Lumped kinetics for supercritical water oxidation of oily sludge

Process Safety and Environmental Protection 8 9 ( 2 0 1 1 ) 198–203

Contents lists available at ScienceDirect

Process Safety and Environmental Protection journal homepage: www.elsevier.com/locate/psep

Short communication

Lumped kinetics for supercritical water oxidation of oily sludge Bao-chen Cui a,b,∗ , Shu-zhi Liu a , Fu-yi Cui b , Guo-lin Jing a , Xian-jun Liu a a

Provincial Key Laboratory of Oil & Gas Chemical Technology, College of Chemistry & Chemical Engineering, Northeast Petroleum University, Daqing 163318, China b State Key Laboratory of Urban Water Resource and Environment, Harbin Institute of Technology, Harbin 150090, China

a b s t r a c t Oxidation of oily sludge in supercritical water was studied in batch reactor under the conditions of reaction temperature from 390 to 450 ◦ C, pressure up to 25 MPa, and time from 1 to 10 min. The oily sludge oxidation undergoes a parallel-consecutive reaction pathways, in which it first decomposes to intermediates of aliphatic ketones, aldehydes and carboxylic acids with conjugated double bonds and via low molecular mass organic acids to the final product carbon dioxide. A 4-lump kinetic model was proposed to describe supercritical water oxidation of oily sludge. The experimental data obtained were used to estimate the six kinetic constants and the corresponding activation energies in the model. The model testing results revealed that the model predictions were in good agreement with the experimental results. The model helps us get good insight into the performance of the batch reactor that would be useful for optimization of supercritical water oxidation of oily sludge. © 2011 The Institution of Chemical Engineers. Published by Elsevier B.V. All rights reserved. Keywords: Oily sludge; Supercritical water; Oxidation; Lumped kinetics; Reaction pathway

1.

Introduction

Supercritical water oxidation (SCWO), occurring in water above water critical point (Tc = 374 ◦ C and Pc = 22.1 MPa) is a promising and efficient process to treat hazardous organic compounds. When exceeding water critical point, the values of density, dielectric constant, and ionic product of water drop down, so supercritical water acts as a non-polar solvent of high diffusivity and excellent transport properties. In the presence of an oxidant such as O2 , the dense, high temperature environment characteristic of the SCWO process results in rapid and complete oxidation of organic species to CO2 , H2 O, and N2 . SCWO is an emission-free and discharge-free alternative to incineration (Mahmood and Elliott, 2006). Oily sludge usually contains plenty of waste oil. This waste oil is a mixture of compounds, mainly hydrocarbons. Advance analyses on waste oil show that it is composed of

40–52 wt% alkanes, 28–31 wt% aromatics, 7–22.4 wt% resins, and 8–10 wt% asphaltenes (Marks et al., 1992). Oxidative degradation of such compounds involves a myriad of successive and simultaneous reactions. This makes it unrealistic to envisage comprehensive kinetic studies based on exhaustive mechanistic considerations. At the other extreme, overall power-law kinetic rate approaches are often too simple to adequately describe the SCWO of complex mixtures. The lumped kinetics approach, which has been widely used in solid-catalyzed wet oxidation (Belkacemi et al., 2000), often offers a suitable trade-off between tedious mechanistic/kinetic formalisms and oversimplified power-law representations. No attempt in studying the lumped kinetics in SCWO of oily sludge has been found in the literature. This article presents, for the first time, a rigorous lumped kinetics framework for SCWO of oily sludge. Our hope upon initiating this investigation, was to provide support for designing SCWO reactors for commercial scale.

∗ Corresponding author at: Provincial Key Laboratory of Oil & Gas Chemical Technology, College of Chemistry & Chemical Engineering, Northeast Petroleum University, Fazhan Road, Daqing 163318, China. Tel.: +86 459 6504758; fax: +86 459 6503502. E-mail address: [email protected] (B.-c. Cui). Received 14 September 2010; Received in revised form 12 January 2011; Accepted 4 February 2011 0957-5820/$ – see front matter © 2011 The Institution of Chemical Engineers. Published by Elsevier B.V. All rights reserved. doi:10.1016/j.psep.2011.02.001

Process Safety and Environmental Protection 8 9 ( 2 0 1 1 ) 198–203

199

Fig. 1 – Schematic diagram of the experimental setup.

