Safety evaluating of Beckmann rearrangement of cyclohexanone oxime in microreactors using inherently safer design concept

Safety evaluating of Beckmann rearrangement of cyclohexanone oxime in microreactors using inherently safer design concept

Chemical Engineering and Processing 110 (2016) 44–51 Contents lists available at ScienceDirect Chemical Engineering and Processing: Process Intensifi...

712KB Sizes 1 Downloads 20 Views

Chemical Engineering and Processing 110 (2016) 44–51

Contents lists available at ScienceDirect

Chemical Engineering and Processing: Process Intensification journal homepage: www.elsevier.com/locate/cep

Safety evaluating of Beckmann rearrangement of cyclohexanone oxime in microreactors using inherently safer design concept J.S. Zhang, K. Wang, C.Y. Zhang, G.S. Luo* The State Key Laboratory of Chemical Engineering, Department of Chemical Engineering, Tsinghua University, Beijing 100084, China

A R T I C L E I N F O

Article history: Received 4 April 2016 Received in revised form 19 August 2016 Accepted 17 September 2016 Available online 19 September 2016 Keywords: Safety evaluating Inherently safer design Beckmann rearrangement Microreactor Risk transfer

A B S T R A C T

Beckmann rearrangement of cyclohexanone oxime is an important step to produce caprolactam. Many researches have been devoted to it by using new organic catalysts or new reactors such as microreactor to improve the process efficiency. In this paper, the process was assessed by an inherently safer design (ISD) methodology based on risk approach to evaluate the safety effects of new organic catalysts and microreactors. The results show that organocatalyzed Beckmann rearrangement in continuous stirred tank reactors is unacceptable from the view of inherently safer concept due to the large amount of use of organic solvent and the large reactor volume. However, the application of microreactors could significantly increase the safety of both traditional and organocatalyzed Beckmann rearrangement processes with making inventories much smaller. The hazard conflicts that would be transferred to other parts of the process due to the application of microreactors were also evaluated by the Likelihood Index of Hazard Conflicts (LIHCs). The high value of LIHC because of microreactors indicates that some factors need to be paid more attention at the early design stage. ã 2016 Elsevier B.V. All rights reserved.

1. Introduction Over the last two decades microreactors have gain rapid development and become a new branch of chemical engineering. Due to the dimension up to several hundreds of micrometers, microreactors have the characteristics of fast mixing, excellent heat and mass transfer and inherent safety [1,2]. A lot of dangerous chemical processes have been performed in microreactors, such as nitration [3], fluorination [4], hydrogenation [5], oxidization [6], and rearrangement [7,8]. Compared to the conventional reactors, the application of microreactors can bring many advantages: higher selectivity, faster reaction rate, better process control and increased safety. However, little research on the safety evaluating of the application of microreactors has been reported [9]. Though it is thought to be inherently safe due to the minimal use of reactants and greatly reduced size of process equipment. The use of microreactors may also bring some risks such as easy blockage of channels and the fast dynamics of the process. In addition, microreactors are often used to handle extreme process conditions and involve with numbering-up of the reactors, which require advanced control system [10]. As a result, it is still desirable to

* Corresponding author. E-mail address: [email protected] (G.S. Luo). http://dx.doi.org/10.1016/j.cep.2016.09.012 0255-2701/ã 2016 Elsevier B.V. All rights reserved.

evaluate the safety of the application of microreactors to find out that how much the microreactor can improve the safety and to indicate the potential hazards and conflicts at the early design stage. The Beckmann rearrangement of cyclohexanone oxime (COX) in oleum (Fig. 1) is an important step in the production of e-caprolactam, the monomer of nylon-6 [11]. Since the rearrangement is very rapid and highly exothermic, it is a serious issue for the process safety. A loop reactor is usually applied for controlling the reaction temperature to avoid the thermal runaway in the reactor. In our previous papers, we tried to carry out the rearrangement reaction in oleum with the microreactor [7,8]. The external circulation was canceled but the cyclohexanone oxime had to be dissolved in the organic inert solvent of n-octane to dilute the reactants and remove reaction heat. Furthermore, we have improved a catalytic system [12] based on trifluoroacetic acid (TFA) developed by Ronchin et al. [13,14] which can avoid the side production of ammonium sulfate. The new process which is performed under much milder conditions brings a new safety problem with the large amount use of trifluoroacetic acid and acetonitrile. To improve the efficiency and safety the microreactor was also applied for the organocatalyzed Beckmann rearrangement [15]. As mentioned above, three new processes have been proposed to carry out the rearrangement of which two are in use of microreactors and one is in use of organic catalyst. This typical

