Skip to content

Literature Synthesis -- Batch 03 Agent 4

Positions 521--560 (40 papers, all 2022 vintage)

Individual Paper Summaries

# Author(s) Year Title (short) Core Finding Method Tags
1 Li et al. 2022 Seismic response of nuclear power stations with pile-raft foundation -- centrifuge tests Soil surface acceleration lower than raft acceleration; seismic response strongly affected by earthquake natural frequency and magnitude Dynamic centrifuge testing (4x3 pile-raft in kaolin clay) centrifuge, pile-raft, seismic, SSI, nuclear
2 Li, Wei, Xu, Qu 2022 Critical acceleration for regional seismic landslide hazard by FE limit analysis FE limit analysis provides efficient critical acceleration maps for regional landslide hazard Finite element limit analysis (FELA) landslide, seismic, hazard, FELA
3 Liu, Yang, Zhang 2022 Review of gassy sediments: mechanical property, disaster simulation, in-situ test Gassy sediments cause submarine landslides, excessive foundation tilting; comprehensive review of mechanical behaviour Literature review gassy sediment, submarine landslide, foundation, review
4 Liu et al. 2022 Dynamic analysis of monopile OWT under sea ice with FSI Sea ice coupled with fluid-structure interaction significantly affects monopile OWT dynamic response Numerical simulation (FSI coupling) monopile, OWT, sea ice, FSI, dynamic
5 Liu, Li, Park 2022 Eighty years of the finite element method: birth, evolution, future Historical overview of FEM from 1941; four eras identified; FEM remains the computational workhorse for engineering Historical review FEM, history, computational mechanics, review
6 Magdy et al. 2022 Comparative study of skirted foundations of different shapes Skirts increase ultimate load up to 5.67x (square) and 8.97x (circular) at skirt depth ratio 1.50; circular outperforms square Physical models + PLAXIS 3D FE skirted foundation, bearing capacity, FEM, experimental
7 Makhoul 2022 Data quality indicators and metrics for SHM Proposed deterministic and probabilistic DQ metrics tailored for SHM; addressed uncertainty in data flow Review + framework proposal SHM, data quality, monitoring, bridge
8 Mebrahtu et al. 2022 Slope stability of deep-seated landslides -- LEM and FEM in Debre Sina, Ethiopia Slopes are unstable; stability depends critically on saturation and seismic load; LEM and FEM safety factors compared Limit equilibrium + FE shear strength reduction slope stability, landslide, LEM, FEM, Ethiopia
9 Momenifar et al. 2022 Physics-informed vector quantized autoencoder for turbulent flow compression VQ-autoencoder achieves CR=85 with MSE O(10^-3), 30% CR improvement over conventional AE; preserves flow statistics Physics-informed deep learning (VQ-VAE) autoencoder, turbulence, data compression, physics-informed
10 Otter et al. 2022 Combined current and wind simulation for floating OWT Dynamic winch actuator with software-in-the-loop can emulate current drag on floating OWT models Physical model testing (SiL) FOWT, current, wind, model testing
11 Pawar et al. 2022 Hybrid AI federated ML/DL for soil component classification ML/DL models classify soil types (sand, clay, silt, peat, chalk, loam); federated approach proposed ML/DL classification soil classification, ML, DL, agriculture
12 Pham, Oh, Ong 2022 UCS of chemical-stabilized clay using gene-expression programming GEP model (R=0.951) predicts UCS; plastic index, clay%, water content negatively affect UCS; cement+slag/lime/FA improves strength Gene-expression programming (GEP) soil stabilization, UCS, GEP, clay
13 Pitcher, Hocking 2022 Risk of SLC-to-block-cave transition via dynamic decision tree + Monte Carlo Decision tree + Monte Carlo recovers ~AUD 20M in valuation and reduces downside risk by 15% on base NPV Monte Carlo + decision tree in DCF mining, block cave, risk, Monte Carlo, decision tree
14 Potentier et al. 2022 Wind turbine blade with root spoilers -- impact on AEP and lifetime Spoilers increase AEP on average but cause severe structural impacts (fatigue) if not designed for BEM aeroelastic simulation (OpenFAST) wind turbine, blade, aerodynamics, fatigue, BEM
15 Pramanick 2022 Review of land use impact on soil physicochemical properties (Wondo Genet) Peer review commentary requesting improved methodology and references Peer review soil properties, land use, review
16 Pula et al. 2022 Failure probability of strip footing on spatially variable soil -- RFELA Combined lower-bound mean + upper-bound std dev from RFELA gives conservative, efficient failure probability Random finite element limit analysis (RFELA) + Monte Carlo reliability, random field, RFELA, foundation, spatial variability
17 Qazi et al. 2022 IoT and AI-enabled smart agriculture: critical review Tutorial + critical review of IoT sensors and AI for precision farming; identifies deployment challenges Review IoT, AI, agriculture, smart farming
18 Quevedo-Reina et al. 2022 Dynamic characterization of OWT on jacket using ANN ANN surrogate reproduces fundamental frequency dependence on system variables including SSI effects ANN surrogate model + FE substructuring OWT, jacket, ANN, SSI, natural frequency
19 Raja, Chiou 2022 Seismic analysis of piles in sand with scouring and water as added mass Fundamental frequency decreases with scour depth; water added mass significantly changes seismic response BNWF + nonlinear time history analysis pile, scour, seismic, added mass, BNWF
20 Razghandi et al. 2022 VAE-GAN for synthetic data generation in smart home VAE-GAN outperforms vanilla GAN for generating synthetic load/PV data distributions VAE-GAN deep learning synthetic data, GAN, VAE, smart grid
