Complex Computational Fluid Dynamics Models Research Articles

Experimental and Numerical Evaluation of a Wildland–Urban Interface Fire Scenario

This paper presents the results obtained from a field fire test, aiming to reproduce a wildland–urban interface scenario to collect relevant information concerning the impact of wildfires on the built environment. The objective was to understand heat transfer mechanisms from forest fires to structures. During the fire test, the temperatures at the exposed face of one building component were monitored, as well as those in the vicinity of that component, using thermal imaging. The detailed characterization of the field test and building component and obtained experimental results of the fire test were then used to develop and validate a complex computational fluid dynamics model (full physics models) using the Fire Dynamics Simulator (FDS). Several numerical models were previously developed to reproduce the behaviour of individual shrubs and trees in fires considering available results in the literature. The developed Computational Fluid Dynamics (CFD) models can accurately reproduce the field test, including the fire spread and the temperature evolution on the surface of the exposed construction component. The obtained maximum temperature in the construction element was 1038 °C, whereas the maximum average temperature was approximately 638 °C. According to the results from the numerical model, the construction element was exposed to a very high heat flux (above 40 kW/m2), indicating direct contact of the flames with the construction element. The use of CFD enables the quantification of the characteristics of the fire and the exposure of structures to fire in the wildland–urban interface (WUI), allowing for the definition of a performance-based design approach for buildings in the WUI. This contributes to developing safe and resilient structures, as well as mitigating and reducing the impacts of wildfires in the built environment.

Toward Data Center Digital Twins via Knowledge-based Model Calibration and Reduction

Computational fluid dynamics (CFD) models have been widely used for prototyping data centers. Evolving them into high-fidelity and real-time digital twins is desirable for the online operations of data centers. However, CFD models often have unsatisfactory accuracy and high computation overhead. Manually calibrating the CFD model parameters is tedious and labor-intensive. Existing automatic calibration approaches apply heuristics to search the model configurations. However, each search step requires a long-lasting process of repeatedly solving the CFD model, rendering them impractical, especially for complex CFD models. This article presents Kalibre , a knowledge-based neural surrogate approach that calibrates a CFD model by iterating four steps of (i) training a neural surrogate model, (ii) finding the optimal parameters through neural surrogate retraining, (iii) configuring the found parameters back to the CFD model, and (iv) validating the CFD model using sensor-measured data. Thus, the parameter search is offloaded to the lightweight neural surrogate. To speed up Kalibre’s convergence, we incorporate prior knowledge in training data initialization and surrogate architecture design. With about ten hours of computation on a 64-core processor, Kalibre achieves mean absolute errors (MAEs) of 0.57°C and 0.88°C in calibrating the CFD models of two production data halls hosting thousands of servers. To accelerate CFD-based simulation, we further propose Kalibreduce that incorporates the energy balance principle to reduce the order of the calibrated CFD model. Evaluation shows the model reduction only introduces 0.1°C to 0.27°C extra errors while accelerating the CFD-based simulations by thousand times.

A Review of the CFD Method in the Modeling of Flow Forces

Hydraulic valves are key components of fluid power systems. They control the flow rate and pressure in hydraulic lines, actuator motion, and direction. Valves that control flow rate or pressure can be divided into two main categories: spool-type valves, where control components are similar to the piston inside a sleeve with control orifices; and seat-type valves, in which a poppet inside a seat opens and closes the flow. Forces induced on valve components during oil flow are crucial to the valve’s operational capabilities. They can be calculated using a formula originating from the momentum conservation equation for a two-dimensional control volume. Increasing demands for flow rate and pressure control accuracy cause flow forces to be calculated much more accurately than when using the analytical formula. Therefore, computational fluid dynamics (CFD) simulations are the only effective tool for their calculation. This paper reviews the CFD approaches used for calculating flow forces inside hydraulic valves. It presents typical approaches used for evaluating flow forces inside hydraulic valves. The oldest and most common are conducted for a fixed position of valve components for defined flow conditions, which do not cover all components of flow forces. The dynamic flow forces can be calculated using more complex CFD models using fluid–structure interaction (FSI) techniques. This paper presents available FSI techniques for the simulation of transient flow forces, mainly for valves whose component position is determined by the forces occurring during oil flow.

