SAM
https://sam.ensam.eu:443
The DSpace digital repository system captures, stores, indexes, preserves, and distributes digital research material.Thu, 01 Dec 2022 17:47:59 GMT2022-12-01T17:47:59ZSome Incipient Techniques For Improving Efficiency in Computational Mechanics
http://hdl.handle.net/10985/6477
Some Incipient Techniques For Improving Efficiency in Computational Mechanics
AMMAR, Amine; CHINESTA, Francisco
This contribution presents a review of different techniques available for alleviating simulation cost in computational mechanics. The first one is based on a separated representation of the unknown fields; the second one uses a model reduction based on the Karhunen-Loève decomposition within an adaptive scheme, and the last one is a mixed technique specially adapted for reducing models involving local singularities. These techniques can be applied in a large variety of models.
Tue, 01 Jan 2008 00:00:00 GMThttp://hdl.handle.net/10985/64772008-01-01T00:00:00ZAMMAR, AmineCHINESTA, FranciscoThis contribution presents a review of different techniques available for alleviating simulation cost in computational mechanics. The first one is based on a separated representation of the unknown fields; the second one uses a model reduction based on the Karhunen-Loève decomposition within an adaptive scheme, and the last one is a mixed technique specially adapted for reducing models involving local singularities. These techniques can be applied in a large variety of models.Advanced parametric space-frequency separated representations in structural dynamics: A harmonic–modal hybrid approach
http://hdl.handle.net/10985/13277
Advanced parametric space-frequency separated representations in structural dynamics: A harmonic–modal hybrid approach
MUHAMMAD HARIS, Malik; BORZACCHIELLO, Domenico; AGUADO, José Vicente; CHINESTA, Francisco
This paper is concerned with the solution to structural dynamics equations. The technique here presented is closely related to Harmonic Analysis, and therefore it is only concerned with the long-term forced response. Proper Generalized Decomposition (PGD) is used to compute space-frequency separated representations by considering the frequency as an extra coordinate. This formulation constitutes an alternative to classical methods such as Modal Analysis and it is especially advantageous when parametrized structural dynamics equations are of interest. In such case, there is no need to solve the parametrized eigenvalue problem and the space-time solution can be recovered with a Fourier inverse transform. The PGD solution is valid for any forcing term that can be written as a combination of the considered frequencies. Finally, the solution is available for any value of the parameter. When the problem involves frequency-dependent parameters the proposed technique provides a specially suitable method that becomes computationally more efficient when it is combined with a modal representation.
Mon, 01 Jan 2018 00:00:00 GMThttp://hdl.handle.net/10985/132772018-01-01T00:00:00ZMUHAMMAD HARIS, MalikBORZACCHIELLO, DomenicoAGUADO, José VicenteCHINESTA, FranciscoThis paper is concerned with the solution to structural dynamics equations. The technique here presented is closely related to Harmonic Analysis, and therefore it is only concerned with the long-term forced response. Proper Generalized Decomposition (PGD) is used to compute space-frequency separated representations by considering the frequency as an extra coordinate. This formulation constitutes an alternative to classical methods such as Modal Analysis and it is especially advantageous when parametrized structural dynamics equations are of interest. In such case, there is no need to solve the parametrized eigenvalue problem and the space-time solution can be recovered with a Fourier inverse transform. The PGD solution is valid for any forcing term that can be written as a combination of the considered frequencies. Finally, the solution is available for any value of the parameter. When the problem involves frequency-dependent parameters the proposed technique provides a specially suitable method that becomes computationally more efficient when it is combined with a modal representation.Parametric solutions involving geometry: A step towards efficient shape optimization
http://hdl.handle.net/10985/10244
Parametric solutions involving geometry: A step towards efficient shape optimization
AMMAR, Amine; HUERTA, Antonio; CHINESTA, Francisco; CUETO, Elias; LEYGUE, Adrien
Optimization of manufacturing processes or structures involves the optimal choice of many parameters (process parameters, material parameters or geometrical parameters). Usual strategies proceed by defining a trial choice of those parameters and then solving the resulting model. Then, an appropriate cost function is evaluated and its optimality checked. While the optimum is not reached, the process parameters should be updated by using an appropriate optimization procedure, and then the model must be solved again for the updated process parameters. Thus, a direct numerical solution is needed for each choice of the process parameters, with the subsequent impact on the computing time. In this work we focus on shape optimization that involves the appropriate choice of some parameters defining the problem geometry. The main objective of this work is to describe an original approach for computing an off-line parametric solution. That is, a solution able to include information for different parameter values and also allowing to compute readily the sensitivities. The curse of dimensionality is circumvented by invoking the Proper Generalized Decomposition (PGD) introduced in former works, which is applied here to compute geometrically parametrized solutions.
