Razvoj surogatnih algebrskih modelov kinetičnih reaktorjev : diplomsko delo univerzitetnega študijskega programa I. stopnje

Munđar, Petra

| | SLO | ENG | Cookies and privacy

Bigger font | Smaller font

First page > Show document

Show document

Title:	Razvoj surogatnih algebrskih modelov kinetičnih reaktorjev : diplomsko delo univerzitetnega študijskega programa I. stopnje
Authors:	ID Munđar, Petra (Author) ID Bogataj, Miloš (Mentor) More about this mentor... ID Nemet, Andreja (Comentor)
Files:	UN_Mundar_Petra_2023.pdf (2,02 MB) MD5: 506E06AC96A5D5626FEBD125BFDCC438
Language:	Slovenian
Work type:	Bachelor thesis/paper
Typology:	2.11 - Undergraduate Thesis
Organization:	FKKT - Faculty of Chemistry and Chemical Engineering
Abstract:	Modeliranje kemijskih kinetičnih reaktorjev na osnovi reakcijske kinetike je zahteven proces in običajno vsebuje vsaj dve diferencialni enačbi (snovna in energijska bilanca). Kadar poteka več reakcij modeli postanejo kompleksni in s tem numerično zahtevni. Take modele najpogosteje rešujemo z uporabo procesnih simulatorjev. Reševanje teh enačb v enačbno orientiranem algebrskem sistemu je težavno, saj moramo sisteme diferencialnih enačb pretvoriti v algebrske. Pretvarjanje le teh pa lahko naredimo na več načinov. V diplomski nalogi smo uporabili tehniko modeliranja, ki temelji na strojnem učenju ti. surogatni modeli, kjer smo iz kompleksnih modelov dobili enostavnejše, ki so primerni za enačbno orinetiran algebraični (EOA) sistem. Za razvoj surogatnih modelov smo uporabili programa ALAMO in Matlab. Cilj diplomske naloge je bil ustvariti matematične modele, ki opisujejo vhodno izhodne podatke in ustvariti zvezne odsekoma linearne modele, ki glede na različne vrednosti statističnega merila najbolje opišejo pridobljene podatke. Za vir podatkov smo uporabili simulacijo kinetičnega reaktorja za sintezo metanola. Simulacijo smo izvedli v programu Aspen Plus; kot podatke smo dobili množinske pretoke sinteznih plinov vzdolž cevnega reaktorja. Ugotovili smo, da lahko kompleksne nelinearne modele snovnega profila komponent v katalitskem cevnem reaktorju zadovoljivo opišemo s sistemom odsekoma zveznih linearnih funkcij in da so ob visokih vrednostih koeficienta determinacije modeli dober približek originalnega nelinearnega modela, ampak ne bodo zadostili masni bilanci. Raziskave razvoja surogatnih modelov so pomembne za področje procesne sistemske tehnike. Optimizacija in sinteza procesov, ki temeljijo na osnovi matematičnega programiranja, z večanjem računske moči in razvojem novih, hitrejših in učinkovitejših algoritmov, ne bodo več temeljila le na sposobnosti reševanja problemov, ampak bo poudarek tudi na napovedovanju natančnosti modelov, zato so raziskave na področju surogatnih modelov tehtne.
Keywords:	Kinetične reakcije, surogatni modeli, odsekoma zvezna linearna regresija, kinetični reaktor, algebrski modeli
Place of publishing:	Maribor
Place of performance:	Maribor
Publisher:	[P. Munđar]
Year of publishing:	2023
Number of pages:	1 spletni vir (1 datoteka PDF (XII, 35 f.))
PID:	20.500.12556/DKUM-85859
UDC:	66.011(043.2)
COBISS.SI-ID:	171885059
Publication date in DKUM:	10.10.2023
Views:	634
Downloads:	71
Metadata:
Categories:	KTFMB - FKKT
:	Copy citation

Average score:	(0 votes)
Your score:	Voting is allowed only for logged in users.
Share:

Hover the mouse pointer over a document title to show the abstract or click on the title to get all document metadata.

Licences

License:	CC BY-NC-ND 4.0, Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International

Link:	http://creativecommons.org/licenses/by-nc-nd/4.0/
Description:	The most restrictive Creative Commons license. This only allows people to download and share the work for no commercial gain and for no other purposes.
Licensing start date:	15.09.2023

Secondary language

Language:	English
Title:	Development of surrogate algebraic models of kinetic reactor
Abstract:	Modeling of chemical kinetic reactors based on reaction kinetics is a challenging process and typically involves at least two differential equations (mass and energy balances). When multiple reactions are involved,in the models they become complex and numerically demanding. Such models are most commonly solved using process simulators. If we want to solve these models in an equation-oriented algebraic system, we need to transform systems of differential equations into algebraic ones. The transformation of these can be done in various ways. In this diploma thesis, we used a modelling technique based on machine learning, i.e. surrogate models, where complex models have been converted into simpler models suitable for Equation-Oriented Algebraic (EOA). We used ALAMO and Matlab programs to develop surrogate models. The aim of the diploma thesis was to create mathematical models that describe input-output data and generate piecewise continuous linearmodels that best describe the obtained data based on various statistical criteria. We used a kinetic reactor simulation for methanol synthesis as the data source. The simulation was performed using the Aspen Plus program, and the result were molar flow rates of synthesis gases along the tubular reactor. We found that complex nonlinear models of component profiles in a catalytic tubular reactor can be satisfactorily described by a system of piecewise continuous linear functions. We also found that at high values of the coefficient of determination, the models are good approximation of the original nonlinear model, but they will not satisfy mass balance. Research into the development of surrogate models is relevant to the field of process systems engineering. In the future, as computational power increases and new, faster and more efficient algorithms are developed, optimisation and synthesis processes based on mathematical programming will no longer be based on problem-solving ability alone but will also focus on the predictive accuracy of models, which is why research into surrogate models is important.
Keywords:	Reaction kinetics, surrogate model, piecewise linear regression, kinetic reactor algebraic model

Comments

Leave comment

You must log in to leave a comment.

Comments (0)

0 - 0 / 0

There are no comments!

Back