Full-field Model Research Articles

An efficient surrogate model for damage forecasting of composite laminates based on deep learning

In this paper, full-field damage forecasting of a laminated composite structure under different low velocity impact (LVI) conditions is realized through the proposed surrogate model, named VQ-SM. First, an efficient surrogate modelling method is proposed based on the advanced Vector Quantised-Variational AutoEncoder (VQ-VAE) proposed by DeepMind. Second, numerical simulation based on the progressive damage model of composite laminates is performed to obtain the training dataset. After training, the performance of VQ-SM is evaluated compared to the surrogate model without a representation learning process. The results show that VQ-SM has better performance with high-precise and good robustness, trained on the small dataset. Finally, the impact damage field of composite laminates is analyzed based on the surrogate model. The proposed surrogate modelling method provides not only the full-field damage forecast model for composite structures, but also an efficient method of improving the performance of the “generative” surrogate model.

Calibrating the Reservoir Model of the Garraf Oil Field

History matching is a significant stage in reservoir modeling for evaluating past reservoir performance and predicting future behavior. This paper is primarily focused on the calibration of the dynamic reservoir model for the Meshrif formation, which is the main reservoir in the Garraf oilfield. A full-field reservoir model with 110 producing wells is constructed using a comprehensive dataset that includes geological, pressure-volume-temperature (PVT), and rock property information. The resulting 3D geologic model provides detailed information on water saturation, permeability, porosity, and net thickness to gross thickness for each grid cell, and forms the basis for constructing the dynamic reservoir model. The dynamic reservoir model integrates a variety of inputs, including well position and trajectory, well completion data, initial reservoir condition, and daily production/injection rates. The validation process involves comparing the original oil reserve derived from the geological model with the one obtained from the dynamic reservoir model. To achieve an accurate history matching, the calibration process has been performed by aligning observed data with simulation results. This involves focusing on production/injection data for each well and pressure measurements for selected wells. Notably, horizontal permeability is identified as a critical parameter in this study, which is adjusted iteratively to achieve a robust match for individual wells and the entire field. Thus, Successful calibration facilitates the subsequent stage and future scenarios allowing for the exploration of different conditions to predict the performance of the Garraf oilfield. This comprehensive approach improves the reliability of reservoir predictions, facilitating well-informed decision-making in reservoir management.

Random distribution of interphase characteristics on the overall electro-mechanical properties of CNT piezo nanocomposite: Micromechanical modeling and Monte Carlo simulation

A phenomenological study is carried out to speculate the statistical impacts of the CNT/polymer interphase on the overall electro-elastic behavior of piezo-polymer nanocomposites by presenting a full-field micromechanical model. The nanocomposite system consists of carbon nanotube (CNT) and PVDF. Various statistical distributions, including Weibull, log-normal, normal, beta, and uniform distributions on the thickness and strength of the interphase are carefully assessed. The results are compared with experimental data, and satisfactory agreements are reported. It is found that, compared to the random distribution of the interphase strength, the statistical distribution of the interphase thickness has more effect on the overall electro-elastic properties. For example, for the effective longitudinal modulus, the overall coefficients of variation are 14 %, 13 %, 13.56, and 10 %, respectively, for the normal, Weibull, beta, and uniform distributions of the thickness compared with the measured experimental values. Also, the effects of the CNT content, aspect ratio, and orientation on the effective electro-elastic properties by considering the various random distributions are fully examined. Moreover, using the Monte Carlo simulation, the probability of not meeting design specification (failure probability) is evaluated at the random distributions of the interphase strength and thickness to identify the optimum CNT content for which the values of the overall properties are maximum. It is obtained that the failure probabilities are different for 5–8 % CNT volume fraction in the distributions of the thickness, and for only 5 VF% CNT in the strength distributions. For other values of the CNT content, the failure probabilities are independent of the distribution of the interphase strength and thickness.

GPU-accelerated full-field modelling of highly dispersive ultrafast optical parametric oscillators

We demonstrate GPU-accelerated modelling of ultrafast optical parametric oscillators (OPOs) via the χ(2) nonlinear envelope equation with 1265× improvement in execution time compared with a CPU-based approach. Incorporating an adaptive step-size algorithm and absorbing boundary conditions, our model is capable of simulating OPOs containing long (>10 mm) nonlinear crystals or significant intracavity dispersion with outputs generated in less than 1 minute, allowing the investigation of systems that were previously computationally prohibitive to explore. We implement real-world parameters such as optical coatings, material absorption, and non-ideal poling domains within quasi-phase matched nonlinear crystals, producing excellent agreement with the spectral tuning behaviour and average power from a previously reported prism-based OPO. Our digital twinning approach provides a low-cost iterative development platform for ultrafast OPOs.