J.S. Zhang et al. / Chemical Engineering and Processing 110 (2016) 44–51

45

Nomenclature

Notation AP AIT C DI DPfe DR dfpro and dfbp EFco, EFph and EFre

F1, F2, F3 and F4

FP FRP Hc DHr INV M Ndf NR, NF and NH NFPA

OT pn1, pn2, pn3, pn4, pn7 and pn8

PP V VP

Atmospheric pressure (kPa) Auto ignition temperature ( C) Concentration of reactant (mol/m3) Damage index Damage potential for fire and explosion Damage radius (m) Design factor for the new process and base process Energy factors for combustion, physical and reaction energy Initial energy factors for combustion, physical and reaction energies Flash point ( C) Fire point ( C) Heat of combustion (kJ/kg) Reaction enthalpy (kJ/mol) Quantity of chemicals involved (tons) Mass of the chemical in use (kg) Total number of the design factor Rankings for reactivity, flammability and health of the chemicals Operating temperature ( C) Penalties for temperature, pressure, quantity of chemical stored, hazardous characteristics, type of reaction and side reaction or decomposition Process pressure (kPa) Volume of reactant (m3) Vapour pressure (kPa)

Abbreviation IRDI Inherent Risk of Design Index ISD Inherently Safer Design ISI Inherent Safety Index I2SI Integrated Inherent Safety Index LIHCpro-i Likelihood Index of Hazard Conflicts LIISD Likelihood Index of Inherently Safer Design LSISD actual Likelihood Score of Inherently Safer Design PFSi Process Factor Score of each design factor QI2SD Quantitative Index of Inherently Safer Design RISI Risk-based Inherent Safety Index SWeHI Safety Weighted Hazard Index TLS Total Likelihood Score

Fig. 1. The Beckmann rearrangement of cyclohexanone oxime to e-caprolactam.

Inherently Safer Design (ISD) aiming to eliminate or minimize the sources of harm by using fewer hazardous chemicals, smaller inventories and milder process conditions have been identified as a reliable technique to design a safer, sustainable and economically viable process plant [16–19]. Inherently safer process means the process is much safer whenever you have chosen any equipment or system. It may be interesting to evaluate the safety of the application of microreactors with the ISD concept. There have been many ISD methods developed, such as Inherent Safety Index (ISI) [20], graphical method [21], Safety Weighted Hazard Index (SWeHI) [22] and Integrated Inherent Safety Index (I2SI) [23]. Rathnayaka et al. [24] further extended the I2SI method, called Risk-based Inherent Safety Index (RISI) to analyze and implement inherent safety throughout the process design life cycle. Tugnoli et al. [25,26] proposed a consequence-based tool to assess the inherent safety of process alternatives using the concept of key performance indicators. However, these tools may ignore the risk transferred due to the change of design. Recently, another approach called Quantitative Index of Inherently Safer Design (QI2SD) [27], which combined the method of Likelihood Index of Hazard Conflicts (LIHCs) [28] and Safety Weighted Hazard Index (SWeHI), has been proposed to indicate the potential hazards and conflicts at the early design stage with the ISD concept. In this work, we try to use the method of QI2SD to evaluate the safety of Beckmann rearrangement of cyclohexanone oxime. The objective is to find the effect of the microreactors and organic catalyst on the safety and identify the inherently safer process option. The potential damage indices of the conventional process and the three new ones were first estimated and then hazard conflicts of the three new ones were evaluated. 2. Methods The QI2SD method includes three steps: quantify inherent hazards, evaluate inherent safety conflicts and rank the ISD alternatives. The details are given as follows. 2.1. Quantify inherent hazards The quantification of hazards is based on the tools of Safety Weighted Hazard Index (SWeHI) [22] and I2SI [23], in which the potential energy is first calculated and then correlated as the potential damage as damage radius (DR). The damage index (DI) is evaluated at last. DI index is calculated as follows. DI = Max (5, Min (100, DR/2))

process is a good model to do the safety assessment to help to understand the safety effect of the application of microreactors and realize the process optimization and process route selection from the view of safety.

(1)

The estimation of DR is according to Eq. (2) to represent 50% probability of fatality or damage, which is the common calculation method [22,23]. DR = 4.76(DPfe)1/3

(2)

46

J.S. Zhang et al. / Chemical Engineering and Processing 110 (2016) 44–51

DPfe = (EFco + EFph + EFre)  pn3  pn4

(3)

where DPfe is the damage potential for fire and explosion; EFco is the energy factor for combustion energy; EFph is the energy factor for physical energy; EFre is the energy factor for reaction energy; pn3 is the penalty for quantity of chemical stored and pn4 is the penalty for the hazardous characteristics. The energy factors, EF are calculated by the initial energy factors and their corresponding penalties. The equations are as follows. EFco = F1  pn1

(4)

EFph = f (F2, F3)  pn2

(5)

EFre = F4  pn7  pn8

(6)

F1 = 0.1  M  Hc/K

(7)

F2 = 1.304  10

3

 PP  V

(8)

V is the volume of reactant; C is the concentration of reactant; DHr is the reaction enthalpy. The estimation equation for each penalty is given in Table 1. In Table 1, OT is the operating temperature; FP is the flash point; FRP is the fire point; AIT is the auto ignition temperature; AP is the atmospheric pressure; NR and NF are NFPA rankings for reactivity and flammability of the chemical respectively [29]; INV is the quantity of chemicals involved, including the chemicals in inventory. From the above equations, we can get the values of DR and DI for a specific unit or process. The DR of 200 m is usually used as the critical value with the assumption that the human population outside the 200 m radius has very low probability to be affected by the hazard. That means that the process could be considered as inherently safe when the DI value is below 100. But it has to be mentioned that the DR critical value of 200 m is only a guideline and it can be modified depending on some factors, such as the size of process plant and the number of the on-site employees. 2.2. Evaluate ISD conflicts This step is to identify potential conflicts when new process is designed. The conflicts are estimated with the Likelihood Index of Hazard Conflicts (LIHCs) [28], which is computed as follows: LIHCpro-i = (1-LIISDpro-i)

