Computational fluid dynamics (CFD) is a valuable tool in designing built environments, enhancing comfort, health, energy efficiency, and safety in both indoor and outdoor applications. Nevertheless, the time required for CFD computations still needs to be reduced for engineering studies. Recent advances in machine learning (ML) techniques offer a promising avenue for developing fast-running data-driven models for physics-related phenomena. As scientific machine learning (SciML) research increasingly focuses on efficiently coupling ML and CFD techniques, this literature review highlights the growing number of applications in the built environment field to accelerate CFD simulations. This work aims to identify emerging trends and challenges in incorporating ML techniques into built environment flow simulations to foster further advancements in this domain. The prevailing approaches are direct surrogate modeling and reduced-order models (ROMs). Both approaches increasingly rely on deep learning architectures based on neural networks. The reviewed studies reported computational time gains of several orders of magnitude in specific scenarios while maintaining reasonable accuracy. However, several challenges remain, such as improving models’ generalizability and interpretability, enhancing methodology scalability, and reducing the computational cost of developing the models. Efforts are underway to address more complex cases with advanced SciML techniques. Notably, incorporating physics into the learning process and hybridizing CFD solvers with data-driven models merit further investigation. The exploration of these approaches represents a crucial step toward the deployment of reliable models that enable fast design for built environment engineering studies.

This paper proposes a new algorithm to solve topology optimization problems for laminar unsteady or turbulent flows. Instead of computing the gradient of the cost function after solving the direct and adjoint (both unsteady) PDE on the full time interval, our algorithm uses averaged physical quantities on a smaller unspecified time interval to define a (steady) Reynolds-Averaged Method (RAM) model which is then used as constraint in an optimization problem to update the design variable. Another feature of the proposed method is that the RAM model can be defined whatever the initial model and CFD turbulence models initially chosen to compute the instantaneous physical quantities. The RAM model involves turbulent quantities such as turbulent kinetic viscosity and turbulent thermal diffusivity are estimated instead of using the concept of ”frozen turbulence”. In contrast with the classical methods built to solve unsteady topology optimization problems, the main advantage of the proposed algorithm is that it updates the design variable by solving an auxiliary steady topology optimization problem. Three configuration cases are studied to illustrate the ability of our algorithm to optimize pressure losses and heat transfer by adding material to smooth the laminar unsteady or turbulent flows. We also calculate the number of required design parameter updates to obtain an optimized design. Thus, our algorithm overcomes three major scientific challenges in solving optimization problems in turbulence, namely leveraging efficient temporal turbulence models or a Direct Numerical Simulation (DNS) model, computational cost and data storage requirements.

Data-driven methods demonstrate considerable potential for accelerating the inherently expensive computational fluid dynamics (CFD) solvers. Nevertheless, pure machine-learning surrogate models face challenges in ensuring physical consistency and scaling up to address real-world problems. This study presents a versatile and scalable hybrid methodology, combining CFD and machine learning, to accelerate long-term incompressible fluid flow simulations without compromising accuracy. A neural network was trained offline using simulated data of various two-dimensional transient buoyant plume flows. The objective was to leverage local features to predict the temporal changes in the pressure field in comparable scenarios. Due to cell-level predictions, the methodology was successfully applied to diverse geometries without additional training. Pressure estimates were employed as initial values to accelerate the pressure-velocity coupling procedure. The results demonstrated an average improvement of 94 first pressure corrector acceleration reached a mean factor of 3, depending on the iterative solver employed. Our work reveals that machine learning estimates at the cell level can enhance the efficiency of CFD iterative linear solvers while maintaining accuracy. Although the scalability of the methodology to more complex cases has yet to be demonstrated, this study underscores the prospective value of domain-specific hybrid solvers for CFD.

