Model based investigation of catalyst fouling in case of special

Subscriber access provided by UNIV OF TEXAS DALLAS

Process Engineering

Model based investigation of catalyst fouling in case of special hydrocracking of sunflower oil and kerosene mixture Omar Péter Hamadi, Tamás Varga, Zoltán Till, Zoltán Eller, and Jen# Hancsók Energy Fuels, Just Accepted Manuscript • DOI: 10.1021/acs.energyfuels.8b04085 • Publication Date (Web): 28 Feb 2019 Downloaded from http://pubs.acs.org on March 5, 2019

Just Accepted “Just Accepted” manuscripts have been peer-reviewed and accepted for publication. They are posted online prior to technical editing, formatting for publication and author proofing. The American Chemical Society provides “Just Accepted” as a service to the research community to expedite the dissemination of scientific material as soon as possible after acceptance. “Just Accepted” manuscripts appear in full in PDF format accompanied by an HTML abstract. “Just Accepted” manuscripts have been fully peer reviewed, but should not be considered the official version of record. They are citable by the Digital Object Identifier (DOI®). “Just Accepted” is an optional service offered to authors. Therefore, the “Just Accepted” Web site may not include all articles that will be published in the journal. After a manuscript is technically edited and formatted, it will be removed from the “Just Accepted” Web site and published as an ASAP article. Note that technical editing may introduce minor changes to the manuscript text and/or graphics which could affect content, and all legal disclaimers and ethical guidelines that apply to the journal pertain. ACS cannot be held responsible for errors or consequences arising from the use of information contained in these “Just Accepted” manuscripts.

is published by the American Chemical Society. 1155 Sixteenth Street N.W., Washington, DC 20036 Published by American Chemical Society. Copyright © American Chemical Society. However, no copyright claim is made to original U.S. Government works, or works produced by employees of any Commonwealth realm Crown government in the course of their duties.

Page 1 of 49 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Energy & Fuels

Model-based investigation of catalyst fouling in case of special hydrocracking of sunflower oil and kerosene mixture Omar Péter Hamadi*, Tamás Varga*, Zoltán Till*, Zoltán Eller**, Jenő Hancsók**

*Department of Process Engineering, University of Pannonia, Veszprém, Hungary

**Department of MOL Hydrocarbon and Coal Processing, University of Pannonia, Veszprém, Hungary

Levenspiel Deactivation Kinetic Model, Eley-Rideal mechanism, competitive adsorption, reaction kinetics, lumped components

ACS Paragon Plus Environment

1

Energy & Fuels 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 2 of 49

Abstract

In this work catalyst deactivation phenomena in case of special hydrocracking of sunflower oil and kerosene mixture was analyzed based on experiments and models. Alternative (bio/waste-originated) fuels are becoming more important to reduce the full life-cycle environmental pollution of transportation. One of the production possibilities of these alternative fuels is the co-processing (i.e. catalytic quality improvement) of fossil and bio-based feedstocks. The huge number of individual chemical components in the system increases the complexity of the investigation. Moreover, from the chemical analysis viewpoint, only some of these can be followed during the experiments. Hence, the models to describe the processes in this system (hydrocracker) are usually based on lumps (i.e. component groups). Experimental data are reported on the special catalytic hydrocracking of sunflower oil and kerosene mixture for the production of highquality aviation fuel (JET). Experiments were carried out in a fixed-bed tubular-reactor over temperatures between 533 and 613 K, pressure ranging from 30 to 70 bar and LHSV of 1.0 h-1 and 2.0 h-1, employing a Pt/H-mordenite catalyst. The objective of our


2

Page 3 of 49 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Energy & Fuels

work is to investigate the existing lumped models that are suitable for describing the experimental data; moreover, while maintaining the possible reaction pathways, model parameters were identified and validated against the measurements. As a result of the analysis of the measurement data, it has been established that in the case of lower liquid load/higher residence time a deactivation phenomenon, the so-called catalyst fouling takes place on the applied catalyst. Three catalyst deactivation models were developed and integrated into the kinetic model: Levenspiel Deactivation Kinetic Model, a simplified Eley-Rideal mechanism, and the last one based on competitive adsorption.

Introduction

The world’s energy supply and demand system is facing several challenges [1]. Firstly, the accessible oil reserves are expected to be depleted, and secondly, the production and application of engine fuels are causing serious environmental pollution [2]. Consequently, the subject of developing alternative fuels has become a very important research area. Biofuels are liquid or gaseous fuels produced from renewable sources [3]. The importance of liquid biofuels is increasing because there is no need to


3

Energy & Fuels 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 4 of 49

change the existing logistic system and the engines; also the safety system is wellknown. The feedstock of liquid biofuel production for JET- and diesel engines are mainly natural/waste fatty acids and their esters (e.g. triglyceride-based edible/non-edible vegetable oils, animal fats, etc.) [4]. In Europe, the most accessible forms of vegetable oils are sunflower and rapeseed oils [1]. A widespread method to model petrochemical systems is using pseudo-components which are the lumps of real components. The essence of this technique is that the components with similar nature (e.g. near boiling points) are lumped together into a fewer number of pseudo-components, and the chemical and thermodynamic changes of the system are investigated by using these lumps. In general, three conditions have to be met to use pseudo-components [5]: Measurability:

The possibility of measuring all the pseudo-components.

Adequacy:

The pseudo-components must have enough detail to determine all product properties of interest.

