1,720,972 research outputs found
A framework for data regression of heat transfer data using machine learning
No full textMachine Learning (ML) algorithms are emerging in various industries as a powerful complement/alternative to traditional data regression methods. A major reason is that, unlike deterministic models, they can be used even in the absence of detailed phenomenological knowledge. Not surprisingly, the use of ML algorithms is being explored also in heat transfer applications. It is of particular interest in systems dealing with complex geometries and underlying phenomena (e.g. fluid phase change, multi-phase flow, heavy fouling build-up). However, heat transfer systems present specific challenges that need addressing, such as the scarcity of high-quality data, the inconsistencies across published data sources, the complex (and often correlated) influence of inputs, the split of data between training and testing sets, and the limited extrapolation capabilities to unseen conditions. In an attempt to help overcome some of these challenges and, more importantly, to provide a systematic approach, this article reviews and analyses past efforts in the application of ML algorithms to heat transfer applications, and proposes a regression framework for their deployment to estimate key quantities (e.g. heat transfer coefficient), to be used for improved design and operation of heat exchangers. The framework consists of six steps: i) data pre-treatment, ii) feature selection, iii) data splitting philosophy, iv) training and testing, v) tuning of hyperparameters, and vi) performance assessment with specific indicators, to support the choice of accurate and robust models. A relevant case study involving the estimation of the condensation heat transfer coefficient in microfin tubes is used to illustrate the proposed framework. Two data-driven algorithms, Deep Neural Networks and Random Forest, are tested and compared in terms of their estimation and extrapolation capabilities. The results show that ML algorithms are generally more accurate in predicting the heat transfer coefficient than a well-known semi-empirical correlation proposed in past studies, where the mean absolute error of the most suitable ML model is 535 [Wm2K-1], compared to the error using the correlation of 1061 [Wm2K-1]. In terms of extrapolation, the selected ML model has a mean absolute error of 1819 [Wm2K-1], while for the correlation is 1111 [Wm2K-1], indicating a disadvantage of the use of semi-empirical models, although the comparison was not entirely suitable, given that the correlation was used as is and no training was done. In addition, feature selection enables simpler models that depend only on features that are potentially most related to the target variable. Special attention is needed however, as overfitting and limited extrapolation capabilities are common difficulties that are encountered when deploying these models
Molecular tracking : a novel approach for multicomponent distillation column design
Get PDFLAUREA MAGISTRALELa progettazione iniziale di unità di distillazione con prelievo laterale, anche detto “side-draw”, è sempre stata un compito impegnativo sia per i professionisti accademici che per quelli industriali a causa della mancanza di metodi semplici e affidabili per determinare il posizionamento del “side-draw” corrispondente alla minima richiesta di energia del sistema. I metodi esistenti si basano su metodiche “trial and error” o su complesse e sofisticate ottimizzazioni, che spesso richiedono anche una validazione industriale dell'unità. In questo contesto pertanto si propone l’utilizzo della metodologia del tracciamento molecolare, strumento semplice e innovativo che permette di progettare unità di distillazione con estrazione laterale che presentano diluizione infinita del componente medio-bollente considerato come impurità. Questo concetto utilizza proprietà termodinamiche del sistema per determinare una funzione che valuta la probabilità che, per ogni stadio dell’unità, una singola molecola di un componente di una miscela si muova verso l’alto o verso il basso dell’unità stessa.
Viene quindi sviluppato un framework per trovare la posizione corrispondente del side-draw sulla colonna. In primo luogo viene simulata una distillazione binaria per i componenti principali della miscela attraverso il metodo della forza motrice (driving force method), anche noto come approccio di efficienza energetica. Poiché tale approccio costringe la colonna a funzionare secondo la sua massima forza motrice, la sola colonna richiede il minimo scambio di energia per il funzionamento. Una dettagliata illustrazione di questa tematica è discussa in questo studio. L'impurità viene quindi introdotta nel feed come componente di ebollizione intermedio e il profilo di probabilità di tale impurità viene generato in base alle proprietà termodinamiche di questo componente e ai flussi molari della colonna. Il prelievo laterale ottimale è quindi posto su un piatto in cui le molecole considerate come impurità tendono a rimanere, invece di spostarsi verso l'alto o verso il basso. I concetti di tracciamento molecolare e forza motrice sono entrambi seguiti da semplici esempi per rappresentare le loro funzionalità.