21 Reale et al. 2022 Spatial variability of Croatian flood embankment via CPT 15 CPTUs over 200 m reveal significant stratigraphy variation; presents method for horizontal correlation in challenging deposits CPT + MASW + random field theory spatial variability, CPT, flood embankment, random field
22 Rehman et al. 2022 Rock mass behaviour and tunnel support via optimized Hoek-Brown parameters RMR and Q-system support recommendations equally efficient; rock bolt length >= 5 m recommended for crown RMR, Q-system, GSI + 2D FEM tunnel, rock mass, RMR, FEM, support design
23 Ren et al. 2022 Multi-column TLP FOWT with tendon failure scenarios Designed TLP for 60 m intermediate depth; analysed dynamic response with broken tendons WAMIT hydrodynamic analysis TLP, FOWT, tendon failure, hydrodynamics
24 Rimoy et al. 2022 Axially cyclic loaded displacement piles in sands -- stability and load-displacement Stable, metastable, unstable cyclic regimes identified; loose fine sand far more susceptible; stable cycling shows no stiffness change over 10,000+ cycles Calibration chamber instrumented pile tests cyclic loading, pile, sand, stability chart, calibration chamber
25 Sandhu et al. 2022 Post-hazard assessment of nuclear piping via deep learning Novel degradation-sensitive feature vector + deep ANN detects location and severity of piping degradation including minor levels Deep ANN + novel feature extraction nuclear, SHM, deep learning, piping, condition assessment
26 Sedehi et al. 2022 Hierarchical Bayesian UQ of FE models using modal data HBM framework; variability across data sets is the dominant uncertainty source, much larger than identification uncertainties Hierarchical Bayesian + EM + Laplace approximation model updating, Bayesian, UQ, FE, modal analysis
27 Seong et al. 2022 Dynamic and monotonic response of monopiles for OWT -- centrifuge testing Liquefaction observed under strong input motions; natural frequency measured via sine sweep; pre/post-earthquake monotonic response compared Dynamic centrifuge testing (dry + saturated sand) monopile, centrifuge, seismic, liquefaction, OWT
28 Shen et al. 2022 Unbonded post-tensioned RC bridge piers under cyclic loading PRC piers limit damage and residual deformation; decreasing PT force enhances energy dissipation but reduces self-centering Parametric experimental campaign bridge pier, self-centering, post-tensioning, seismic, experimental
29 Sheng et al. 2022 Landslide susceptibility via frequency ratio + C5.0 decision tree Combined frequency ratio and C5.0 decision tree provides efficient LSP model Frequency ratio + C5.0 decision tree landslide, susceptibility, decision tree, ML
30 Sibaii et al. 2022 BIM for geotechnical data from investigations Product Data Template for boreholes; visual programming scripts import geotechnical data into BIM and generate 3D subsurface models BIM + visual programming + IFC BIM, geotechnical, borehole, data management
31 Stuyts et al. 2022 Semi-structured database for back-analysis of OWT monopile foundation stiffness Cloud-based serverless application for parametric geotechnical/structural data retrieval; natural frequency underestimated by 5-15% in design Cloud database + API + digital twin monopile, database, back-analysis, digital twin, OWT
32 Sorum et al. 2022 Wind and soil model influences on fatigue uncertainty of monopile OWT Choice of soil-structure interaction model (macro-element vs p-y) and wind coherence model significantly influences predicted fatigue Comparative numerical study monopile, fatigue, SSI, wind model, uncertainty
33 Taleb, Guemidi 2022 Influence of fissured material on tunnel stability Weakness plane orientations at 45-60 deg (alpha1) and 110-135 deg (alpha2) are most critical for tunnel stability 2D FEM (OPTUMG2), Mohr-Coulomb tunnel, fissured, FEM, stability
34 Tehrani et al. 2022 Machine learning and landslide studies: recent advances and applications ML/DL increasingly used for landslide detection, susceptibility mapping, and temporal forecasting; critical evaluation of data-driven approaches Review ML, deep learning, landslide, review
35 Tom et al. 2022 Review of methodologies to assess bridge safety during/after floods Recommends sonar-equipped remote vessels, drone PIV, holistic risk-based assessment tools for bridge scour monitoring during floods Literature review + DOT interviews bridge, scour, flood, monitoring, risk assessment
36 Wang et al. 2022 3D particle FEM for simulating soil flow with elastoplasticity Novel 3D PFEM with implicit framework permits large time steps; mixed quadratic-linear element avoids volumetric locking 3D PFEM (Hellinger-Reissner) PFEM, large deformation, soil flow, landslide, 3D
37 Wang et al. 2022 CFD validation of moored DeepCwind semisubmersible in irregular waves CFD captures low-frequency slow-drift well but underpredicts pitch resonance; discrepancies linked to incident wave fidelity CFD + OpenFAST validation FOWT, CFD, semisubmersible, validation, irregular waves
38 Wardana et al. 2022 Missing air pollutant data estimation via spatiotemporal convolutional autoencoder Autoencoder with 1D convolutions achieves up to 65% RMSE improvement over univariate methods; exploits spatial correlation among nearby stations Spatiotemporal convolutional autoencoder autoencoder, missing data, air quality, spatiotemporal
39 Wei et al. 2022 LSTM-autoencoder anomaly detection for indoor air quality time series LSTM-autoencoder achieves 99.50% accuracy for CO2 anomaly detection in schools LSTM-autoencoder LSTM, autoencoder, anomaly detection, IAQ, time series
40 Widiyati, Winartha 2022 Risk assessment of dropped/dragged anchor on offshore pipeline (DNV RP F107/F111) Both pipelines show medium risk; 50-80 mm concrete coating recommended for mitigation DNV RP F107/F111 risk framework offshore pipeline, anchor, risk assessment, DNV

