025)

Academic fraud report · Geng Detector

Summary

This report presents a 'Highly Suspicious' verdict on a 2025 Energy journal article (DOI: 10.1016/j.energy.2025.138554) by Zhangyuan Wang et al. concerning flapping-foil energy harvesting. Five key concerns are identified. First, the literature review appears to contain severe cross-domain citation errors: references [14] and [15] in the introduction are network-science papers (random walks on core-periphery structures; community detection algorithms), while reference [42] is a 5G telecommunications waveform paper, none of which relate to hydrodynamics. Second, Table 2 reports implausibly tight standard deviations (e.g., 27.5% ± 0.2% efficiency) for Re = 6000 tow-tank experiments, where much larger scatter is expected. Third, the claimed '48.2% cost reduction' conflate low-cost CFD samples (620) with expensive physical experiments (321), inflating rather than reducing total cost. Fourth, Figure 6(a) uses a distinct colorbar for an inset, a known visualization-manipulation technique that exaggerates local differences. Fifth, a Pearson r = 0.487 is described as 'strong' statistical evidence, which overstates a moderate correlation. Limitations: this is an AI-assisted review; final determination of misconduct requires institutional investigation.

Verdict

🟠 Highly Suspicious — five independent concerns spanning citation integrity, statistical plausibility, cost-accounting logic, figure manipulation, and over-interpretation. AI-assisted assessment; institutional confirmation required.

Key findings

Cross-domain citation hallucination: References [14] and [15] (network-science papers on random walks and Louvain community detection) and reference [42] (a 5G waveform paper) cited in a hydrodynamic context.
Implausibly tight experimental scatter: Standard deviations as low as ± 0.2% efficiency in tow-tank tests at Re = 6000 (Table 2).
Illogical 'cost reduction' arithmetic: Cost computed as (620 CFD samples − 321 experimental samples)/620 ≈ 48.2%, conflating cheap compute samples with expensive physical tests.
Colorbar manipulation in Figure 6(a): Inset uses a different color scale than the main panel, exaggerating local contrast.
Overstated correlation: Pearson r = 0.487 labeled as 'strong statistical evidence'.

Evidence highlights

Reference [14]: Rossa F, et al., *Sci Rep* 2013 — random walks on core-periphery network structures.
Reference [15]: Blondel VD, et al., *J Stat Mech* 2008 — community-detection algorithm (Louvain).
Reference [42]: Ibrahim M, et al., *IEEE Commun Lett* 2019 — 5G time-frequency warped waveforms.
Table 2 efficiency values: 27.5 ± 0.2%, 25.2 ± 0.6%, 25.4 ± 0.3%, 19.4 ± 0.8% (Re = 6000).
Cost arithmetic: (620 − 321)/620 ≈ 48.2%.
Figure 6(a): main panel and inset use different colorbars for the same quantity η.
Pearson r = 0.487 between efficiency and angle of attack, described as 'strong'.

Notes

DOI: 10.1016/j.energy.2025.138554
Authors: Zhangyuan Wang, Wenhua Xu, Xuliang Han, Ruipeng Li, Jiaye Gong, Weicheng Cui, Dixia Fan.
Recommended actions (per original report): request raw experimental data to verify Table 2 SDs; raise citation concerns on PubPeer; consider a formal editorial complaint regarding colorbar manipulation and cost-logic fallacies.
This automated review is advisory only; some flagged items (e.g., very low SDs) could reflect idealised test conditions not described in the paper, and citation errors could be typographical. Confirmation requires institutional investigation.