Wed, 01 Jan 2014 00:00:00 GMThttp://hdl.handle.net/10985/102442014-01-01T00:00:00ZAMMAR, AmineHUERTA, AntonioCHINESTA, FranciscoCUETO, EliasLEYGUE, AdrienOptimization of manufacturing processes or structures involves the optimal choice of many parameters (process parameters, material parameters or geometrical parameters). Usual strategies proceed by defining a trial choice of those parameters and then solving the resulting model. Then, an appropriate cost function is evaluated and its optimality checked. While the optimum is not reached, the process parameters should be updated by using an appropriate optimization procedure, and then the model must be solved again for the updated process parameters. Thus, a direct numerical solution is needed for each choice of the process parameters, with the subsequent impact on the computing time. In this work we focus on shape optimization that involves the appropriate choice of some parameters defining the problem geometry. The main objective of this work is to describe an original approach for computing an off-line parametric solution. That is, a solution able to include information for different parameter values and also allowing to compute readily the sensitivities. The curse of dimensionality is circumvented by invoking the Proper Generalized Decomposition (PGD) introduced in former works, which is applied here to compute geometrically parametrized solutions.Microscopic modelling of orientation kinematics of non-spherical particles suspended in confined flows using unilateral mechanics
http://hdl.handle.net/10985/13304
Microscopic modelling of orientation kinematics of non-spherical particles suspended in confined flows using unilateral mechanics
SCHEUER, Adrien; ABISSET-CHAVANNE, Emmanuelle; CHINESTA, Francisco; KEUNINGS, Roland
The properties of reinforced polymers strongly depend on the microstructural state, that is, the orientation state of the fibres suspended in the polymeric matrix, induced by the forming process. Understanding flow-induced anisotropy is thus a key element to optimize both materials and process. Despite the important progresses accomplished in the modelling and simulation of suspensions, few works addressed the fact that usual processing flows evolve in confined configurations, where particles characteristic lengths may be greater than the thickness of the narrow gaps in which the flow takes place. In those circumstances, orientation kinematics models proposed for unconfined flows must be extended to the confined case. In this short communication, we propose an alternative modelling framework based on the use of unilateral mechanics, consequently exhibiting a clear analogy with plasticity and contact mechanics. This framework allows us to revisit the motion of confined particles in Newtonian and non-Newtonian matrices. We also prove that the confined kinematics provided by this model are identical to those derived from microstructural approaches
Mon, 01 Jan 2018 00:00:00 GMThttp://hdl.handle.net/10985/133042018-01-01T00:00:00ZSCHEUER, AdrienABISSET-CHAVANNE, EmmanuelleCHINESTA, FranciscoKEUNINGS, RolandThe properties of reinforced polymers strongly depend on the microstructural state, that is, the orientation state of the fibres suspended in the polymeric matrix, induced by the forming process. Understanding flow-induced anisotropy is thus a key element to optimize both materials and process. Despite the important progresses accomplished in the modelling and simulation of suspensions, few works addressed the fact that usual processing flows evolve in confined configurations, where particles characteristic lengths may be greater than the thickness of the narrow gaps in which the flow takes place. In those circumstances, orientation kinematics models proposed for unconfined flows must be extended to the confined case. In this short communication, we propose an alternative modelling framework based on the use of unilateral mechanics, consequently exhibiting a clear analogy with plasticity and contact mechanics. This framework allows us to revisit the motion of confined particles in Newtonian and non-Newtonian matrices. We also prove that the confined kinematics provided by this model are identical to those derived from microstructural approachesKinetic theory of colloidal suspensions: morphology, rheology, and migration
http://hdl.handle.net/10985/10262
Kinetic theory of colloidal suspensions: morphology, rheology, and migration
GRMELA, Miroslav; MAITREJEAN, Guillaume; CHINESTA, Francisco; AMMAR, Amine
Smoluchowski kinetic equation governing the time evolution of the pair correlation function of rigid sphericalparticles suspended in a Newtonian fluid is extended to include particle migration. The extended kinetic equation takes into account three types of forces acting on the suspended particles: a direct force generated by an interparticle potential, hydrodynamic force mediated by the host fluid, and the Faxén-type forces bringing about the across-the-streamline particle migration. For suspensions subjected to externally imposed flows, the kinetic equation is solved numerically by the proper generalized decomposition method. The imposed flow investigated inthe numerical illustrations is the Poiseuille flow. Numerical solutions provide the morphology (the pair correlation function), the rheology (the stress tensor), and the particle migration.
