9HSTFMG*afjiga+

ISBN 978-952-60-5986-0 (printed) ISBN 978-952-60-5985-3 (pdf) ISSN-L 1799-4934 ISSN 1799-4934 (printed) ISSN 1799-4942 (pdf) Aalto University School of Chemical Technology Department of Biotechnology and Chemical Technology www.aalto.fi

BUSINESS + ECONOMY ART + DESIGN + ARCHITECTURE SCIENCE + TECHNOLOGY CROSSOVER DOCTORAL DISSERTATIONS

Aalto-D

D 19

5/2

014

Aarne Sundberg

Micro-scale D

istillation and Microplants in P

rocess Developm

ent A

alto U

nive

rsity

Department of Biotechnology and Chemical Technology

Micro-scale Distillation and Microplants in Process Development

Aarne Sundberg

DOCTORAL DISSERTATIONS

Aalto University publication series DOCTORAL DISSERTATIONS 195/2014

Micro-scale Distillation and Microplants in Process Development

Aarne Sundberg

A doctoral dissertation completed for the degree of Doctor of Science (Technology) to be defended, with the permission of the Aalto University School of Chemical Technology, at a public examination held at the lecture hall Ke2 of the school on the 30th of January 2015 at 12.

Aalto University School of Chemical Technology Department of Biotechnology and Chemical Technology Chemical Engineering Research Group

Supervising professor Professor Ville Alopaeus Thesis advisor D. Sc. (Tech) Petri Uusi-Kyyny Preliminary examiners Professor Juha Tanskanen, University of Oulu, Finland Professor Ilkka Turunen, Lappeenranta University of Technology, Finland Opponent Reader Eva Sørensen, London's Global University, United Kingdom

Aalto University publication series DOCTORAL DISSERTATIONS 195/2014 © Aarne Sundberg ISBN 978-952-60-5986-0 (printed) ISBN 978-952-60-5985-3 (pdf) ISSN-L 1799-4934 ISSN 1799-4934 (printed) ISSN 1799-4942 (pdf) http://urn.fi/URN:ISBN:978-952-60-5985-3 Unigrafia Oy Helsinki 2014 Finland

Abstract Aalto University, P.O. Box 11000, FI-00076 Aalto www.aalto.fi

Author Aarne Sundberg Name of the doctoral dissertation Micro-scale Distillation and Microplants in Process Development Publisher School of Chemical Technology Unit Department of Biotechnology and Chemical Technology

Series Aalto University publication series DOCTORAL DISSERTATIONS 195/2014

Field of research Chemical Engineering Research Group

Manuscript submitted 29 April 2014 Date of the defence 30 January 2015

Permission to publish granted (date) 28 October 2014 Language English

Monograph Article dissertation (summary + original articles)

Abstract The purpose of this work was to evaluate the feasibility of using micro scale units for the

process development projects. The work consisted of three tasks; the measurement of thermodynamic equilibria, the development of a micro distillation column, and the use of microplants in process development. In the first task new thermodynamic data was measured; 361 vapor-liquid equilibria points for 10 binary mixtures, 83 vapor pressure points for pure compounds, 27 excess enthalpy of mixing, and 27 excess volume of mixing. The results were validated against literature values. Phase equilibria measurements were additionally validated using thermodynamic consistency, and by comparing the total pressure at equimolar composition. The results of phase equilibria measurements were modeled using Soave-Redlich-Kwong equation of state, and Hayden-O'Connel equation to calculate the vapor phase fugacity, and Non-Random Two Liquid, Wilson, Universal Quasi-Chemical and Universal Quasi-Chemical Functional Group models to calculate liquid phase relative activity coefficients. Results obtained from modeling were compared against literature values with good accuracy. A novel micro distillation column was developed based on a planar design using open-cell metal foam as column packing. Operation of the column was tested both as a heat-pipe and as part of a modular distillation unit. In the heat-pipe operation the column operation was controlled via surface temperatures. In the modular operation reboiler and condenser were located outside the column, which is the approach more commonly used in the chemical industry. The quality of data obtained from a microplant for the process development was evaluated using etherification as a model reaction. For this purpose the dynamic change in stream compositions, the reaction equilibrium, and the separation of product from feedstock was measured and modeled. The modeled results corresponded accurately with the experiments. This indicates that when the essential process steps are included in the trial plant the process model can be rigorously validated based on data obtained from micron scale piloting.

Keywords micro distillation, microplant, process development

ISBN (printed) 978-952-60-5986-0 ISBN (pdf) 978-952-60-5985-3

ISSN-L 1799-4934 ISSN (printed) 1799-4934 ISSN (pdf) 1799-4942

Location of publisher Helsinki Location of printing Helsinki Year 2014

Pages 122 urn http://urn.fi/URN:ISBN:978-952-60-5985-3

Tiivistelmä Aalto-yliopisto, PL 11000, 00076 Aalto www.aalto.fi

Tekijä Aarne Sundberg Väitöskirjan nimi Mikromittakaavan tislaus ja koelaitteisto prosessikehityksessä Julkaisija Kemian tekniikan korkeakoulu Yksikkö Biotekniikan ja kemian tekniikan laitos

Sarja Aalto University publication series DOCTORAL DISSERTATIONS 195/2014

Tutkimusala Kemian laitetekniikan laboratorio

Käsikirjoituksen pvm 29.04.2014 Väitöspäivä 30.01.2015

Julkaisuluvan myöntämispäivä 28.10.2014 Kieli Englanti

Monografia Yhdistelmäväitöskirja (yhteenveto-osa + erillisartikkelit)

Tiivistelmä Työn tarkoituksena oli tutkia mikromittakaavan koelaitteiston käyttökelpoisuus

prosessikehityksen kannalta. Työ jakaantui kolmeen eri osaan; termodynaamisen tasapainon mittauksiin, mikrotislaimen kehittämiseen ja prosessikehitykseen mikromittakaavan koelaitteistolla. Termodynaamisen tasapainon mittauksissa mitattiin kymmenelle kahden komponentin seokselle 361 neste-höyry–tasapainopistettä, 83 höyrynpainepistettä puhtaille komponenteille, sekä 27 eksessientalpiapistettä ja eksessitilavuuspistettä. Mittausten oikeellisuus varmistettiin vertaamalla tuloksia kirjallisuusarvoihin. Faasitasapainomittausten oikeellisuus varmistettiin lisäksi termodynaamisen konsistenssin perusteella ja vertaamalla mittausten kokonaispainetta ekvimolaarisessa pitoisuudessa. Faasitasapainomittauksen tulokset mallinnettiin käyttäen Soave-Redlich-Kwong tilanyhtälöä ja Hayden-O'Connel yhtälöä höyryfaasin fugasiteettien laskemiseen sekä Non-Random Two Liquid, Wilson, Universal Quasi-chemical ja Universal Quasi-chemical Functional group Activity Coefficients malleja nestefaasin suhteellisten aktiivisuuskertoimien laskentaan. Mallinnuksen oikeellisuus todennettiin vertaamalla aktiivisuuskerroinmalleilla laskettuja tasapainopisteitä kirjallisuudessa esitettyihin tuloksiin. Uuden tyyppinen mikrotislain kehitettiin perustuen levymäiseen rakenteeseen, jossa käytettiin avosoluista metallivaahtoa täytekappaleena. Mikrotislainta tutkittiin kahdella erilaisella operointitavalla; lämpöputkityyppisellä ja modulaarisella. Lämpöputkityyppisessä operoinnissa tislaimen toimintaa säädettiin pintalämpötilan avulla. Modulaarisessa operointitavassa pohjakiehutin ja höyryn lauhdutin sijaitsivat tislaimen ulkopuolella, mikä on yleisempi ratkaisu teollisen mittakaavan tislaimissa. Mikromittakaavan koelaitteistosta saatavan tiedon käyttökelpoisuutta prosessikehityksessä tutkittiin käyttäen eetteröintiä mallireaktiona. Työssä mitattiin ja mallinnettiin virtojen koostumuksen muuttumista, reaktiotasapainoa ja tuotteiden erotusta lähtöaineista. Mittaustulokset korreloivat suurella tarkkuudella prosessimallin kanssa. Tämä osoittaa, että kun olennaiset prosessivaiheet on sisällytetty koelaitteistoon, voidaan mikromittakaavan laitteistosta saatavaa tietoa hyödyntää prosessimallien validointiin. Avainsanat mikrotislaus, mikroplant, prosessikehitys

ISBN (painettu) 978-952-60-5986-0 ISBN (pdf) 978-952-60-5985-3

ISSN-L 1799-4934 ISSN (painettu) 1799-4934 ISSN (pdf) 1799-4942

Julkaisupaikka Helsinki Painopaikka Helsinki Vuosi 2014

Sivumäärä 122 urn http://urn.fi/URN:ISBN:978-952-60-5985-3

Preface The work described in this thesis was carried out at the Laboratory of Chemical Engineering, Helsinki University of Technology (later known as Aalto University), between August 2007 and December 2012. The financial support from the Fortum Foundation, from the Finnish Funding Agency for Technology and Innovation, and from the Academy of Finland is gratefully acknowledged. I am most grateful to my instructor Petri Uusi-Kyyny for his guidance and support throughout the preparation of this thesis. It is doubtful, that this thesis would have ever been completed without his continuing support. I thank professors Juhani Aittamaa and Ville Alopaeus for their help. I give my warmest thanks to all my colleagues in the Laboratory of Chemical Engineering, especially to Minna Pakkanen and Lasse Westerlund. My humble gratitude and my love to Pia, who supported me even when my experimental work was not progressing as well as I expected. Finally, my love to Adela, Zsuzsa, and Mauri. Aarne Sundberg

List of publications This thesis is based on the following publications, which are referred to in the text by their Roman numerals. I. Sundberg, Aarne; Uusi-Kyyny, Petri; Pakkanen, Minna; Alopaeus,

Ville. Vapor–Liquid Equilibrium for Tetrahydrothiophene + n-Butane, + trans-2-Butene, + 2-Methylpropane, and + 2-Methylpropene. J. Chem. Eng. Data. 2009, 54, 1311–1317.

II. Sundberg, Aarne; Uusi-Kyyny, Petri; Pakkanen, Minna; Alopaeus,

Ville. Vapor−Liquid Equilibrium for Methoxymethane + Methyl Formate, Methoxymethane + Hexane, and Methyl Formate + Methanol, J. Chem. Eng. Data. 2011, 56, 2634–2640.

