Skip to content

Literature Synthesis: Batch 02 Agent 4 (Positions 321--360)

All 40 papers are from 2020. Topics span offshore wind engineering, geotechnical foundations, deep learning for geoscience, structural health monitoring, and soil-structure interaction.

Paper-Level Extractions

# Author(s) Year Title (short) Core Finding Method Tags
1 Mooney et al. 2020 Acoustic impacts of OWF on fisheries All OWF lifecycle phases produce underwater noise impacting marine life; data deficient for most species/life-stages Review of field measurements offshore-wind, environmental, acoustics
2 Muhling et al. 2020 Predictability of species distributions under novel conditions Species distribution model accuracy degrades under unprecedented environmental conditions in the California Current SDM, GAM, random forest ecology, ML, climate
3 Naji et al. 2020 Review of integral abutment bridges with SSI Backfill material and thermal loads dominate IAB performance; compressible backfill enhances service life FE review, parametric soil-structure-interaction, bridge
4 Neupane & Seok 2020 Bearing fault detection using CWRU dataset with DL Comprehensive review of DL methods on the CWRU benchmark; CNN-based models dominate accuracy Review, CNN/RNN/AE variants fault-detection, deep-learning, benchmark
5 Ng et al. 2020 Training sample size effect on DL for soil property prediction CNN outperforms PLSR/Cubist only above ~2000 samples; CNN accuracy does not plateau unlike classical ML CNN, PLSR, Cubist, VIS-NIR spectroscopy deep-learning, soil, sample-size
6 Olsen & Haun 2020 Numerical model for soil slides at reservoir banks 3-D Navier-Stokes + limit-equilibrium model captures bank failure during flushing; results sensitive to cohesion 3-D CFD + geotechnical, field validation numerical, slope-stability, reservoir
7 Ouabel et al. 2020 Settlement estimation via pressuremeter FEM FEM with Mohr-Coulomb and Cam-Clay using pressuremeter data gives reliable settlement estimates for shallow foundations FEM, pressuremeter geotechnical, settlement, foundation
8 Paucar & Gutierrez 2020 Anchor wall system for deep excavation in Lima conglomerate FEM-predicted lateral displacements match inclinometer readings; validates geomechanical parameters for Lima conglomerate FEM, field monitoring excavation, geotechnical, monitoring
9 Peyghami et al. 2020 Reliability of modern power-electronic power systems New reliability framework needed for DG/microgrid-dominated grids incorporating local reliability and cyber-physical aspects Review, reliability framework power-systems, reliability, smart-grid
10 Pham et al. 2020 GA-optimized DNN for pile bearing capacity GA-DLNN hybrid using feature selection outperforms standard DNN; optimal architecture search improves pile capacity prediction (472 test records) GA + DNN, 472-pile database deep-learning, pile, bearing-capacity
11 Rahimi et al. 2020 TMD for structural vibration control -- state of the art Review of passive/active/semi-active/hybrid TMDs; gap in nonlinear analysis and natural-frequency tuning methods Review vibration-control, TMD, structural
12 Rahman et al. 2020 Rayleigh damping of submerged floating tunnel Modal analysis via ANSYS yields alpha=0.946, beta=0.00022 for SFT; damping is significant despite slenderness FEA (ANSYS), modal analysis dynamics, SFT, damping
13 Razouki & Kuttah 2020 Safety factor for strip footings on gypsum-rich soils Soaking causes dramatic strength loss; SF=8 on unsoaked basis needed to guarantee SF=3 under soaked conditions Triaxial tests, Terzaghi bearing capacity geotechnical, gypsum, bearing-capacity
14 Ren et al. 2020 Physics-based NN for seismic full waveform inversion SWINet embeds wave equation in NN forward pass; avoids need for labeled velocity models; faster convergence than FWI Physics-informed NN (PyTorch) PINN, seismic, inversion
15 Rentschler et al. 2020 Dynamic inter-array cables for floating OWT Lazy-wave umbilical outperforms catenary shape; generalized design chart for first-pass cable sizing across water depths Hydrostatic + parametric study offshore-wind, cable, floating
16 Roach et al. 2020 IEC 61400-3-1 applied to Massachusetts OWE area Control system failure and electrical network loss are critical DLCs; hurricane loading significant for US East Coast FAST simulation, IEC standard offshore-wind, design-standard, monopile
17 Schulz et al. 2020 Linear vs DL scaling on brain images vs ML benchmarks For brain imaging (~10k subjects), linear models match deep/kernel models; nonlinearities inaccessible at current sample sizes Linear, kernel, DNN on UKBB deep-learning, sample-size, neuroimaging
18 Seguini & Nedjar 2020 Dynamic probabilistic analysis of shear-deformable pipeline Soil spatial variability greatly affects seismic pipe response; stochastic FEM + Monte Carlo on Winkler-Pasternak soil Stochastic FEM, Monte Carlo pipeline, probabilistic, soil-variability
19 Sevieri & De Falco 2020 Dynamic SHM for concrete gravity dams via Bayesian inference Polynomial chaos + Bayesian updating enables real-time SHM of dams with low computational cost Bayesian, polynomial chaos, OMA SHM, dam, Bayesian