Tue, 01 Jan 2013 00:00:00 GMThttp://hdl.handle.net/10985/102622013-01-01T00:00:00ZGRMELA, MiroslavMAITREJEAN, GuillaumeCHINESTA, FranciscoAMMAR, AmineSmoluchowski kinetic equation governing the time evolution of the pair correlation function of rigid sphericalparticles suspended in a Newtonian fluid is extended to include particle migration. The extended kinetic equation takes into account three types of forces acting on the suspended particles: a direct force generated by an interparticle potential, hydrodynamic force mediated by the host fluid, and the Faxén-type forces bringing about the across-the-streamline particle migration. For suspensions subjected to externally imposed flows, the kinetic equation is solved numerically by the proper generalized decomposition method. The imposed flow investigated inthe numerical illustrations is the Poiseuille flow. Numerical solutions provide the morphology (the pair correlation function), the rheology (the stress tensor), and the particle migration.On the solution of the heat equation in very thin tapes
http://hdl.handle.net/10985/14857
On the solution of the heat equation in very thin tapes
PRULIERE, Etienne; CHINESTA, Francisco; AMMAR, Amine; LEYGUE, Adrien; POITOU, Arnaud
This paper addresses two issues usually encountered when simulating thermal processes in forming processes involving tape-type geometries, as is the case of tape or tow placement, surface treatments, / The first issue concerns the necessity of solving the transient model a huge number of times because the thermal loads are moving very fast on the surface of the part and the thermal model is usually non-linear. The second issue concerns the degenerate geometry that we consider in which the thickness is usually much lower than the in-plane characteristic length. The solution of such 3D models involving fine meshes in all the directions becomes rapidly intractable despite the huge recent progresses in computer sciences. In this paper we propose to consider a fully space-time separated representation of the unknown field. This choice allows circumventing both issues allowing the solution of extremely fine models very fast, sometimes in real time.
Sun, 01 Jan 2012 00:00:00 GMThttp://hdl.handle.net/10985/148572012-01-01T00:00:00ZPRULIERE, EtienneCHINESTA, FranciscoAMMAR, AmineLEYGUE, AdrienPOITOU, ArnaudThis paper addresses two issues usually encountered when simulating thermal processes in forming processes involving tape-type geometries, as is the case of tape or tow placement, surface treatments, / The first issue concerns the necessity of solving the transient model a huge number of times because the thermal loads are moving very fast on the surface of the part and the thermal model is usually non-linear. The second issue concerns the degenerate geometry that we consider in which the thickness is usually much lower than the in-plane characteristic length. The solution of such 3D models involving fine meshes in all the directions becomes rapidly intractable despite the huge recent progresses in computer sciences. In this paper we propose to consider a fully space-time separated representation of the unknown field. This choice allows circumventing both issues allowing the solution of extremely fine models very fast, sometimes in real time.Non-intrusive Sparse Subspace Learning for Parametrized Problems
http://hdl.handle.net/10985/18435
Non-intrusive Sparse Subspace Learning for Parametrized Problems
BORZACCHIELLO, Domenico; AGUADO, José Vicente; CHINESTA, Francisco
We discuss the use of hierarchical collocation to approximate the numerical solution of parametric models. With respect to traditional projection-based reduced order modeling, the use of a collocation enables non-intrusive approach based on sparse adaptive sampling of the parametric space. This allows to recover the low-dimensional structure of the parametric solution subspace while also learning the functional dependency from the parameters in explicit form. A sparse low-rank approximate tensor representation of the parametric solution can be built through an incremental strategy that only needs to have access to the output of a deterministic solver. Non-intrusiveness makes this approach straightforwardly applicable to challenging problems characterized by nonlinearity or non affine weak forms. As we show in the various examples presented in the paper, the method can be interfaced with no particular effort to existing third party simulation software making the proposed approach particularly appealing and adapted to practical engineering problems of industrial interest.