Different Experimental and Numerical Models to Analyse Emptying Processes in Pressurised Pipes with Trapped Air

In hydraulic engineering, some researchers have developed different mathematical and numerical tools for a better understanding of the physical interaction between water flow in pipes with trapped air during emptying processes, where they have made contributions on the use of simple and complex models in different application cases. In this article, a comparative study of different experimental and numerical models existing in the literature for the analysis of trapped air in pressurised pipelines subjected to different scenarios of emptying processes is presented, where different authors have develope, experimental, one-dimensional mathematical and complex computational fluid dynamics (CFD) models (two-dimensional and three-dimensional) to understand the level of applicability of these models in different hydraulic scenarios, from the physical and computational point of view. In general, experimental, mathematical and CFD models had maximum Reynolds numbers ranging from 2670 to 20,467, and it was possible to identify that the mathematical models offered relevant numerical information in a short simulation time on the order of seconds. However, there are restrictions to visualise some complex hydraulic and thermodynamic phenomena that CFD models are able to illustrate in detail with a numerical resolution similar to the mathematical models, and these require simulation times of hours or days. From this research, it was concluded that the knowledge of the information offered by the different models can be useful to hydraulic engineers to identify physical and numerical elements present in the air–water interaction and computational conditions necessary for the development of models that help decision-making in the field of hydraulics of pressurised pipelines.

Measurement of a radial flow profile with eddy current flow meters and deep neural networks *

Eddy current flow meters (ECFMs) measure flows of conductive fluids. Recent interest in ECFMs has increased due to applications in advanced nuclear reactors. ECFMs are well suited for such applications, as they can provide non-invasive measurements of flow in fluids that are often difficult to measure. Traditionally, ECFMs are operated using an alternating current at a single frequency, limiting ECFMs to measure average fluid velocities, blockages, or voids. We expand the capabilities of ECFMs by measuring the fluid radial velocity profile of liquid mercury. To accomplish this, we made several ECFM sensitivity measurements at a range of frequencies. Different frequencies vary the electromagnetic skin depth of the device. By adjusting frequencies, we probed the fluid velocity at various radial locations and constructed a flow-velocity profile. The relationship between the ECFM measurements and velocity profile is nonlinear and requires solving an inverse problem. Using electromagnetic finite-element simulations to train a deep neural network (DNN), we created a model that provides a stable general relationship between the sensitivity measurements of an ECFM and the fluid velocity profile. Using ECFM measurements of liquid mercury, our DNN model calculates a flow profile that agrees well with computational fluid dynamics (CFD) simulations. This technique has potential to improve flow monitoring for optimization, safe operation of conductive fluid loops, and/or validating complex CFD models.

Development of Compartment Models for Diagnostic Purposes

The importance of recognizing the presence of process faults and resolving these faults is continuously increasing parallel to the development of industrial processes. Fault detection methods which are both robust and sensitive help to recognize the presence of faults in time to avoid malfunctions, financial loss, environmental damage or loss of human life. In the literature, the use of various model-based fault detection methods has gained a considerable degree of popularity. Methods usually based on black-box models, data-based techniques or models using symbolic logic, e.g.\ expert systems, have become widespread. White-box models, on the other hand, have been applied less despite their considerable robustness because of multiple reasons. Firstly, their complexity and the relatively vast amount of technological and modelling knowledge needed to construct them for industrial systems. Secondly, their large computational demand which makes them less suitable for online fault detection. In this study, the aim was to resolve these problems by developing a method to simplify the complex Computational Fluid Dynamics models employed to describe the equipment used in the chemical industry into less complex model structures. These simpler structures are Compartment Models, a type of white-box model which breaks down a complex system into smaller units with idealized behaviour. In the case of a small number of compartments, the computational load of such models is not significant, therefore, they can be employed for the purposes of online fault detection while providing an accurate representation of the system. For the purpose of identifying the compartmental structure, fuzzy logic was employed to create a model which approximates the real behaviour of the system as accurately as possible. Our future objective is to explore the possibility of combining this model with various diagnostic methods (expert systems, Bayesian networks, parity relations, etc.) and derive robust tools for the purpose of fault detection.

Computational Fluid Dynamics Based Mixing Prediction for Tilt Pad Journal Bearing TEHD Modeling—Part II: Implementation With Machine Learning