F3 = 1.0  10

3

 1/(OT + 273)  (PP

VP)2  V

(9)

F4 = V  C  DHr/3.148

(10)

where F1, F2, F3 and F4 are the initial energy factors for combustion, physical and reaction energies; pn1 is the penalty for temperature; pn2 is the penalty for pressure; pn7 is the penalty for type of reaction; pn8 is the penalty for side reaction or decomposition; M is the mass of the chemicals in the reactor; Hc is the heat of combustion; PP is the process pressure; VP is the vapour pressure;

(11)

where pro-i is the process-i; LIHC is the Likelihood Index of Hazard Conflicts for process-i and LIISD is Likelihood Index of Inherently Safer Design for process-i. The LIISD is estimated to represent the changes in hazard magnitude with the selected IS principles as described in Table 2. Therefore, the likelihood of risk change can be captured in the main process unit and other related site-process units such as the auxiliary units, chemicals storages and transportation at the early design stage.

Table 1 Detailed equations for each penalty [22]. Penalty

Estimation equation

pn1 pn2 pn3 pn4 pn7

=If(OT > FP, if(OT < FRP, 1.45; if(OT < 0.75  AIT, 1.75; 1.95)), 1.1) =If(VP > AP, IF(VP < PP, 1 + 0.6  (/PP-VP)/PP, F = F2 + F3; 1 + 0.4  (VP PP)/PP, F = F2), 1 + 0.2  (PP VP)/PP, F = F3) =If(Max(NR, NF) = 4, 0.01  INV  1000 + 1, If(Max(NR, NF) = 3; 0.007  INV  1000 + 1, If(Max(NR, NF) = 2, 0.005  INV  1000 + 1.05; 0.002  INV  1000 + 1.02))) =1 + 0.25  (NR + NF) Oxidation, 1.60; electrolysis, 1.20; nitration, 1.95; polymerization, 1.50; pyrolysis, 1.45; halogenation, 1.45; aminolysis, 1.40; esterification, 1.25; hydrogenation, 1.35; sulfonation, 1.30; alkylation, 1.25; reduction,1.10 Autocatalytic reaction, 1.65; non-autocatalytic reaction (above normal), 1.45; non-autocatalytic reaction (below normal), 1.20

pn8

Table 2 The factors considered in LIISD [27]. Inherent safety principles

Target characteristics

Target process safety factors

Substitution

Quality of materials used or produced Quantity of process inventory Operating and safe limit conditions

Hazardous of substances; NFPA ranking on flammability, reactivity and toxicity for feed, product and byproduct

Minimization Moderation

Volume; percent accumulated in vessel and intermediate storage, amount of gas release, concentration Thermal runaway

Fire and explosion Simplification

Easiness in the design and operating

Controllability—basic requirement Controllability—technical requirement Complexity on overall process unit and plant

Temperature effect: adiabatic temperature rise, time to maximum rate of runaway Pressure effect: vapor pressure, amount of solvent evaporated Temperature effect: flash point, flammability limits Pressure effect: fraction liquid vaporized, pressure build-up Basic controls in flow, temperature, pressure, level, etc Advanced control measures: emergency cooling, quenching and flooding, depressurisation Number of vessels, auxiliary units, complexity in maintenance

J.S. Zhang et al. / Chemical Engineering and Processing 110 (2016) 44–51

The calculation of LIISD is achieved by the ratio of the actual Likelihood Score of Inherently Safer Design (LSISDact) to the maximum LSISDmax as shown in: LIISDpro-i = LSISDact/LSISDmax

(12)

Table 4 Guidelines for the range of risk level [27]. IRDI

Risk level

Design Description

76– 200

High

LSISDact is derived from the summation of Total Likelihood Score (TLS) with all IS principles. The TLS for each principle is estimated by adding the Process Factor Score of each design factor as illustrated in Eqs. (10) and (11), respectively:

- Design option is highly critical - Redesign process is required

26– 75

Medium

- Design option is critical - Redesign may be required - Technical measures may be required

LSISDact = TLSsub + TLSmin + TLSmod + TLSsim

(13)

0–25

Low

- Design option is inherently safer - Additional risk reduction may not required - Follow standard process safety management

TLSj = SPFSi

(14)

where the subscripts j, i, sub, min, mod and sim refer to principle j, process factor scrore i, substitute, minimize, moderate and simplify, respectively. LSISDmax is calculated as follows: LSISDmax = Ndf  10

as ‘High’ and the process is highly critical. The IRDI of 26–75 is considered as ‘Medium’ and the process is critical. The IRDI of 0–25 is considered as ‘Low’ and the process is inherently safe.