2.

Experimental

2.1.

Apparatus and procedure

The experiments were conducted in a laboratory-scale, batch reactor with a volume capacity of 650 ml designed to a maximum temperature and pressure of 450 ◦ C and 40 MPa. Fig. 1 shows a schematic diagram of the laboratory-scale, batch reactor system used in this work. All wetted parts, from the pumps to the condenser, were made of stainless steel (1Cr18Ni9Ti). The stirrer was used to keep from the formation of char resulting from thermal cracking of oily sludge at the walls of the reactor. The electric furnace was used as heater. The heating wires of electric furnace were placed below and around the reactor. The reaction temperature was monitored directly using thermocouple (inserted inside the reactor) and controlled within 1 ◦ C by a temperature controller. Before the experiment, the reactor was loaded with oily sludge and deionized water to bring the total volume of liquid to 95–130 ml. Then, nitrogen gas was used to purge the reactor for 10 min. After purging, the reactor was heated about 65 min. Upon reaching the reaction condition of reaction temperatures from 390 to 450 ◦ C and pressures from 25 MPa, reaction temperature remained stable within 1 ◦ C, then the specified amount of H2 O2 was pressured and fed into the reactor using a syringe pump within 2 s at room temperature. Hydrogen peroxide was used as a source of oxygen. The initial concentration of H2 O2 is 18148 mg l−1 and O2 excess is 427%. The reaction was conducted for a given reaction time from 1 to 10 min. After the reaction, the sample valve of reactor was opened and the effluent was cooled rapidly in a shell and tube heat exchanger and then depressurized to ambient condition. The product stream was then separated into liquid and vapor phases. The liquid products were collected in a graduated cylinder. Gaseous samples were collected with sample tubes.

2.2.

Materials and analytical methods

Hydrogen peroxide (analysis purity, 35 wt% aqueous solution) was used as a source of oxygen. The oily sludge was obtained from a crude oil storage tank in the 4th oil production plant of Daqing oil fields. The oily sludge sample contains 58.5 wt% oil, 23.2 wt% water, and 18.3 wt% inorganic particle. The oil

contains 37.3 wt% alkanes, 25.4 wt% aromatic hydrocarbons, 36.3 wt% resins, and 1.1 wt% asphaltines. Analysis of total organic carbon (TOC) in oily sludge and in water was performed respectively using a TOC analyzer (SSM5000A, Shimadzu) and a TOC analyzer (TOC-VCPH, Shimadzu). Determination of UV254 was performed by a spectrophotometry (UV-2550, Shimadzu). A pH meter (PHS-25, Shanghai Precision Scientific Instrument Co., Ltd.) was used to determine pH. A gas chromatograph (SP3420, Beijing Beifenruili Analytic Instrument Co., Ltd.) equipped with a flame ionization detector (FID) was used to analyze the acetic acid content. The column used was 3 m long with 3 mm diameter packed with 80–100 mesh GDX-103 + 2% H3 PO4 . Samples of 1 ␮l were injected into a hydrogen carrier flow (30 ml/min). The acetic acid was added as external standard for calculation of concentrations. Gas samples were analyzed using a gas chromatograph (GC-14C, Shimadzu) with a thermal conductivity detector (TCD). The carbon molecular sieve column (TDX-01) was used to separate the carbon monoxide and carbon dioxide from gaseous samples. Argon was used as the carrier gas. The system was calibrated with a standard gas mixture containing H2 , O2 , N2 , CO, CO2 and CH4 .

3.

Results and discussion

Table 1 displays the experimental conditions and results obtained from oxidation experiments. As it is shown in Table 1, that CO2 and CO account for higher percentage of total mass balance. CO2 and CO can be the only gaseous products detected without regard for trace amounts of H2 and CH4 . The yield of CO is one order of magnitude higher than that of acetic acid. It had been proved that main refractory intermediate is not acetic acid but carbon monoxide in supercritical water oxidation of oily sludge (Cui et al., 2009). The sum of yields of CO2 , CO and acetic acid is not equal to 100%. The yield of acetic acid is lowest among liquid products. It is found that carbon content in acetic acid is significantly lower than TOC in liquid product. This shows that acetic acid is not only liquid product. In Figs. 2 and 3, UV254 and pH of liquid products are shown as a function of reaction time at different temperatures. Fig. 2 indicates that stronger absorption at 254 nm appears to liquid products. UV254 decreases with reaction time. It was found by Onwudili and Williams (2006, 2007a) that PAHs such as pyrene, phenanthrene, naphthalene, fluorene and biphenylene can