III. Sundberg, Aarne; Laavi, Helena; Kim, Y.; Uusi-Kyyny, Petri; Pokki,

Juha-Pekka; Alopaeus, Ville. Vapor-Liquid Equilibria, Excess Enthalpy, and Excess Volume of binary mixtures containing an alcohol (1-butanol, 2-butanol, or 2-methyl-2-butanol), and 2-ethoxy-2-methylbutane, J. Chem. Eng. Data. 2012, 57, 3502-3509.

IV. Sundberg, Aarne; Uusi-Kyyny, Petri; Alopaeus, Ville. Microscale

distillation, Russ. J. Gen. Chem. 2012, 82, 2079-2087.

V. Sundberg, Aarne; Uusi-Kyyny, Petri; Alopaeus, Ville. Novel micro-distillation column for process development. Chem. Eng. Res. Des. 2009, 87, 705-710.

VI. Sundberg, Aarne; Uusi-Kyyny, Petri; Jakobsson, Kaj; Alopaeus, Ville.

Control of reflux and reboil flow rates for milli and micro distillation. Chem. Eng. Res. Des. 2012, 91, 753-760.

VII. Sundberg, Aarne; Uusi-Kyyny, Petri; Alopaeus, Ville. The use of

Microplants in Process Development – Case Study of the Etherification of 2-Ethoxy-2-methylbutane. Chem. Eng. Process. 2013, 74, 75-82.

The author’s contribution to the publications

I-II. The author evaluated the results, modeled the systems, and wrote the manuscript together with co-authors. III. The author evaluated the results, modeled the systems, and wrote the manuscript together with co-authors. The author synthetized and

purified the 2-ethoxy-2-methylbutane used in all the experiments. IV. The author made the literature survey, analysed the results, and

wrote the manuscript. V-VII. The author designed and constructed the experimental system,

planned and performed the experiments, evaluated and modeled the results, and wrote the manuscript.

Table of contents 1 Introduction .................................................................................................................................. 1

1.1 Objectives and Scope ................................................................................................................. 2

2 Components of interest ................................................................................................................ 3

2.1 2-Ethoxy-2-methylbutane ........................................................................................................... 3 2.2 Tetrahydrothiophene .................................................................................................................. 3 2.3 Methoxymethane ........................................................................................................................ 3 2.4 Methyl formate ........................................................................................................................... 4

3 Phase equilibria ............................................................................................................................ 5

3.1 Vapor Pressure .......................................................................................................................... 5 3.2 Vapor-liquid equilibria .............................................................................................................. 5 3.3 Static total pressure method ...................................................................................................... 9 3.4 Recirculation method ............................................................................................................... 10 3.5 Excess enthalpy and excess volume ......................................................................................... 11 3.6 Quality of measured data ......................................................................................................... 11 3.7 Use of VLE models in process modeling .................................................................................. 12

4 Distillation ................................................................................................................................... 14

4.1 Miniaturization of distillation columns .................................................................................... 14 4.2 Novel micro-distillation column .............................................................................................. 17 4.3 Effect of design parameters and metal foam properties .......................................................... 23

5 Trial Plants .................................................................................................................................. 27

5.1 Microplant ............................................................................................................................... 28 5.2 Case Study – Etherification of 2-ethoxy-2-methylbutane ......................................................... 28 5.2.1 Reaction Kinetics ..................................................................................................................... 30 5.2.2 Distillation Model .................................................................................................................... 31 5.2.3 Process Model ......................................................................................................................... 31 5.2.4 Accumulation of Impurities ...................................................................................................... 33 5.2.5 Mass Balance ........................................................................................................................... 33

6 Conclusions ................................................................................................................................. 35 7 Literature cited ........................................................................................................................... 36

1

1 Introduction Process development can be seen as a series of consequential scale-up tasks leading to a full-scale process. These tasks are usually conducted via a number of intermediate scale trial plants, where the process feasibility is evaluated before committing to higher expenses [1]. The aim of these trial plant operations is to reduce the risks associated with the scale-up. This potential for cost savings was understood already in the early 20th century by pioneers, such as Leo Hendrik Baekeland, who in 1903 build and used trial plants for testing of electrolytic processes. A widely quoted maxim of his was “Commit your blunders on a small scale and make your profits on a large scale.” [2]. In the early days of chemical engineering, the scale-up was based mainly on the physical similarity of the units. This approach did not guarantee a successful scale-up, and led in general to very modest scale-up ratios; the use of three intermediate steps was recommended between the laboratory stage and the commercial plant [3]. Even in the 90’s, some rules of thumb scale-up factors were in the range of 5 to 15 [4], while in the modern process development, even scale-up values of 100 000 are sometimes met [5, 6]. The advanced use of modeling has reduced the degree of uncertainty related to the scale-up, but it has not completely eliminated the need for experimental work, as a model can be only as good as the data it is based on. As less than 5% of physical properties for compounds can be found in data banks, and even less for mixtures, trace compounds are routinely omitted from modeling work or modeled as ideal mixtures leading to incorrect simulations [7]. Additionally, compounds below the routinely reported detection limits occasionally accumulate in the recycle stream causing malfunctions [8]. According to Kister, incorrect simulations and intermediate component accumulations caused over 7% of tower malfunctions [9]. Thus experimental work is still needed to improve the accuracy of models, and to validate models. As part of this work, the vapor-liquid equilibria, and excess enthalpy of mixing for selected binary systems was measured and modeled. As most reactions have less than complete conversion of basic materials into products, the unreacted feed is separated and recycled back into the reactor to improve the yield. These recycling streams, which are commonly an economic necessity, may raise new problems due to accumulation of impurities. A by-product or an impurity in the feed, originally below the detection limit in a batch wise operation, could increase to harmful concentration with a recycle. The integrated trial plant with all the recycling

2

streams is usually the first stage where the characteristic problems of positive feedback and accumulation of impurities are encountered. [1, 5] In the model-based approach all process units are modeled separately based on the physical and chemical phenomena occurring within, such as reaction kinetics and phase equilibria. The unit models are verified against stand-alone experiments, or combined directly into a process model according to a process flow diagram. This process model is then validated against data obtained from an integrated trial plant operation to ascertain that all the important aspects of the process were taken into account [8]. The combination of modeling and experimental work makes it possible to use scale-up ratios up to 100 000 [5], as demonstrated by Lievo et al. [6]. This scale-up ratio is often limited by the miniaturization limits of corresponding process units, especially the separation step [1]. The smallest commercially available distillation columns exceed the flow rate offered by micro reactors by a factor of 100. For this reason a novel micron scale distillation column suitable for use in process development was developed and characterized. The feasibility of using the column as part of integrated trial plant was evaluated using the modeling approach. The process was modeled based on phase equilibria, reaction equilibria, reaction kinetics, distillation column separation, and the liquid holdup of the system. The model was evaluated against the experimental results with good correspondence, thus proving the feasibility of microplants.

1.1 Objectives and Scope This thesis had three related objectives, as detailed below. To measure and publish new phase equilibria data for selected binary mixtures. This objective was met in publications [I, II, III]. To describe the operation and characteristics of micro-scale distillation, with focus on the planar packed column type. This objective was met in publications [IV, V, VI]. To evaluate the feasibility of using microplants for process development projects in the chemical process industry. This objective was met in publications [VII].

3

2 Components of interest

2.1 2-Ethoxy-2-methylbutane The 2-ethoxy-2-methylbutane (tert-amyl-ethyl-ether, TAEE) is a gasoline additive to improve the octane rating and to improve the combustion of gasoline. TAEE is produced from 2-methylbut-2-ene or 2-methylbut-1-ene and ethanol. When ethanol used for etherification comes from renewable sources, the use of TAEE releases less carbon emissions than fossil based fuels. An advantage of TAEE over ethanol is that TAEE reduces the vapor pressure of gasoline blends, while ethanol significantly increases it. For this reason, the amount of ethanol in blends is capped in many countries, while at the same time the use of other bio-based fuel components is encouraged by legislation.[10, 11, 12, 13] In this work the synthetization of TAEE was used as the demonstration case for the evaluation of the micro plant [VII], and the phase equilibria for this tertiary system was used to characterize the separation of the distillation column [VI]. Additionally, the vapor-liquid equilibria, excess enthalpy of mixing, and volume of mixing for binary mixtures containing TAEE and butanols were measured and modeled [III].

2.2 Tetrahydrothiophene Liquefied petroleum gas (LPG) is a mixture of hydrocarbon gases. It consists mainly of propane and butane, which both form flammable mixture with air [14]. To mitigate the risk of LPG explosion caused by an undetected leak, legislation requires the addition of odorants into LPG [15]. The most commonly used odorant compound is tetrahydrothiophene (thiophane, THT). It can be detected by its characteristic strong sweet smell [16]. To evaluate the amount of THT needed to obtain the required THT concentration in the gas phase, the vapor-liquid equilibria of binary mixtures containing THT and butanes was measured and modeled [I].

2.3 Methoxymethane Methoxymethane is a colourless, odourless and gaseous in standard temperature and pressure. It is used as propellant and substitute for LPG especially in China [17]. Methoxymethane is produced from synthesis gas via methanol dehydration step, or via single-step processes [18]. To improve

4

the accuracy of process models the vapor-liquid equilibria for binary mixtures of methoxymethane and methyl formate, and methoxymethane and hexane, was measured and modeled [II].

2.4 Methyl formate Methyl formate or formamide is an intermediate in the formic acid synthesis. It is formed by the carbonylation of methanol with carbon monoxide. The formed methyl formate is hydrolyzed to methanol and formic acid, which is then used in textile and leather industry in dying, or as an intermediate in the chemical and pharmaceutical industry [19]. To improve the accuracy of process models the vapor-liquid equilibria for the binary mixture of methyl formate and methanol was measured and modeled [II].

5

3 Phase equilibria

3.1 Vapor Pressure Pressure exerted by vapor against liquid or solid in thermodynamic equilibrium is called the vapor pressure. It is specific for each compound. It is independent of system pressure but depends strongly and in a nonlinear manner on the system temperature, as presented in Figure 1. For applications that require accurate vapor pressure correlation over a wide temperature range the most commonly used equation is the Antoine's equation. In this work the DIPPR 101 modification of the Antoine’s equation (1) [20] was used as the vapor pressure correlation.