Accuracy:

Producing fuels from different feedstocks described with


4

Page 5 of 49 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Energy & Fuels

the same pseudo-components results products which can be characterized using the same pseudo-components. Based on our review of the literature, most of the models contain three, four, five and six pseudo-components in case of modeling vegetable oil hydrocracking, as shown in Table 1. Table 1. Possible pseudo-components

3 pseudo-components

4 pseudo-components

5 pseudo-components

6 pseudo-components

Triglycerides (TG)

Triglycerides (TG)

Triglycerides (TG)

Triglycerides (TG)

Organic liquid product

Organic liquid product

Light

(OLP)

(OLP)

(C5-C8)

Gas and Coke (GC)

Gas (G)

Middle (C9-C14)

Gasoline (GO)

Coke (C)

Heavy (C15-C18)

Kerosene (K)

Oligomerized (>C18)

Diesel (D)

Gas (G)

Coke (C) [1], [3], [6], [7],[8]

[1], [3], [6], [8]–[10],

[11]–[13]

[1], [3], [6]

A total of thirteen mechanisms were found in the literature. Ten out of these thirteen mechanisms involve the formation of coke or some oligomerized by-product as a pseudo-component; two contain oligomerization reactions, while all mechanisms


5

Energy & Fuels 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 6 of 49

contain only first-order and non-reversible reactions. The models with 3, 4 and 6 pseudo-components were applied to model the conversion of palm oil into biofuels. The model with 6 pseudo-components predicted the yield and conversion of gasoline fraction with an error of 10% [1]. The model with 4 pseudo-components was able to predict the thermal cracking of waste cooking oil with a maximum error of 23% [9]. The hydrocracking process of triglycerides was described by models containing 5 pseudocomponents (with different possible reaction pathways) with an error of 23% and 43% [11]. Three models were developed to determine the kinetics of a continuous bio-oil catalytic reactor, concentrations calculated with the best model at 450 and 500 °C resulted in a difference of 6.71% and 6.18% compared to the experimental data [8]. There are several reasons behind catalyst deactivation which has a significant impact on the efficiency of the production process. The physical and chemical mechanisms of deactivation can be divided into four groups: poisoning, coking or fouling, sintering and phase transportation [14]. Catalyst fouling is a physical process in which some species from the reaction mixture adsorb on the catalyst surface and reversibly block the active sites or pores [15].


6

Page 7 of 49 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Energy & Fuels

In fundamental studies, the deactivation is approached with an empirical function of time [16]. In further studies, the deactivation is assumed to be the function of kinetic variables like temperature and concentration of reactants, for the same reason that the main reaction rate depends upon these kinetic variables [17], [18]. The relationship between the empirical and fundamental aspects has been established [16], [19]. Some authors developed models in which the analysis of fouling are restricted to power law model relations [20]–[22]. Expecting non-selective deactivation, a correlation for estimating the activity coefficients has been established [23]. Nonintegral order rate expression, the Langmuir-Hinshelwood mechanism was used to model the fouling process with typified rate expression [24], and this model was also extended with the case of combined series/parallel fouling [25]. A wide variety of liquid fuels can be produced from renewable sources (methanol, biodiesel, etc.) [3], [6]. In our work, the focus is on the production of JET fuel from a mixture of sunflower oil and unrefined kerosene fraction. During the special hydrocracking of this feedstock mixture, the triglycerides of sunflower oil are transformed

to

alkanes

with

different

carbon

numbers;

in

addition,

the


7

Energy & Fuels 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 8 of 49

hydrodesulfurization of unrefined kerosene fraction and the partial hydrogenation of aromatic compounds take place by consecutive and parallel reactions. As the results of these reactions, JET fuel with higher hydrogen content in its molecular structure with excellent burning characteristics can be obtained [26]. The main goal of this work is to develop a mathematical model to establish the occurring physical (fouling) and chemical changes. To achieve this goal four models were applied, the first is a simple lumped model without any deactivation mechanism. The second one is an empirical model called Levenspiel’s Deactivation Kinetic Model (LDKM), the parameters of this model do not have any physical meaning. The other two is based on the catalyst fouling (deposits). The first one of these is a simple decomposition reaction model on the surface; the second one is based on the strength of the adhesive bonds. The unknown parameters of all the proposed fouling models were identified based on measurements and finally, the increase in model complexity is evaluated.


8

Page 9 of 49 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Energy & Fuels

Experimental setup

A fixed-bed tubular reactor (diameter: 29 mm, length: 700 mm) was applied to perform the special hydrocracking reactions of the investigated oil mixture. The feedstock was the mixture of kerosene and sunflower oil in a mass ratio of 3:7. In the experimental system the temperature, the pressure and the LHSV can be controlled, so the hydrocracking experiments were performed at 533-613 K, 30-70 bar, 1.0-2.0 h-1 employing Pt/H-mordenite catalyst (Pt content: 0.45%, specific area, BET: 451 m2/g, micropore volume: 0.18 ml/g, Si/Al ratio from XRF: 19 mol/mol, acid sites by ammonia TPD: 0.82 mmol/g, particle diameter: 1.4 mm). The measurements were performed after reaching a steady state condition.

Experimental results

The experimental results are shown in Figures 1-2. In general, it can be observed that the total product yield is directly proportional to the increase in temperature, meaning that the total product yield is increased due to the increase in temperature (in case of 30


9

Energy & Fuels 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 10 of 49

bar and LHSV 1.0 h-1 the total product yields are: 17.6% at 533 K, 18% at 553 K, 20.8% at 573 K, 32.9 at 593 K and 44.6% at 613 K). However, by increasing the pressure, mainly the proportions of gas and gasoline fractions are increasing, while the proportion of the diesel fraction is decreasing. The proportion of conversions measured in case of different residence times are shown in Fig 3. In this figure, every point is a result of the following division and shows how the residence time affects the conversion (Eq. 1).