Topology optimization is widely used to design heat exchangers and may involve different expressions of objective functions to increase heat exchange. This work proposes a new thermal objective function based on the local orientation of velocity and temperature gradient fields. The latter is then defined as the cosine of these two vector fields and thus our approach has connections with the field synergy principle. The cosine objective function is compared with a more classical one in a multi-objective optimization framework whose resolution is done with the adjoint method. Our results reveal that the cosine objective function lead to results that are comparable with those obtained with the classical cost function and may be used by designers to look for optimized design by taking into account the synergy of the fields. A study of the field synergy principle reveals that it is only an accurate indicator of heat exchange in some cases discussed in this article.

Anaerobic digestion is a widely used process for organic waste treatment and biogas production. It is a complex process involving four main biochemical reactions, four microbial groups, and hydrodynamics phenomena. Research focuses on the development of numerical models based on mass transfer and biochemical reaction kinetics, adapted to a specific waste. However, parameters such as mass transfer, fluid rheology, and temperature strongly influence the physicochemical and biological reactions. These parameters are considered in computational fluid dynamic models. In this paper, we present a critical analysis of advances in computational fluid dynamic modeling of anaerobic digesters. This review addresses the work contributions, the technologies of bioreactors as well as the meshing approaches in order to model the mechanical mixing. A detailed analysis is also presented on the evaluation of the effects of waste rheology and total solid content on the flow. Specific studies on biofilm, high anaerobic digesters, multiphase studies, coupling of biochemical and computational fluid dynamic models, turbulence modeling are addressed in this review. In conclusion, computational fluid dynamic models provide a local description of the process and constitute a promising tool to design anaerobic digestion units and optimization the anaerobic digestion process (reduction of dead zone, fluid homogenization, etc.). Thermal modeling is mainly useful in cold countries or those experiencing significant temperature variations. Few studies concern the coupling of biochemical and computational fluid dynamic models, allowing to consider both the bio-physicochemical aspects and the flow field. Future research should focus on microbiological aspects, as well as the full integration of biochemical reactions in computational fluid dynamic models.

Several engineering problems result in a PDE-constrained optimization problem that aims at finding the shape of a solid inside a fluid which minimizes a given cost function. These problems are categorized as Topology Optimization (TO) problems. In order to tackle these problems, the solid may be located with a penalization term added in the constraints equations that vanishes in fluid regions and becomes large in solid regions. This paper addresses a TO problem for anisothermal flows modelled by the steady-state incompressible Navier–Stokes system coupled to an energy equation, with mixed boundary conditions, under the Boussinesq approximation. We first prove the existence and uniqueness of a solution to these equations as well as the convergence of its finite element discretization. Next, we show that our TO problem has at least one optimal solution for cost functions that satisfy general assumptions. The convergence of discrete optimum toward the continuous one is then proved as well as necessary first order optimality conditions. Eventually, all these results let us design a numerical algorithm to solve a TO problem approximating solids with piecewise constant thermal diffusivities also refered as multi-materials. A physical problem solved numerically for varying parameters concludes this paper.

Anaerobic digestion is a widely used process for waste treatment and energy production. This natural process takes place in a controlled environment, anaerobic digesters. Mixing is one of the main operating parameters. The understanding of the flows during the agitation of the medium is crucial for the optimization of the process yield. In fact, the mass and heat transfers are enhanced by the agitation. However, the complex biochemical reactions can be inhibited with overly vigorous agitation. A detailed and in-depth understanding of the phenomena occurring during agitation requires modeling studies. In this chapter, we propose a general approach, based on computational fluid mechanics (CFD), to analyze the mechanical mixing of an anaerobic reactor. We apply this work to the anaerobic digestion of the sugarcane vinasse, which is a liquid waste generated during the production of alcohol. The single-phase Reynolds-averaged Navier-Stokes (RANS) simulations of mechanical agitation of Newtonian fluids for different rotational speeds are presented. The equations system is closed with the standard k-epsilon turbulence model. The flow field is analyzed with the velocity profiles, the Q and Lambda2 fields, the pressure and the vorticity.