(15)

where Ndf is the total number of the design factor (df) for a specific process. PFS of the process-i for one IS principle is estimated as follows: PFSi = Max [ 10, (1

PFSi = Min [10,(1

47

dfpro/dfbp)  10] if dfpro > dfbp

(16)

dfbp/dfpro)  ( 10)] if dfpro < dfbp

(17)

where the subscripts i refers to design factor i, dfpro is the design factor for the new process and dfbp is the design for the base process. There are four principles during the estimation of df value. The Likelihood Score for substitute, minimize and moderate principles is estimated using the actual value for each design factor dfpro and dfbp. But the estimation of Likelihood Score for simplification principle needs to apply the guidelines as shown in Table 3. The application of indices in the table is based on the process factors in the design, which can be obtained by basic design calculations or expert judgements. 2.3. Rank ISD alternatives The Inherent Risk of Design Index (IRDI) is at last developed to capture the potential of the risk transfer and illustrate the criticality of the hazards. IRDIpro-i = DIpro-i  LIHCpro-i

(18)

The Inherent Risk of Design Index (IRDI) is categorized to ‘High’, ‘Medium’ and ‘Low’ using the Dow F&E Index [30] as the benchmark. The guidelines for the three risk levels are shown in Table 4. The Maximum of IRDI is 200 as the maximum of DI and LIHC are 100 and 2, respectively. So the IRDI of 76–200 is considered as ‘High’ and redesign is required. The IRDI of 76–200 is considered Table 3 Guidelines for the estimation of simplification principle [27]. Description

Index value

Essential very important Important Not important but required Required Requirement is moderate Good if available Requirement does not affect process Not required

10 9 8 7 6 5 4 3 1–2

2.4. Description of the Beckmann rearrangement Beckmann rearrangement of cyclohexanone oxime is an important step to producee-caprolactam, the monomer for nylon-6. The conventional process is shown in Fig. 2(a). A large amount of rearrangement mixture including COX, oleum and caprolactam, is recycled by a pump to remove the reaction heat. The mass fraction of oleum in the mixture is about 54 wt% and the others are unreacted COX and caprolactam. COX is mixed with the rearrangement mixture using the mixer in the rearrangement reactor [11]. The rearrangement reaction in microreactors using oleum (Process 1) is shown in Fig. 2(b) [7]. COX is dissolved in noctane as the continuous phase (mass fraction 15 wt%) and oleum is dispersed in the octane solutions as microdroplets in the micromixer. The rearrangement reaction proceeds in the delay loop connecting directly to the micromixer. The mixture is then separated in a split phase tank. During the process, there is no external circulation. Fig. 2(c) shows the organocatalyzed Beckmann rearrangement with the trifluoroacetic acid as the catalyst in continuous stirred tank reactors (Process 2) [12]. COX (15 wt%), acetonitrile (25 wt%) and TFA (60 wt%) are mixed in the reactor. There is no external circulation too as the reaction is slow and mild. The organocatalyzed Beckmann rearrangement in microreactors is shown in Fig. 2(d) (Process 3) [15]. COX which is dissolved in acetonitrile mixes with TFA in the micromixer. The rearrangement reaction is then finished in the delay loop. The composition of mixture in the delay loop is the same as that in the reactor of Process 2. The process pressure has to be controlled at 0.6 MPa with a back pressure valve. As the index adopted here is subjected to size of the plant, the capacity of the process is assumed to be 70 thousand tons per year. The topical process parameters for the four processes are listed in Table 5. The reaction enthalpies of the reaction in the oleum and organic catalyst are 254 [31] and 169 [15] kJ/mol, respectively. Although the rearrangement reaction in the oleum is very fast and can be finished with seconds at industrial conditions, long residence time (40 min) is still needed due to the limited mixing performance and poor heat transfer in conventional reactors [11]. Due to excellent mixing and enhanced heat transfer, the residence time in the microreactors (Process 1) can be decreased to less than 40 s [7]. To ensure the conversion, the residence time in Process 1 is assumed to be 1 min. The organocatalyzed Beckmann rearrangement in the trifluoroacetic acid is much milder and no limitations of mixing or heat transfer exist. As a result, the residence time is determined by the kinetics. As reported in our previous studies, the reaction at the operating temperature can be finished within

48

J.S. Zhang et al. / Chemical Engineering and Processing 110 (2016) 44–51

Fig. 2. The schematic overview of the four processes for Beckmann rearrangement.

60 min and 2 min for Process 2 and 3, respectively [12,15]. The working hours for the plant is assumed to be 8000 h. Thus the capacity of the process can be calculated and then the reactor volume required can be determined with the results of residence time. It should be noted that volume of reactor can represent the process productivity, which means that the small the reactor is, the more productive the process is. From the reactor volume, we can conclude that the Process 1 and 3 with microreacors are much more efficient and Process 2 has the smallest productivity. In this paper, the microreactors are assumed to have the same capacity with the conventional reactors. So of course the microreactors need to be scale up. But only the reactor volume is required for the safety evaluation, which can be obtained as mentioned above. So here we will not talk about the scaling-up of the microreactors. 3. Results and discussion 3.1. Quantify inherent hazards Table 6 shows the DR and DI calculation process for the conventional rearrangement process. There is no guideline for estimating pn7 of Beckmann rearrangement. It is assumed as 1.45 according to the characteristic of the rearrangement. The Damage Radius (DR) is determined to be 188 and Damage Index (DI) is 94, which is very close to the corresponding tolerable limit of 200 and 100. Though it can be accepted for the production, the process is not so inherently safe. It can easily exceed the tolerable limit if the process capacity is increased. We can find that energy factors for combustion (EFco) and reaction energies (EFre) are the main reason