𝑐𝑜𝑛𝑣𝑒𝑟𝑠𝑖𝑜𝑛𝑝𝑟𝑜𝑝. =

𝑐𝑜𝑛𝑣𝑒𝑟𝑠𝑖𝑜𝑛𝐿𝐻𝑆𝑉 = 1 ℎ ―1 𝑐𝑜𝑣𝑒𝑟𝑠𝑖𝑜𝑛𝐿𝐻𝑆𝑉 = 2 ℎ ― 1

(1)

It can be seen that the effect of residence time is not considerable. Despite our expectation that the doubled residence time should have a major effect on the conversion, the two times higher residence time increases the conversion with only 1% to 14% (e.g. at 30 bar and 533 K the conversion increased with 9.61% ). Reaching chemical equilibrium could cause such effect; however, in this case, this is not an adequate explanation. Here, the mass transfer of hydrogen from the gas phase, the mass transfer of all the other substances from the liquid phase, and the hydrocracking reactions on active sites of the catalyst take place in the system. From these, the hydrocracking reactions were considered as the rate dependent steps in the overall


10

Page 11 of 49 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Energy & Fuels

process based on some corresponding works [27], [28]. But the chemical equilibrium of these reactions is extremely shifted towards the production of smaller, but hydrogen saturated molecules in case the temperature and the hydrogen concentration are high enough. In addition, the maximal product yield is lower than 55.8% which means that chemical equilibrium in case of these reactions cannot have a significant effect on the reaction rates. The minor effect of the residence time on conversion is more likely the result of some catalyst deactivation process, in which the big oil molecules cover the macro-pores of the catalyst due to the low fluid velocity. The maximum of total product yield is at 613 K, 70 bar and LHSV = 1.0 h-1 (55.8%); the maximum of kerosene yield is at 613 °C, 70 bar and LHSV = 2.0 h-1 (53.5%). It should be taken into consideration that, as mentioned earlier, the feedstock contains a considerable amount of kerosene (30%).


11

Energy & Fuels

40 20 0

533 [K] G GO K D TG 30 bar

60

50 bar 573 [K]

m/m%

50 bar

70 bar

60 m/m%

20 30 bar

50 bar 593 [K]

70 bar

30 bar

50 bar

70 bar

60

20 30 bar

40

0

70 bar

40

0

553 [K]

60 m/m%

m/m%

60

m/m%

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 12 of 49

40 20 0

613 [K]

40 20 0

30 bar

50 bar

70 bar

Figure 1. Experimental data, LHSV = 1.0 h-1


12

Page 13 of 49

40 20 0

533 [K] G GO K D TG 30 bar

60

50 bar 573 [K]

m/m%

50 bar

70 bar

60 m/m%

20 30 bar

50 bar 593 [K]

70 bar

30 bar

50 bar

70 bar

60

20 30 bar

40

0

70 bar

40

0

553 [K]

60 m/m%

m/m%

60

m/m%

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Energy & Fuels

40 20 0

613 [K]

40 20 0

30 bar

50 bar

70 bar

Figure 2. Experimental data, LHSV = 2.0 h-1


13

Energy & Fuels

533 [K] 553 [K] 573 [K] 593 [K] 613 [K]

1.14 1.12

Propotion of conversion [-]

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 14 of 49

1.1 1.08 1.06 1.04 1.02 1 20

30

40

50 Pressure [bar]

60

70

80

Figure 3.Proportion of conversions at LHSV = 1.0 h -1 divided with conversion at LHSV = 2.0 h-1


14

Page 15 of 49 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Energy & Fuels

The developed models

Reaction kinetic model – M1

Since the experimental data contain the same pseudo-components, the reaction mechanism based on 6 pseudo-components was investigated during the modeling of the system [6]. During the experiments, the rate of coke formation was not monitored (assuming that there was no coke formation), so in the model, the residue is the remaining triglycerides and the coke. In our analyses, models with fewer pseudocomponents were also tested, these models only were able to describe the system with significant errors. Moreover, it can be suggested that the more pseudo-components present in the model, i.e. with increasing the model complexity, the higher model accuracy can be achieved. Hence, only the results obtained with the model based on the reaction mechanism shown in Figure 4 are the part of this work. The following differential equations describe changes in the concentration of pseudo-components along the reactor: 𝑑𝐶𝑇𝐺 𝑑𝜏

= ― 𝑘0𝐶𝑇𝐺

(2)


15

Energy & Fuels 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

𝑑𝐶𝐷

= 𝑘11𝐶𝑇𝐺 ― 𝑘5𝐶𝐷

(3)

= 𝑘12𝐶𝑇𝐺 + 𝑘51𝐶𝐷 ― 𝑘6𝐶𝐾

(4)

= 𝑘13𝐶𝑇𝐺 + 𝑘52𝐶𝐷 + 𝑘61𝐶𝐾 ― 𝑘7𝐶𝐺𝑂

(5)

= 𝑘21𝐶𝑇𝐺 + 𝑘53𝐶𝐷 + 𝑘62𝐶𝐾 + 𝑘71𝐶𝐺𝑂 ― 𝑘4𝐶𝐺

(6)

= 𝑘22𝐶𝑇𝐺 + 𝑘54𝐶𝐷 + 𝑘63𝐶𝐾 + 𝑘72𝐶𝐺𝑂 + 𝑘4𝐶𝐺

(7)

𝑑𝜏 𝑑𝐶𝐾 𝑑𝜏 𝑑𝐶𝐺𝑂 𝑑𝜏 𝑑𝐶𝐺 𝑑𝜏 𝑑𝐶𝐶 𝑑𝜏

Page 16 of 49

𝑘0 = 𝑘1 + 𝑘2

(8)

𝑘2 = 𝑘21 + 𝑘22

(9)

𝑘1 = 𝑘11 + 𝑘12 + 𝑘13

(10)

𝑘5 = 𝑘51 + 𝑘52 + 𝑘53 + 𝑘54

(11)

𝑘6 = 𝑘61 + 𝑘62 + 𝑘63

(11)

𝑘7 = 𝑘71 + 𝑘72

(12)

The reactor model was defined as a plug flow reactor; the longitudinal changes were identified with the residence time. Despite the experiments were carried out at different residence times, temperatures and pressures, as the result of the identification process all the unknown kinetic parameters were given for each pressure separately. The pressure dependence can only be investigated if hydrogen is also considered as a separate component in the kinetic model. The identified reaction rates are higher for lower liquid load and lower for higher liquid load, as can be expected. To describe the


16

Page 17 of 49 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Energy & Fuels

effect of catalyst fouling on the overall reaction rate, three models (M2 – M4) were developed and integrated into the kinetic model and all the unknown parameters of these models are identified based on measurements. Levenspiel’s Deactivation Kinetic Model (LDKM) – M2

LDKM is an empirical model originally developed to describe the long-term deactivation of the catalyst [29]. Here, in Eq. 13-15, the elapsed time is considered as the residence time in the reactor. With this approach, the catalyst fouling can be estimated by the addition of the catalytic activity (a) to the kinetic model (Eq. 16), increasing the number of unknown parameters by two (the deactivation function, Ψd* and residual activity, as), d is the deactivation order. Eq. 16 is the general form of Eq. 26, except the reactions in which the coke fraction is involved in the reaction.