The objective of this study is to highlight the effect of porosity variation in a topology optimization process in the field of fluid dynamics. Usually a penalization term added to momentum equation provides to get material distribution. Every time material is added inside the computational domain, there is creation of new fluid-solid interfaces and apparition of gradient of porosity. However, at present, porosity variation is not taken account in topology optimization and the penalization term used to locate the solid is analogous to a Darcy term used for flows in porous media. With that in mind, in this paper, we first develop an original one-domain macroscopic model for the modelling of flow through spatially varying porous media that goes beyond the scope of Darcy regime. Next, we numerically solve a topology optimization problem and compare the results obtained with the standard model that does not include effect of porosity variation with those obtained with our model. Among our results, we show for instance that the designs obtained are different but percentages of reduction of objective functional remain quite close (below 4\% of difference). In addition, we illustrate effects of porosity and particle diameter values on final optimized designs.

Indoor flow behaviour can strongly impact safety in the case of a fire event. Indeed, when air inlet and outlet are correctly placed, stratification allows a fresh air layer to be maintained in the room that can be favourable for the egress. However, in the case of a naturally ventilated building, the presence of the wind will interact with the indoor flow pattern. Hence, fire smoke extraction in naturally ventilated buildings can strongly be influenced by the wind. Indeed, when wind opposed buoyancy, a change in the ratio between buoyancy forces and wind forces can impact the internal flow pattern resulting in a mixed ventilation mode, which is unfavourable for egress, in the case of a fire event. Flow pattern inside a room and their transitions are then a safety purpose. In this paper, natural ventilation of a singular room with two asymmetrical and opposed openings was studied experimentally when wind opposes buoyancy. The buoyant source was generated by an injection of an air/he mix. Varying wind and injection conditions, the ventilation regimes change as well as the indoor flow dynamic. Three ventilation regimes are experimentally observed depending on the balance between J and Fr :buoyancy-driven, bidirectional and wind-driven. From dimensional analysis, we shown that the ventilation regimes can be described via the Froude number, based on the injection flow rate Fr, and the momentum flux ratio J. in the literature, transitions between these regimes is still difficult to estimate as far as it depends on the assumption made on the initial indoor flow pattern (layered or fully mixed). Hence, an experimental investigation on the transition from each extreme regime (buoyancy-driven or wind-driven) to the bi-directional one, has been conducted. The transitions are found to follow a power low in the form Fr ∝ J3/4. A discussion on the inner flow pattern for this bi-directional regime is also proposed and a focus on how the knowledge of the behaviour of the inflowing flow contributes to improve the modelling is made.

This paper deals with the finite element approximation of the Darcy–Brinkman- Forchheimer equation, involving a porous media with spatially-varying porosity, with mixed boundary condition such as inhomogeneous Dirichlet and traction boundary conditions. We first prove that the considered problem has a unique solution if the source terms are small enough. The convergence of a Taylor–Hood finite element approximation using a finite element interpolation of the porosity is then proved under similar smallness assumptions. Some optimal error estimates are obtained if the solution to the Darcy–Brinkman–Forchheimer model is smooth enough. We end this paper by providing a fixed-point method to solve the discrete non-linear problems and with some numerical experiments to make more precise the smallness assumptions on the source terms and to illustrate the theoretical convergence results.

Building energy simulation (BES) and Airflow network (AFN) programs generally incorporate wind pressure coefficients (C p ) estimated from secondary sources, namely data bases or analytical models. As these coefficients are influenced by a wide range of parameters, it is difficult to obtain reliable C p data. This leads to uncertainties in BES-AFN models results, especially for naturally ventilated building studies, where air change rate which strongly depends on C p , is a key value for thermal comfort and energy consumption results. This study focuses on naturally ventilated buildings in tropical climate and presents an alternative approach to estimate wind pressure coefficient. Computational fluid dynamics (CFD) simulations are performed to calculate wind-driven airflow rates at building opening level. Then wind pressure coefficient difference ΔC p is calculated from large opening equation to be used as input data in BES-AFN program. Numerical simulations are performed for various wind directions on a typical cross-ventilated isolated building and a more complex building with opposite large openings. CFD results of wind pressure coefficient difference and airflow rate are compared to those obtained from AFN model using two different C p sources. The results show that the calculated values vary greatly depending on the method used and highlight that an accurate estimation of wind pressure coefficient is a key parameter for evaluating natural ventilation in buildings.

