Computational methods applied to chemical engineering

Abstract

[eng] Computational tools, and more specifically Computational Fluid Dynamics (CFD) have seen a rapid development and increase in use in recent years thanks to the exponential growth that computing power experiences every few years. These techniques have been historically implemented by physicists and aerospace engineers to unravel the complex turbulence phenomena and aid in the design of new aircrafts or fast land vehicles. Nowadays, these tools are starting to see use in all kinds of engineering environments, including chemical engineering. This thesis is based on the evaluation of the usefulness and accuracy of various methods of computer simulation applied to chemical and materials engineering units. The thesis is divided into various Chapters, each studying different units chosen for having a characteristic phenomenon occurring on them. In Chapters 5-10 CFD is used in different scenarios that are fundamental of chemical engineering. Chapter 11 is based on aiding in the development of new nanomaterials by using various computational tools and techniques, that range from the atomic scale to a full unit. Chapter 5 is dedicated to heat transfer. In this Chapter different configurations of heat exchangers are studied via CFD with increasing geometrical complexity. First, a simple double pipe heat exchanger is simulated, and the results of various turbulent modelling techniques are compared against semiempirical correlations used historically by chemical engineers. Next, a similar study is conducted for tubes with a passive heat transfer enhancement insert: the twisted tape. These inserts induce turbulence by creating a swirling motion inside the tube. For these units there are still some correlations and experimental data to compare them against. Finally, a new type of heat exchanger is created combining various passive heat transfer enhancement techniques: the twisted tape and an extended corrugated surface. The geometry of this new unit is optimized based on the geometrical parameters of the system and the efficiency of the heat exchanger is maximized. In Chapter 6, other important transport phenomenon is studied: mass transfer. For such task an atmospheric gas release is studied with CFD. Numerous experimental points of gas dispersion on plain terrain are reproduced with CFD and other classical modelling tools like Gaussian plume model (GPM) or the software ALOHA (Areal Locations of Hazardous Atmospheres) developed by the National Oceanic and Atmospheric Administration (NOAA). All these tools are compared against each other and the experimental points to assess their accuracy in this scenario. Additionally, Aloha and CFD tools are compared in a case with obstacles, where the differences between the two may be more extreme than when studying simple terrain. Chapter 7 revolves around momentum transfer, which is especially important in agitation of highly viscous fluids like the ones found in polymerization reactors. Here, a batch agitated reactor is studied with different fluid viscosities and the relationship between the agitation Reynolds number (dependant on the viscosity) and the power required to agitate is found. Moreover, different agitator blade angles are tested to discern which one better

suits the specific operational requirements (more power efficiency, enhanced heat transfer or faster product mixing among others). A study of multiphase flow is conducted in Chapter 8, where partially filled pipes or canals are studied with CFD. This type of flows have been historically modelled with the Manning equation, which relates slope angle, fluid flow, depth and velocity. With CFD an analysis of the Manning coefficients has been performed for simple half-filled pipes. The flow stabilization is analysed at different entrance conditions at the inlet of the pipe. Additionally, an elbow geometry is studied and its effect on the flow structures that are created as the fluid passes through it. Lastly, Chapter 9 is a compilation of different studies performed in the scope of the European project “Thermodust”, which aims at creating new dust nanomaterials with various thermal properties. For this part, various types of simulations are performed, as the process of producing these nanomaterials must be understood from an industrial perspective but also taking into account the interatomic interactions of the powder. For studying the production process, performed in a planetary mill, this unit is studied by Discrete Element Method (DEM) different conditions found in experimental. Feedback with experimental procedures is provided both ways: simulations used experimental feedback to validate and create the particle comminution modelling and experimental then uses simulation feedback to scale and improve the process efficiency and powder quality. To study the collision between particles inside the mill, Finite Element Analysis is conducted with the software LS-DYNA from Ansys. Finally, to better understand the materials interactions, different Molecular Dynamics simulations (MD) are performed. These simulations are realized accurately thanks to the use of a neural network that was trained using Density Functional Theory (DFT). The simulations performed with DFT for the training of the neural network is an ab initio technique that estimates different atomic interactions present on the system with quantum mechanics. This thesis looks over a vast variety of themes in chemical engineering and analyses them via simulation tools like CFD. The studied cases are closely examined thanks to these microscopical simulation methods, and they are optimized and improved when possible. The sum of the techniques detailed in this thesis and many more, constitutes a clear advancement in engineering design and analysis that will keep growing in importance and use especially for chemical and materials engineering, where empirical correlations that are decades old are still the preferred design choice.

[cat] Les eines computacionals estan transformant la investigació en diversos camps, inclosa l’Enginyeria Química. Aquesta tesi explora l’aplicació de mètodes computacionals avançats en processos i operacions unitàries d’aquesta disciplina, centrant-se en la Dinàmica de Fluids Computacional (CFD). A través de CFD, s’analitzen interaccions de fluids a petita escala en problemes fonamentals de l’Enginyeria Química com intercanviadors de calor, sistemes de dispersió de gasos, reactors de tanc agitat i fluxos a canal obert, aportant nous coneixements sobre fenòmens de transport i optimització de sistemes. Un dels focus principals és l’estudi d’intercanviadors de calor, essencials en la transferència d’energia. S’analitzen diverses configuracions amb CFD, des d’un intercanviador de doble tub fins a dissenys innovadors amb cintes trenades i superfícies corrugades, optimitzant l’eficiència tèrmica. També s’investiga la dispersió de gasos mitjançant CFD, comparant els resultats amb models tradicionals com el Model de Ploma Gaussiana i el programari ALOHA de la NOAA, destacant la capacitat del CFD per millorar la predicció en escenaris reals. L’estudi de la transferència de moment es focalitza en reactors de polimerització amb fluids altament viscosos. S’analitza com la viscositat influeix en el nombre de Reynolds i el consum energètic, provant diferents angles de pales per optimitzar la mescla industrial. Per aquest tema també es simulen fluxos multifàsics en canonades parcialment plenes i canals oberts, comparant els resultats amb l’equació de Manning per validar els models hidràulics clàssics envers el CFD. Finalment, la tesi contribueix al projecte europeu Thermodust, que investiga nanomaterials amb propietats tèrmiques avançades. Es fan servir tècniques com la Dinàmica Molecular (MD) i el Mètode d’Elements Discrets (DEM) per estudiar la síntesi i interaccions a escala atòmica, validant resultats experimentals i millorant processos. L’ús d’una xarxa neuronal basada en la Teoria del Funcional de la Densitat (DFT) permet modelar amb precisió aquestes interaccions, connectant la química computacional amb aplicacions industrials. Aquesta tesi demostra com eines computacionals avançades, com CFD, DEM i MD, milloren la comprensió i optimització de sistemes complexos en Enginyeria Química i de Materials, impulsant el disseny basat en dades i l’eficiència industrial.