―

(

𝑑(𝜏) = 1 + 𝑎𝑠 +

𝛹𝑑(𝜏) =

𝛹𝑑∗

∙

(

𝑑𝑎 = 𝛹𝑑 ∙ 𝑎𝑑 𝑑𝜏

1 ― 𝑎𝑠 1 + (1 ― 𝑎𝑠) ∙ 𝛹𝑑∗ ∙ 𝜏 1 ― 𝑎𝑠

(13)

)[ ( )(

1 + (1 ― 𝑎𝑠) ∙ 𝛹𝑑∗ ∙ 𝜏

∙ 𝛹𝑑∗ ∙ 𝜏 ∙

2

∙ 𝑎𝑠 +

2 ―1

1 ― 𝑎𝑠 1 + (1 ― 𝑎𝑠) ∙ 𝛹𝑑∗ ∙ 𝜏 1 ― 𝑎𝑠

1 + (1 ― 𝑎𝑠) ∙ 𝛹𝑑∗ ∙ 𝜏

)

)]

(14)

―𝑑(𝜏)

(15)


17

Energy & Fuels 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

𝑑𝑐𝑖 𝑑𝜏

𝑛

=𝑎 ∙

Page 18 of 49

𝑚

∑𝑘 ∙ ∏𝑐 𝑙

𝑙=1

𝑘,𝑙

(16)

𝑘=𝑖

Simplified Eley-Rideal mechanism of catalyst deactivation– M3

The second deactivation model is based on the catalyst fouling caused by one of the considered pseudo-components, the TG fraction. In this model, the catalyst is also considered as a chemical component and interacts with only this one pseudocomponent (i.e. the catalyst concentration influences all the reaction rates through the activity, but only one component can reduce it, according to Eq. 18). Therefore, the adhesively bonded component must be assumed as a pseudo-component. The component-mass balance is defined with Eq. 16 and with the following equations:

𝑎= 𝑑𝑐𝑐𝑎𝑡. 𝑑𝜏

=0 𝑐𝜏𝑐𝑎𝑡

= ― 𝑘𝑇𝐺 + 𝑐𝑎𝑡. ∙ 𝑐𝑐𝑎𝑡. + 𝑘𝑇𝐺 ― 𝑐𝑎𝑡. ∙ 𝑐𝑇𝐺&𝑐𝑎𝑡.

𝑑𝑐𝑇𝐺&𝑐𝑎𝑡. 𝑑𝜏

𝑐𝑐𝑎𝑡.

= 𝑘𝑇𝐺 + 𝑐𝑎𝑡. ∙ 𝑐𝑐𝑎𝑡. ― 𝑘𝑇𝐺 ― 𝑐𝑎𝑡. ∙ 𝑐𝑇𝐺&𝑐𝑎𝑡.

(17) (18) (19)

Fouling based on the strength of adhesive bonds and the number of active sites – M4


18

Page 19 of 49 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Energy & Fuels

In this model, all the components are able to bind to active sites of the catalyst at the same time. However, the extents of the adhesions of pseudo-components are different. The number of occupied active sites by each pseudo-component is based on the considered molecular size of components. The size of pseudo-components (Ai) is estimated with Eq. 20-22, assuming that the molecules have (quasi) spherical shapes. Table 2 contains the material data at 298 K. The temperature dependence of the density is neglected in the developed models.

𝑉𝑖 = 𝑟𝑖 = 𝐴𝑖 =

𝑀𝑖

(20)

𝜌𝑖 3 ∙ 𝑉𝑖

(21)

4∙𝜋 𝑟𝑖2 ∙ 𝜋

(22)

𝑀𝑖

Table 2. The density and the molecular weight of pseudo-components

298 K

TG

D

K

GO

G

Mi [g/mol]

839.1

222.7

178.1

106.1

33.48

ρ [g/cm3]

0.9182

0.8355

0.7934

0.7369

0.0814

The strength of adhesive bonds and reaction rate constants are the unknown parameters in M4 that should be identified based on the experimental data. The catalyst


19

Energy & Fuels 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 20 of 49

activity in this model is proportional to the surface concentration and the size of the pseudo-components, so the higher the surface concentration and the bigger the size of the pseudo-component, the more active sites are occupied by the pseudo-component. The surface concentration (ci) and the activity (a) can be calculated based on Eq. 23-24. In fact, the Ki values are the equilibrium constants for the distribution of adsorbates between the surface and the liquid phase. The component mass balance is the same as in the case of M2 (Eq. 16). The temperature dependence of Ki values was also approached with the van’t Hoff equation.

𝑐𝑖 = 𝐾𝑖 ∙ 𝐶𝑖 𝑎=

∑ 𝑖

𝐴𝑐𝑎𝑡 ― 𝐴𝑖 ∙ 𝑐𝑖

(23) (24)

𝐴𝑐𝑎𝑡

All the introduced models were solved in MATLAB 2017b using the Runge-Kutta method. The reaction rate constants were identified simultaneously at all temperatures (the temperature dependence was estimated with the Arrhenius equation). Thus, the objective function was the square error between the calculated and measurement data for different components, liquid loads at all temperatures:


20

Page 21 of 49 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Energy & Fuels

𝑓(𝑥 ) = 𝑛

∑∑ ∑(

)

𝑦𝑒𝑥𝑝 ― 𝑦𝑚𝑜𝑑𝑒𝑙

𝑇 𝑐𝑜𝑚𝑝𝐿𝐻𝑆𝑉

𝑦𝑚𝑎𝑥 𝑒𝑥𝑝

2

(25)

To determine the minimum of the objective functions in different cases NOMAD algorithm [30] and MATLAB was applied.