Abstract Reynolds based thermo-elasto-hydrodynamic (TEHD) simulations of tilting pad journal bearings (TPJBs) generally provide accurate results; however, the uncertainty of the pad’s leading edge thermal boundary conditions causes uncertainty of the results. The highly complex thermal-flow mixing action between pads (BPs) results from the oil supply nozzle jets and geometric features. The conventional Reynolds approach employs mixing coefficients (MCs), estimated from experience, to approximate a uniform inlet temperature for each pad. Part I utilized complex computational fluid dynamics (CFD) flow modeling to illustrate that temperature distributions at the pad inlets may deviate strongly from being uniform. The present work retains the uniform MC model but obtains the MC from detailed three-dimensional CFD modeling and machine learning, which could be extended to the radially and axially varying MC case. The steps for implementing an artificial neural network (ANN) approach for MC regression are provided as follows: (1) utilize a design of experiment step for obtaining an adaptable training set, (2) conduct CFD simulations on the BP to obtain the outputs of the training set, (3) apply an ANN learning process by Levenverg–Mardquart backpropagation with the Bayesian regularization, and (4) couple the ANN MC results with conventional TEHD Reynolds models. An approximate log fitting method provides a simplified approach for MC regression. The effectiveness of the Reynolds TEHD TPJB model with ANN regression-based MC distributions is confirmed by comparison with CFD based TEHD TPJB model results. The method obtains an accuracy nearly the same as the complete CFD model, but with the computational economy of a Reynolds approach.

Surrogate-based model parameter optimization based on gas explosion experimental data

ABSTRACTThis article presents a method for optimizing the predictive capabilities of a computational fluid dynamics (CFD) tool used for consequence analysis in the process industries. A procedure for surrogate-based optimization of empirically determined sub-grid model parameters is implemented. The objective is to yield the best fit between certain outputs of a CFD model and corresponding experimental values. The present study investigates whether the optimization approach is applicable for improving the predictions of gas explosions when optimizing (i) only for experiments involving similar physical phenomena and (ii) for a wider range of experiments including symmetric modules as well as realistic offshore modules. Employing optimization may detect model limitations, supporting the further development of complex CFD models to represent a wider range of physical phenomena. The optimization process significantly improves the prediction of the maximum overpressure and pressure impulse at specific monitor points. Online supplemental data for this article can be accessed athttps://doi.org/10.1080/0305215X.2018.1450399.

Heap Leaching: Modelling and Forecasting Using CFD Technology

Heap leach operations typically employ some form of modelling and forecasting tools to predict cash flow margins and project viability. However, these vary from simple spreadsheets to phenomenological models, with more complex models not commonly employed as they require the greatest amount of time and effort. Yet, accurate production modelling and forecasting are essential for managing production and potentially critical for successful operation of a complex heap, time and effort spent in setting up modelling tools initially may increase profitability in the long term. A brief overview of various modelling approaches is presented, but this paper focuses on the capabilities of a computational fluid dynamics (CFD) model. Advances in computational capability allow for complex CFD models, coupled with leach kinetic models, to be applied to complex ore bodies. In this paper a comprehensive hydrodynamic CFD model is described and applied to chalcopyrite dissolution under heap operating conditions. The model is parameterized against experimental data and validated against a range of experimental leach tests under different thermal conditions. A three-dimensional ‘virtual’ heap, under fluctuating meteorological conditions, is simulated. Continuous and intermittent irrigation is investigated, showing copper recovery per unit volume of applied leach solution to be slightly increased for pulse irrigation.

Effects of Endcaps and Spacer Pads on Unsteady Fluid Forces and Flow Through a 37-Element CANDU Fuel Bundle

Unsteady fluid forces acting on CANDU fuel bundles inside a pressure tube are an important source that induces bundle vibration during operation. In this study, a comprehensive three-dimensional computational fluid dynamics (CFD) model is presented to investigate unsteady coolant flow through a 37-element CANDU fuel bundle in a pressure tube. The CFD model includes all fuel elements with endcaps and spacer pads and two endplates. Large eddy simulation is employed to solve the complex CFD model. Time histories and frequency spectra of unsteady fluid forces acting on individual components and the fuel bundle are predicted. Fluid velocities and flow development after the upstream endplate and spacer pads are also investigated. The effects of the spacer pads and endcaps on the mean value and standard deviation of unsteady fluid forces and flow characteristics are predicted.

An adaptive-grid model for dynamic simulation of thermocline thermal energy storage systems

The advent of the smart grid requires both reliable, cost-effective energy storage solutions and the ability to accurately and efficiently simulate these systems. Thermocline thermal energy storage, where heat is stored over a temperature gradient, is a promising energy storage technology. However, because of the complexities of thermal stratification, accurate, yet simple models are difficult to identify and generally require complex computational fluid dynamics (CFD) models. In this work, a novel, adaptive-grid one-dimensional model for representing thermocline thermal energy storage systems is presented and validated using experimental data. Compared to standard, fixed-grid one-dimensional models, the adaptive-grid model improves accuracy and decreases the required number of state variables and equations by using a high-resolution grid in the region of thermal stratification and larger, variable-volume, isothermal nodes to represent the ends of the tank. Experimental data shows that the adaptive-grid model is more accurate than other one-dimensional models because it minimizes the effect of numerical diffusion and better predicts system behavior when temperature inversion is introduced into the system. The model presented has applicability in both cold and hot thermal energy storage systems and is advantageous over CFD models when many repeated simulations are required, such as in optimization, control, or prediction of long-term performance.