20 Shao et al. 2020 Sub-synchronous oscillation in direct-drive wind farms with VSC-HVDC SSO driven by grid-side converter and rectifier controllers; eigenvalue analysis clarifies mechanism Eigenvalue, time-domain simulation wind-farm, oscillation, power-electronics
21 Sharif et al. 2020 DEM-based CPT method for rotary-installed pile requirements DEM simulation of CPT linked to installation torque/force for rotary piles; enables CPT-based design DEM, CPT correlation geotechnical, pile-installation, DEM
22 Silva & Tsuha 2020 Model-scale deep helical piles in very dense sand Installation torque correlates with helix bearing capacity; deep embedment effects quantified experimentally 1-g model tests helical-pile, sand, installation
23 Song et al. 2020 Real-time hybrid simulation for monopile OWT Conceptual RTHS framework partitions OWT into physical (foundation) and numerical (tower+turbine) substructures RTHS framework, FAST offshore-wind, monopile, hybrid-testing
24 Stapelfeldt et al. 2020 Drainage regime effect on suction caisson cyclic loading Pore-fluid drainage is the key factor governing cyclic response; installation pumping rate less influential Centrifuge testing, CPT suction-caisson, cyclic, centrifuge
25 Stieng & Muskulus 2020 Reliability-based design optimization of OWT support structures Factorized uncertainty decouples reliability from optimization; gradient-based RBDO feasible with modest extra cost RBDO, Gaussian process surrogate offshore-wind, reliability, optimization
26 Sun et al. 2020 Multilevel DL for county-level corn yield Coupled CNN-LSTM integrating time-series + constant data outperforms single-model approaches for yield prediction CNN + LSTM, remote sensing deep-learning, agriculture, yield
27 Tarpo 2020 Stress estimation of offshore structures (PhD thesis) Modal-based stress estimation from OMA data enables fatigue assessment of offshore structures without full instrumentation OMA, modal expansion SHM, offshore, stress-estimation
28 Tartakovsky et al. 2020 PIDNNs for subsurface flow parameters (Garbled PDF -- insufficient text extracted) PINN PINN, subsurface, constitutive
29 Tavakoli et al. 2020 Autoencoder-based time series clustering Two-stage method (volatility-label creation + autoencoder) achieves 87.5% clustering accuracy on 70+ stock indices Autoencoder, K-means deep-learning, time-series, clustering
30 Tonk et al. 2020 Flat oyster settlement on OWF scour protection Eco-friendly scour protection in Borssele V supports oyster settlement; dual ecological-engineering benefit Field experiment, Borssele V offshore-wind, ecology, scour-protection
31 Tsang et al. 2020 Geotechnical seismic isolation with rubber-soil mixtures Rubber-soil mixture layer reduces seismic demand on structures; validated by centrifuge testing Centrifuge testing seismic-isolation, centrifuge, geotechnical
32 Vasilchuk & Yashnov 2020 Diagnostics of bridge scour by dynamic parameters Natural frequency shift detects and locates scour at bridge piers; validated by FEM and field accelerometers FEM (Midas Civil), field sensors scour, SHM, bridge, frequency
33 Wahab et al. 2020 Condition assessment of aged fixed offshore platforms Pushover analysis dominant for reassessment; reserve strength quantifies residual capacity for life extension/decommissioning Review, pushover analysis offshore, decommissioning, assessment
34 Wang & Yin 2020 DEM micro-mechanics of caisson in sand with particle breakage Particle breakage significantly affects caisson capacity and load-displacement in sand DEM with breakage caisson, DEM, particle-breakage
35 Wang & Giaralis 2020 Top-storey softening + TMDI for vortex-shedding mitigation Top-storey softening reduces required TMDI mass while improving peak acceleration under vortex shedding Optimal tuning, stochastic wind vibration-control, TMDI, tall-building
36 Wines 2020 Sensitivity of slope stability to geotechnical inputs Rock density, strength parameters and pore pressures have greatest influence on FoS; 3-D modelling essential 3DEC, parametric study slope-stability, sensitivity, numerical
37 Winkelmann et al. 2020 Reliability-based cliff stability via combined RSM Combined RSM (PEM + MC + FEM) efficiently estimates failure probability for multi-layered cliff RSM, Monte Carlo, PEM, FEM probabilistic, slope, reliability
38 Wu et al. 2020 Bearing capacity of pipes buried in sand Bearing capacity formulas adequate if using friction angle corrected for non-associativity; loose backfill offers no benefit over medium-dense Physical model tests pipeline, bearing-capacity, sand
39 Yamashkin et al. 2020 GeoSystemNet for remote sensing with scarce labels Geosystem-based augmentation improves DL classification by 5-9% in label-scarce conditions CNN + geosystem augmentation deep-learning, remote-sensing, augmentation
40 Yang et al. 2020 TCT for scour reduction around monopile Tidal current turbine placed upstream reduces monopile scour while generating electricity; dual benefit CFD + laboratory tests scour, monopile, tidal-energy