21

Energy & Fuels 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 22 of 49

Results

The kinetic model found in literature resulted in significant differences between the calculated and measured data, which is due to the fact that the reported kinetic parameters were identified in a different catalyst system (HZSM-5 and CMZ20 catalysts and 673 – 723 K) [6]. Hence, the results accordingly showed that these constants are not suitable for describing the processes taking place in the investigated system.


22

Page 23 of 49 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Energy & Fuels

5 TG

TG 4.5 4 3.5

D

K

D

K 3 2.5 2

GO

G

C

GO

G

C 1.5 1

Published reaction rate constants

Reaction rate constants in case of M1

0.5 0

Figure 4. The considered kinetic schemes with published and identified reaction rate constants at 673 K Since the reaction rate constants in the literature are not suitable for the catalyst used in our system based on some preliminary simulations, the identification of the kinetic constants is necessary. In the first step - because no coke formation is assumed - the coking reaction steps were eliminated from the considered reaction mechanism, therefore the number of the possible reaction steps are decreased to 10. The results at 30 bar are plotted in correlation charts shown in Fig 5.


23

Energy & Fuels

LHSV = 1 h-1 Gasoline

LHSV = 1 h-1 Petroleum

60 533 K

experimental [%]

experimental [%]

LHSV = 1 h-1 Gas

40 20 10

50

20 30 40 model [%] R2 = 0.9713

50

573 K

40 30 20 10 20 30 40 model [%] R2 = 0.94785

experimental [%]

10

LHSV = 2 h-1 Gas

40

50

LHSV = 1 h-1 Diesel

LHSV = 1 h-1 TG

60 553 K 40 20

60

10

experimental [%]

experimental [%]

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 24 of 49

20 30 40 model [%] R2 = 0.97156

50

60

40 593 K 30 20 10 5

10

15 20 25 30 35 model [%] R2 = 0.91573

40

613 K

30 20 10 10

20 30 model [%] R2 = 0.83879

40



LHSV = 2 h-1 Diesel

LHSV = 2 h-1 TG

Figure 5. Correlation charts in case of M1, at 30 bar

The identified parameters are collected in Table 3. In case of all pressures the model is more inaccurate at higher temperatures. Hence, it worth to investigate, how the accuracy is changing with the integration of different catalyst deactivation models.


24

Page 25 of 49 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Energy & Fuels

Table 3. The identified kinetic parameters for M1 at different pressures

TG → D

TG → K

TG → GO

TG → G

D→K

D → GO

D→G

K→ GO

K→ G

GO→ G

30 bar

A0 [1/min] Ea [kJ/kg]

1.725∙10 3

3.868∙10 4

9.336∙1011

2.604∙10-1

2.881

4.616∙105

7.732∙109

0

1.781∙10-1

0

0

1.461∙105

1.217∙104

1.827∙104

8.349∙104

1.210∙105

0

4.008∙103

0

0

50 bar A0 [1/min]

1.106∙104

Ea [kJ/kg]

4.683∙104

1.406∙101 7

2.044∙105

6.950∙102

1.400∙101

4.288∙102

2.511∙1025

0

1.511∙101

0

0

4.836∙104

2.364∙104

3.303∙104

3.058∙105

0

2.110∙104

0

0

70 bar

A0 [1/min]

5.557∙103

Ea [kJ/kg]

4.362∙104

1.013∙101 5

1.743∙105

3.627∙10-1

1.421∙102

1.014∙103

0

0

6.531∙106

0

0

4.881∙103

3.189∙104

3.910∙104

0

0

8.477∙104

0

0

To prove that optimal parameters were calculated, and to investigate the weights of each reaction pathways, a sensitivity analysis was performed. A small perturbation was made in identified the reaction rate constants; the results are shown in Fig. 6. The analysis indicates that optimal parameters were found, and it can be establish that the greater the change in the value of the objective function as a result of the perturbation, the bigger the weight of the reaction pathway. For example, the TG to D and the K to G can be the most dominating reactions based on the reaction kinetic constants.


25

Energy & Fuels

2.6

TG to K

TG to GO

TG to G

30bar

2.5 2.4 2.3 -10

0 Perturbation [%]

Value of objective function

-20

2.1

D to K


TG to D


1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

10

20

Page 26 of 49

D to GO 2.4

D to G

K to GO

K to G

GO to G

50 bar

2.3 2.2 2.1 -20

-10

0 Perturbation [%]

10

20

70 bar

2 1.9 1.8 -20

-10

0 Perturbation [%]

10

20

Figure 6. Sensitivity analysis of identified parameters of M1 model

In the case of M2, the results appear to be in line with the assumption that deactivation is present on the catalyst. The value of the coefficient of determination (R2) increased at almost all temperature and pressure. The previously identified parameters were used to calculate the reaction rate constants and the temperature dependence of the parameters of LDKM model was also estimated with the Arrhenius equation. The deactivation model was used to identify the deactivation on the catalyst continuously at all time, but in the model, the catalyst activity changed only when the LHSV value changed from 1.0 h-1 to 2.0 h-1. That means the catalyst bed can be separated into two


26

Page 27 of 49 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Energy & Fuels

parts related to the activity and in each part, a constant activity was considered to characterize the effect of catalyst fouling. The identified parameters for M2 are summarized in Table 4.