A macroscopic viscoelastic viscoplastic constitutive model for porous polymers under multiaxial loading conditions

A macroscopic constitutive model, the Porous Eindhoven Glass Polymer (Porous EGP) model, is presented to describe the deformation behavior of cavitated rubber toughened polymers under multiaxial loading conditions. It is shown that the proposed macroscopic constitutive model is able to describe the non-linear pre-yield regime, strain rate dependence, post-yield behavior (strain softening and hardening) and void evolution for loading conditions ranging from shear to equi-triaxial (pure triaxial) tension and compression. The Porous EGP model is a combination of a well established non-linear viscoelastic viscoplastic model, the Eindhoven Glassy Polymer (EGP) model, and the modified Gurson model. The Gurson model is adopted to determine the equivalent stress and plastic rate of deformation tensor making it depending on the void volume fraction, deviatoric and hydrostatic stress. The macroscopic constitutive model is developed based on the response of realistic 3D representative volume elements (RVEs) containing randomly positioned mono-disperse inclusions. The constitutive behavior of the matrix phase in this full-field model is described by the EGP model, and the cavitated inclusions are idealized as voids. Their response is studied for a range of void volume fractions, multiaxial loading conditions, strain rates and thermodynamic states. The yield behavior of the heterogeneous material depends non-linearly on the macroscopic hydrostatic stress. This response is well captured with the proposed macroscopic constitutive model.

Prediction of fracture evolution in the TiN/Al thin films based on a full-field modelling approach

The classical approach to fracture simulation in a thin film for bioengineering applications neglects the underlying complex morphology of the deposited layer, leading to inaccurate numerical predictions. That may carry high risks during product development and its further exploitation, especially in biomedical applications. Therefore, here, we demonstrate the new approach to numerical investigation of the fracture evolution in pulsed laser-deposited (PLD) titanium nitride (TiN) thin films under loading conditions based on the full-field model of layer columnar morphology. We use the TiN film deposited on the aluminium (Al) substrate as a case study. Based on the nanoindentation test, we first determine the flow stress characteristics of the Al substrate and TiN/Al structure. An inverse analysis technique allowed us to precisely recalculate measured load–displacement values into the required stress–strain curve. Then, we developed a full-field model based on the digital material representation (DMR) concept for further investigation. The DMR was generated to replicate major morphological features of the deposited thin film identified under transmission electron microscop (TEM). Finally, we incorporated the TiN/Al full-field model into the finite element analysis with a cohesive zone approach for local analysis of fracture behaviour. Inverse analysis based on the developed direct model and experimental measurements allowed us to identify the parameters of the fracture model. As a result, we proved that the full-field model based on the digital material representation concept could be used for reliable predictions of local fracture development along the morphological features of the deposited thin film.

Embryo-to-lamella transition of grain boundary twins in magnesium

A combined experimental and computational analysis is performed to investigate the less commonly studied embryo-to-lamella transition of deformation twins in magnesium. This work aims to understand the structural variables controlling the embryo-to-lamella transition from grain boundaries. Statistical analysis of hundreds of early-stage twins in the lightly deformed microstructure reveals a prevailing wedge shape, with a much thicker base along the grain boundary (GB) where they originate and a thinner tip terminating in the crystal. The analysis also shows that the GB base is super thick and identifies a minimum GB twin thickness among all early-stage twins that is about one micron. A crystal plasticity-based full-field model is employed to calculate the driving forces to migrate the boundary of a three-dimensional GB twin embryo. The stress analysis, considering a full range of embryo shapes and neighboring grain orientations, indicate that the twin embryo is most likely going to form a wedge shape when it first propagates. The calculations predict that the thickness of the embryo at the GB needs to be significantly larger than its length into the crystal in order to propagate into the crystal. The analysis finds that the more aligned the twin embryo variant is with basal slip in the neighboring grain, the thinner the twin embryo needed for propagation.