Inoltre, due unità di distillazione a estrazione laterale sono progettate per i due casi studio di miscele ternarie ideali con struttura di tracciamento molecolare. Al fine di confrontare il metodo del tracciamento molecolare con altri metodi esistenti, la posizione del prelievo laterale viene trovata anche con il metodo basato sulla forza motrice classica. Le configurazioni di ciascun metodo sono confrontate tra loro in termini di funzioni di ribollimento. La configurazione progettata dal tracciamento molecolare rappresenta la minore energia richiesta per il ribollitore della colonna, elemento che influisce direttamente sui costi operativi dell'attività di separazione. Alla fine del progetto vengono condotte un'analisi di incertezza sull'errore di calcolo relativo alla volatilità e una valutazione economica delle possibili alternative progettuali di colonne di distillazione con prelievo laterale per i due casi studio. L'analisi mostra come il framework sviluppato dall'autore sia in grado di proporre lo stesso risultato anche con un'incertezza nella termodinamica del sistema in termini di relativa volatilità delle molecole. Da ultimo, la distillazione a prelievo laterale risulta economicamente più adatta per una miscela ternaria ideale con tracce di componente medio bollente.The early stage design of side-draw distillation units has always been a challenging task both for academic and industrial practitioners due to a lack of simple and reliable methods to determine the side-draw location corresponding to the minimum energy demand of the system. Existing methods are based on trial and error or complex mathematical optimization, which mostly become sophisticating and tedious and often also require a validated simulation of the unit. In this work, a novel, simple framework, similar in concept to conventional methods is proposed to design side-draw distillation units with infinite dilution of impurities of middle boiling component using the concept of molecular tracking. This concept is based on a probability function highly correlated to thermodynamic properties of the system, which evaluates that how probable is for a single molecule of a component in the mixture moving upward/downward on each stage in the unit. A systematic framework is developed to find the corresponding location of side-draw on a column. First, a binary distillation is designed for the key components in the mixture using driving force-based method well-known as an energy efficient approach. Since driving force-based method forces the column to operate at maximum driving force, the “column only” requires the minimum energy exchange for operation. An in detail illustration of which is discussed in this study. Then, the impurity is introduced in the feed as the middle boiling component and the probability profile of such impurity is generated based on the thermodynamic properties of this component and molar flows of the column. Thereafter, the side-draw is located on a tray, in which the molecules of the impurity prefer to stay, instead of moving either upward or downward. The concept of molecular tracking and driving force are both followed by simple examples to represent their functionalities. Moreover, two side-draw distillation units are designed for the two case studies of ideal ternary mixtures with molecular tracking framework. In order to compare molecular tracking with other existing methods, the side-draw location of each case is also found by classical driving force-based method. The configurations of each method are compared to each other in terms of reboiler duties. The configuration designed by molecular tracking represents lower energy required for the reboiler of the column, which directly affects operating costs of the separation task. At the end of the work, an uncertainty analysis on relative volatility miscalculation and an economic evaluation of the possible design alternatives of side-draw distillation for the two case studies are carried out. The uncertainty analysis illustrates that the framework developed by the author is capable of proposing the same result even with an uncertainty in system’s thermodynamics in terms of relative volatility. Moreover, the side-draw distillation is economically more suitable for an ideal ternary mixture with trace amounts of middle boiling component
An integrated multi-scale modeling framework for flocculation processes
No full textThe significance of the biochemical industry in the production of various chemical, biochemical, pharmaceutical, food, and many more products have become clearer in the past two decades and it is expected that this industrial sector attracts more attention and investment compared to the chemical industry in near future to move towards a more sustainable industry. While the fundamental knowledge is well-established in the chemical industry, this is not the case in the biochemical sector. Concerning the lack of fundamental and causal understanding in bioprocesses, the industry most often resorts to heuristics or the so-called recipe-based approaches in their daily operations. Although these approaches have been successful in addressing some of the challenges in the bioprocess industry, they are not ideal and in most cases are time-consuming and it is likely that lead to unnecessary product losses. On another note, most of the problems in both industries have a multi-scale nature and cover a wide range of spatio-temporal scales from atomic to ecological scales that seem to be too complex to be resolved with recipe-based approaches. To tackle these associated challenges and problems, systematic approaches are preferable to heuristics since they provide a causal understanding of the physical phenomena and provide the opportunity for further process optimization and