200

Process Safety and Environmental Protection 8 9 ( 2 0 1 1 ) 198–203

Table 1 – The experimental conditions and results obtained from oxidation experiments. Temperature (◦ C)

Time (min)

390 390 390 390 410 410 410 410 430 430 430 430 450 450 450 450 a b

Product yield (%)a

TOC conversion (%)

1 3 5 10 1 3 5 10 1 3 5 10 1 3 5 10

56.3 60.3 61.7 63.1 62.4 67.3 69.6 72.9 66.9 71.4 75.9 80.0 75.5 78.9 82.6 88.0

CO2

CO

Acetic acid

44.9 46.4 48.5 51.5 48.1 51.5 55.3 58.6 50.6 56.2 62.1 64.9 54.2 62.9 68.9 72.7

9.0 8.4 7.2 4.6 14.0 13.0 10.1 8.0 21.5 17.5 12.7 10.5 30.4 22.7 17.4 15.1

2.18 1.90 1.39 1.16 1.67 1.58 1.21 0.83 1.26 0.96 0.69 0.42 0.49 0.42 0.37 0.12

Carbon balance (%)b

97.6 94.5 94.0 93.0 99.7 97.2 95.8 93.7 105.2 102.3 98.9 95.4 109.1 106.7 103.7 99.8

Calculated in terms of carbon. Carbon balance = [100 − TOC conversion] + CO2 yield + CO yield.

0.4 390 410 430 450

UV254

0.3

0.2

0.1

0.0

0

2

4

6

8

10

12

Reaction time (min) Fig. 2 – The relationship curves between UV254 and reaction time. form aliphatic acids, ketones and aldehydes. It is possible that petroleum constituents in oily sludge can be converted into the compounds with conjugated double bond, benzene ring and carbonyl group. Decomposition of these compounds results in a decrease of UV254 as the reaction proceeds. For alkanes in petroleum, the oxidative decomposition of hex-

8 390 410 430 450

7

pH

6

5

4

3

0

2

4

6

8

10

12

Reaction time (min)

Fig. 3 – The relationship curves between pH and reaction time.

adecane and eicosane showed a pattern of hydroxylation and further oxidation to form aliphatic ketones, aldehydes and carboxylic acids (Onwudili and Williams, 2007b). It is shown from Fig. 3 that pH declines at first rapidly in the early stages, indicating organic acids formed. Then pH climbs back gradually due to oxidative degradation of organic acids. It is here suggested that the oily sludge oxidation undergoes a parallelconsecutive reaction pathways, in which it first decomposes oxidatively to intermediates of aliphatic acids, ketones and aldehydes with conjugated double bonds and via low molecular mass organic acids and CO to the final product carbon dioxide. From the standpoint of describing the characteristics of SCWO of oily sludge, it is more reasonable to treat the intermediates as separate lumps from oily sludge. Based on the above mentioned findings, it was concluded that many intermediates except acetic acid and carbon monoxide are formed by SCWO of oily sludge. The intermediates are oxidized eventually to acetic acid, then be further oxidized to carbon dioxide and water. On the other hand, lumped model should be as simple as possible from the view of parameter estimation experiment because the complicated model needs the great bulk of laboratory work and sometimes even cannot be realized. In order to develop a lumped kinetic model capable of predicting organic carbon reduction, all organic species except acetic acid present in solution are grouped together forming a smaller number of “pseudo” species, and lumped by TOC. Acetic acid has been identified as one of the most refractory intermediates in the oxidation of many complex organics in both supercritical and subcritical water oxidation (Li et al., 1991; Holgate et al., 1995; Shanableh and Gloyna, 1991). Oxidation of acetic acid is considered as a critical, rate-limiting step in the total oxidation of organics to carbon dioxide and water due to the high stability of acetic acid. As it is shown in Table 1, the presence of CO in the gaseous effluent is always significant. CO is generated by specific pathways in the final steps of the oxidation mechanism (Brock and Savage, 1995). Portela et al. (2001) proposed CO as the main refractory intermediate. Alternatively, CO2 is considered as the end product in the oxidation process of a complex waste. On the other hand, Levec and Smith (1976) assumed the formation of interme-

201

Process Safety and Environmental Protection 8 9 ( 2 0 1 1 ) 198–203

CO, C2

0

k2

1-n n-1 -1

lnk (mol l s )

k1 CO2, C4

Feedstock, C1 k6

k5

k3

k1 k2

-1

k4

-2

k3 k4

-3

k5

-4

k6

-5 -6

Acetic acid, C3 -7

Fig. 4 – Reaction net of four lumped kinetics model.