27

Energy & Fuels 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 28 of 49

Table 4. The identified parameters of M2

as

Ψ d*

30 bar

A0 [1/min]

8.746·10-2

6.144·101

Ea [kJ/kg]

1.566·104

1.499·104

50 bar

A0 [1/min]

3.103·10-2

5.614·101

Ea [kJ/kg]

1.102·104

1.444·104

70 bar

A0 [1/min]

4.417·10-2

2.206·101

Ea [kJ/kg]

1.495·104

1.075·104

Also, in these cases, the previously identified (M1) kinetic parameters were used and corrected with the catalyst activity. As for M3 and M4, the initial activity can be more than 1. When the kinetics parameters were identified (M1), the best solution for both residence time was found (metaheuristic), so the activity above 1 can be expected as the correction of the pre-exponential factors (β) at higher liquid velocity. Thus Eq. 17 and 24 have been changed as the following:

𝑎= 𝑎=

∑ 𝑖

𝑐𝑐𝑎𝑡.

∙𝛽 =0 𝑐𝜏𝑐𝑎𝑡 𝐴𝑘𝑎𝑡 ― 𝐴𝑖 ∙ 𝑐𝑖 𝐴𝑘𝑎𝑡

(26) ∙𝛽

(27)


28

Page 29 of 49 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Energy & Fuels

The previously mentioned β parameter was added to the unknown parameters at 30 bar, and the identified value was set as a constant at 50 and 70 bar. In case M3 β = 3.16, as for M4 β = 1.27. The identified parameters for M3 and M4 are collected in Table 5 and 6. The conclusion can be made from the parameters of Table 6, that the temperature dependence of the K value is not significant at the investigated temperature and pressure range, and the dependency decreases at higher pressure. Moreover the assumption in M3 (the TG fraction can bind the active sites mostly) is proven since the maximum value of the equilibrium constants is at TG in case off all temperature and pressure. Table 5. The identified parameters for M3

kTG+cat.

kTG-cat.

30 bar

A0 [1/min]

8.876·101

1.949·102

Ea [kJ/kg]

3.161·104

4.475·104

50 bar

A0 [1/min]

4.009·101

4.812·101

Ea [kJ/kg]

3.276·104

4.287·104

70 bar

A0 [1/min]

1.212·100

1.936·10-1


29

Energy & Fuels 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Ea [kJ/kg]

Page 30 of 49

3.338·104

9998·104

Table 6. The identified parameters for M4

KTG

KD

KK

KGO

KG

1.000·100

1.061·10-1

6.761·103

2.584·103

30 bar

A0 [1/min]

9.990·10-1

Ea [kJ/kg]

1.343·103

9.274·10- 9.993·101

1

9.792·103 6.342·103

50 bar

A0 [1/min]

9.983·10-1

Ea [kJ/kg]

1.713·103

2.702·10- 5.665·10- 4.689·10-1 2.026·10-1 1

1

1.028·103 2.492·103

2.049·103

-4.848·100

70 bar 8.382·10- 7.718·10- 3.195·10-1 1.737·10-1

A0 [1/min]

6.948·10-1

Ea [kJ/kg]

-4.975·10-1 7.126·100 3.832·103 -4.948·100 -4.907·100

2

1

In Fig 7 the coefficients of determination in the case of all models, temperatures and pressures are compared. The objective function values are shown in Fig 8. As can be seen, the accuracy of models is proportional to their complexity. The estimates provided by M3 and M4 tend to be reliable, but the best result was achieved with M4 (Fig 9.). It


30

Page 31 of 49

can be established that the objective and R2 values are more dependent on temperature than the pressure. M2

M3

0.95

0.9

0.9

2

0.95

0.85

0.8 520

M4

50 bar

1

R

2

M1

30 bar

1

R

0.85

540

560

580

T [K]

600

620

0.8 520

540

560

T [K]

580

600

620

70 bar

1

2

0.95

R

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Energy & Fuels

0.9

0.85

0.8 520

540

560

T [K]

580

600

620

Figure 7. Coefficients of determination (R2) in case of M1-M4 at all temperatures and pressures


31

Energy & Fuels

2.5

M1 M2 M3 M4

2

objective value

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 32 of 49

1.5

1

0.5

0

30 bar

50 bar

70 bar

Figure 8. Objective values in case of M1-M4


32

Page 33 of 49



experimental [%]

60 533 K 40 20 10

50

20 30 40 model [%] R2 = 0.99276

50

573 K

40 30 20 10 20 30 40 model [%] R2 = 0.97981

experimental [%]

10

40

LHSV = 1 h-1 TG

40 20 10

20 30 40 model [%] R2 = 0.98402

50

60

40 593 K 30 20 10 5

10

15 20 25 30 35 model [%] R2 = 0.96443

40

613 K

30 20 10 10

LHSV = 2 h-1 Gas

50

LHSV = 1 h-1 Diesel

60 553 K

60

experimental [%]

experimental [%]

LHSV = 1 h-1 Gas

experimental [%]

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Energy & Fuels


20 30 model [%] R2 = 0.8568 LHSV = 2 h-1 Petroleum

40 LHSV = 2 h-1 Diesel

LHSV = 2 h-1 TG

Figure 9. Correlation charts in case of M4, at 30 bar


33

Energy & Fuels 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 34 of 49

Conclusions In this work, heterocatalytic reactor models were developed and tested to establish the occurring physical (fouling) and chemical changes. In the kinetic model, the components with similar nature were lumped into pseudo-components, and the process that leads to the chemical transformation was approached with first order non-reversible reactions. The performance of the (kinetic) model was further improved by integrating three different catalyst deactivation models. The first one is an empirical model (LDKM), where the catalyst activity is the function of the residence time. The second one is a simplified ER mechanism, where only one component is able to reduce the catalytic activity. The last model is based on the strength of the adhesive bonds and the number of active sites. The activity in this model means the number of the unblocked active sites which depends on the surface concentration and the size of the pseudocomponents. It has been established that the accuracy of models is proportional to their complexity, so the best results were provided by M4. Besides, based on the improvement in the accuracy of the developed models it can be stated, that due to the low flow rate catalyst fouling has a significant effect on conversion. Moreover it has


34

Page 35 of 49 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Energy & Fuels

been stated that the TG fraction can mostly bind the active sites, and the adsorption ability of the molecules in all fractions is not depending on temperature significant, and the dependency decreases at high pressure.