This paper deals with a multi-objective topology optimization problem in an asymmetrically heated channel, based on both pressure drop minimization and heat transfer maximization. The problem is modeled by assuming steady-state laminar natural convection flow. The incompressible Navier-Stokes equations coupled with the convection-diffusion equation, under the Boussinesq approximation, are employed and are solved with the finite volume method. In this paper, we discuss some limits of classical pressure drop cost function for buoyancy-driven flow and, we then propose two new expressions of objective functions: the first one takes into account work of pressure forces and contributes to the loss of mechanical power while the second one is related to thermal power and is linked to the maximization of heat exchanges. We use the adjoint method to compute the gradient of the cost functions. The topology optimization problem is first solved for a Richardson (Ri) number and Reynolds number (Re) set respectively to Ri ∈{100,200,400} and Re = 400. All these configurations are investigated next in order to demonstrate the efficiency of the new expressions of cost functions. We compare two types of interpolation functions for both the design variable field and the effective diffusivity. Both interpolation techniques have pros and cons and give slightly the same results. We notice that we obtain less isolated solid elements with the sigmoid-type interpolation functions. Then, we choose to work with the sigmoid and solve the topology optimization problem in case of pure natural convection, by setting Rayleigh number to {3 × 10 3 ,4 × 10 4 ,5 × 10 5 }. In all considered cases, our algorithm succeeds to enhance one of the phenomenon modeled by the proposed cost functions without deteriorating the other one. The optimized design obtained suppresses any reversal flow at the exit of the channel. We also show that the thermal exchanges are improved by computing the Nusselt numbers and bulk temperature. We conclude that the new expressions of objective functions are well suited to deal with natural convection optimization problem in a vertical channel.

Topology optimization for fluid flow aims at finding the location of a porous medium minimizing a cost functional under constraints given by the Navier-Stokes equations. The location of the porous media is usually taken into account by adding a penalization term alpha u, where alpha is a kinematic viscosity divided by a permeability and u is the velocity of the fluid. The fluid part is obtained when alpha = 0 while the porous (solid) part is defined for large enough a since this formally yields u = 0. The main drawback of this method is that only solid that does not let the fluid to enter, that is perfect solid, can be considered. In this paper, we propose to use the porosity of the media as optimization parameter hence to minimize some cost function by finding the location of a porous media. The latter is taken into account through a singular perturbation of the Navier-Stokes equations for which we prove that its weak-limit corresponds to an interface fluid-porous medium problem modeled by the Navier-Stokes-Darcy equations. This model is then used as constraint for a topology optimization problem. We give necessary condition for such problem to have at least an optimal solution and derive first, order necessary optimality condition. This paper ends with some numerical simulations, for Stokes flow, to show the interest of this approach.

This paper proposes a new interpolation technique based on density approach to solve topology optimization problems for heat transfer. Natural convection forces are dominated as Richardson number is equal to 2.8. Problems are modeled under the assumptions of steady-state laminar flow using the incompressible Navier-Stokes equations coupled to the convection-diffusion equation through the Boussinesq approximation. The governing equations are discretized using finite volume elements and topology optimization is performed using adjoint sensitivity analysis. Material distribution and effective conductivity are interpolated by two sigmoid functions respectively h(tau)(alpha) and k(tau)(alpha) in order to provide a continuous transition between the solid and the fluid domains. Comparison with standard interpolation function of the literature (RAMP function) shows a smaller transition zone between the fluid and the solid thereby, avoiding some regularization techniques. In order to validate the new method, numerical applications are investigated on some geometric configurations from the literature, namely the single pipe and the bend pipe. Lastly, as two new parameters are introduced thanks to the interpolation functions, we study their impact on results of the optimization problem. The study shows that the proposed technique is a viable approach for designing geometries and fluid-porous media interfaces are well-defined. (C) 2018 Elsevier Ltd. All rights reserved.