1. 38 1. 40 1. 42 1.44

1.46 -1

1.48

1. 50 1. 52 1. 54

-1

1000 T (K )

diates of CO during the catalytic wet-air oxidation of acetic acid. In this way, a four-lump kinetic model including feedstock lump, acetic acid, CO, and CO2 is proposed here. Reaction net can be represented in Fig. 4. The reaction order assigned to the four lumps based on the oxidative abilities of different lump pseudo species. Since CO, acetic acid and CO2 are simple compounds, it was argued that a first order should be given to CO, acetic acid and CO2 . However, since feedstock lump contains a mixture of several thousands compounds of widely different properties, it was suggested that a n order should be assigned to the feedstock lump. Based on these assumptions, the following equations can be considered: (a) Oily sludge consumption rate

1.36

Fig. 5 – Arrhenius plots for SCWO of oily sludge.

The kinetic parameters to be determined are reaction order and vector k = [k1 , k2 , k3 , k4 , k5 , k6 ]. The reaction order for oily sludge in SCWO is 1.405, basing on COD (Cui et al., 2009). Therefore, the reaction order of feedstock should be between 1 and 2, namely, n = 1.1, 1.2, . . ., 2.0. Parameter identification rests on solving the nonlinear least-squares problem U by minimizing the sum of squared errors (SSE):

⎧ n m 2 ⎪ ⎨ SSE(k- , n) = i=1 j=1 (Y¯ ij − Yij ) U

dC1 = −(k1 + k2 + k3 )Cn1 dt

⎪ ⎩

(6)

k1 , k2 , k3 , k4 , k5 , k6 ≥ 0 n = 1.0, 1.1, . . . , 2.0

(1)

dC2 = k2 Cn1 + k6 C3 − k4 C2 dt

(b) CO formation rate

(c) Acetic acid formation rate

(d) CO2 formation rate

(2)

dC3 = k3 Cn1 − k5 C3 − k6 C3 (3) dt

dC4 = k1 Cn1 + k4 C2 + k5 C3 dt

(4)

where ki (i = 1, 2, . . ., 6) the reaction rate coefficient assuming an Arrhenius law:

 E  i

ki = Ai exp −

(5)

RT

where Ai was the pre-exponential factor ((mol l−1 )1−n s−1 ), Ei the activation energy (kJ mol−1 ), R the universal gas constant (8.314 J mol−1 K−1 ) and T the temperature (K).

where Yij is the jth state variable from the ith experimental run in the kinetic data set P ={Pi ; i = 1, n ; Pi ∈  n }.The bar stands for the model-predicted quantities. The reaction rate constants were estimated by a multi variable non-linear least squares, combining a Runge–Kutta integration algorithm (ode45 function) with nonlinear least-squares (lsqnonlin function) provided in optimization toolbox of MATLAB. The identification results showed that the reaction order of feedstock lump is 1.3. Arrhenius parameters (pre-exponential factor and activation energy) have been estimated by linear regression of ln(ki ) versus 1/T from the slopes and intercept of Arrhenius plots in Fig. 5. The results showed that a linear relationship of ln(ki ) versus 1/T was obtained and correlation coefficients in Table 2 are between 0.9648 and 0.9925. The kinetic parameters obtained listed in Table 2. Further, in Table 3, the parameters obtained in the present work showed significant difference from those of the others. The reason for the difference has not yet been clarified, but it may be

Table 2 – Parameters of the four lumped kinetics model. Parameters

k1 k2 k3 k4 k5 k6 a

Reaction rate constant [(mol l−1 )1−n s−1 ]a 390 ◦ C

410 ◦ C

430 ◦ C

450 ◦ C

0.0192 0.0481 0.00193 0.0103 0.00819 0.0051

0.0387 0.0701 0.00338 0.0313 0.0335 0.0079

0.0578 0.135 0.00449 0.0727 0.0768 0.0132

0.087 0.189 0.00964 0.126 0.245 0.0301

For k1 , k2 and k3 , n = 1.3, for k4 , k5 and k6 , n = 1.