Tue, 01 Jan 2019 00:00:00 GMThttp://hdl.handle.net/10985/184352019-01-01T00:00:00ZBORZACCHIELLO, DomenicoAGUADO, José VicenteCHINESTA, FranciscoWe discuss the use of hierarchical collocation to approximate the numerical solution of parametric models. With respect to traditional projection-based reduced order modeling, the use of a collocation enables non-intrusive approach based on sparse adaptive sampling of the parametric space. This allows to recover the low-dimensional structure of the parametric solution subspace while also learning the functional dependency from the parameters in explicit form. A sparse low-rank approximate tensor representation of the parametric solution can be built through an incremental strategy that only needs to have access to the output of a deterministic solver. Non-intrusiveness makes this approach straightforwardly applicable to challenging problems characterized by nonlinearity or non affine weak forms. As we show in the various examples presented in the paper, the method can be interfaced with no particular effort to existing third party simulation software making the proposed approach particularly appealing and adapted to practical engineering problems of industrial interest.Manifold embedding of heterogeneity in permeability of a woven fabric for optimization of the VARTM process
http://hdl.handle.net/10985/13821
Manifold embedding of heterogeneity in permeability of a woven fabric for optimization of the VARTM process
LOPEZ, Elena; CHINESTA, Francisco; SURESH, Advania
In Vacuum Assisted Resin Transfer Molding (VARTM), fabrics are placed on a tool surface and a Distribution Media (DM) is placed on top to enhance the flow in the in-plane direction. Resin is introduced from one end and a vacuum is applied at the other end to create the pressure gradient needed to impregnate the fabric with resin before curing the resin to fabricate the composite part. Heterogeneity in through the thickness permeability of a woven fabric is one of the causes for the variability in the quality of the final composite part fabricated using the VARTM process. The heterogeneity is caused by the varying sizes of pinholes which are meso-scale empty spaces between woven tows as a result of the weaving process. The pinhole locations and sizes in the fabric govern the void formation behavior during impregnation of the resin into the fabric. The pinholes can be characterized with two parameters, a gamma distribution function parameter α and Moran's I (MI). In this work, manifold embedding methods such as t – Distributed Stochastic Neighborhood Embedding (t_SNE) and Principal Component Analysis (PCA) are used to visually characterize fabrics of interest with the two variables, α and MI, through the reduction of dimensionality. To demonstrate the manifold embedding method, a total of 450 training sample data with ranges of α from 1 to 3 and MI from 0 to 0.5 were used to create a map in three-dimensional space for ease of visualization and characterization. The method is validated with a plain-woven fabric sample in a testing step to show that the two parameters of the fabric are identified with its corresponding α and MI using these machine learning algorithms. Numerical flow simulations were carried out for varying α, MI, and DM permeability, and the results were used to predict final void percentage. The quick online identification of the fabric parameters with machine learning algorithms can instantly provide expected variability in void formation behavior that will be encountered in a VARTM process.
Mon, 01 Jan 2018 00:00:00 GMThttp://hdl.handle.net/10985/138212018-01-01T00:00:00ZLOPEZ, ElenaCHINESTA, FranciscoSURESH, AdvaniaIn Vacuum Assisted Resin Transfer Molding (VARTM), fabrics are placed on a tool surface and a Distribution Media (DM) is placed on top to enhance the flow in the in-plane direction. Resin is introduced from one end and a vacuum is applied at the other end to create the pressure gradient needed to impregnate the fabric with resin before curing the resin to fabricate the composite part. Heterogeneity in through the thickness permeability of a woven fabric is one of the causes for the variability in the quality of the final composite part fabricated using the VARTM process. The heterogeneity is caused by the varying sizes of pinholes which are meso-scale empty spaces between woven tows as a result of the weaving process. The pinhole locations and sizes in the fabric govern the void formation behavior during impregnation of the resin into the fabric. The pinholes can be characterized with two parameters, a gamma distribution function parameter α and Moran's I (MI). In this work, manifold embedding methods such as t – Distributed Stochastic Neighborhood Embedding (t_SNE) and Principal Component Analysis (PCA) are used to visually characterize fabrics of interest with the two variables, α and MI, through the reduction of dimensionality. To demonstrate the manifold embedding method, a total of 450 training sample data with ranges of α from 1 to 3 and MI from 0 to 0.5 were used to create a map in three-dimensional space for ease of visualization and characterization. The method is validated with a plain-woven fabric sample in a testing step to show that the two parameters of the fabric are identified with its corresponding α and MI using these machine learning algorithms. Numerical flow simulations were carried out for varying α, MI, and DM permeability, and the results were used to predict final void percentage. The quick online identification of the fabric parameters with machine learning algorithms can instantly provide expected variability in void formation behavior that will be encountered in a VARTM process.Review on the Brownian Dynamics Simulation of Bead-Rod-Spring Models Encountered in Computational Rheology
http://hdl.handle.net/10985/19135
Review on the Brownian Dynamics Simulation of Bead-Rod-Spring Models Encountered in Computational Rheology
CRUZ, C.; CHINESTA, Francisco; RÉGNIER, G.