Layout Optimization Algorithms for the Offshore Wind Farm with Different Densities Using a Full-Field Wake Model

To decrease the power deficit of a wind farm caused by wake effects, the layout optimization is a feasible way for the wind farm design stage. A suitable optimization algorithm can significantly improve the quality and efficiency of the optimization process. For exploring the high-performance algorithms under different layout densities, a comparison is conducted by optimizing the layout of a real offshore wind farm with five algorithms, namely two population-based algorithms and three single-point algorithms. Wake effects are considered by a full-field wake model. A penalty function is proposed for the population-based algorithms to handle the constraint violations. Different iterations and constraints of the layout density are applied in the optimization. The random search has the best optimization results in all the cases and the control of the feasibility check is necessary for this algorithm. More iterations can advance the optimization results. The density constraint greatly affects the computational cost of the random search, which is significantly increased under the strict constraint. Except under the strict constraint, the random search has the best performance of optimization efficiency. A combination of the pattern search and random search is recommended when the strict constraint is applied in the layout optimization.

Layout optimization of an offshore floating wind farm deployed with novel multi-turbine platforms with the self-adaptive property

To deploy offshore wind in deeper waters, floating offshore wind confronts simultaneously significant opportunities and challenges. For dealing with the power deficit caused by the wake effect, a layout optimization method has been proposed for wind farms under variable wind conditions using the full-field wake model with the capacity to deploy multi-turbine platforms. The joint distribution of wind speed and direction is utilized to consider the variable wind conditions. The layout of an offshore wind farm, comprised of the floating platforms with multiple wind turbines and the wind-orientated ability, i.e., the self-adaptive property, is optimized under different wind conditions. Different numbers of turbines on a platform and platforms in the wind farm are deployed, respectively. The optimization results confirm that the platform with the self-adaptive property has a remarkable advantage to improve the efficiency of the wind farm. More turbines on a platform and less platforms can obtain higher efficiency. The distribution of wind speed and wind direction impacts the optimized layout and power output of the wind farm, simultaneously. The platforms arrange parallel lines which are perpendicular to the wind direction with more frequency and higher speed in the optimized wind farm. The concentration of the wind direction and the orthogonal angle between the wind and the long edge of the wind farm can increase the power output of the wind farm. However, the distribution of the wind speed has the dominant influence on the power output.

A non-contact inspection method of tile debonding using tuned acoustic wave and laser doppler vibrometer

The debonding of tiles from walls is currently a critical issue in modern cities. This study develops a non-contact acoustic approach integrating an effective numerical model for accurate inspection of tile debonding from a wall, with the aid of a directional sound source and a laser doppler vibrometer (LDV). Typical engineering tile debonding cases are characterised and analysed using the non-contact inspection method. Numerical modelling based on full-field modelling is performed to simulate and predict the inspection process. In both experiments and simulations, the acoustic wave with a sweeping frequency band is generated by the sound source and directed to the tiles. Activated vibration signals are captured at multiple points on the tile surface and then processed, and an out-of-surface velocity map of the tile is plotted to identify the debonding area, where good agreements between the numerical and experimental results can be seen. The results show that different shapes of the debonding areas under the tiles can be well detected, proving the effectiveness of the method.

Microplasticity and macroplasticity behavior of additively manufactured Al-Mg-Sc-Zr alloys: In-situ experiment and modeling

Understanding and controlling the performance of additively manufactured aluminum alloys containing scandium (Sc) and zirconium (Zr) elements heavily relies on knowledge of their microplasticity and macroplasticity behavior. However, this aspect has received very little attention. In this investigation, we examined the microplasticity and macroplasticity behavior of additively manufactured Al-Mg-Sc-Zr alloys before and after aging, using in-situ synchrotron X-ray diffraction and full-field crystal plasticity modeling. Our study provides a quantitative assessment of the transitions from elasticity to microplasticity and then to macroplasticity and analyzes the development of the initial microstructure, particularly the dislocations. We constructed crystal plasticity fast-Fourier-transform models based on dislocation densities. The predicted evolutions of macroscopic stress-strain curves, lattice strains, and dislocation densities agree with in-situ measurements. The present findings provide deep insights into controlling the performance of AM Al-Mg-Sc-Zr alloys. Besides, the micromechanical model developed in this investigation paves the way for predicting the microplasticity and macroplasticity behavior of various metallic materials.