1. CONSENSUS

  • Soil-structure interaction dominates geotechnical performance. Across foundations (IABs, suction caissons, piles, pipelines), the interaction between structure and surrounding soil -- especially pore pressure regime, drainage, and spatial variability of soil properties -- is the primary driver of system response (Naji, Stapelfeldt, Seguini, Wu, Wines).
  • Deep learning requires large datasets to outperform classical methods. Both Ng et al. and Schulz et al. independently demonstrate that CNN/DL models only surpass linear or conventional ML above a threshold sample size (~1000--2000). Below that threshold, simpler models are competitive or superior.
  • Physics-informed approaches improve DL generalization. Ren (SWINet) and Tartakovsky (PIDNNs) embed governing equations into neural networks, reducing dependence on labeled data and improving physical consistency -- a converging theme across geophysics and subsurface domains.
  • OWT foundation design must account for combined, multi-phase loading. Roach, Stieng, Song, and Stapelfeldt all emphasize that realistic load modeling (hurricane, cyclic, wind+wave coupling) and uncertainty quantification are essential for monopile and suction caisson design.
  • Scour at foundations remains a critical concern for offshore structures. Yang, Tonk, and Vasilchuk address scour from different angles (mitigation via TCT, eco-protection, SHM detection), confirming its status as a persistent engineering problem.