Kinetic theory is a mathematical framework intended to relate directly the most relevant characteristics of the molecular structure to the rheological behavior of the bulk system. In other words, kinetic theory is a micro-to-macro approach for solving the flow of complex fluids that circumvents the use of closure relations and offers a better physical description of the phenomena involved in the flow processes. Cornerstone models in kinetic theory employ beads, rods and springs for mimicking the molecular structure of the complex fluid. The generalized bead-rod-spring chain includes the most basic models in kinetic theory: the freely jointed bead-spring chain and the freely-jointed bead-rod chain. Configuration of simple coarse-grained models can be represented by an equivalent Fokker-Planck (FP) diffusion equation, which describes the evolution of the configuration distribution function in the physical and configurational spaces. FP equation can be a complex mathematical object, given its multidimensionality, and solving it explicitly can become a difficult task. Even more, in some cases, obtaining an equivalent FP equation is not possible given the complexity of the coarse-grained molecular model. Brownian dynamics can be employed as an alternative extensive numerical method for approaching the configuration distribution function of a given kinetic-theory model that avoid obtaining and/or resolving explicitly an equivalent FP equation. The validity of this discrete approach is based on the mathematical equivalence between a continuous diffusion equation and a stochastic differential equation as demonstrated by Itô in the 1940s. This paper presents a review of the fundamental issues in the BD simulation of the linear viscoelastic behavior of bead-rod-spring coarse grained models in dilute solution. In the first part of this work, the BD numerical technique is introduced. An overview of the mathematical framework of the BD and a review of the scope of applications are presented. Subsequently, the links between the rheology of complex fluids, the kinetic theory and the BD technique are established at the light of the stochastic nature of the bead-rod-spring models. Finally, the pertinence of the present state-of-the-art review is explained in terms of the increasing interest for the stochastic micro-to-macro approaches for solving complex fluids problems. In the second part of this paper, a detailed description of the BD algorithm used for simulating a small-amplitude oscillatory deformation test is given. Dynamic properties are employed throughout this work to characterise the linear viscoelastic behavior of bead-rod-spring models in dilute solution. In the third and fourth part of this article, an extensive discussion about the main issues of a BD simulation in linear viscoelasticity of diluted suspensions is tackled at the light of the classical multi-bead-spring chain model and the multi-bead-rod chain model, respectively. Kinematic formulations, integration schemes and expressions to calculate the stress tensor are revised for several classical models: Rouse and Zimm theories in the case of multi-bead-spring chains, and Kramers chain and semi-flexible filaments in the case of multi-bead-rod chains. The implemented BD technique is, on the one hand, validated in front of the analytical or exact numerical solutions known of the equivalent FP equations for those classic kinetic theory models; and, on the other hand, is control-set thanks to the analysis of the main numerical issues involved in a BD simulation. Finally, the review paper is closed by some concluding remarks.
Sun, 01 Jan 2012 00:00:00 GMThttp://hdl.handle.net/10985/191352012-01-01T00:00:00ZCRUZ, C.CHINESTA, FranciscoRÉGNIER, G.Kinetic theory is a mathematical framework intended to relate directly the most relevant characteristics of the molecular structure to the rheological behavior of the bulk system. In other words, kinetic theory is a micro-to-macro approach for solving the flow of complex fluids that circumvents the use of closure relations and offers a better physical description of the phenomena involved in the flow processes. Cornerstone models in kinetic theory employ beads, rods and springs for mimicking the molecular structure of the complex fluid. The generalized bead-rod-spring chain includes the most basic models in kinetic theory: the freely jointed bead-spring chain and the freely-jointed bead-rod chain. Configuration of simple coarse-grained models can be represented by an equivalent Fokker-Planck (FP) diffusion equation, which describes the evolution of the configuration distribution function in the physical and configurational spaces. FP equation can be a complex mathematical object, given its multidimensionality, and solving it explicitly can become a difficult task. Even more, in some cases, obtaining an equivalent FP equation is not possible given the complexity of the coarse-grained molecular model. Brownian dynamics can be employed as an alternative extensive numerical method for approaching the configuration distribution function of a given kinetic-theory model that avoid obtaining and/or resolving explicitly an equivalent FP equation. The validity of this discrete approach is based on the mathematical equivalence