Ea (kJ/mol)

98.4 94.7 101 167 219 116

Pre-exponential factor [(mol l−1 )1−n s−1 ]a 1.18 × 106 1.35 × 106 1.80 × 105 1.59 × 1011 1.72 × 1015 5.91 × 106

Correlation coefficient

0.9854 0.9848 0.9648 0.9855 0.9925 0.9679

202

Process Safety and Environmental Protection 8 9 ( 2 0 1 1 ) 198–203

Table 3 – Parameters for the SCWO of CO and acetic acid. Pre-exponential factor (s−1 )

Compounds

Ea (kJ/mol)

Reference

CO

1.59 × 10 3.16 × 108 107.25 –

167 134 120 238

This work Holgate et al. (1992) Helling and Tester (1988) Helling and Tester (1987)

Acetic acid

1.72 × 1015 19.8 × 1010 109.9 4.4 × 1011 9.3 × 1010

219 308 168 182 172.2

This work Savage and Smith (1995) Meyer et al. (1995) Krajnc and Levec (1997) Maharrey and Miller (2001)

11

0.10

1.0

0.8

0.08

0.8

0.08

0.6

0.06

0.6

0.06

0.4

0.04

0.4

0.04

0.2

0.02

0.2

0.02

0.00

0.0

1.0

0.10

2

4

6

8

C/C0

10

0

2

4

6

8

10

0.00

Reaction time (min)

Reaction time(min) 1.0

1.0

0.10

0.10

430

450 0.08

0.6

0.06

0.6

0.06

0.4

0.04

0.4

0.04

0.2

0.02

0.2

0.02

0.00

0.0

0.0

0

2

4

6

8

10

C/C0

0.8

CAcH/C0

0.08

C/C0

0.8

0

Reaction time(min)

2

4

6

8

10

CAcH/C0

0

CAcH /C0

C/C0 0.0

CAcH/C0

410

390

0.00

Reaction time (min)

TOC(Model prediction)

CO2(Model prediction)

CO(Model prediction)

AcH(Model prediction)

TOC(Experimental)

CO2(Experimental)

CO(Experimental)

AcH(Experimental)

Fig. 6 – Evolution of organic carbon observed experimentally and predicted by lumped kinetic model. 1.0

0.03

Predicted C/C0

Predicted C/C0

0.8

0.6

TOC

0.4

0.02

0.01

Acetic acid

CO 2 0.2

0.0

CO 0.0

0.2

0.4

0.6

Experimental C/C 0

0.8

1.0

0.00 0.00

0.01

0.02

0.03

Experimental C/C 0

Fig. 7 – Parity plots for experimental C/C0 and those predicted of organic carbon by lumped kinetic model.

Process Safety and Environmental Protection 8 9 ( 2 0 1 1 ) 198–203

owing to not only the difference in the reaction conditions, such as temperature range, pressure and feed concentration, but also the difference in the pure component and the mixture. In the presence of mixture, the reaction pathways of CO and acetic acid are significantly different when compared with the pure component. Some component in oily sludge probably contributed to acceleration/retardation effects in oxidation of CO and acetic acid. Once those kinetic parameters are known, the evolution of the groups included in the reaction scheme in Fig. 4 can be predicted by using follow procedure: At first, the kinetic parameters are substituted into the lumped kinetics model. The values of C1 , C2 , C3 and C4 at different temperatures have been determined by solving numerically the differential equation using Runge–Kutta integration algorithm by ode45 function in optimization toolbox of MATLAB. Fig. 6 shows those predictions made by the lumped kinetics model and experimental results at different temperatures. As can be seen, the model fits reasonably well the evolution observed for the four groups included in the simplified reaction scheme in Fig. 4. Fig. 7 shows parity plots for experimental yields and those predicted of organic carbon by lumped kinetic model. It is clear that the model provides a good correlation of the experimental data.

4.