EDTTCTBAP lnln (1)

Figure 1. Saturated Vapor Pressure p, for Temperature T, of 2-Ethoxy-2-methylbutane. [21], ;[22], ;[23], ; [24], —; [III], . The solid line represents the DIPPR 101 modification of the Antoine's equation [20]. The vapor pressure of pure compounds was measured at the beginning of every experiment in [I-III] both to verify the success of purification and gas removal steps, and to improve the accuracy of the vapor-liquid model. Overall, 83 new vapor pressure points were reported in this work.

3.2 Vapor-liquid equilibria For a pure compound in thermodynamic equilibrium the two phases may coexist only if the system pressure equals the vapor pressure. For a system containing two or more components the pressure range where more than one

LIQUID PHASE

VAPOR PHASE

6

phase may coexist in equilibrium is wider, as demonstrated in Figure 2 for a mixture of 2-ethoxy-2-methylbutane and 1-butanol.

Figure 2. Experimental vapor-liquid equilibrium for pressure p, liquid-phase mole fraction x,( ); and gas-phase mole fraction y,( ); at a temperature of 358.17 K, for the binary system of 2-ethoxy-2-methylbutane (1) + 1-butanol (2). The solid line represents the Wilson model [25]. [III]. If the system pressure is below both lines presented in Figure 2, the system is in vapor phase. As the system pressure increases, vapor begins to condensate. The point where the first drop appears is called the dew point and it forms the lower line. Above the lower line the system both liquid and vapor phase may exist. As the pressure is further increased, the upper line is reached. It is called the bubble point and it is characterized as the point where the first bubble appears within the liquid phase as the pressure is decreased. Above the upper line the system is in the liquid phase. At any pressure, the concentration of the liquid phase and the vapor phase can be evaluated from these two lines. The concentration of the two phases differ from each other depending on the volatility of the compounds, with the lower boiling (light) component being of higher concentration in the vapor phase, and the higher boiling (heavy) component being of higher concentration in the liquid phase. A number of separation methods are based on this phenomenon, which is called vapor-liquid equilibria (VLE). To optimize the industrial processes accurate knowledge of the VLE behaviour is required. As measuring the actual VLE for complex mixtures at various process conditions would be an overwhelming task, the VLE behaviour is estimated with the help of modeling. The experimental data measured and the process conditions studied in this work were in sub-critical conditions both for pressure and temperature. In these conditions the liquid phase can be accurately modeled using correlative liquid activity coefficient models, such as the local composition models: Wilson model

LIQUID PHASE

VAPOR PHASE

AZEOTROPE

7

[25], Non-Random Two Liquid model [26], and Universal Quasi-chemical model [27], or estimated with Universal Quasi-Chemical Functional Group Activity Coefficients model (Dortmund) [28]. The vapor phase can be modeled with the cubic Soave-Redlich-Kwong equation of state [29], or with the Hayden - O’Connell equation [30, 31]. Binary systems consisting of hydrocarbons with nearly equal molar mass can also be modeled with the ideal liquid and gas assumption with acceptable deviation. In [I] the vapor-liquid equilibrium for tetrahydrothiophene with C4 components was measured with the static total pressure apparatus. The systems presented in [I] had not been published earlier. The four systems were modeled using local composition models for the liquid phase and the Soave-Redlich-Kwong equation of state for the vapor phase, as presented in Figure 3 for Wilson model.

Figure 3. Experimental pressure, and vapor and liquid phase equilibrium composition at 318 K in mole fractions of tetrahydrothiophene + n-butane,

; + 2-methylpropane, ; + 2-methylpropene, ; + trans-2-butene, . The solid line represents values calculated with Wilson model. Deviation from the experimental results was small for all three local composition models. The systems were also described using the Universal Quasi-Chemical Functional Group Activity Coefficients model (Dortmund) with the Soave-Redlich-Kwong equation of state, but it was left out of the manuscript as calculation using parameters for similar chemical groups did not yield acceptable results; parameters for tetrahydrothiophene had not been published. All four systems exhibited positive deviation from the Raoult’s Law without azeotropic behaviour. In [II] the vapor-liquid equilibrium for the binary systems of methoxymethane + methyl formate, methoxymethane + hexane, and methyl

8

formate + methanol was measured with the static total pressure method. The systems containing methoxymethane had not been published earlier. The three systems were modeled with local composition models together with Soave-Redlich-Kwong equation of state with good accuracy. The capability of Universal Quasi-Chemical Functional Group Activity Coefficients model (Dortmund) to describe these systems was slightly less accurate than that of the local composition models, especially for the binary system of methoxymethane + methyl formate. The Wilson model, based on the experimental data for methyl formate + methanol, was compared against the works published by Kozub et al. [32], Polak and Lu[33], Zharov and Pervukhin [34], and Zheng et al. [35] with good accuracy, as presented in Figure 4. Comparison against the work by Sohlo and Holma [36] was the least accurate. This was not surprising, as their work lacked the experimental points and pressure correlation parameters, which made recalculation less accurate.

Figure 4. Vapor and liquid phase equilibrium composition in mole fractions of methyl formate + methanol. [II] at 358.94 K, ; [34] at 318.15 K, ; [32] at 303.15 K, ; [33] at 298.15 K, . Lines represent values calculated with the Wilson model in [II]. All three systems presented in [II] exhibited positive deviation from the Raoult’s Law. The methyl formate + methanol system showed pressure maximum azeotropic deviation, which was calculated with the Wilson model to methyl formate concentration of 0.937 at the temperature of 358.94 K. Regression of data published by Zheng et al. [35] showed similar behaviour, which supports our observations. In [III] the vapor-liquid equilibria for the systems of 2-ethoxy-2-methylbutane + 1-butanol, + 2-butanol, and + 2-methyl-2-butanol was

9

measured using a flow recirculation still. The binary systems were modeled with the local composition models with good accuracy, as presented in Figure 5. The accuracy of Universal Quasi-Chemical Functional Group Activity Coefficients model (Dortmund) estimation was also fair.

Figure 5. Experimental (Vapor + Liquid) Equilibrium for Pressure p, Liquid-Phase Mole Fraction x, and Gas-Phase Mole Fraction y, at Temperature of 358 K. 2-Ethoxy-2-methylbutane (1) + 1-Butanol (2), x, , y, ; 2-Ethoxy-2-methylbutane (1) + 2-Butanol (2), x, , y, ; 2-Ethoxy-2-methylbutane (1) + 2-Methyl-2-butanol (2), x, , y, . The solid lines represent the Wilson model, the dashed lines represent Universal Quasi-Chemical Functional Group Activity Coefficients model (Dortmund). [III].

3.3 Static total pressure method Experimental data for [I] and [II] was obtained with the static total pressure method. This method is characterized by the injection of known samples of pure compounds into a closed vessel, also known as equilibrium cell, which is equipped with temperature and pressure probes as presented in Figure 6.

Figure 6. Static total pressure apparatus.

10

The static total pressure apparatus used in this work has been described in detail by Uusi-Kyyny et al. [37]. After injection the cell is heated or cooled to a predetermined temperature, followed by thorough mixing of the compounds. After mixing the cell is left undisturbed until a steady state is achieved. This is observed from temperature and pressure measurements, at which point the measurements are recorded. The data point thus obtained consists of the total concentration within the cell, temperature, and pressure at equilibrium. Concentration of the two phases can then be reduced from the experimental data by the Barker method [38] using models presented in 3.2. The quality of data is evaluated based on the ability of the model to describe data within the uncertainty of the pressure measurement. In this work this data fitting was performed using VLEFIT-software [39].

3.4 Recirculation method The recirculation method, which was used in [III], is based on the partial vaporization of liquid sample and the analyzation of concentration in both phases and the measurement of system temperature and pressure. The advantage of this method is that the experimental data obtained can be checked for thermodynamic consistency. The recirculation method can be used for pure compounds, binary mixtures, and for mixtures consisting of 3 or more compounds. The recirculation still used in this work was based on the work by Yerazunis et al. [40], with modifications described by Uusi-Kyyny et al.[41], as presented in Figure 7. The apparatus and the experimental setup have been described by Uusi-Kyyny et al. [41], and by Sapei et al. [42].

Figure 7. Recirculation still.

11

3.5 Excess enthalpy and excess volume The excess enthalpy and excess volume of mixing describe how much a system deviates from the ideal mixture. In the modeling of vapor-liquid equilibria the excess enthalpy is used in the extrapolation of the temperature dependency of liquid phase activity model outside the experimental range by using the Gibbs–Duhem equation [43]. A calorimeter equipped with static mixing cell was used to measure the excess enthalpy of binary mixtures followed by a densimeter used to measure the excess volume [III].

3.6 Quality of measured data Thermodynamic consistency of measured data can be evaluated only if the experimental data contains at least one degree of freedom. This condition was fulfilled in the measurements with the recirculation still, as the concentration of both phases together with temperature and pressure were determined. Thus the data measured with the recirculation still in [III] was tested with the integral test [44], the infinite dilution test [45], and the point test [44]. The measured data passed these tests and were considered to be of good quality. For the static total pressure apparatus the quality of the data was evaluated based on how well the measured pure compound vapor pressure agreed with the literature values, and how well the vapor pressure of the two ends of the composition range coincided at the equimolar composition. In the two works [I, II] where static total pressure apparatus was used, the vapor pressure agreed with the literature values within the uncertainty of the measurement, and the two sides of the isothermal VLE data coincided. Additionally, the model was able to describe the data below the experimental uncertainty, which also indicates that the data were of good quality. A summary of VLE measurements reported in this work is presented in Table 1. Table 1. Summary of vapor-liquid equilibria measurements. Ref. Type Compound 1 Compound 2 T/K Pts