Machine learning-enabled identification of micromechanical stress and strain hotspots predicted via dislocation density-based crystal plasticity simulations

The present work uses a full-field crystal plasticity model with a first principles-informed dislocation density (DD) hardening law to identify the key microstructural features correlated with micromechanical fields localization, or hotspots, in polycrystalline Ni. An ensemble learning approach to machine learning interpreted with Shapley additive explanation was implemented to predict nonlinear correlations between microstructural features and micromechanical stress and strain hotspots. Results reveal that regions within the microstructure in the vicinity of grain boundaries, higher Schmid factors, low slip transmissions and high intergranular misorientations, are more prone to being micromechanical hotspots. Additionally, under combined loading and large plastic deformations, slip transmissions take precedence over intergranular misorientations in formation of both strain and stress hotspots. The present work demonstrates a successful integration of physics-based crystal plasticity with DD-based hardening into machine learning models to reveal the microscale features responsible for the formation of local stress and strain hotspots within the grains and near the grain boundaries, as function of applied deformation states, grain morphology/size distribution, and microstructural texture, providing insights into micromechanical damage initiation zones in polycrystalline metals.

Multiscale identification and NTFA reduction of the elasto-viscoplastic polycrystalline behaviour of uranium dioxide (UO2) for wide range of loading conditions

This numerical study presents a micromechanical modelling of the behaviour of uranium dioxide fuel (UO2), a polycrystalline ceramic used in nuclear power plants, for loading condition typical of a reactivity initiated accident (RIA). Above a specific temperature, UO2 has an elasto-viscoplastic behaviour with strain hardening depending on both temperature and macroscopic strain rate (T,ε¯˙).In a first step, via a full-field approach, an inverse identification of the local single crystal evolution law is performed based on experimental data obtained at the macroscopic scale [4,5]. Beyond the macroscopic behaviour of the fuel pellet, it is sometimes necessary to go down in scale to study local phenomena. A first approach would be to perform a Finite Element Method squared to two (FEM2) resolution [37]. It consists to solve a thermomechanical problem on a first mesh defined at the pellet scale where at each integration point is solved another thermomechanical problem on a Representative Volume Element. The full-field approach guarantees access to the local field at the expense of relatively high computational time. It is not conceivable within industrial codes.In a second step it appeared necessary to replace the second full-field resolution by a model reduction method based on the Nonuniform Transformation Field Analysis (NTFA) approach. This method allows access to macroscopic quantities and local fields while drastically reducing the computational time. The NTFA model [21] with tangent second order (TSO) approximation [24,21] is developed and applied to the problem at hand. Two decompositions of the hardening variable are studied. It is taken as in the reference [21], as uniform per grain or taken as decomposed, like modes associated with the viscoplastic strain tensor. Then, work is done to integrate the specific loading conditions, i.e. temperatures and strain rates of the RIA, into the NTFA model while keeping the number of internal variables as low as possible. Finally, the results from the experimental, full-field and NTFA TSO models are compared. It is checked that the numerical results are not deteriorated either macroscopically or locally at the macroscopic and local scales for uniaxial and triaxial loading.

By Visualizing the Deformation with Mechanoluminescent Particles, Additive Manufacturing Offers a Practical Alternative to Stress and Strain Simulation

The use of stress–strain analysis in structural design or mechanical components is critical for avoiding or investigating structural failures. In the case of complicated designs, mathematical full-field stress modeling produces imprecise predictions. Experimental analysis can be used as a replacement for mathematical modeling, but with the use of currently available strain gauges, it is cumbersome and impossible in the case of moving parts. Mechanoluminescent materials transform mechanical energy into visible light and can be used as a replacement for strain gauges to monitor strain/stress. Three-dimensional printing technology has made major advances in terms of additive manufacturing. In this article, we describe a method to produce an ML 3D print. The fabricated samples are precise and versatile and satisfy the need for easy and non-destructible spatial stress analysis. A 3D printed photopolymer sample with SrAl2O4: Eu, Dy particle addition only to the final layers was tested, and the number of layers was optimized. It was determined that the optimal number of layers for easy detection is in the range of 10 to 20 layers. It opens the possibility for the real-time evaluation of complex uneven forces on complex parts, thus having a good potential for commercialization.