2. DEBATES

  • When does deep learning add value over classical ML? Ng et al. argue CNN is clearly superior above ~2000 training samples for spectral soil data; Schulz et al. counter that for brain-imaging phenotypes, linear models match DL even at ~10k samples. The discrepancy hinges on whether the underlying data relationships are sufficiently nonlinear -- a question that remains domain-specific and unresolved.
  • Deterministic vs. probabilistic foundation design. Stieng and Winkelmann advocate reliability-based methods (RBDO, RSM+MC) that explicitly propagate uncertainty, while most practicing codes (Roach, Razouki) rely on deterministic safety factors. The practical threshold where probabilistic methods become cost-justified vs. conservative SFs is contested.
  • Passive vs. active vibration control. Rahimi et al. document that passive TMDs are widely adopted due to simplicity, yet active/hybrid TMDs offer superior performance under broadband excitation. Wang & Giaralis propose a middle path (structural modification + TMDI) that reduces hardware mass. No consensus exists on optimal cost-performance trade-off.

3. GAPS

  • Acoustic impact data for most marine species across OWF lifecycle phases are missing. Mooney et al. explicitly flag data deficiency for non-charismatic taxa and for survey/decommissioning phases.
  • No experimental validation of physics-informed NNs on real-world (non-synthetic) geophysical data. Ren's SWINet and Tartakovsky's PIDNNs are demonstrated on numerical benchmarks only.
  • Suction caisson behavior under combined V-H-M cyclic loading in sand is poorly characterized. Stapelfeldt's centrifuge tests address vertical cycling only; lateral and moment cycling in dense sand remain sparse.
  • Reliability-based optimization has not been applied to floating OWT foundations. Stieng demonstrates RBDO for fixed monopiles; extension to floating systems (spar, semi-sub) with mooring uncertainty is absent.
  • Bridge scour SHM lacks validated thresholds for intervention. Vasilchuk shows frequency-based detection is feasible but does not propose quantitative alert criteria tied to structural safety margins.

4. METHODS INVENTORY

Category Methods Used Papers
Numerical (FEM/FDM) Mohr-Coulomb, Cam-Clay, Midas Civil, 3DEC, ANSYS, FAST Ouabel, Paucar, Olsen, Wines, Vasilchuk, Rahman, Roach
Discrete Element Method DEM with particle breakage, DEM-CPT correlation Sharif, Wang & Yin
Centrifuge / physical model 1-g model tests, geotechnical centrifuge Stapelfeldt, Tsang, Wu, Silva
Deep learning CNN, LSTM, autoencoder, GA-DNN, PLSR, Cubist Ng, Pham, Sun, Tavakoli, Neupane, Yamashkin
Physics-informed NN SWINet, PIDNNs Ren, Tartakovsky
Probabilistic / reliability Bayesian inference, Monte Carlo, RSM, RBDO, stochastic FEM Sevieri, Winkelmann, Stieng, Seguini
Review / synthesis Systematic literature review Mooney, Naji, Neupane, Rahimi, Wahab, Peyghami
Field monitoring Inclinometer, accelerometer, OMA Paucar, Vasilchuk, Tarpo, Tonk

5. BENCHMARKS AND DATASETS

  • CWRU bearing dataset -- de facto standard for machinery fault detection DL benchmarks (Neupane & Seok).
  • Brazilian VIS-NIR-SWIR soil spectra library -- 12,044 samples / 4,251 sites used to benchmark CNN vs. PLSR/Cubist (Ng et al.).
  • OC3 Monopile reference model -- used for RBDO demonstration (Stieng & Muskulus).
  • IEC 61400-3-1 DLCs + Massachusetts OWE metocean data -- reference site for US offshore standard application (Roach et al.).
  • 472-pile driven pile database -- static load test records for GA-DNN bearing capacity model (Pham et al.).
  • EuroSAT dataset -- satellite image classification benchmark augmented via geosystem approach (Yamashkin et al.).
  • Bodendorf reservoir (Austria) -- field measurement baseline for bank-failure numerical model (Olsen & Haun).
  • Borssele V wind farm -- ecological scour protection field experiment (Tonk et al.).