The study’s ultimate goal consists of modelling and optimising the co-digestion of particulate substrates like sewage sludge, biowaste and vinasse from sugar cane. Given these substrates, the medium is highly heterogeneous and has particular physical characteristics and behaviour. Moreover, anaerobic digestion could be lead in wet-mode or in dry-mode, thus we want to develop a model valid for both configurations. In this context, the fluid flow modelling is of interest. Indeed, the contact between substrates and microorganisms has a significant influence on the anaerobic digestion process. It is directly correlated to the flow inside the digester. The research question is therefore the following: Is it possible to predict the spread of microorganisms inside the digester for both liquid and solid digestion? In this paper, we present the context of the work and the mathematical model based on Navier-Stokes equations with the Boussinesq approximation. Large-eddy simulation (LES) is used for the turbulence modelling. The sliding mesh method is used to characterise the impeller rotation. In further work, we will present the simulation results validated with the literature data. The differential equations will be solved on OpenFOAM software using the finite volume method. The flow will be simulated for both Newtonian (vinasse) and non-Newtonian fluids (manure) and at different total solid (TS) rates for non-Newtonian fluids. The obtained results will be the velocity and temperature profiles within the digester. It will hence be possible to optimise the process through both the medium physical properties (TS rate and viscosity) and the digester configuration. We will also consider the multiphase flow. Alongside numerical simulations, experiments are carried out on the substrates studied in order to estimate the methane production. In this poster, we present the results of experiments on the vinasse and the development of the CFD model.

Topology optimization for fluid flow aims at finding the location of a porous medium minimizing a cost functional under constraints given by the Navier–Stokes equations. The location of the porous media is usually taken into account by adding a penalization term [Formula: see text], where [Formula: see text] is a kinematic viscosity divided by a permeability and [Formula: see text] is the velocity of the fluid. The fluid part is obtained when [Formula: see text] while the porous (solid) part is defined for large enough [Formula: see text] since this formally yields [Formula: see text]. The main drawback of this method is that only solid that does not let the fluid to enter, that is perfect solid, can be considered. In this paper, we propose to use the porosity of the media as optimization parameter hence to minimize some cost function by finding the location of a porous media. The latter is taken into account through a singular perturbation of the Navier–Stokes equations for which we prove that its weak-limit corresponds to an interface fluid-porous medium problem modeled by the Navier–Stokes–Darcy equations. This model is then used as constraint for a topology optimization problem. We give necessary condition for such problem to have at least an optimal solution and derive first order necessary optimality condition. This paper ends with some numerical simulations, for Stokes flow, to show the interest of this approach.

In tropical context and particularly in Reunion Island, promoting passive design is compulsory, but cannot be achieved without considering fire safety. This article aims at evaluating the interactions that exist between fire safety and thermal comfort in buildings, using variance based and moment independent methods. Two stategies are proposed: a first one, dissociated, where fire safety assessment and thermal comfort are evaluated separately and a second combining both considerations. Input parameters that influence the most the selected output indices are investigated for each strategy. A normalized index combining thermal comfort and fire safety is suggested introducing a new methodology. Difference between the two strategies is highlighted.

This article consists in a numerical study of the influence of thermal stratification and surface radiation on laminar airflow induced by natural convection in vertical, asymmetrically heated channels. Several cases are investigated to spotlight their influence on fluid dynamics and thermal quantities. Thermal stratification is obtained by a weak gradient of temperature outside of the channel, and then the temperature at the bottom end of the channel is considered as a function of time. Significant effects on vertical velocities, mass flow, and flow structure are shown. Surface radiation is also considered but appears less predominant than thermal stratification for the selected conditions of this article. The impact on heat transfer is also evaluated for each studied configuration. It is observed that local and mean Nusselt numbers weakly increase for the investigated cases.