Modeling Thermoelasticity of HCP single crystals using a nonlocal discrete approach

This paper presents full-field modeling Thermoelasticity of hexagonal close-packed (HCP) single crystals using reformulations of conventional continuum theories based on a novel nonlocal lattice particle method (LPM). In LPM, grain domain is modeled as an assembly of regularly packed material particles according to the atomic lattice. The interaction between material particles is nonlocal, e.g., for mechanical problems, a material particle interacts with neighboring material particles up to certain distance via bonds and the interaction between two material particles depends on the deformation states of not only the two material particles but also all their neighbors. To determine the nonlocal interaction between material particles, a top-down approach based on the classical continuum theories is developed. Equivalency assumptions of energy, heat transfer rate and thermal strain between LPM and its continuum counterpart are made to consistently determine the material particle interaction for mechanical, thermal, and thermal–mechanical coupling problems, respectively. To capture the crystallographic orientation, a lattice rotation scheme that is equivalent to the coordinate transformation used in the classical continuum theories is adopted. Numerical studies regarding validity of the proposed nonlocal discrete reformulation and accuracy of the model prediction are conducted for all three types of analysis. Good agreements are observed between the results from the developed reformulations and the conventional continuum theories.

Technology Focus: Heavy Oil (April 2023)

_ Heavy oils are characterized by high-density, high-viscosity, and high-heavy-fraction components. One of the more common characteristics of heavy oil reservoirs is a low primary recovery factor, which is mainly because of unfavorable mobility ratios between oil and water, negligible solution drives, and faster decline of reservoir pressures because of relatively low oil compressibility. Most of the technologies that apply to heavy oil reservoirs need to address the mobility ratio or viscous forces before any flooding. Given the complex nature of heavy oil reservoirs, paper SPE 209328 present a comprehensive review of the latest technologies and work flows developed for heavy oil reservoir management. Although secondary oil production methods such as waterflooding, water-alternating-gas injection, and gas drive might not be efficient in recovery from heavy oil reservoirs, the applicability and success of these methods can be analyzed on a reservoir basis. The offtake and throughput management philosophy for heavy oil waterflooding is not the same as for classical light oil, and it could play a major role in recovery efficiency. The key point is to understand the nature of heavy oil and reservoir heterogeneity regarding their fluid and rock properties. Recent technology advances in hardware and software have enabled the industry with tools that allow achieving full-field high-resolution simulation modeling; therefore, performing a realistic reservoir simulation is achievable. This might decrease the loss of millions of dollars before field applications. Paper SPE 210185 presents an example of taking advantage of recent simulation technologies and presents a method and work flow under subsurface uncertainty for a large heavy oil field with complex geological characteristics, a long production history, and many wells. The work-flow design is applicable for other fields with similar characteristics and delivery objectives. Paper SPE 203783 discusses the results of a recovery-optimization study conducted in a heavy oil field in south Oman aimed at improving the understanding of deep bottom-up water injection. Recommended additional reading SPE 205449 First Successful Controlled Dumpflood in Deepwater Gulf of Mexico Results in Promising Incremental Rate and Recovery by Bilal A. Hakim, Talos Energy, et al. URTeC 2021-5077 Heavy Oil Polymer Enhanced Oil Recovery in the Challenging Alaskan Arctic—It Works! by A. Dandekar, University of Alaska Fairbanks, et al. SPE 212387 Evaluation of the Oil Recovery and Economic Benefit of the First-Ever Polymer flood Field Pilot To Enhance the Recovery of Heavy Oils on Alaska’s North Slope Using Machine-Assisted Reservoir Simulation by Cody Douglas Keith, University of Alaska Fairbanks

Sensitivity of local cyclic deformation in lath martensite to flow rule and slip system in crystal plasticity

The prediction of the cyclic deformation behaviour in lath martensite-based high-strength steels requires constitutive models that reflect the local stress and strain fields as accurately as possible. At the same time, the constitutive models should act as efficiently as possible in order to achieve the required high number of cycles in a finite time. Only few research works have studied the sensitivity of the local cyclic deformation in lath martensite to the power law-based flow rule (Hutchinson or Chaboche–Cailletaud) and the active body-centred cubic (bcc) slip systems ({110}〈111〉 and {112}〈111〉) in the crystal plasticity finite element method (CPFEM). This paper, therefore, aims to provide some guidance in the selection of suitable flow rule and slip systems. Based on full-field micromechanical modelling of a medium-carbon steel under symmetric strain-controlled cyclic loading, it can be shown that the two most commonly used flow rules according to Hutchinson and Chaboche–Cailletaud are equally capable of predicting the local stress and strain distributions within the hierarchical martensitic microstructure. However, using the Hutchinson flow rule increases the computational performance for the quasi-rate-independent problem considered here. The local distributions found differ strongly from those in the parent austenitic microstructure. If plastic deformation is assumed not only on the slip systems {110}〈111〉, as often done, but also on the {112}〈111〉 type, a redistribution of the bimodal distributed local stresses occurs at a significantly lower stress level. The unimodal distributed local strains are less affected by this. In addition, it is found that slightly different critical resolved shear stress (CRSS) values for both slip system types influence the local stress and strain distributions less severely than the additional plastic slip activation in the material.