Conclusions

A novel 4-lump kinetic model for oxidation of oily sludge in supercritical water including feedstock, acetic acid, CO, and CO2 lumps is proposed. The kinetic parameters are estimated for each involved reaction. The products yields predicted by the proposed model are consistent with experimental results. But there is much work to be done in order to check the suitability of this model for different oily sludges and various reaction conditions.

Acknowledgements The China National Petroleum Corporation is gratefully acknowledged for financial support in the form of CNPC Innovation Fund (No.05E7050). This research was also supported by the Foundation for Youth Academic Backbone from Colleges and Universities of Heilongjiang Province (No. 1153G002).

References Belkacemi, K., Larachi, F., Sayari, A., 2000. Lumped kinetics for solid-catalyzed wet oxidation: a versatile model. J. Catal. 193, 224–237.

203

Brock, E.E., Savage, P.E., 1995. Detailed chemical kinetics model for supercritical water oxidation of C1 compounds and H2 . AIChE J. 41, 1874–1888. Cui, B.C., Cui, F.Y., Jing, G.L., Xu, S.L., Huo, W.J., Liu, S.Z., 2009. Oxidation of oily sludge in supercritical water. J. Hazard. Mater. 165, 511–517. Helling, R.K., Tester, J.W., 1987. Oxidation kinetics of carbon monoxide in supercritical water. Energy Fuels 5, 417–423. Helling, R.K., Tester, J.W., 1988. Oxidation of simple compounds and mixtures in supercritical water:carbon monoxide, ammonia and ethanol. Environ. Sci. Technol. 11, 1319–1324. Holgate, H.R., Meyer, J.C., Tester, J.W., 1995. Reactors, kinetics and catalysis glucose hydrolysis and oxidation in supercritical water. AIChE J. 41, 637–648. Holgate, H.R., Webley, P.A., Tester, J.W., Helling, R.K., 1992. Carbon monoxide oxidation in supercritical water: the effects of heat transfer and the water-gas shift reaction in observed kinetics. Energy Fuels 5, 586–597. Krajnc, M., Levec, J., 1997. The role of catalyst in supercritical water oxidation of acetic acid. Appl. Catal. B: Environ. 2, 93–103. Levec, J., Smith, J.M., 1976. Oxidation of acetic acid solutions in a trickle-bed reactor. AIChE J. 22, 159–168. Li, L.X., Chen, P.S., Earnest, F.C., 1991. Generalized kinetic model for wet oxidation of organic compounds. AIChE J. 37, 1687–1691. Maharrey, S.P., Miller, D.R., 2001. Quartz capillary microreactor for studies of oxidation in supercritical water. AIChE J. 5, 1203–1211. Mahmood, T., Elliott, A., 2006. A review of secondary sludge reduction technologies for the pulp and paper industry. Water Res. 40, 2093–2112. Marks, R.E., Field, S.D., Wojtanowicz, A.K., Britenbeck, G.A., 1992. Biological treatment of petrochemical wastes for removal of hazardous polynuclear aromatic hydrocarbon constituents. Water Sci. Technol. 25, 213–220. Meyer, J.C., Marrone, P.A., Tester, J.W., 1995. Acetic acid oxidation and hydrolysis in supercritical water. AIChE J. 9, 2108–2121. Onwudili, J.A., Williams, P.T., 2006. Flameless incineration of pyrene under sub-critical and supercritical water conditions. Fuel 85, 75–83. Onwudili, J.A., Williams, P.T., 2007a. Reaction mechanisms for the decomposition of phenanthrene and naphthalene under hydrothermal conditions. J. Supercrit. Fluids 39, 399– 408. Onwudili, J.A., Williams, P.T., 2007b. Reaction mechanisms for the hydrothermal oxidation of petroleum derived aromatic and aliphatic hydrocarbons. J. Supercrit. Fluids 43, 81–91. Portela, J.R., Nebot, E., Martıˇınez de la Ossa, E., 2001. Generalized kinetic models for supercritical water oxidation of cutting oil wastes. J. Supercrit. Fluids 21, 135–145. Savage, P.E., Smith, M.A., 1995. Kinetics of acetic acid oxidation in supercritical water. Environ. Sci. Technol. 1, 216–221. Shanableh, A., Gloyna, E.F., 1991. Supercritical water oxidation-wastewaters and sludges. Water Sci. Technol. 23, 389–395.