for high damage potential. The initial energy factor for physical energy (EFph) can be ignored. From the calculation equation for energy factors, the inherent risk is mainly due to the large mass of chemical in use, large reactor volume, high reaction enthalpy and reactant concentration. If any scenario of failure occurs such as pump for external circulation, the system temperature would be uncontrolled and may result in loss of containment such as fire and explosion. Therefore, new inherently safer design alternatives are also required to reduce the potential energy of the rearrangement reaction. Based on the above method, the DR and DI values for Processes 1, 2 and 3 are calculated as shown in Table 7. Similar to the results in conventional rearrangement, energy factors for combustion (EFco) and reaction energies (EFre) contribute most to the calculation of DR and DI. The initial energy factor for physical energy (EFph) can be also ignored. As shown in Table 7, Processes 1 and 3 show a significant reduction of DR and DI, which is much smaller than the tolerable limit. The results indicate that the application of microreactors can greatly improve the inherent safety for Beckmann rearrangement. From the data shown in Table 7, we can find that energy factors for combustion (EFco) and reaction energies (EFre) are dramatically decreased by about 50 times in Processes 1 and 3. This is because that the mass of reactants and reactor volume is greatly reduced by about 5–10 times in microreactors than that in conventional reactors. This is the main reason for the greatly reduced values of DR and DI. For Process 1, plenty of octane is used (about 5.6 times of COX amount in use), which can greatly increase the possibility of fire and explosion. The DR and DI would be greatly increased if the process is still carried

Table 5 The topical process data for the four processes [7,11,12,15]. Process parameters

Conventional process

Process 1

Process 2

Process 3

Chemicals

Oleum, COX (46 wt %) 9.5 m3 40 min 90–125  C 101 kPa 254 kJ/mol

Oleum, COX (15 wt% in octane), octane (85 wt %) 1 m3 1 min 90  C 101 kPa 254 kJ/mol

TFA (60 wt%), COX (15 wt%), acetonitrile 53 m3 60 min 70  C 101 kPa 169 kJ/mol

TFA (60 wt%), COX (15 wt%), acetonitrile 1.8 m3 2 min 110  C 606 kPa 169 kJ/mol

Reactor volume Residence time Temperature Process pressure Reaction enthalpy

J.S. Zhang et al. / Chemical Engineering and Processing 110 (2016) 44–51

49

Table 6 The DR and DI calculation process for the conventional rearrangement reaction [32]. Parameters

value

Factors

value

Mass of the chemical in use, M, kg Heat of combustion, Hc, kJ/kg Process pressure, PP, kPa Volume of the unit, V, m3 Flash point of the chemical, FP,  C Fire point of the chemical, FRP,  C Autoignition temperature, ATI,  C Operating temperature, OT,  C NFPA rank for reactivity, NR NFPA rank for flammability, NF NFPA rank for health, NH Quantity involved, INV, tons

1.3  104 1.47  104 101.3 9.5 93 100 265 90–125 2 1 3 100

6.07  103 1.25 1.11 10 6 1.11 10 6 4.16  103 1.75 1 1.7 1.75 1.45 1.65 1.06  104

Reactant concentration, C, mol/m3

5.43  103

Vapour pressure, VP, kPa

1

Enthalpy of reaction, DHr, kJ/mol

254

Type of reaction

1.45

Side reaction

1.65

F1 (0.1  M  Hc/K) F2 (1.304  10 3  PP  V) F3 (1.0  10 3  1/(T + 273)  (PP-VP)2  V) F (f(F2, F3)) F4 (V  C  DHr/K) pn1 pn2 pn3 pn4 pn7 pn8 EF combustion (F1  pn1) EF physical (f (F2, F3)  pn2) EF reaction (F4  pn7  pn8) Damage Potential (DP) ((EFco + EFph + EFre)  pn3  pn4) Damage Radius (DR) (4.76(DPfe)1/3) Damage Index (DI) (Max (5, Min (100, DR/2)))

Table 7 The DR and DI calculation results for other processes. Factors

Process 1

Process 2

Process 3

F1 F2 F3 F F4 pn1 pn2 pn3 pn4 pn7 pn8 EF combustion EF physical EF reaction DP DR DI

1.41 103 0.13 1.10  10 7 1.10  10 7 1.04  102 1.75 1.1 3.38 2.25 1.45 1.45 2.47  103 1.25  10 7 2.19  102 2.04  104 130 65

2.32  104 7.00 6.18  10 6 6.18  10 6 4.17  103 1.75 1.1 3.03 1.75 1.25 1.2 4.06  104 6.53  10 6 6.26  103 2.49  105 299 100