SYNTHESIS

CONSENSUS

  1. FEM dominance in geotechnics persists but demands extensions. Liu (2022) traces 80 years of FEM supremacy. Across this batch, FEM (including FELA, PFEM, SSR) is the default numerical tool for foundations (#6, #8, #16, #22, #33, #36). No paper advocates replacing FEM entirely; rather, extensions (particle FEM, limit analysis, random fields) address its known weaknesses in large deformation and probabilistic analysis.

  2. Soil spatial variability must be modelled probabilistically. Reale (#21) and Pula (#16) both apply random field theory to geotechnical problems. The community agrees that deterministic single-value soil parameters are insufficient for reliability assessment of foundations and embankments; the scale of fluctuation (horizontal and vertical) is the critical parameter.

  3. Machine learning is a supplement, not a replacement, for physics-based models. Tehrani (#34), Sheng (#29), Quevedo-Reina (#18), Pham (#12), and Pawar (#11) all deploy ML/DL but frame them as surrogates or complements to physical understanding. Physics-informed constraints (Momenifar #9) improve compression fidelity.

  4. Scour degrades foundation performance and must be coupled with dynamic loads. Raja (#19), Tom (#35), Seong (#27), and Liu (#4) converge on the finding that scour reduces natural frequency and lateral stiffness, with combined scour-plus-seismic or scour-plus-ice loading being under-studied.

  5. Monopile natural frequency is systematically underestimated in design. Stuyts (#31) reports 5-15% underestimation; Sorum (#32) shows that SSI model choice (macro-element vs p-y) significantly affects fatigue predictions. Back-analysis via monitoring data is converging as standard practice.

DEBATES

  1. Appropriate SSI model fidelity for monopile fatigue. Sorum (#32) finds that the macro-element model and p-y model yield materially different fatigue damage estimates. The community has not converged on which model is "correct" for design -- macro-element models capture coupling better, but p-y curves are entrenched in standards (DNV).