[I] VLE (zi, T, p) n-Butane Tetrahydrothiophene 318 347

24 24

[I] VLE (zi, T, p) 2-Methylpropane Tetrahydrothiophene 318 347

24 24

[I] VLE (zi, T, p) 2-Methylpropene Tetrahydrothiophene 318 347

24 24

[I] VLE (zi, T, p) Trans-2-Butene Tetrahydrothiophene 318 347

24 24

12

[II] VLE (zi, T, p) Methoxymethane Methyl Formate 308 24

[II] VLE (zi, T, p) Methoxymethane Hexane 308 336

24 24

[II] VLE (zi, T, p) Methyl Formate Methanol 359 24

[III] VLE (xi, yi, T, p) 2-Ethoxy-2-methylbutane

1-Butanol 358 21

[III] VLE (xi, yi, T, p) 2-Ethoxy-2-methylbutane

2-Butanol 358 28

[III] VLE (xi, yi, T, p) 2-Ethoxy-2-methylbutane

2-Methyl-2-butanol 358 24

For the excess enthalpy (HE) measurements in [III] the evaluation of data quality was based on the calibration of the calorimeter with known samples. For the excess volume (VE) measurements in [III] the evaluation of data quality was based on modeling with the Redlich-Kister model [46]. The model could describe the experimental data with good accuracy even with one parameter so the data was considered to be of good quality. A summary of HE and VE measurements reported in this work is presented in Table 2. Table 2. Summary of excess enthalpy (HE) and excess volume (VE) measurements. Ref. Type Compound 1 Compound 2 T / K Pts [III] HE, VE 2-Ethoxy-2-

methylbutane 1-Butanol 298 9

[III] HE, VE 2-Ethoxy-2-methylbutane

2-Butanol 298 9

[III] HE, VE 2-Ethoxy-2-methylbutane

2-Methyl-2-butanol 298 9

3.7 Use of VLE models in process modeling In [V] the separation capability of various distillation columns was characterized with hexane + cyclohexane and n-hexane + cyclohexane + heptane systems. As both systems behaved in nearly ideal manner, they were modeled using original Universal Quasi-Chemical Functional Group Activity Coefficients model for the liquid phase and ideal gas model for the vapor phase. In [VI] the distillation column was characterized with methyl formate + methanol, 2-methylbut-2-ene + ethanol + 2-ethoxy-2-methylbutane, and n-hexane + cyclohexane systems. The two first systems were modeled using Non-random Two Liquid Model as the correlative liquid activity coefficient

13

model, and Soave-Redlich-Kwong equation of state as the vapor phase fugacity coefficient model. The n-hexane + cyclohexane system was modeled using Soave-Redlich-Kwong equation of state for both phases. In [VII] the liquid phase was modeled using Wilson model as the correlative liquid activity coefficient model, and the vapor phase was modeled using Soave modification of Redlich-Kwong equation of state as the vapor phase fugacity coefficient model. The literature data used for modeling in [VI] and [VII] is presented in Table 3. Table 3. Phase equilibria data used in the modeling. Binary system Referenc

e Methyl formate + methanol [III]a Methyl formate + methanol [32]a Methyl formate + methanol [33]a Methyl formate + methanol [35]a Methyl formate + methanol [34]a 2-Methylbut-2-ene + 2-Ethoxy-2-methylbutane [47]b 2-Methylbut-1-ene + 2-Ethoxy-2-methylbutane [47]b Ethanol + 2-Methylbut-1-ene [48]a Ethanol + 2-Methylbut-1-ene [47]b Ethanol + 2-Methylbut-2-ene [47]b Ethanol + 2-Ethoxy-2-methylbutane [49]a Ethanol + 2-Ethoxy-2-methylbutane [50]a Ethanol + 2-Ethoxy-2-methylbutane [51]b Ethanol + 2-Ethoxy-2-methylbutane [47]b Ethanol + 2-Ethoxy-2-methylbutane [52]a Ethanol + Pentane [53]a Ethanol + Pentane [54]a Ethanol + Pentane [55]c Ethanol + Pentane [47]b a Vapor-liquid equilibria data b Azeotropic conditions data c Excess enthalpy data

14

4 Distillation According to Fair [56] distillation is the most commonly used separation method in the chemical engineering accounting for 95% in volume of liquid separations. This role is based on the fact that distillation is the only separation method capable of separating a multi-component mixture into all its constituents in a single column. Distillation, or rectification, is based on the generation of a two–phase system, the mass transfer between the phases, and the separation of phases. The principle is similar to the alternative separation methods such as absorption, adsorption, and extraction, except that distillation does not require additional components for creating the separation. Instead, the addition and removal of heat is used to generate vapor phase for the two-phase system. [57, 58] In a two-phase vapor-liquid system with more than one component, both phases are driven towards a state of equilibrium. At equilibrium, both phases have the same temperature and total pressure. However, with the exception of azeotropic conditions, in equilibrium the liquid and vapor phases have different concentration. This separation of components is the key phenomena behind distillation, which makes knowledge of vapor-liquid equilibria essential for the design of distillation processes. [57, 58] The separation by one step evaporation is rarely sufficient for the needs of the chemical industry. The separation can be improved by stacking these separation steps together in a counter current cascade to form a fractionating column, with vapor flow driven by pressure difference towards the condenser, and liquid by gravitational force towards the reboiler. The efficient mass transfer between the two phases is achieved either by bubbling vapor through the liquid phase on the column plates or allowing the liquid to flow on packing surfaces in contact with the vapor. [57, 58]

4.1 Miniaturization of distillation columns Distillation is one of the most significant unit operations in the chemical industry and is an essential part in downstream processing of most chemical processes. In piloting, where the process model is validated against experimental results, the overall cost depends heavily on the scale of trial plant. Thus, strong drive exists towards miniaturization of trial plants. [1] To realize the potential of the miniaturization of trial plants into laboratory scale, all process units must be available at the same scale. For distillation the smallest commercially available column packing has a diameter of 30

15

mm, requiring throughput of over one kilogram per hour [VI], which is ten times the flow rate specified for microplants [59]. Although a need for a continuous micro scale distillation exists, relatively little research has been published on the topic, as discussed in [IV], by Ziogas et al.[60], by Kenig et al. [61], and by Lam et al. [62]. As the column dimensions are reduced by the use of miniaturization, the significance of gravitation is reduced in lieu of surface forces. To increase the surface forces and to realize micron scale distillation requires the use of novel column designs and packing materials. An often overlooked challenge is how to compensate for the heat losses. This problem is magnified by the high surface temperature of the column combined with the high surface-to-volume ratio. This makes rigorous evaluation of heat balance and the accurate heating control critical issues for the steady operation of these units. The earliest work recognized as micro distillation was published by Seok and Hwang [63]. The column presented in the work was made of a circular pipe with porous glass sinter placed on the inside walls of the column to facilitate liquid flow. The column was operated in a horizontal orientation with the ends of the column heated and cooled via column walls, in a similar manner to the operation of a heat-pipe. At the hot end liquid evaporated and was driven by pressure difference towards the cold end. At the cold end vapor condensed and flowed through the porous wick towards the hot end driven by capillary forces. As products were withdrawn as liquid from the column no auxiliary reboiler or condenser was required. As a downside the operation of the column required separate removal of incondensable gases from the column prior to and possibly during its use. A similar design was later patented by TeGrotenhuis and Stenkamp [64]. The design was evaluated by Zheng et al. [65] and by Huang et al.[66]. The design employed a rectangular channel with the wicking material placed only at the bottom of the channel. As mass transfer took place in a vertical direction the width of the channel could in theory be freely increased to achieve desired throughput without affecting separation, which was an improvement over the earlier design. Additional advantage of the planar design was the easy scaling up using numbering-up method. In [V] this approach was further improved by using open-cell metal-foam as column packing. An approach first published by Find and Hampe [67] was based on the falling film principle. In their work a downward flowing liquid film was formed in the annulus of two vertical heated pipes, which caused partial

16

evaporation of liquid. The vapor was removed from the top of the column thus creating counter current liquid-vapor contact. However, the control of heating and flow profile proved to be problematic. An improved design employing planar shape with a wicking structure and multiple separation sections in vertical orientation was patented by Tonkovich et al.[68], with experimental results and simulation published by Arora et al. [69]. A more conventional approach towards the miniaturization was published by Ziogas et al. [70, 60] presenting a column design closely resembling a 2-dimensional slice of a plate column with changeable plate inserts. The column was heated directly through column walls, which made finding the optimal temperature profile for different systems tedious. An even greater disadvantage of the heating method was that quantifying of heat losses, and thus the calculation of heat balance, was not possible. However, the device was suitable for analytical separations and removal of solvents. A design utilizing planar design with pillar structures at both sides of the channel was presented by Lam et al. [71]. The channel was fabricated using photolithographic procedure on a silicon wafer to achieve large specific surface area. The column was operated in a countercurrent flow with relatively good separation. In subsequent work by Foerster et al. [72] the column was fitted with Raman spectroscopy based in situ composition measurement, which enabled the determination of compositions in real time without disturbing the process. MacInnes et al. [73] presented a spiral micro channel design fabricated from glass layers. It was operated with rotational motion to create a thin layer of liquid within the unit, which allowed for rapid contacting time. However, fluidic connections into and out of the rotating unit as well as the scaling up issues remained unsolved. A number of micro scale single stage distillation unit designs have been published by Cypes et al.[74], and by Sotowa and Kusakabe [75] employing a heated channel section for the partial vaporization of the liquid followed by the separation of phases. Variation of this was published by Hartman et al.[76], and by Wootton and deMello [77] utilizing carrier gas in addition to evaporation to reduce the demands on the temperature control. However, due to concurrent flow of vapor and liquid the maximum separation achievable by these units is one separation stage. The separation could in theory be increased by connecting multiple units to process the product from the previous steps, but as the separation of phases with these units require accurate control of differential pressures, this has yet to be demonstrated.

17

4.2 Novel micro-distillation column The time required for process to reach steady state depends in linear manner on the residence time. To minimize the residence time, either feed rate must be maximized, or holdup minimized. As the purpose of the experimental work in process development is to obtain sufficient data to validate the process model, minimization of holdup is the preferred option. For a distillation unit this requires the miniaturization of both the column and the auxiliary units. As distillation units at micro scale were not commercially available, the only choice was to manufacture the column. This required the use of novel design and new column packings. The design of the column was inspired by the flat shape commonly used in microfluidic applications. The planar shape has a large surface to volume ratio which allows for rapid heating and cooling of the unit, accurate temperature control, and large contact area for mass transfer. Additionally, it was easy to manufacture. The general shape of the column, which was used for all distillation columns developed and characterized in [V] and [VI] is presented in Figure 8.

Figure 8. Micro-distillation column [VI]. The inside of the unit formed a single planar channel that was filled with metal foam. The purpose of the metal foam was to facilitate liquid flow and to increase mass transfer between vapor and liquid. [V] Three different designs were used to evaluate the column. The internal height of the designs varied from (2 to 5) mm, internal length from (140 to 500) mm, and internal width from (30 to 100) mm. The columns were operated in a horizontal orientation with heating and cooling through the column walls, as presented in Figure 9.