7.76  102 1.42 2.10  10 7 1.42 1.42  102 1.75 1.36 3.03 1.75 1.25 1.2 1.36  103 1.94 2.13  102 8.34  103 97 48

out in conventional reactors. Because of microreactors, the DR and DI are decreased to 130 and 65, much smaller than the tolerable limit. For Process 2, the DR and DI values are as large as 299 and 100, respectively. Therefore, though the process is slow and mild, organocatalyzed Beckmann rearrangement in continuous stirred tank reactors is unacceptable from the view of inherently safer concept. The large amount of organic solvent and the large reactor volume are the main reasons. With the use of microreactors in Process 3, the DR and DI values are reduced to 97 and 48, respectively. As a result, the microreactor is an effective way to significantly increase the safety of Beckmann rearrangement by using fewer hazardous chemicals and smaller inventories. 3.2. Evaluate ISD conflicts In this step, potential conflicts for the three new processes are identified through the application of LIHC. Table 8 shows the calculation process of LIHC. The LIHC values for the three processes are all higher than the conventional value of 1, indicating that some hidden hazards or instability could occur when the new process is

1.33  10

6

9.95  103 6.12  104 188 94

applied. The LIHC values for Processes 1 and 2 are 1.16 and 1.14, respectively. Process 3 has the largest LIHC value of 1.31, demonstrating a significant likelihood of hazard transferred. To find different contributions for the conflicts, values of Total Likelihood Score (TLS) for the four principles are compared as shown in Fig. 3. For the substitution principle, all the values of Total Likelihood Score (TLS) are increased due to the application of flammable solvent of octane and acetonitrile. Processes 2 and 3 demonstrate a smaller value as TFA is not as reactive as oleum. For the minimization principle, only TLS of Process 2 is increased while those for Processes 1 and 3 are decreased. The results are mainly from the change of reactor volume. The unite volume of Process 2 is increased by about 5 times while those for Processes 1 and 3 are decreased by about 10 and 5 times, respectively. For the moderation principle, risk of fire and explosion are all increased for the three processes due to the application of organic solvent. The risk of reactor runaway and chemicals decomposition are reduced for Processes 2 and 3 as the Beckmann rearrangement in TFA is slow and less exothermic. Process 3 has the largest TLS value due to the high operating temperature and pressure. The rigorous conditions in microreactors aiming to better utilize physical resources and reduce plant inventory will increase the robustness of plant. That means some disturbances such as flow and concentration fluctuations in time and space may endanger the stability of the process and increase the risk of fire and explosion. The result alerts us that special attention should be given to the protection systems for fire and explosion at the early design stage. The simplification principle especially for complexity to overall plant contributes most for the increase of TLS value of the Processes 1 and 3 due to the application of microreactors. As mentioned above, in microreactors more rigorous conditions are usually adopted and the channels are easy to be blocked by impurities due to the small dimension. As a result, the microreactors require high frequency of maintaining auxiliary units and advanced controls in order to ensure the reliability [33]. Besides, numbering-up of microreactors is usually required to increase the plant capacity, which would greatly increase the complexity of the plant and the control system. All these factors could bring high likelihood of

50

J.S. Zhang et al. / Chemical Engineering and Processing 110 (2016) 44–51

Table 8 The calculation process of LIHC. Process safety factors Substitution: characteristics of hazardous substance Flammable Explosive Reactive Toxicity TLSsub (SPFSi) Minimization: Quantity of process inventory Unite volume Feed and product vessels TLSmin (SPFSi) Moderation: Criticality of operating conditions Temperature (runway, decomposition) Adiabatic temperature rise Time to max rate Temperature (fire and explosion) Boiling point Flash point Lower flammability limit Pressure Operating pressure Vapor pressure TLSmod (SPFSi)

Process 1

Process 2

Process 3

15

10.0 5.0 10.0 0 5

10.0 5.0 10.0 0.0 5

10.0 0 10

10.0 6.7 16.7

10.0 5.0 0 0

10 0.0

10.0 3.3 6.7

5.0 10.0

7.5 10.0 6.0 0

5.0 10.0

10.0 10.0 6.0

10.0 10.0 6.0

6.7 17.7

10.0 6.7 27.7

0 3.3 16.8

Simplification: Complexity of unites and overall process Complexity 1: Controllability-basic control requirements Temperature Pressure Flow Level

6.7 4.0 6.7 3.3

6.7 0 3.3 0

2.5 10.0 6.7 3.3

Complexity 2: Controllability-advance safety control requirements Forced dilution system Blast wall Depressurization Quenching and flooding

2.5 0 3.3 3.3

10.0 3.3 0 6.0

2.5 0.0 6.7 3.3

0.0 6.0 2.0 10.0 4.7 1.14

10.0 10.0 10.0 10.0 51.7 1.31

Complexity 3: Complexity to overall plant Frequency in maintenance of auxiliary units, advanced controls Process extension- numbering-up of the unites Strength of equipment Resistant material TLSsim (SPFSi) LIHC(1 (TLSsub + TLSmin + TLSmod + TLSsim)/(10  Ndf))

hazards transferred. The facts have been pointed out by Luyben [34] that intensified continuous reactor process would put the process stability in jeopardy and highly dependent on the instrument which could fail and lead to an explosion. In addition,

10.0 10.0 0 0 18.2 1.16

resistant material is required because of the use of corrosive TFA in Processes 2 and 3. And stronger strength of equipment is required due to the higher operating pressure in Process 3. As a result, Process 3 has the highest likelihood of hazard transferred, which alerts the designer that the design of Process 3 must be further evaluated based on the above factors. In general, the moderation and simplification principles contribute to most of the risk transferred. The designer should pay more attention to the factors based on the two principles and try to minimize the hazard of potential likelihood. The results also demonstrate that the DI alone would not fully indicate the actual hazards in the process. The estimation of LIHC is important to find the new hazard transferred.