Nonlinear finite element free and forced vibrations of cellular plates having lattice-type metamaterial cores: a strain gradient plate model approach

The main objective of this paper is to develop a theoretically and numerically reliable and efficient methodology based on combining a finite element method and a strain gradient shear deformation plate model accounting for the nonlinear free and forced vibrations of cellular plates having lattice-type metamaterial cores. The proposed model provides a modelling framework computationally more efficient than framework of the conventional three-dimensional elasticity for the simulation of the underlying nonlinear dynamics of cellular plates with microarchitectures. The corresponding governing equations follow Mindlin’s strain gradient elasticity theory including the micro-inertia effect applied to the first-order shear deformation plate theory along with the nonlinear von Kármán kinematics. Standard and higher-order computational homogenization methods determine the classical and strain gradient material constants, respectively. A higher-order C1-continuous 6-node finite element is adopted for the discretization in the spatial domain, and an arc-length continuation technique along with time-periodic discretization is implemented to solve the resulting nonlinear time-dependent problem. Through a set of comparative studies with 3D full-field models as reliable validation references for triangularly prismatic example cores, the accuracy and efficacy of the proposed dimension reduction methodology are demonstrated for a diverse range of problem parameters for analyzing the large-amplitude dynamic structural response. The results show that the proposed strain gradient plate model can accurately capture the microarchitecture effects of cellular plates with low computational costs, which signifies a relevant contribution particularly to the computational analysis of nonlinear dynamics.

Full Field Model Describing Phase Front Propagation, Transformation Strains, Chemical Partitioning, and Diffusion in Solid–Solid Phase Transformations

AbstractA novel mathematical formulation is presented for describing growth of phase in solid‐to‐solid phase transformations and it is applied for describing austenite to ferrite transformation. The formulation includes the effects of transformation eigenstrains, the local strains, as well as partitioning and diffusion. In the current approach the phase front is modeled as diffuse field, and its propagation is shown to be described by the advection equation, which reduces to the level‐set equation when the transformation proceeds only to the interface normal direction. The propagation is considered as thermally activated process in the same way as in chemical reaction kinetics. In addition, connection to the Allen–Cahn equation is made. Numerical tests are conducted to check the mathematical model validity and to compare the current approach to sharp interface partitioning and diffusion model. The model operation is tested in isotropic 2D plane strain condition for austenite to ferrite transformation, where the transformation produces isotropic expansion, and also for austenite to bainite transformation, where the transformation causes invariant plane strain condition. Growth into surrounding isotropic austenite, as well as growth of the phase which has nucleated on a grain boundary are tested for both ferrite and bainite formation.

Coupled chemo-mechanical modeling of point-defect diffusion in a crystal plasticity fast Fourier transform framework

Below the yield strength and at moderate-to-high homologous temperatures, the inelastic deformation of metals is mostly governed/rate-controlled by vacancy diffusion-mediated processes. As a function of grain size, stress, temperature and dislocation content, vacancies (or atoms) can adopt preferential diffusion paths across grain interiors, along grain boundaries, or towards and along dislocations, resulting in climb and self-climb. In the steady state and under constant load, grain boundary and grain bulk vacancy diffusion-mediated plasticity have been described in seminal works by Coble and by Nabarro and Herring, respectively. Yet, the interplay between all aforementioned potential diffusion pathways has not been comprehensively mapped. This work presents a thermodynamically-consistent full-field model integrated within a voxel-based elasto-viscoplastic fast Fourier transform framework, which considers the coupling between the diffusion-mediated plasticity mechanisms. In the proposed approach, the kinetics and kinematics of plastic deformation due to vacancy diffusion along grain boundaries and grain bulk, as well as the exchange between grain boundaries and bulk are described explicitly. A homogenization approach at the voxel level is further introduced to simultaneously consider bulk and grain boundary diffusion in a numerically efficient fashion. The new formulation predicts the expected strain rate dependencies and the scaling of the steady-state creep rate with respect to grain size, temperature, and stress. The model predicts the transition from grain bulk to grain boundary-dominated diffusion with reduction in grain size, a significant step towards capturing transitions in deformation behavior without any phenomenological or ad-hoc adjustments.