between a continuous diffusion equation and a stochastic differential equation as demonstrated by Itô in the 1940s. This paper presents a review of the fundamental issues in the BD simulation of the linear viscoelastic behavior of bead-rod-spring coarse grained models in dilute solution. In the first part of this work, the BD numerical technique is introduced. An overview of the mathematical framework of the BD and a review of the scope of applications are presented. Subsequently, the links between the rheology of complex fluids, the kinetic theory and the BD technique are established at the light of the stochastic nature of the bead-rod-spring models. Finally, the pertinence of the present state-of-the-art review is explained in terms of the increasing interest for the stochastic micro-to-macro approaches for solving complex fluids problems. In the second part of this paper, a detailed description of the BD algorithm used for simulating a small-amplitude oscillatory deformation test is given. Dynamic properties are employed throughout this work to characterise the linear viscoelastic behavior of bead-rod-spring models in dilute solution. In the third and fourth part of this article, an extensive discussion about the main issues of a BD simulation in linear viscoelasticity of diluted suspensions is tackled at the light of the classical multi-bead-spring chain model and the multi-bead-rod chain model, respectively. Kinematic formulations, integration schemes and expressions to calculate the stress tensor are revised for several classical models: Rouse and Zimm theories in the case of multi-bead-spring chains, and Kramers chain and semi-flexible filaments in the case of multi-bead-rod chains. The implemented BD technique is, on the one hand, validated in front of the analytical or exact numerical solutions known of the equivalent FP equations for those classic kinetic theory models; and, on the other hand, is control-set thanks to the analysis of the main numerical issues involved in a BD simulation. Finally, the review paper is closed by some concluding remarks.On the effective conductivity and the apparent viscosity of a thin rough polymer interface using PGD‐based separated representations
http://hdl.handle.net/10985/19486
On the effective conductivity and the apparent viscosity of a thin rough polymer interface using PGD‐based separated representations
AMMAR, Amine; GHNATIOS, Chady; DELPLACE, Frank; BARASINSKI, Anais; DUVAL, Jean-Louis; CUETO, Elias; CHINESTA, Francisco
Composite manufacturing processes usually proceed from preimpregnated preforms that are consolidated by simultaneously applying heat and pressure, so as to ensure a perfect contact compulsory for making molecular diffusion possible. However, in practice, the contact is rarely perfect. This results in a rough interface where air could remain entrapped, thus affecting the effective thermal conductivity. Moreover, the interfacial melted polymer is squeezed flowing in the rough gap created by the fibers located on the prepreg surfaces. Because of the typical dimensions of a composite prepreg, with thickness orders of magnitude smaller than its other in-plane dimensions, and its surface roughness having a characteristic size orders of magnitude smaller than the prepreg thickness, high-fidelity numerical simulations for elucidating the impact of surface and interface roughness remain today, despite the impressive advances in computational availabilities, unattainable. This work aims at elucidating roughness impact on heat conduction and the effective viscosity of the interfacial polymer squeeze flow by using an advanced numerical strategy able to reach resolutions never attained until now, a sort of numerical microscope able to attain the scale of the smallest geometrical detail.
Wed, 01 Jan 2020 00:00:00 GMThttp://hdl.handle.net/10985/194862020-01-01T00:00:00ZAMMAR, AmineGHNATIOS, ChadyDELPLACE, FrankBARASINSKI, AnaisDUVAL, Jean-LouisCUETO, EliasCHINESTA, FranciscoComposite manufacturing processes usually proceed from preimpregnated preforms that are consolidated by simultaneously applying heat and pressure, so as to ensure a perfect contact compulsory for making molecular diffusion possible. However, in practice, the contact is rarely perfect. This results in a rough interface where air could remain entrapped, thus affecting the effective thermal conductivity. Moreover, the interfacial melted polymer is squeezed flowing in the rough gap created by the fibers located on the prepreg surfaces. Because of the typical dimensions of a composite prepreg, with thickness orders of magnitude smaller than its other in-plane dimensions, and its surface roughness having a characteristic size orders of magnitude smaller than the prepreg thickness, high-fidelity numerical simulations for elucidating the impact of surface and interface roughness remain today, despite the impressive advances in computational availabilities, unattainable. This work aims at elucidating roughness impact on heat conduction and the effective viscosity of the interfacial polymer squeeze flow by using an advanced numerical strategy able to reach resolutions never attained until now, a sort of numerical microscope able to attain the scale of the smallest geometrical detail.