Table 9 The risk ranking for the three processes.

Fig. 3. The values of Total Likelihood Score (TLS) for the four principles.

Process

DI

LIHC

IRDI

Risk Level

Process 1 Process 2 Process 3

65 100 48

1.16 1.14 1.31

75 114 63

Medium High Medium

J.S. Zhang et al. / Chemical Engineering and Processing 110 (2016) 44–51

3.3. Rank the three processes Finally, the risk ranking for the three processes is shown in Table 9. The estimated IRDI for Process 2 show the high risk level of 114. That means organocatalyzed Beckmann rearrangement in continuous stirred tank reactors is highly dangerous and redesign or technical safety measure are highly required. The risk level for Processes 1 and 3 are 75 and 63, respectively, which is medium. It means that the process is inherently safer but some technical measures may be required. The results demonstrate that the application of microreactor can significantly increase the safety of Beckmann rearrangement. However, the high value of LIHC because of microreactors reminds us that some factors need to be paid more attention at the early design stage. For the process in microreactors with more rigorous conditions, more protection systems such as blast wall and forced quenching and flooding should be used. Meanwhile, the advantage brought by microreactors is that the reactor volume is much lower due to the fewer chemicals. To increase the robustness of plant, some buffer unit can be added to suppress the rangeability of parameters. For the risk transferred in the simplification principle, more reliable control system for microreactors should be developed and more precise instruments should be used at the early stage of process design. 4. Conclusions In this work, the Quantitative Index of Inherently Safer Design (QI2SD), which combined the method of Likelihood Index of Hazard Conflicts (LIHCs) and Safety Weighted Hazard Index (SWeHI), is applied for Beckmann rearrangement of cyclohexanone oxime to evaluate the effect of microreators on the process safety. The damage index (DI) is reduced from a high value of 94 in conventional process to lower ones of 65 and 48 in microreactors. The results show that the microreactors can greatly improve the inherent safety by using fewer chemicals and smaller inventories. Furthermore, with the Likelihood Index of Hazard Conflicts (LIHCs), potential hazards and conflicts transferred due to the microreactors are found to be focused on the moderation and simplification principle, including increasing the robustness of plant, higher requirement to maintain auxiliary units and more complicated control system. But after comparing several processes, the application of microreactor still can be considered as an inherently safer option for the Beckmann rearrangement. In the future more work can be carried out to evaluate the safety of microreactors with more other methods. Acknowledgements We gratefully acknowledge the supports of the National Natural Science Foundation of China (91334201, U1463208, 21506110) and China Postdoctoral Science Foundation (2015M570111) on this work. References [1] M.N. Kashid, L. Kiwi-Minsker, Microstructured reactors for multiphase reactions: state of the art, Ind. Eng. Chem. Res. 48 (2009) 6465–6485. [2] R.L. Hartman, K.F. Jensen, Microchemical systems for continuous-flow synthesis, LabChip 9 (2009) 2495–2507. [3] L. Ducry, D.M. Roberge, Controlled autocatalytic nitration of phenol in a microreactor, Angew. Chem. Int. Ed. 44 (2005) 7972–7975. [4] P. Löb, H. Löwe, V. Hessel, Fluorinations, chlorinations and brominations of organic compounds in micro reactors, J. Fluorine Chem. 125 (2004) 1677–1694.