control.The objective of this dissertation is to propose a systematic approach in the form of a multi-scale framework for modeling flocculation processes in the scope of process systems engineering discipline with the main focus on tools integration. The flocculation process has been selected as one of the challenging processes in various chemical and biochemical industries for downstream product purification. The process has a multi-scale nature with complex phenomena including agglomeration and breakage of particles concerning the system’s condition. Considering the status quo of this process, it is an excellent candidate to address some of the corresponding challenges by providing a multi-scale modeling framework. The multi-scale modeling framework developed in this work attempts to predict the future state of the flocculation processes by incorporating a priori physics-based knowledge and data-driven approaches spanning from nano-scale properties of the system to macro-scale properties. The first-principles models used in this work are the population balance model, molecular modeling, and computational chemistry, while deep neural networks are the data-driven modeling components in the proposed framework. The deep neural network is used to compensate for the gap of knowledge in the kinetics of the population balance model.To develop the multi-scale modeling framework, a bottom-up approach is used to compare various model integrations. The bottom-up approach shows the impact of first-principles model incorporation at various scales on the performance of the integrated model. For the hybrid multi-scale model, molecular modeling methods and computational chemistry calculations are employed to estimate the properties essential for the process kinetics from non-observable scales. Eventually, an efficient integrated scheme of these models with the previously developed hybrid model is suggested to predict the size distribution of particles in a future time horizon of the process. The application of the developed modeling framework at each step is demonstrated on a batch-mode shear-induced laboratory-scale flocculation of silica particles in water with pH as the main process variable
Monitoring, analysis and modeling of the flocculation process
No full textLAUREA MAGISTRALELa flocculazione consiste in un processo di aggregazione di particelle colloidali al fine di formare particelle più grandi e pesanti che possono sedimentare o essere rimosse mediante processi di filtrazione. Pertanto, la flocculazione rappresenta uno dei processi principali nella purificazione primaria nei downstream degli impianti di bio-produzione per la purificazione dei prodotti e rimozione di biomassa e detriti cellulari, con varie applicazioni nel trattamento delle acque reflue, nell'industria alimentare, nei processi farmaceutici. Il meccanismo di flocculazione è intrinsecamente complicato poiché si svolge su scale di lunghezza diversa partendo dalla nanoscala fino oltre alla microscala, il che porta a una mancanza di conoscenza nella modellazione, nel controllo e nella progettazione di questo processo. La modellizzazione della flocculazione può rappresentare un primo passo verso un suo migliore controllo e design, che comporterebbe una riduzione dei costi e di perdita del prodotto.
In questo progetto la modellizzazione del processo di flocculazione viene eseguita attraverso l'analisi e l'implementazione del principale set di equazioni differenziali ordinarie che descrive tale processo, le equazioni del bilancio di popolazione. Verrà innanzitutto costruito un modello matematico, basato su uno studio di letteratura, che sarà validato con i dati della letteratura. Successivamente, verrà eseguito un monitoraggio del processo mediante analisi dell'immagine su un sistema flocculante costituito da nanoparticelle di silice in acqua demineralizzata. L'obiettivo principale della caratterizzazione sperimentale è analizzare la flocculazione, comprendere quali parametri hanno impatto maggiore e raccogliere dati. Successivamente, verrà eseguita la stima dei parametri cinetici dell'esperimento, utilizzando metodi di minimizzazione non lineare dei minimi quadrati. Prima di applicare il metodo ai dati raccolti in laboratorio, la stima dei parametri verrà eseguita con i dati noti della letteratura, al fine di convalidare il metodo di minimizzazione.
Come risultati, sia la modellazione della flocculazione sia il metodo di minimizzazione sono stati validati con successo. Il monitoraggio della flocculazione della silice in acqua ha mostrato come il processo dipenda fortemente dal pH della soluzione. La flocculazione più intensa è stata osservata a pH intorno a 2, a causa della compressione del doppio strato di cariche che circondano le particelle, portando alla loro destabilizzazione e conseguente aggregazione.
Il tentativo di stima dei parametri cinetici del suddetto sistema flocculante ha mostrato una forte correlazione tra i due parametri da stimare, che ha portato alla stima di uno dei parametri con una precisione del 92%, mentre la stima del secondo parametro è risultata non altrettanto accurata.
Di conseguenza, a causa della forte correlazione tra i parametri, è stata eseguita infine una riduzione dei parametri, ipotizzando che il sistema allo stato stazionario venga raggiunto dopo qualche tempo.Flocculation is the process of colloidal particles aggregating in order to form bigger and heavier particles which can sediment or can be removed by filtration. Thus, the flocculation process represents one of the most important primary purification steps in downstream bio-manufacturing operations for the purification of products and removal of biomass and cell debris, with various applications in water/wastewater treatment, in food industry, in pharmaceutical production.