The selection of the most appropriate airflow model for designing indoor air sensor systems

Current indoor air sensor system design is mostly by intuition or experience rather than by design. Sensor systems that are intended to ensure the safety and well-being of building occupants should be systematically designed and their performance evaluated. The literature shows that selection of an airflow model for sensor system design has been either out of convenience (thus choosing a simpler multizone model) or for proven accuracy (thus choosing a more complex computational fluid dynamics (CFD) model). In this study, the most appropriate airflow model for use in designing indoor air sensor systems was selected without compromising design objectives for 12 test zones. The test zones differed in the diffuser location, furniture presence and location, and airtightness level. Simulated contaminant concentrations from multizone, zonal, and CFD models were used to design sensor systems that either minimized detection time or occupant exposure. In lieu of experimental data, the performance of the sensor systems designed using multizone and zonal model data were benchmarked using CFD data. The sensor systems designed using CFD data were considered the “best-performing” in lieu of actual performance data. It was found that multiple-sensor systems designed using multizone model (the simplest of the three airflow models tested) data performed just as well as the best-performing sensor systems. For 1-sensor systems, common engineering practices resulted in performance comparable to the best-performing ones. These results applied even though the test zones were not considered well-mixed, and furniture and infiltration was modeled.

Comparison of sensor systems designed using multizone, zonal, and CFD data for protection of indoor environments

Sensors that detect chemical and biological warfare agents can offer early warning of dangerous contaminants. However, current sensor system design is mostly by intuition and experience rather than by systematic design. To develop a sensor system design methodology, the proper selection of an indoor airflow model is needed. Various indoor airflow models exist in the literature, from complex computational fluid dynamics (CFD) to simpler approaches such as multizone and zonal models. Airflow models provide the contaminant concentration data, to which an optimization method can be applied to design sensor systems. The authors utilized a subzonal modeling approach when using a multizone model and were the first to utilize a zonal model for systematic sensor system design. The objective of the study was to examine whether or not data from a simpler airflow model could be used to design sensor systems capable of performing just as well as those designed using data from more complex CFD models. Three test environments, a small office, a large hall, and an office suite were examined. Results showed that when a unique sensor system design was not needed, sensor systems designed using data from simpler airflow models could perform just as well as those designed using CFD data. Further, only for the small office did the common engineering sensor system design practice of placing a sensor at the exhaust result in sensor system performance that was equivalent to one designed using CFD data.

Comparison of Axisymmetric and Three-Dimensional Models for Gas Uptake in a Single Bifurcation During Steady Expiration

Reactive gas uptake is predicted and compared in a single bifurcation at steady expiratory flow in terms of Sherwood number using an axisymmetric single-path model (ASPM) and a three-dimensional computational fluid dynamics model (CFDM). ASPM is validated in a two-generation geometry by comparing the average gas-phase mass transfer coefficients with the experimental values. ASPM predicted mass transfer coefficients within 20% of the experimental values. The flow and concentration variables in the ASPM were solved using Galerkin finite element method and in the CFDM using commercial finite element software FIDAP. The simulations were performed for reactive gas flowing at Reynolds numbers ranging from 60 to 350 in both symmetric bifurcation for three bifurcation angles, 30 deg, 70 deg, and 90 deg, and in an asymmetric bifurcation. The numerical models compared with each other qualitatively but quantitatively they were within 0.4-8% due to nonfully developed flow in the parent branch predicted by the CFDM. The radially averaged concentration variation along the axial location matched qualitatively between the CFDM and ASPM but quantitatively they were within 32% due to differences in the flow field. ASPM predictions compared well with the CFDM predictions for an asymmetric bifurcation. These results validate the simplified ASPM and the complex CFDM. ASPM predicts higher Sherwood number with a flat velocity inlet profile compared to a parabolic inlet velocity profile. Sherwood number increases with the inlet average velocity, wall mass transfer coefficient, and bifurcation angle since the boundary layer grows slower in the parent and daughter branches.