35

Energy & Fuels 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 36 of 49

Nomenclature TG

concentration of Triglycerides

[g/g]

CD

concentration of Diesel

[g/g]

CK

concentration of Kerosene

[g/g]

CGO

concentration of Gasoline

[g/g]

CG

concentration of Gas

[g/g]

ki

reaction rate constant

[1/min]

a

activity

[-]

Ψd

deactivation function

[-]

Ψ*d

deactivation function*

[-]

d

deactivation order

[-]

as

residual activity

[-]

t

time

[min]

Ccat

catalyst concentration

[g/g]

CTG&cat.

catalyst+ Triglycerides (adsorbed TG)

[g/g]

kTG+cat.

reaction rate constant for adsorbing

[1/min]

kTG-cat.

reaction rate constant for desorbing

[1/min]

Vi

volume of pseudo-components

[cm3]

ρi

density of pseudo-components

[g/cm3]

Mi

Average molar mass of pseudo-components

[g/mol]

ri

radius of pseudo-components

cm

Ai

surface of pseudo-components

cm2/g

Acat.

surface of the catalyst

cm2/g

Ki

Extent of the adhesion of pseudo-components

[-]

ci

surface concentration of pseudo-components

[g/g]

yexp

measurement concentrations

[g/g]


36

Page 37 of 49 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Energy & Fuels

ymodel

calculated concentrations

[g/g]

ymaxexp

maximum of the measurement concentrations

[g/g]

A0

Pre-exponential factor

[1/min]

Ea

Activation Energy

[kJ/kg]

T

Temperature

[K]

β

Correction of pre-exponential factor

[-]

AUTHOR INFORMATION Tamás Varga, PhD

Omar Péter Hamadi

Associate professor

MSc student

University of Pannonia, Hungary

University of Pannonia, Hungary

Department of Process Engineering

Department of Process Engineering

10 Egyetem Str.

10 Egyetem Str.

H-8200 Veszprém, P.O.B 158

H-8200 Veszprém, P.O.B 158

e-mail: [email protected]

e-mail: [email protected]


37

Energy & Fuels 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 38 of 49

Acknowledgment

The authors acknowledge the financial support of the project of the Economic Development and Innovation Operative Programme of Hungary, GINOP-2.3.2-15-201600053: Development of liquid fuels having high hydrogen content in the molecule (contribution to sustainable mobility) and the project of Széchenyi 2020 under the EFOP-3.6.1-16-2016-00015: University of Pannonia’s comprehensive institutional development program to promote Smart Specialization Strategy. The Project is supported by the European Union and co-financed by Széchenyi 2020.

References

[1] N. Taufiqurrahmi and S. Bhatia, ‘Catalytic cracking of edible and non-edible oils for the production of biofuels’, Energy & Environmental Science, vol. 4, no. 4, p. 1087, 2011. [2] S. P. Srivastava and J. Hancsók, Fuels and fuel-additives. Hoboken, New Jersey: Wiley, 2013. [3] Y. K. Ong and S. Bhatia, ‘The current status and perspectives of biofuel production via catalytic cracking of edible and non-edible oils’, Energy, vol. 35, no. 1, pp. 111–119, Jan. 2010. [4] J. Hancsók, D. Sági, and J. Valyon, ‘Diesel fuel blending components from mixture of waste animal fat and light cycle oil from fluid catalytic cracking’, Journal of Environmental Management, vol. 223, pp. 92–100, Oct. 2018. [5] G. Astarita and S. I. Sandler, Eds., Kinetic and thermodynamic lumping of multicomponent mixtures: proceedings of an ACS Symposium on Kinetic and Thermodynamic Lumping of Multicomponent Mixtures, Atlanta, GA, April 15, 1991. Amsterdam ; New York: Elsevier, 1991.


38

Page 39 of 49 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Energy & Fuels