51

[5] G.S. Calabrese, S. Pissavini, From batch to continuous flow processing in chemicals manufacturing, AIChE J. 57 (2011) 828–834. [6] J.H. Park, C.Y. Park, H.S. Song, Y.S. Huh, G.H. Kim, C.P. Park, Green diacetoxylation of alkenes in a microchemical system, Org. Lett. 15 (2013) 752–755. [7] J.S. Zhang, K. Wang, Y.C. Lu, G.S. Luo, Beckmann rearrangement in a microstructured chemical system for the preparation of e-caprolactam, AIChE J. 58 (2012) 925–931. [8] J.S. Zhang, K. Wang, X.Y. Lin, Y.C. Lu, G.S. Luo, Intensification of fast exothermic reaction by gas agitation in a microchemical system, AIChE J. 60 (2014) 2724– 2730. [9] O. Klais, F. Westphal, W. Benaissa, D. Carson, J. Albrecht, Guidance on safety/ health for process intensification including MS design. Part III: risk analysis, Chem. Eng. Technol. 33 (2010) 444–454. [10] J.A. Moulijn, A. Stankiewicz, J. Grievink, A. Górak, Process intensification and process systems engineering: a friendly symbiosis, Comput. Chem. Eng. 32 (2008) 3–11. [11] K.T. Zuidhof, M.H.J.M. de Croon, J.C. Schouten, Beckmann rearrangement of cyclohexanone oxime to e-caprolactam in microreactors, AIChE J. 56 (2010) 1297–1304. [12] J.S. Zhang, A. Riaud, K. Wang, Y.C. Lu, G.S. Luo, Beckmann rearrangement of cyclohexanone oxime to e-caprolactam in a modified catalytic system of trifluoroacetic acid, Catal. Lett. 144 (2014) 151–157. [13] L. Ronchin, A. Vavasori, On the mechanism of the organocatalyzed Beckmann rearrangement of cyclohexanone oxime by trifluoroacetic acid in aprotic solvent, J. Mol. Catal. A: Chem. 313 (2009) 22–30. [14] L. Ronchin, A. Vavasori, M. Bortoluzzi, Organocatalyzed Beckmann rearrangement of cyclohexanone oxime by trifluoroacetic acid in aprotic solvent, Catal. Commun. 10 (2008) 251–256. [15] J.S. Zhang, C. Dong, C.C. Du, G.S. Luo, Organocatalyzed Beckmann rearrangement of cyclohexanone oxime in a microchemical system, Org. Process Res. Dev. 19 (2015) 352–356. [16] R. Srinivasan, S. Natarajan, Developments in inherent safety: a review of the progress during 2001–2011 and opportunities ahead, Process Saf. Environ. 90 (2012) 389–403. [17] M.H. Ordouei, A. Elkamel, G. Al-Sharrah, New simple indices for risk assessment and hazards reduction at the conceptual design stage of a chemical process, Chem. Eng. Sci. 119 (2014) 218–229. [18] Y. Lee, S. Lee, S. Shin, G. Lee, J. Jeon, C. Lee, C. Han, Risk-based process safety management through process design modification for gas treatment unit of gas oil Separation plant, Ind. Eng. Chem. Res. 54 (2015) 6024–6034. [19] P. Gangadharan, R. Singh, F. Cheng, H.H. Lou, Novel methodology for inherent safety assessment in the process design stage, Ind. Eng. Chem. Res. 52 (2013) 5921–5933. [20] A.M. Heikkilä, M. Hurme, M. Järveläinen, Safety considerations in process synthesis, Comput. Chem. Eng. 20 (1996) S115–S120. [21] J.P. Gupta, D.W. Edwards, A simple graphical method for measuring inherent safety, J. Hazard. Mater. 104 (2003) 15–30. [22] F. Khan, T. Husain, S. Abbasi, Safety weighted hazard index (SWeHI)—a new, user-friendly tool for swift yet comprehensive hazard identification and safety evaluation in chemical process industries, Process Saf. Environ. 79 (2001) 65– 80. [23] F.I. Khan, P.R. Amyotte, Integrated inherent safety index (I2SI): a tool for inherent safety evaluation, Process Saf. Prog. 23 (2004) 136–148. [24] S. Rathnayaka, F. Khan, P. Amyotte, Risk-based process plant design considering inherent safety, Saf. Sci. 70 (2014) 438–464. [25] A. Tugnoli, V. Cozzani, G. Landucci, A consequence based approach to the quantitative assessment of inherent safety, AIChE J. 53 (2007) 3171–3182. [26] A. Tugnoli, F. Santarelli, V. Cozzani, Implementation of sustainability drivers in the design of industrial chemical processes, AIChE J. 57 (2011) 3063–3084. [27] R. Rusli, A.M. Shariff, F.I. Khan, Evaluating hazard conflicts using inherently safer design concept, Saf. Sci. 53 (2013) 61–72. [28] R. Rusli, A.M. Mohd Shariff, Qualitative assessment for inherently safer design (QAISD) at preliminary design stage, J. Loss Prev. Proc. 23 (2010) 157–165. [29] NFPA, Hazardous Chemical Data, National Fire Protection Association code325M, New Jersey, 1991. [30] AIChE, Dow’s Fire & Explosion Index Hazard Classification Guide, 7th ed., Wiley-American Institute of Chemical Engineers, New York, 1994. [31] N.T. Zuidhof, M.H.J.M. de Croon, J.C. Schouten, J.T. Tinge, Beckmann rearrangement of cyclohexanone oxime to e-caprolactam in a microreactor, Chem. Eng. Technol. 35 (2012) 1–6. [32] C.L. Yaws, Chemical Properties Handbook: Physical, Thermodynamic, Environmental, Transport, Safety, and Health Related Properties for Organic and Inorganic Chemicals, McGraw-Hill, New York, 1999. [33] N.M. Nika9 cevi c, A.E.M. Huesman, P.M.J. Van den Hof, A.I. Stankiewicz, Opportunities and challenges for process control in process intensification, Chem. Eng. Process. 52 (2012) 1–15. [34] W.L. Luyben, D.C. Hendershot, Dynamic disadvantages of intensification in inherently safer process design, Ind. Eng. Chem. Res. 43 (2004) 384–396.