Gas Uptake in a Three-Generation Model Geometry During Steady Expiration: Comparison of Axisymmetric and Three-Dimensional Models

Mass transfer coefficients were predicted and compared for uptake of a formaldehyde–air gas system using an axisymmetric single path model (ASPM) and a three-dimensional computational fluid dynamics model (CFDM) in three-generation model geometry at steady expiratory flow. The flow and concentration fields in the ASPM were solved using Galerkin's finite-element method and in the CFDM using a commercial finite-element software, FIDAP. Numerical results were compared for two different inlet flow rates, wall mass transfer coefficients, and bifurcation angles. The mass transfer coefficients variation with bifurcation unit from the ASPM and CFDM compared qualitatively and quantitatively closely at all flows and lower wall mass transfer coefficients for both 40° and 70° bifurcation angles. However, at higher wall mass transfer coefficients, quantitatively they were within 40% for both the bifurcation angles. Also, at higher flow and wall mass transfer coefficients, they were off qualitatively for a 70° bifurcation angle although the uptake compared qualitatively. This is due to the normalization of uptake within a bifurcation unit with the average of inlet and outlet average concentrations. Both CFDM and ASPM predict the same trends of increase in mass transfer coefficients with inlet flow and wall mass transfer coefficients. Also, the local values of the mass transfer coefficients compared closely at all conditions. These results validate the simplified ASPM and the complex CFDM. Mass transfer coefficients increase with bifurcation angles and with a flat inlet velocity profile compared to a parabolic velocity profile since the flow is non-fully developed and hence, the uptake increases.

Gas Uptake in a Three-Generation Model Geometry with a Flat Inlet Velocity During Steady Inspiration: Comparison of Axisymmetric and Three-Dimensional Models

Mass transfer coefficients were predicted and compared for uptake of reactive gas system using an axisymmetric single-path model (ASPM) with experimentally predicted values in a two-generation geometry and with a three-dimensional computational fluid dynamics model (CFDM) in a three-generation model geometry at steady inspiratory flow with a flat inlet velocity profile. The flow and concentration fields in the ASPM were solved using Galerkin's finite element method and in the CFDM using a commercial finite element software FIDAP. ASPM predicted average gas phase mass transfer coefficients within 25% of the experimental values. Numerical results in terms of overall mass transfer coefficients from the two models within each bifurcation unit were compared for two different inlet flow rates, wall mass transfer coefficients, and bifurcation angles. The overall mass transfer coefficients variation with bifurcation unit from the ASPM and CFDM compared qualitatively and quantitatively closely at lower wall mass transfer coefficients for both 40° and 70° bifurcation angles. But at higher wall mass transfer coefficients, quantitatively they were off in the range of 2–10% for 40° bifurcation angle and in the range of 4–15% for 70° bifurcation angle. Both CFDM and ASPM predict the same trends of increase in mass transfer coefficients with inlet flow, wall mass transfer coefficients, and during inspiration compared to expiration. Higher mass transfer coefficients were obtained with a flat velocity profile compared to a parabolic velocity profile using ASPM. These results validate the simplified ASPM and the complex CFDM.

Numerical Simulation and Controller Development for Energy Transfer in Imperfectly Mixed Fluids

This paper describes the possibility of using Computational Fluid Dynamics (CFD) models for developing Model Predictive Controller (MPC) algorithms. In this work a complex CFD model is combined with a reduced order Data-Based Mechanistic (DBM) formulation so that the low-order model can be used for control applications. To demonstrate the methodology, a Single Input-Single Output (SISO) ventilated installation is used. Initially, a CFD model is used for the installation. At the inlet a step rise in air temperature is applied and temperature responses at 36 monitoring positions are extracted. Then a reduced order model is formulated for the test case and model parameters are identified using the data generated by the CFD model. Finally the reduced order CFD model is used to develop an MPC algorithm. The developed algorithm is found to be robust for disturbance effects and capable of tracking the desired reference temperature trajectory very well.

Combining CFD and data-based mechanistic (DBM) modelling approaches

This paper outlines a methodology for the reduction of a complex computational fluid dynamics (CFD) model using data-based mechanistic (DBM) modelling technique. For this study a full scale, adiabatic walled ventilated installation was used. Primarily, a CFD model that is three dimensional, transient and turbulent was proposed and solved by a commercial package CFX 4.3 in order to obtain informative data that will be used to build a DBM model. At the inlet a step rise in temperature was applied and the ventilation rate is kept steady. Turbulence is modelled by the widely used standard k– ϵ model. Temperature responses at 36 monitoring positions inside the installation were extracted out of the simulation. In the second stage of the work, for the same ventilation system, a DBM model was formulated from an energy balance equation. The simplified refined instrumental variable (SRIV) algorithm is employed as model parameter identification tool. The obtained highly accurate model parameters for all monitoring positions demonstrate the possibility of reducing complex CFD models using identification technique.