18

Figure 9. Heat-pipe type operation of a planar micro-distillation column in a horizontal orientation. [V]. In this heat-pipe type operation liquid at the hot end evaporates and the formed vapor is driven by pressure difference towards the cold end. At the cold end the vapor condenses into liquid and is drawn by capillary forces towards the hot end. The advantage of this operating mode is that the column does not require pumps or external fluidic modules to operate. [V] This counter-current flow of vapor and liquid together with good mass transfer between the phases caused the light and heavy components to be enriched at the two ends of the columns. This phenomenon is known as distillation [58]. The concentration change over the column was measured with gas chromatography from samples drawn via the fluidic connections. The minimum number of theoretical stages required for the separation of a binary system was calculated with Fenske equation (2) [78] and McCabe-Thiele method as presented in Figure 10.

avg

B

B

D

D

xx

xx

NTSlog

11log

(2)

NTS equals number of theoretical stages required at total reflux xD equals mole fraction of the volatile component in the distillate xB equals mole fraction of the volatile component in the bottom product

avg equals the average relative volatility of components

19

Figure 10. McCabe-Thiele method for the separation of n-Hexane(1) + Cyclohexane. For multicomponent systems the separation was simulated with a flow-sheet process simulator, Flowbat [79]. To simulate total reflux, reboiler was used as the feed plate with the reflux ratio set to 80. For all practical purposes, the whole feed was removed from the reboiler as bottom product. Number of theoretical stages was obtained from the simulation. Simulated values corresponded well with experimental results, as presented in Figure 11.

Figure 11. Distillation of ethanol + 2-methylbut-2-ene + 2-ethoxy-2-methylbutane.

, ethanol; , 2-methylbut-2-ene; , 2-ethoxy-methylbutane. The solid lines represent simulated values. The column was manufactured from metal with relatively high thermal conductivity causing the column wall temperature to change in a linear manner from hot to cold. As the column temperature varied between the boiling points of the two products, this approach was acceptable for mixtures of close boiling components. Due to limited temperature difference between the column and the fluid the capacity of column section intended for vaporization of liquid was very poor. This led to extremely small mass

20

and heat fluxes within the column causing it to be susceptible to even small variations in the heat loss as caused by even small changes in the temperature or humidity of the surrounding air. However, when experiments were successful, the separation of the column at this operating mode was very high. [V] However, it was not very practical to operate the column with such a small capacity. It was also impossible to accurately measure column internal flows or energy balance, and the column temperature had to be optimized for each feed mixture separately, making operation of the column very tedious. As product outlets were from the liquid phase, air could not be vented outside the unit, decreasing the vapor phase concentration and hindering vapor flow, which limited the repeatability of results. [V] For these reasons the column was later fitted with auxiliary reboiler and condenser as presented in Figure 12. The setup for the improved distillation unit is presented in Figure 13 [VI].

Figure 12. Micro channel heat exchanger design, cut-out of the heater (2) with electrical heating cartridges (1), cut-out of the heat exchanger (3), channel structure in the finished unit (4) [VI].

Figure 13. Flow diagram of the distillation column with external reboiler and condenser [VI].

FEED

VENT

BOTTOMS

DISTILLATE

21

With the auxiliary devices the unit temperature could be selected without affecting column temperature profile. Thus the evaporation capacity of the reboiler could be greatly increased over that of heat-pipe type operation. Additional advantages were the possibility to optimize heating and cooling separately, to measure mass flows, and to analyse stream compositions and temperatures. From these measurements the rigorous calculation of mass and energy balances became possible, which enabled the modeling of column heat losses. As measurement of column wall temperature profile was beyond our means, the heat losses were modeled as a function of difference between column shell and fluid boiling point temperatures, as presented in Figure 14. The model was mainly used for the optimization of column shell temperature.

Figure 14. Heat losses from the distillation column as a function of column wall (shell) temperature and fluid boiling point difference. The column mass flow measurements allowed the characterization of column separation for different mass fluxes, as presented in Figure 15.

22

Figure 15. Separation as function of mass flux. n-Hexane + Cyclohexane -system,

; Methyl formate + Methanol, ; 2-Methylbut-2-ene + Ethanol + 2-Ethoxy-2-methylbutane, [VI]. In horizontal operation the liquid flowed on the bottom of the channel. Advantage gained from this mode is the elimination of weeping, which limits the minimum flow rate of vapor for columns operated in a vertical orientation. This allowed the operation of column at flow rates significantly below the operation of commercial column packings. It was observed, that at mass fluxes below 0.1 Pa½ the separation of the column increased in almost linear manner. [VI] Based on the experimental column temperature and concentration profiles the heat losses from the column were calculated for the heat-type operation. Calculated heat losses were translated to column mass flux with heat of vaporization. The calculation process was repeated for modular operation with known mass fluxes entering and leaving the column. Both results are presented in Figure 16.

Figure 16. Simulated mass flux inside the column, solid line represent heat-pipe type operation at average F-factor of 0.02 Pa½, dashed line represents modular operation at average F-factor of 0.16 Pa½.

23

As observed from Figure 16, the mass flux for the heat-pipe type operation was extremely low. Close to evaporation section it reached 0.05 Pa½ but decreased close to zero at both ends. As presented in Figure 15 the separation improved with decreasing mass flux, which would explain the high separation observed for the heat-pipe operation.

4.3 Effect of design parameters and metal foam properties Three different channel geometries and four different column packings were tested in [V] and [VI] as summarized in Table 4. Table 4. Summary of conditions and results for distillation at mass flux of 0.02 Pa½.

# Ref. Channel height (mm)

Pore size (mm)

Foam thickness (mm)

Packing volume ratio a

Specific surface area (m2/m3)

HETP b

(mm)

Mode of operation

1 [V] 2 0.15 0.5 25% 15500 51 Heat-pipe c 2 [V] 3.5 0.4 1.6 46% 5400 17 Heat-pipe 3 [V] 3.5 0.6 2 57% 2800 13 Heat-pipe 4 [V] 5 0.6 3 60% 2800 25 Heat-pipe 5 [VI] 5 0.6 5 100% 2800 30 Distillation

a channelofVolumepackingofVolumeratiovolumePacking

b Height to a theoretical stage (HETP). Smaller number represents better separation. c Omitted from further analysis, see explanation below. The purpose of metal foam was to facilitate liquid flow from the cold end towards the hot end. As the column was operated in a vertical orientation, the effect of gravitational force on liquid flow was negligible compared to capillary forces. For metal foams with 0.4 mm and 0.6 mm pore size the capillary forces were sufficiently high for the transportation of liquid, as reported by Sundberg [80]. The resistance caused by the metal foam depends on the permeability, which depends on the porosity and type of interconnected pore structure of the foam. Foams manufactured by Recemat had high porosity, as reported by Khayargoli et al. [81]. This made them suitable for use as column packing. The metal foam sample manufactured by Mitsubishi consisted of pores that were open only towards the surface of foam. This caused liquid to flow

24

around the metal foam making it unsuitable for use as column packing, as indicated by low separation. For this reason case 1 was omitted from further analysis. Although mass flux within the column for a heat–pipe type operation could not be directly measured, a value of 0.02 Pa½ was estimated based on the heat losses as presented in Chapter 4.2. For ease of comparison, separation for case 5 was extrapolated to same mass flux, based on results presented in Figure 15. The causes and effects identified based on these cases are summarized in Table 5. Table 5. Summary of column geometries and foam properties, and their effects on column operation. Variable increases Effect Channel height HETP b increases Channel length NTS c increases Channel width Mass flux (capacity) increases Packing volume ratio a HETP b increases Foam pore size Capillary forces weaken [80] Foam porosity Permeability increases [81] Level of interconnections Permeability increases [81] Foam thickness HETP b and mass flux increase Specific surface area No effect observed

a channelofVolumepackingofVolumepackingofRatio

.

b Height to a theoretical plate c Number of theoretical stages Increasing the channel height increases the cross sectional area of the column and thickness of vapor and liquid layers while keeping the mass transfer surface between liquid and vapor phases constant, thus decreasing the effective specific surface area. Thicker fluid layers leads to slower diffusion rate which decreases the driving force between the phases and eventually leads to reduced mass transfer. This ultimately leads to longer channel length requirement for separation equal to a single equilibrium stage, i.e. height equivalent to a theoretical plate (HETP). In this work increasing the channel height from 3.5 mm to 5 mm decreased the separation by 85% as presented in Figure 17.

25

Figure 17. Effect of channel height on separation for cases 2 to 5. Increasing width of a channel increases the cross-sectional area of the channel. For a column with a high width to height ratio, this only slightly increases the effective specific surface area. Thus it is assumed that effect on separation would be small. Increasing channel width also increases cross-sectional area of the channel which leads to increased mass flux within the channel. However, as channel width in the successful experiments was constant these assumptions could not be validated. Increasing foam thickness leads to thicker liquid layer and to slower diffusion and mass transfer in the liquid phase. As increasing the metal foam thickness resulted in decreased separation, as presented in Figure 18, it was estimated that mass transfer resistance was higher in the liquid phase. As diffusion rates in liquid are significantly lower than in vapor, this result was not totally unexpected.

Figure 18. Effect of metal foam thickness on separation for cases 2 to 5. As the thickness of metal foam is increased, the level of liquid layer is initially increased. At certain point the liquid level falls below the surface of metal foam. This leads to a situation, where vapor fills the upper part of metal foam and only part of the vapor layer is in contact with the liquid layer. Diffusion from the contact surface to bulk of vapor would then occur through tortuous pathways via interconnected pore structure. In the liquid phase decrease in diffusion by tortuous pathways is offset by the increased mixing caused by flow through the pores resulting in better

26

overall diffusion compared to unobstructed laminar flow. These assumptions are supported by results presented in Figure 19.

Figure 19. Effect of the packing volume ratio on separation for cases 2 to 5. The pore size of foam seemed to decrease separation only slightly as presented in Figure 20. However, this result remains inconclusive.

Figure 20. Effect of pore size on separation for cases 2 to 5. Most surprising was the effect of specific surface area, which seemed to have a slight adverse effect on separation as presented in Figure 21. The most likely explanation for this behavior is that as fluid flow did not conform to the surface of the pores, but instead flowed as a bulk layer the mass transfer surface was equal to the cross-section of the column. Thus the separation was not directly a function of the packing specific surface.