[6] Y.-S. Ooi, R. Zakaria, A. R. Mohamed, and S. Bhatia, ‘Catalytic Cracking of Used Palm Oil and Palm Oil Fatty Acids Mixture for the Production of Liquid Fuel: Kinetic Modeling’, Energy & Fuels, vol. 18, no. 5, pp. 1555–1561, Sep. 2004. [7] Z. D. Yigezu and K. Muthukumar, ‘Biofuel production by catalytic cracking of sunflower oil using vanadium pentoxide’, Journal of Analytical and Applied Pyrolysis, vol. 112, pp. 341–347, Mar. 2015. [8] S. Sunarno, R. Rochmadi, P. Mulyono, M. Aziz, and A. Budiman, ‘Kinetic Study of Catalytic Cracking of Bio-oil over Silica-alumina Catalyst’, BioResources, vol. 13, no. 1, Jan. 2018. [9] H. F. Meier, V. R. Wiggers, G. R. Zonta, D. R. Scharf, E. L. Simionatto, and L. Ender, ‘A kinetic model for thermal cracking of waste cooking oil based on chemical lumps’, Fuel, vol. 144, pp. 50–59, Mar. 2015. [10] B. Periyasamy, ‘Reaction pathway analysis in thermal cracking of waste cooking oil to hydrocarbons based on monomolecular lumped kinetics’, Fuel, vol. 158, pp. 479–487, Oct. 2015. [11] M. Anand and A. K. Sinha, ‘Temperature-dependent reaction pathways for the anomalous hydrocracking of triglycerides in the presence of sulfided Co–Mo-catalyst’, Bioresource Technology, vol. 126, pp. 148–155, Dec. 2012. [12] A. K. Sinha et al., ‘Development of Hydroprocessing Route to Transportation Fuels from Non-Edible Plant-Oils’, Catalysis Surveys from Asia, vol. 17, no. 1, pp. 1–13, Mar. 2013. [13] A. A. Forghani, M. Jafarian, P. Pendleton, and D. M. Lewis, ‘Mathematical modelling of a hydrocracking reactor for triglyceride conversion to biofuel: model establishment and validation: Analysis of a hydrocracking reactor for triglyceride conversion’, International Journal of Energy Research, vol. 38, no. 12, pp. 1624–1634, Oct. 2014. [14] P. Forzatti, ‘Catalyst deactivation’, Catalysis Today, vol. 52, no. 2–3, pp. 165–181, Sep. 1999. [15] C. H. Bartholomew and R. J. Farrauto, Fundamentals of industrial catalytic processes, 2nd ed. Hoboken, N.J: Wiley, 2006. [16] B. W. Wojciechowski, ‘THE KINETIC FOUNDATIONS AND THE PRACTICAL APPLICATION OF THE TIME ON STREAM THEORY OF CATALYST DECAY’, Catalysis Reviews, vol. 9, no. 1, pp. 79–113, Jan. 1974. [17] C. G. Rudershausen and C. C. Watson, ‘Variables affecting activity of molybdena-alumina hydroforming catalyst in aromatization of cyclohexane’, Chemical Engineering Science, vol. 3, no. 3, pp. 110–121, Jun. 1954. [18] E. Wolf, ‘Kinetics of deactivation of a reforming catalyst during methylcyclohexane dehydrogenation in a diffusion reactor’, Journal of Catalysis, vol. 46, no. 2, pp. 190–203, Feb. 1977. [19] S. Szepe and O. Levenspiel, ‘Proceedings of the 4th European Symposium on Chemical Reaction Engineering’, in Chemical reaction engineering: proceedings, 1st ed., European Federation of Chemical Engineering, Ed. Oxford, New York: Pergamon Press, 1971. [20] L. Jossens, ‘Fouling of a platinum-rhenium reforming catalyst using model reforming reactions’, Journal of Catalysis, vol. 76, no. 2, pp. 265–273, Aug. 1982. [21] M. Pacheco, ‘On the development of a catalyst fouling model’, Journal of Catalysis, vol. 88, no. 2, pp. 400–408, Aug. 1984. [22] M. Pacheco, ‘On a general correlation for catalyst fouling’, Journal of Catalysis, vol. 86, no. 1, pp. 75–83, Mar. 1984.


39

Energy & Fuels 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 40 of 49

[23] I. Pitault, D. Nevicato, M. Forissier, and J.-R. Bernard, ‘Kinetic model based on a molecular description for catalytic cracking of vacuum gas oil’, Chemical Engineering Science, vol. 49, no. 24, pp. 4249–4262, 1994. [24] C. Chu, ‘Effect of Adsorption on Fouling of Catalyst Pellets’, Industrial & Engineering Chemistry Fundamentals, vol. 7, no. 3, pp. 509–514, Aug. 1968. [25] E. K. T. Kam, P. A. Ramachandran, and R. Hughes, ‘The effect of film resistances on the fouling of catalyst pellets—I Pseudo-steady state analysis’, Chemical Engineering Science, vol. 32, no. 11, pp. 1307–1315, 1977. [26] Z. Eller, Z. Varga, and J. Hancsók, ‘Advanced production process of jet fuel components from technical grade coconut oil with special hydrocracking’, Fuel, vol. 182, pp. 713–720, Oct. 2016. [27] G. Félix, J. Ancheyta, and F. Trejo, ‘Sensitivity analysis of kinetic parameters for heavy oil hydrocracking’, Fuel, vol. 241, pp. 836–844, Apr. 2019. [28] C. E. Galarraga, C. Scott, H. Loria, and P. Pereira-Almao, ‘Kinetic Models for Upgrading Athabasca Bitumen Using Unsupported NiWMo Catalysts at Low Severity Conditions’, Industrial & Engineering Chemistry Research, vol. 51, no. 1, pp. 140–146, Jan. 2012. [29] A. Monzón, E. Romeo, and A. Borgna, ‘Relationship between the kinetic parameters of different catalyst deactivation models’, Chemical Engineering Journal, vol. 94, no. 1, pp. 19–28, Jul. 2003. [30] S. Le Digabel, ‘Algorithm 909: NOMAD: Nonlinear Optimization with the MADS Algorithm’, ACM Transactions on Mathematical Software, vol. 37, no. 4, pp. 1–15, Feb. 2011.


40

Page 41 of 49 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Energy & Fuels

Figure 1. Experimental data, LHSV = 1.0 h-1 508x254mm (96 x 96 DPI)


Energy & Fuels 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Figure 2. Experimental data, LHSV = 2.0 h-1 508x254mm (96 x 96 DPI)


Page 42 of 49

Page 43 of 49 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Energy & Fuels

Figure 3.Proportion of conversions at LHSV = 1.0 h -1 divided with conversion at LHSV = 2.0 h-1 508x254mm (96 x 96 DPI)


Energy & Fuels 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Figure 4. The considered kinetic schemes with published and identified reaction rate constants at 673 K 370x238mm (96 x 96 DPI)


Page 44 of 49

Page 45 of 49 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Energy & Fuels

Figure 5. Correlation charts in case of M1, at 30 bar 508x254mm (96 x 96 DPI)


Energy & Fuels 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Figure 6. Sensitivity analysis of identified parameters of M1 model 508x254mm (96 x 96 DPI)


Page 46 of 49

Page 47 of 49 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Energy & Fuels

Figure 7. Coefficients of determination (R2) in case of M1-M4 at all temperatures and pressures 508x254mm (96 x 96 DPI)


Energy & Fuels 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Figure 8. Objective values in case of M1-M4 508x254mm (96 x 96 DPI)


Page 48 of 49

Page 49 of 49 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Energy & Fuels

Figure 9. Correlation charts in case of M4, at 30 bar 508x254mm (96 x 96 DPI)


Model based investigation of catalyst fouling in case of special

Recommend Documents