The flocculation mechanism is intrinsically complicated since it takes place across different length scales, which leads to a lack of knowledge in modelling, control and design of this process. Modelling of the flocculation may represent a first step towards a better control and design of the process, which would result in a reduction of costs, of the times and of the product loss.
In this project the modelling of the flocculation process is performed through the analysis and implementation of the main set of ordinary differential equations describing such process, the Population Balance Equations. First, a model will be built, based on literature study, and it will be validated against literature data. Subsequently, a process monitoring through image analysis will be performed in laboratory on a flocculant system made by silica nanoparticles in demineralized water. The main aim of the experimental characterization is to analyze the flocculation by understanding which parameters has major impact on it and collect data. Later, an attempt is made to estimate the kinetic parameters of the above-mentioned system, based on non-linear least squares minimization and curve fitting methods.
Before applying the method to the collected data, the parameters estimation is performed with the well- known literature data, in order to validate the minimization method.
As results, both flocculation modelling and minimization method are successfully validated. The monitoring of the flocculation of silica in water showed that the process is strongly depending on the pH of the solution. The agglomeration of particles is was observed at pH around 2, due to the compression of the double layer of charges that surround the particles, which leads to their destabilization and consequent aggregation.
The attempt the kinetic parameters estimation of the above-mentioned flocculant system showed a strong correlation between the two parameters to be estimated, which lead to one parameter being well estimated with an accuracy of 92% and the other one being not well estimated. As a consequence, a reduction of the parameters was performed by considering that the system reaches the steady state after a certain time
Going Beyond Counting First Authors in Author Co-citation Analysis
Get PDFThe present study examines one of the fundamental aspects of author co-citation analysis (ACA) - the way co-citation
counts are defined. Co-citation counting provides the data on which all subsequent statistical analyses and mappings
are based, and we compare ACA results based on two different types of co-citation counting - the traditional type that
only counts the first one among a cited work's authors on the one hand and a non-traditional type that takes into
account the first 5 authors of a cited work on the other hand. Results indicate that the picture produced through this non-traditional author co-citation counting contains more coherent author groups and is therefore considerably clearer. However, this picture represents fewer specialties in the research field being studied than that produced through the traditional first-author co-citation counting when the same number of top-ranked authors is selected and analyzed. Reasons for these effects are discussed
Variations on the Author
Get PDF“Variations on the Author” discusses two of Eduardo Coutinho’s recent films (Um Dia na Vida, from 2010, and Últimas Conversas, posthumously released in 2015) and their contribution to the general question of documentary authorship. The director’s filmography is characterized by a consistent yet self-effacing form of authorial self-inscription: Coutinho often features as an interviewer that rather than express opinions propels discourses; an interviewer that is good at listening. This mode of self-inscription characterizes him as an author who is not expressive but who is nonetheless markedly present on the screen. In Um Dia na Vida, however, Coutinho is completely absent form the image, while Últimas Conversas, on the contrary, includes a confessional prologue that moves the director from the margins to the center of his films. This article examines the ways in which these works stand out in the filmography of a director who offers new insights into the notion of cinematic authorship
Appropriate Similarity Measures for Author Cocitation Analysis
Get PDFWe provide a number of new insights into the methodological discussion about author cocitation analysis. We first argue that the use of the Pearson correlation for measuring the similarity between authors’ cocitation profiles is not very satisfactory. We then discuss what kind of similarity measures may be used as an alternative to the Pearson correlation. We consider three similarity measures in particular. One is the well-known cosine. The other two similarity measures have not been used before in the bibliometric literature. Finally, we show by means of an example that our findings have a high practical relevance.information science;Pearson correlation;cosine;similarity measure;author cocitation analysis
Dispelling the Myths Behind First-author Citation Counts
Get PDFWe conducted a full-scale evaluative citation analysis study of scholars in the XML research field to explore just how different from each other author rankings resulting from different citation counting methods actually are, and to demonstrate the capability of emerging data and tools on the Web in supporting more realistic citation counting methods. Our results contest some common arguments for the continued
use of first-author citation counts in the evaluation of scholars, such as high correlations between author rankings by first-author citation counts and other citation
counting methods, and high costs of using more realistic citation counting methods that are not well-supported by the ISI databases. It is argued that increasingly available digital full text research papers make it possible for citation analysis studies to go beyond what the ISI databases have directly supported and to employ more
sophisticated methods
- …