Figure 21. Effect of specific surface area on separation for cases 2 to 5.

27

5 Trial Plants Before committing to a commercial scale, the feasibility of a process is evaluated based on a process model. The process model combines the unit models depicting the behaviour of the chemical steps necessary to synthesize and separate the desired products. As the reliability of these models affects the success of the sc§ale-up, the risks are minimized by validating the models against experimental results in the piloting phase. [1, 82] The size of the trial plant depends on the type of process steps used and the maximum acceptable scale-up ratios related to them. In general, larger scale-up ratios are possible for gas and liquid phase operations than for operations involving solids. Depending on the novelty of the process steps used, part or even the whole process may need to be tested prior to committing to commercial scale. The advantage of testing the whole process in a trial plant is that all aspects related to process can be tested and verified, thus minimizing the risks related to scale-up. The downside is the significant cost of construction, feedstock, waste disposal, and labour. Because of these costs, work on a pilot plant is postponed until sufficient certainty of the process feasibility is obtained. The delay caused by piloting itself is usually 3 to 4 years.[1, 4, 83, 84] In situations where the operation of the full scale process unit can be described accurately by unit models and only the physical phenomena such as vapor-liquid equilibria, the chemical reaction, or the accumulation of impurities needs to be validated, even a micro scale unit can be used. For novel processes involving exothermic reactions with heat transfer limitation or fast reactions with mass transfer limitation the small scale piloting could result in incorrect conclusions. [85] However, if all the process units are based on well-established technology, it might be sufficient to validate the process model and to study the time dependent phenomena with an integrated trial plant in a laboratory scale. An integrated trial plant is defined as a plant consisting of all the recycle paths relevant to the production along with all the necessary process units. These process units should be scaled-down versions of the commercial scale units, which may limit the maximum acceptable scale-up ratio obtained from the trial plant. Alternatively, the level of miniaturization available may limit the smallest scale achievable for the trial plant. Depending on the capacity these integrated trial plants are named as a pilot plant, miniplant, and microplant, in decreasing size scale. [1, 4, 83, 59, 82, 86, 87, 88]

28

5.1 Microplant A microplant is often defined as being hundred times smaller than a miniplant. Where the cost of constructing a pilot plant can be up to one tenth of the commercial scale, the construction cost of a microplant is very modest, with the highest costs accrued from the analytics and instrumentation. The microplant requires very small space and is typically found inside a fume hood with good venting, which practically eliminates risk of exposure to emissions. The microplant is constructed using small diameter tubing, from (1.5 to 6) mm in diameter, and if possible, the same process units that were used during the initial laboratory testing. It is characterized as having a production rate of a gram per hour, or a throughput of 100 cm3×h-1. As the amount of feed required in the process is limited, the costs related to chemicals, and product and waste disposal are minimal. Despite the small size, the microplant is an integrated trial plant, consisting of all the recycling paths required for assessing the accumulation of impurities in the process, and the effect of positive feedback on the catalyst activity and selectivity.[1, 8, 84] As the small size makes the process inherently safer, it is convenient to automatize the process to as high degree as possible in order to minimize the operator workload. With a computer based process control combined with automated sampling and analysis the process can be operated even unmanned outside the normal working hours, achieving significant savings in labour costs. Additionally, this may allow the operation of multiple microplants in parallel thus further speeding up the entry into market. [1, 8, 82, 84].

5.2 Case Study – Etherification of 2-ethoxy-2-methylbutane The applicability of using microplants in process development was assessed using the synthesis of 2-ethoxy-2-methylbutane as a case study. A laboratory-scale integrated trial plant, microplant, was constructed for the purpose, as presented in Figure 22.

29

Figure 22. Laboratory-scale integrated trial plant for the synthesis of 2-ethoxy-2-methylbutane. The main components of the microplant were the small scale distillation unit, and a tubular reactor filled with Amberlyst 16 catalyst. Additionally, four micro gear pumps, a syringe pump used both as the feed pump and the feed container, two heat exchangers, two preheaters, one reboiler, and three mass flow controllers were used. The column temperature was monitored with 16 temperature probes connected to computer based control software. The flow chart for the process is presented in Figure 23. The total liquid hold-up of the process was 50 cm3. [VII]

Figure 23. Flow chart of the etherification microplant [VII].

30

The process was constructed using compression type fittings and flexible polytetrafluoroethylene tubing, which enabled rapid modifications to the plant. Additional advantage in using the polytetrafluoroethylene tubing was the possibility of observing visually the liquid flow, which was useful especially during the start-ups. [VII]

5.2.1 Reaction Kinetics 2-ethoxy-2-methylbutane is formed by etherification of 2-methylbut-1-ene or 2-methylbut-2-ene and ethanol. Additionally 2-methylbut-1-ene can isomerise into 2-methylbut-2-ene, as presented in Figure 24. All the three reactions are reversible. The reaction kinetics data for the etherification over Amberlyst 16 catalyst was published by Linnekoski et al.[13]. The liquid phase activity in the reactor was calculated from the activity coefficients estimated with original Universal Quasi-Chemical Functional Group Activity Coefficients model [86]. The reactor was simulated with a plug-flow model using a steady-state simulator, Flowbat [79]. The reactor model described the results of the once-through etherification with good correlation, which was accepted as the validation of the kinetics model. The feed consisted of an equimolar mixture of ethanol and 2-methylbut-2-ene. Catalyst deactivation was below level of uncertainty, thus it was not accounted for in the process model. [VII]

Figure 24. Etherification and isomerisation reaction paths for 2-methylbutenes [VII].

31

5.2.2 Distillation Model The phase equilibrium for the system of 2-methylbut-1-ene + 2-methylbut-2-ene + ethanol + 2-ethoxy-2-methylbutane was modeled using Wilson model [25] as the correlative liquid activity coefficient model and Soave modification of Redlich-Kwong equation of state [29] as the vapor phase fugacity coefficient model. The Wilson model parameters were regressed based on the literature data. The phase equilibria model was used to model the distillation with Murphree vapor phase efficiency set to unity. The number of theoretical stages was estimated based on the correlation between the flooding factor and the height of a theoretical stage, as presented in Figure 15. The distillation model was validated against the experimental results of once-through separation of reactor product, as presented in Figure 11.

5.2.3 Process Model As the distillation model presented in section 5.2.2 and the reactor model presented in section 5.2.1 were modeled as steady state processes, they were combined with continuous-flow stirred-tank models to account for the liquid hold-up inside the microplant, as presented in Figure 25.

Figure 25. Block diagram of the process model [VII]. The reactor volume was evaluated based on reactor volume, and packed catalyst void fraction. The distillation column volume was evaluated experimental from the amount of liquid required to reach steady state with total reflux. The total hold-up of the system was 50 cm3. Feed flow rate was kept constant for the duration of each experiment. For different experiments, feed flow rates between 8.4 gh-1 and 16.4 gh-1 were used, equal to 12 cm3h-1 and 23.5 cm3h-1. Thus the residence time in the process varied from 4 to 2 hours. The process model was validated experimentally against the stream concentrations as presented in Figure 26, Figure 27, and Figure 29. The

FEED REACTION MODEL

REACTOR (10 cm3)

DISTILLATION UNIT (40 cm3)

DISTILLATIONMODEL

BOTTOM PRODUCT

32

correspondence was very good. No accumulation of impurities was detected.

Figure 26. Reactor product concentration for etherification with recycle of unreacted feed.

Figure 27. Reflux concentration for etherification with recycle of unreacted feed.

2-Methylbutenes

2-Ethoxy-2-methylbutane

Ethanol

2-Methylbutenes

Ethanol

2-Ethoxy-2-methylbutane

2-Ethoxy-2-methylbutane

Ethanol

2-Methylbutenes

33

Figure 28. Bottoms concentration for etherification with recycle of unreacted feed.

5.2.4 Accumulation of Impurities To observe the accumulation of impurities in the microplant, an inert impurity was deliberately added into the equimolar feed mixture. n-Pentane, which is a linear hydrocarbon with a boiling point above that of the product, 2-ethoxy-2-methylbutane, was chosen as the inert component. Based on the boiling point of the components present in the system, n-pentane was expected to accumulate in the recycle stream. This assumption was verified experimentally. [VII] The process was operated continuously for 43 hours using feed containing 0.022 mole fractions of the impurity. The impurity concentration in the process at the end of the experiment was 0.15, as presented in Figure 29. The experiment compared well against both the process model and the extrapolation model presented by Wörz [5].

Figure 29. Accumulation of impurity (pentane) in the system. Experimental, ; Process model, +. The dashed line represents the extrapolation method [5].

5.2.5 Mass Balance Mass loss from the process was 4%, which was caused by evaporation from the bottom product container as verified by collection of the product into closed container in a separate experiment. This did not affect concentration analysis when sampled directly from the process streams, but it did alter the concentration of bottom product, which was collected before sampling. Estimated decrease for the pentane concentration in the bottom product was 0.021 mole fractions in 16 hours. However, as bottom part of the distillation

0

0.05

0.1

0.15

0.2

0 5 10 15Number of recycles

xPe

ntan

e

34

column accounted for a small fraction of the overall hold-up, the estimated error in the overall pentane accumulation was 0.003 mole fractions, which was close to the uncertainty of the concentration analysis. Thus the results were estimated to be of good quality.

35

6 Conclusions In this work three different but linked objectives were completed. In the first step, novel vapor-liquid equilibrium data for ten binary systems was measured and modeled. A total of 83 new vapor pressure points, 361 new vapor-liquid equilibrium points, and 27 new excess enthalpy and excess volume points were reported. The measurements were verified with thermodynamic consistency tests, by evaluating pure component vapor pressures against literature values, and by the coincidence of measurements at equilibrium concentration. The data was found to be of good quality. The phase equilibria were rigorously modeled using correlative liquid activity coefficient models with equation of state model. The results of this step allows for the creation of more accurate process models. In the second step, a novel micron scale continuous distillation column was designed, constructed, characterized and modeled. Design parameters affecting the separation of the column and mass flux within were analysed and their effect explained. A distillation unit comprising of modular heating and heat exchangers was constructed and its operation validated using phase equilibria models, separation models, and energy and mass balance. The mass flux within the column was calculated based on the unit model; thus the internal reflux ratio was rigorously calculated, which was unprecedented in this size scale. In the third step, the results from the previous two steps were combined. An integrated trial plant with tubular reactor for the synthesis of 2-ethoxy-2-methylbutane, and micron scale distillation unit for the recycle of unreacted feed was demonstrated. The liquid holdup of the system was below 50 cm3, making the process inherently safe. The residence time of the system was 2 hours for a single recycle, thus allowing the evaluation of slower dynamic effects such as accumulation of impurities within reasonable time. The results were validated using modeling, thus proving the feasibility of performing pilot plant experiments on micron scale as part of process development projects. As with all great works, this one also leaves issues unfinished. The design parameters discussed in chapter 4.3 were based on limited amount of experiments. It would be useful to verify the assumptions using a number of different designs and column packings. For the distillation column and the microplant, the real challenge would be to run experiments both in high pressure and in very high vacuum. These additional experiments would also help in getting the results of this work widely used.