  2. Probabilistic vs deterministic geotechnical design. While Pula (#16) and Reale (#21) advocate full probabilistic treatment via random fields, most practice-oriented papers (#6, #22) use deterministic FEM. The gap between research and practice remains wide.

  3. ML interpretability in geohazard assessment. Tehrani (#34) notes that ML models show promising predictive performance for landslides but lack physical interpretability. Whether "black-box" predictions are acceptable for safety-critical decisions (nuclear #25, bridge scour #35) remains contentious.

  4. Role of centrifuge vs numerical modelling. Li (#1) and Seong (#27) rely on centrifuge testing as ground truth for SSI problems, whereas Sorum (#32) and Quevedo-Reina (#18) trust numerical/surrogate models. Neither camp fully validates against field-scale data.

GAPS

  1. Combined multi-hazard loading on foundations. Few papers address simultaneous scour + seismic + ice + cyclic wind loading. Raja (#19) and Liu (#4) each tackle two-hazard combinations, but the full multi-hazard envelope for OWT monopiles is unexplored.

  2. Field-scale validation of RFELA and RFEM. Pula (#16) demonstrates RFELA efficiency but validation is against other numerical benchmarks, not field failure data. No paper in this batch provides field-calibrated random field parameters for offshore foundations.

  3. Digital twin integration for geotechnical assets. Stuyts (#31) pioneers a cloud database for monopile back-analysis, and Sibaii (#30) proposes BIM workflows for geotechnical data, but no paper connects monitoring data to real-time geotechnical model updating in a closed loop.

  4. Autoencoder applications in geotechnical/structural monitoring. Autoencoders appear for air quality (#38, #39), turbulence (#9), and smart grid (#20), but none in this batch applies them to geotechnical or structural health monitoring sensor data -- a clear transfer opportunity.

  5. Cyclic degradation models for offshore piles in variable soils. Rimoy (#24) provides calibration-chamber data for cyclic pile behaviour in sand, but extension to layered or cohesive soils, and to field-scale OWT monopiles under millions of low-amplitude cycles, is missing.

METHODS

Method Papers Maturity
FEM (standard, PLAXIS, OPTUMG2) #6, #8, #22, #33 Mature, industry-standard
FELA / RFELA #2, #16 Emerging for probabilistic geotechnics
3D PFEM #36 Novel, benchmarked but not field-validated
Dynamic centrifuge testing #1, #27 Mature for SSI research
Calibration chamber pile tests #24 Mature for cyclic pile research
ANN / ML surrogates #11, #12, #18, #25, #29, #34 Rapidly growing; interpretability concern
Physics-informed autoencoders #9 Novel, demonstrated for CFD data
LSTM-autoencoder #39 Proven for time-series anomaly detection
Spatiotemporal convolutional AE #38 Proven for imputation tasks
Hierarchical Bayesian model updating #26 Rigorous but computationally demanding
Random field + Monte Carlo #16, #21 Standard for probabilistic geotechnics
BEM aeroelastic simulation #14 Mature for wind turbine design
CFD (RANS/LES) for FOWT #37 Maturing; validation gaps remain
BIM for geotechnical data #30 Early adoption phase
Cloud database + API #31 Emerging for wind farm digital twins
DNV risk framework #40 Industry standard for offshore pipelines

BENCHMARKS

Benchmark / Dataset Used By Notes
DeepCwind OC5 semisubmersible Wang (#37) Multi-participant CFD validation; publicly available
NREL 5 MW reference turbine Sorum (#32), Potentier (#14), Ren (#23) De facto standard for OWT research
Hostun sand (calibration chamber) Rimoy (#24) Imperial College pile test programme
Fraction E sand (Cambridge) Seong (#27) Cambridge centrifuge standard sand
Shanxi kaolin clay Li (#1) Centrifuge programme for nuclear SSI
Dunedin CO2 school dataset Wei (#39) Real-world IAQ time series from NZ schools
Carrapateena mine (OZ Minerals) Pitcher (#13) Case study for block cave DCF risk
Croatian flood embankment CPT Reale (#21) 15 CPTUs over 200 m, unique dataset
Debre Sina landslide complex Mebrahtu (#8) Ethiopian Rift margin, deep-seated slides

Synthesised 2026-04-17. 40 papers, positions 521-560.