36

7 Literature cited [1] G. H. Vogel, “Process Development,” in Ullmann's Encyclopedia of

Industrial Chemistry, Wiley-VCH Verlag GmbH, 2011. [2] C. F. Kettering, “Biographical Memoir of Leo Hendrik Baekeland,”

Biographical Memoirs, vol. XXIV, 1946. [3] J. Whiting, “The Commercial Development of Chemical Processes,”

8th Intl. Congress of Applied Chemistry, vol. 21, pp. 203-215, 1912. [4] J. G. Lowenstein, “The Pilot Plant,” Chem. Eng., vol. 23, pp. 62-75,

1985. [5] O. Wörz, “Process Development via a Miniplant,” Chem. Eng. Process,

vol. 34, pp. 261-268, 1995. [6] P. Lievo, M. Almark, V.-M. Purola, A. Pyhälahti and J. Aittamaa,

“Miniplant - Effective Means to Develop and Design new Processes,” in AIChE Annual Meeting, Indianapolis, 2002.

[7] C. Helling, H. Fröhlich, J. Eggersglüss and J. Strube, “Fundamentals towards a modular microstructured production plant,” Chemie Ingenieur Technik, vol. 84, no. 6, pp. 892-904, 2012.

[8] H. G. Vogel, in Process Development, Wiley-VCH Verlag GmbH, 2005, pp. 279-303.

[9] H. Z. Kister, “Can we believe the simulation results?,” Chemical Engineering Progress, no. October, p. 52, 2002.

[10] Clean Fuels Development Coalition, “Oxygenates Fact Book,” 2013. [Online]. Available: http://www.cleanfuelsdc.org/pubs/documents/oxyfactbook.pdf. [Accessed 11 September 2013].

[11] European Fuel Oxygenates Association, “Legislation: drivers for change of the fuel ether market,” 2013. [Online]. Available: http://www.efoa.eu/en/legislation.aspx. [Accessed 11 September 2013].

[12] M. Di Girolamo, M. Brianti, M. Conte and M. Marchionna, “Octane Enchancers,” in Ullmann's Encyclopedia of Industrial Chemistry, Wiley-VCH Verlag GmbH, 2009.

[13] J. A. Linnekoski, P. Kiviranta-Pääkkönen and O. Krause, “Simultaneous Isomerization and Etherification of Isoamylenes,” Ind. Eng. Chem. Res., vol. 38, pp. 4563-4570, 1999.

[14] D. P. Nolan, Handbook of Fire and Explosion Protection Engineering, William Andrew Publishing/Noyes, 1996.

[15] U.S. Department of Labor, Occupational Safety & Health Administration, “Standard 1910.110 - Storage and handling of liquefied petroleum gases,” 2007.

37

[16] J. Swanston, “Thiophene,” in Ullmann's Encyclopedia of Industrial Chemistry, Wiley-VCH Verlag GmbH, 2006.

[17] International DME Association, “Frequently Asked Questions,” 2013. [Online]. Available: http://www.aboutdme.org/index.asp?bid=234#Q2.1. [Accessed 11 September 2013].

[18] M. Müller and U. Hübsch, “Dimethyl Ether,” in Ullmann's Encyclopedia of Industrial Chemistry, Wiley-VCH Verlag GmbH, 2000.

[19] W. Reutemann and H. Kieczka, “Formic Acid,” in Ullmann's Encyclopedia of Industrial Chemistry, Wiley-VCH Verlag GmbH, 2011.

[20] Design Institute for Physical Properties, DIPPR Project 801 - Full Version, New York: Design Institue for Physical Property Data/AIChE, 2005.

[21] E. Reboul and P. Truchot, “Ueber einen mit dem Aethylamyläther isomeren Aether: das Amylenäthylat, und Beobachtungen bezüglich der Bildung der gemischten Aether.,” Justus Liebigs Ann. Chem., vol. 144, no. 2, pp. 241-245, 1867.

[22] W. V. Steele, A. Nguyen, S. E. Knipmeyer and R. D. Chirico, Final Report. DIPPR Project 821. Vapor Pressure of Organic Chemicals of Industrial Interest. The 1991 Project Results National Institute for Petroleum and Energy Research, Bartlesville, 1992.

[23] M. A. Krähenbühl and J. Gmehling, “Vapor Pressures of Methyl tert-Butyl Ether, Ethyl tert-Butyl Ether, Isopropyl tert-Butyl Ether, tert-Amyl Methyl Ether, and tert-Amyl Ethyl Ether,” J. Eng. Chem. Data, vol. 39, no. 4, pp. 759-762, 1994.

[24] S. P. Verevkin, E. L. Krashnykh, T. V. Vasiltsova and A. Heintz, “Determination of Ambient Temperature Vapor Pressures and Vaporization Enthalpies of Branched Ethers,” J. Chem. Eng. Data, vol. 48, pp. 591-599, 2003.

[25] G. M. Wilson, “Vapor-Liquid Equilibrium. XI. A New Expression for the Excess Free Energy of Mixing,” J. Am. Chem. Soc., vol. 86, pp. 127-130, 1964.

[26] H. Renon and J. M. Prausnitz, “Local Compositions in Thermodynamic Excess Functions for Liquid Mixtures,” AIChE J., vol. 14, pp. 135-144, 1978.

[27] D. S. Abrams and J. M. Prausnitz, “Statistical Thermodynamics of Liquid Mixtures: A New Expression for the Excess Gibbs Energy of Partly or Completely Miscible Systems,” AIChE J., vol. 21, pp. 116-128, 1975.

38

[28] U. Weidlich and J. Gmehling, “A Modified UNIFAC Model. 1. Prediction of VLE, Excess Enthalpy, and Activity at Infinity Dilution,” Ind. Eng. Chem. Res., vol. 26, pp. 1372-1381, 1987.

[29] G. Soave, “Equilibrium Constants from a Modified Redlich–Kwong Equation of State,” Chem. Eng. Sci., vol. 57, pp. 1197-1203, 1972.

[30] J. G. Hayden and J. P. O'Connell, “Generalized Method for Predicting Second Virial Coefficients,” Ind. Eng. Chem. Process. Des. Dev., vol. 14, pp. 209-216, 1975.

[31] J. M. Prausnitz, T. F. Anderson, E. A. Grens, C. A. Eckert, R. Hsieh and J. P. O’Connell, Computer Calculations for Multicomponent Vapor-Liquid and Liquid-Liquid Equilibria, New Jersey: Prentice-Hall Inc, 1980.

[32] N. Kozub, H. Schuberth and E. Leibnitz, “Phasengleichgewichtsuntersuchungen dampfförmig-flüssig am system methanol/methylformiat unter verwendung einer gaschromatographischen analysenmethod,” J. Prakt. Chem., vol. 17, pp. 282-292, 1962.

[33] J. Polak and B. C.-Y. Lu, “Excess Gibbs free energies and excess volumes of methyl formate + methanol and methyl formate + ethano lat 298.15 K,” J. Chem. Thermodyn., vol. 4, pp. 469-476, 1972.

[34] V. T. Zharov and O. K. Pervukhin, “Structure of the vapor-liquid equilibrium diaghrams of reactive systems II. Methanol - formic acid - methyl formate system,” Zh. Fiz. Khim., vol. 46, pp. 1970-1973, 1972.

[35] J. Zeng, C. Fu and W.-L. Hu, “Study of the VLE formethyl formate - methanol system,” Tianrangi Huagong, vol. 25, pp. 54-56, 2000.

[36] J. Sohlo and H. Holma, “Calculation of a ternary component distillation,” Kem. Teollisuus, vol. 8, pp. 346-350, 1973.

[37] P. Uusi-Kyyny, J.-P. Pokki, M. Laakkonen, J. Aittamaa and L. S, “Vapor liquid equilibrium for the binary systems 2-methylpentane+ 2-butanol at 329.2 K and n-hexane + 2-butanol at 329.2 and 373.2 K with a static apparatus,” Fluid Phase Equilib., vol. 201, pp. 343-358, 2002.

[38] J. A. Barker, “Determination of Activity Coefficients from Total Pressure Measurements,” Aust. J. Chem., vol. 6, pp. 207-210, 1953.

[39] J. Aittamaa and J.-P. Pokki, User Manual of Program VLEFIT, Espoo: Helsinki University of Technology, 2003.

[40] S. Yerazunis, J. D. Plowright and F. M. Smola, “Vapor-Liquid Equilibrium Determination by a New Apparatus,” AIChE J., vol. 10, pp. 660-665, 1964.

[41] P. Uusi-Kyyny, J.-P. Pokki, J. Aittamaa and S. Liukkonen, “Vapor-Liquid Equilibrium for the Binary Systems of 3-Methylpentane + 2-Methyl-2-propanol at 331 K and + 2-Butanol at 331 K,” J. Chem. Eng.

39

Data, vol. 46, pp. 754-758, 2001. [42] E. Sapei, A. Zaytseva, P. Uusi-Kyyny, K. I. Keskinen and J. Aittamaa,

“Vapor-liquid equilibrium for binary system of thiophene + 2,2,4-trimethylpentane at 343.15 and 353.15 K and thiophene + 2-ethoxy-2-methylpropane at 333.15 and 343.15 K,” Fluid Phase Equilib., vol. 261, pp. 115-121, 2007.

[43] S. M. Walas, Phase equilibria in Chemical Engineering, Oxford, U.K.: Butterworth-Heinemann, 1985.

[44] J. Gmehling and U. Onken, Vapor-Liquid Data Collection, DECHEMA Chemistry Data Series, Part 1, vol. 1, Frankfurt/Main: DECHEMA, 1977.

[45] K. Kojima, H. Moon and K. Ochi, “Thermodynamic consistency test of vapor-liquid equilibrium data: Methanol + water, benzene + cyclohexane, and ethyl methyl ketone + water,” Fluid Phase Equilibria, vol. 56, pp. 269-284, 1990.

[46] O. Redlich and A. T. Kister, “Algebraic representation of thermodynamic properties and the classification of solutions,” Ind. Eng. Chem., vol. 40, pp. 345-358, 1948.

[47] J. Gmehling, J. Menke, J. Krafczyk and K. Fischer, Azeotropic Data, 2nd ed., Weinheim, Germany: Wiley-VCH, 2004.

[48] A. Verrazzi and I. Kikic, “Vapor–liquid equilibrium in binary systems ethanol + C4 and C5 hydrocarbons,” J. Chem. Eng. Data, vol. 43, pp. 949-953, 1998.

[49] A. Arce, A. J. A, E. Rodil and A. Soto, “Isobaric vapor–liquid equilibria for systems composed by 2-ethoxy-2-methylbutane, methanol or ethanol and water at 101.32 kPa,” Fluid Phase Equilib., vol. 233, pp. 9-18, 2005.

[50] R. C. Everson and W. Jansen, “Isobaric vapor–liquid equilibria of binary and ternary mixtures of heavy ethers, isoamylenes, and alcohols. 2. Tertiary amyl ethyl ether, 2-methyl-1-butene, and ethanol,” J. Chem. Eng. Data, vol. 47, pp. 247-250, 2001.

[51] J. Gmehling, J. Menke, J. Krafczyk and K. Fischer, “A data bank for azeotropic data—status and applications,” Fluid Phase Equil., vol. 103, pp. 51-76, 1995.

[52] A. Heine, K. Fischer and J. Gmehling, “Various thermodynamic properties for binary systems with tertiary ethers,” J. Chem. Eng. Data, vol. 44, pp. 373-378, 1999.

[53] N. Ishii, “The volatility of fuels containing ethyl alcohol. VII. Total and partial partialvapor pressures of mixtures of ethyl alcohol and pentane,” J. Soc. Chem. Ind. Jpn., vol. 38, pp. 705-707, 1935.

[54] S. W. Campbell, R. A. Wilsak and G. J. Thodos, “(Vapor + liquid)

40

equilibrium behavior of (n-pentane + ethanol) at 372.7, 397.7, and 422.6 K,” J. Chem. Themodyn., vol. 19, pp. 449-460, 1987.

[55] S. G. Collins, J. J. Christensen, R. M. Izatt and R. W. Hanks, “The excess enthalpies of 10 (n-pentane + an n-alkanol) mixtures at 298.15 K,” J. Chem. Thermodyn., vol. 12, pp. 609-614, 1980.

[56] J. R. Fair, “Distillation: King in Separations,” Chem. Processing, vol. 9, pp. 23-31, 1990.

[57] J. Stichlmair, “Distillation, 1. Fundamentals,” in Ullmann's Encyclopedia of Industrial Chemistry, Wiley-VCH Verlag GmbH, 2012.

[58] J. D. Seader and E. J. Henley, Separation Process Principles, John Wiley & Sons, 2006.

[59] A. Behr, W. Ebbers and N. Wiese, “Miniplants - Ein Beitrag zur inhärenten Sicherheit?,” Chem. Ing. Tech., vol. 72, pp. 1157-1166, 2000.

[60] A. Ziogas, V. Cominos, G. Kolb, H.-J. Kost, B. Werner and V. Hessel, “Development of a Microrectification Apparatus for Analytical and Preparative Applications,” Chem. Eng. Technol., vol. 35, pp. 58-71, 2012.

[61] E. Y. Kenig, Y. Su, A. Lautenschleger, P. Chasanis and M. Grünewald, “Micro-separation of fluid systems: A state-of-the-art review,” Sep. Purif. Technol., vol. 120, pp. 245-263, 2013.

[62] K. F. Lam, E. Sorensen and A. Gavriilidis, “Review on gas-liquid separations in microchannel devices,” Chem. Eng. Res. Des., vol. 91, pp. 1941-1953, 2013.

[63] D. R. Seok and S. Hwang, “Zero-gravity distillation utilizing the heat pipe principle (micro-distillation),” AIChE J., vol. 31, no. 12, pp. 2059-2065, 1985.

[64] W. E. TeGrotenhuis and V. S. Stenkamp, “Conditions for fluid separations in microchannels, capillary-driven fluid separations, and laminated devices capable of separating fluids”. USA Patent 6875247 B2, 2005.

[65] F. Zheng, V. S. Stenkamp, W. E. TeGrotenhuis, X. Huang and D. L. King, “Microchannel Distillation of JP-8 Jet Fuel for Sulfur Contect Reduction,” AIChE Annual Meeting, San Francisco, 2006.

[66] X. Huang, D. A. King, F. Zheng, V. S. Stenkamp, W. E. TeGrotenhuis, B. Q. Roberts and D. L. King, “Hydrodesulfurization of JP-8 fuel and its microchannel distillate using steam reformate,” Cat. Today, vol. 136, pp. 291-300, 2008.

[67] H. Fink and M. J. Hampe, “Designing and constructing micro plants,” IMRET 3, Berlin, 2000.

41

[68] A. L. Tonkovich, W. W. Simmons, L. J. Silva, D. Qiu, S. T. Perry, T. Yuschak, T. P. Hickey, R. Arora, A. Smith, R. D. Litt and P. Neagle, “Distillation process using microchannel technology”. USA Patent 7610775 B2, 2009.

[69] R. Arora, A. L. Tonkovich, M. Fanelli, L. Silva, J. Marco, D. Weidert, J. Marco, P. Neagle and T. Hickey, “Separation of Ethane-Ethylene in High Flux Microchannels,” IMRET-10, New Orleans, 2008.

[70] A. Ziogas, H. Kost and B. Werner, “Micro-rectification column for the thermal separation of fluids”. Germany Patent WO2009/080591 A1, 2009.

[71] K. F. Lam, E. Sorensen and A. Gavriilidis, “Towards an understanding of the effects of operating conditions on separation of microfluidic distillation,” Chem. Eng. Sci., vol. 66, pp. 2098-2106, 2011.

[72] M. Foerster, K. F. Lam, E. Sorensen and A. Gavriilidis, “In situ monitoring of microfluidic distillation,” Chem. Eng. J., vol. 227, pp. 13-21, 2013.

[73] J. M. MacInnes, J. Ortiz-Osorio, P. J. Jordan, G. H. Priestman and R. W. K. Allen, “Experimental demonstration of rotating spiral microchannel distillation,” Chem. Eng. J., vol. 159, pp. 159-169, 2010.

[74] S. Cypes, S. H. Bergh and D. Hajduk, “Microscale flash separation of fluid mixtures”. USA / International Patent WO 2007/033335 A1, 2007.

[75] K. I. Sotowa and K. Kusakabe, “Development of micro distillation device in Proceeding of the 4th ECCE Conference,” Granada, 2003.

[76] R. L. Hartman, H. R. Sahoo, B. C. Yen and K. F. Jensen, “Distillation in microchemical systems using capillary forces and segmented flow,” Lab Chip, vol. 9, pp. 1843-1849, 2009.

[77] R. C. R. Wootton and A. J. deMello, “Continuous laminar evaporation: micron-scale distillation,” Chem. Commun., pp. 266-267, 2004.

[78] M. R. Fenske, “Fractionating of Straight-Run Pennsylvania Gasoline,” Ind. Eng. Chem., vol. 24, pp. 482-485, 1932.

[79] K. K. Keskinen and J. Aittamaa, Web FLOWBAT User's Instruction manual, 2012.

[80] A. Sundberg, “Master's Thesis: The use of Micro Technology in Process Development,” 2007.

[81] P. Khayargoli, V. Loya, L. P. Lefebvre and M. Medra, “The impact of microstructure on the permeability of metal foams,” CSME Forum, pp. 220-228, 2004.

[82] R. P. Palluzi, “Pilot Plants,” in Ullmann's Encyclopedia of Industrial Chemistry, Wiley-VCH Verlag GmbH, 2000.

42

[83] J. A. Jones, J. Bomben, D. Marynowski, C. Murray, R. Phillips, R. Roberts, D. Semray, R. Swidler, C. Witham and J. Zuegel, “Tips for Justifying Pilot Plants,” Chem. Eng., vol. 100, pp. 136-143, 1993.

[84] S. Mannan, “Pilot Plants,” in Lee's Loss Prevention in the Process Industries, 3rd ed., Elsevier, 2005.

[85] N. Kockmann, Transport Phenomena in Micro Process Engineering, Berlin: Springer-Verlag, 2008.

[86] A. Fredenslund, J. Gmehling, M. L. Michelsen, P. Rasmussen and J. M. Prausnitz, “Computerized design of multicomponent distillation columns using the UNIFAC group contribution method for calculation of activity coefficients,” Ind. Eng. Chem. Process. Des. Dev., vol. 4, pp. 450-462, 1977.

[87] T. Ouni, P. Lievo, P. Uusi-Kyyny, C. Dell'Era, K. Jakobsson, A. Pyhälahti and J. Aittamaa, “Practical methodology for distillation design using miniplant,” Chem. Eng. Technol., vol. 29, pp. 104-112, 2006.

[88] L. A. Robbins, “The Miniplant Concept,” Chem. Eng. Progress, vol. 9, pp. 45-47, 1979.

9HSTFMG*afjiga+

ISBN 978-952-60-5986-0 (printed) ISBN 978-952-60-5985-3 (pdf) ISSN-L 1799-4934 ISSN 1799-4934 (printed) ISSN 1799-4942 (pdf) Aalto University School of Chemical Technology Department of Biotechnology and Chemical Technology www.aalto.fi

BUSINESS + ECONOMY ART + DESIGN + ARCHITECTURE SCIENCE + TECHNOLOGY CROSSOVER DOCTORAL DISSERTATIONS

Aalto-D

D 19

5/2

014

Aarne Sundberg

Micro-scale D

istillation and Microplants in P

rocess Developm

ent A

alto U

nive

rsity

Department of Biotechnology and Chemical Technology

Micro-scale Distillation and Microplants in Process Development

Aarne Sundberg

DOCTORAL DISSERTATIONS