Paraffin wax fractionation: state of the art vs. supercritical fluid fractionation

Paraffin wax fractionation: state of the art vs. supercritical fluid fractionation

J. of Supercritical Fluids 27 (2003) 39 /54 www.elsevier.com/locate/supflu Paraffin wax fractionation: state of the art vs. supercritical fluid frac...

330KB Sizes 1 Downloads 40 Views

J. of Supercritical Fluids 27 (2003) 39 /54 www.elsevier.com/locate/supflu

Paraffin wax fractionation: state of the art vs. supercritical fluid fractionation J.C. Crause, I. Nieuwoudt * Department of Chemical Engineering, University of Stellenbosch, Private Bag X1, Matieland 7602, South Africa Received 8 April 2002; received in revised form 19 July 2002; accepted 23 July 2002

Abstract Supercritical fluid fractionation of paraffin wax is compared with current state of the art processes i.e. short path distillation (SPD) and static crystallisation. Detailed cost analyses of the different processes are made and compared. From these cost analyses it appears that SPD is the cheaper fractionation process for light paraffin wax. However, supercritical fluid fractionation appears to be the more attractive option for the fractionation of medium to heavy paraffin waxes. # 2002 Elsevier Science B.V. All rights reserved. Keywords: Paraffin wax; Supercritical fluid fractionation; Short path distillation; Static crystallisation; Economic evaluation

1. Introduction Paraffin wax is present in some crude oil reserves and is also produced in the Fischer / Tropsch process. Paraffin wax is widely used in applications such as treatment of paper and cardboard, printing inks, as well as in food, cosmetic and pharmaceutical industries. Depending on the intended application of a wax, it needs to be deoiled and fractionated to ensure that the properties are tailored for the particular application. Some of the current processes used to deoil and fractionate the waxes include wax sweating [1], recrystallisation [2], warm-up [3] and spray de-

oiling [4], and short path distillation (SPD) processes [5,6]. Previously supercritical fluid fractionation of liquid paraffins [7] and high molecular weight n alkanes [5] have been investigated. In these studies the technical feasibility of supercritical fractionation of paraffins has been investigated. In this study supercritical fluid deoiling and fractionation of paraffin wax (SCFE) has been investigated and compared (from an economical perspective) with the current state-of-the-art processes, namely SPD and static crystallisation.

2. Problem description * Corresponding author. Fax: /27-21-8082059 E-mail address: [email protected] (I. Nieuwoudt).

A refined wax must have an oil content of less than 0.5 wt.%. The oil content is defined as the

0896-8446/02/$ - see front matter # 2002 Elsevier Science B.V. All rights reserved. PII: S 0 8 9 6 - 8 4 4 6 ( 0 2 ) 0 0 1 8 5 - 7

40

J.C. Crause, I. Nieuwoudt / J. of Supercritical Fluids 27 (2003) 39 /54

percentage of the wax that is soluble in methyl / ethyl /ketone (MEK) at /32 8C. Since no models exist to accurately calculate the amount of MEK solubles, in this study the n -paraffins lighter than n -C20 will be defined as oil components. The composition profile of the feed wax is shown in Fig. 1. The feed wax consists of nparaffins from n -C11 to n -C36. Although this wax composition does not correspond to any particular wax feed, the n-paraffin distribution should qualitatively correspond to a soft wax cut from a Fischer /Tropsch plant. This wax has to be refined to less than 0.5 wt.% material lighter than n -C20. From an economic perspective it is important to maximise the wax recovery. In this study a minimum recovery of 70% will be imposed as an additional constraint. Flow sheets for static crystallisation, SPD and SCFE are proposed that can achieve both the specified product purity and recovery. Detailed cash flow analyses of these plants are done to compare their profitability. (The costs associated with utilities are listed in Table 1.) The internal rate of return (IRR) and the pay pack time for each plant is calculated. These values are used to determine the relative profitability of the different plants.

3. Supercritical fluid extraction Near critical fluid processing of petroleum using technologies such as the propane deasphalting

Table 1 Cost of utilities at battery limit of process (from [23]) Utility

US$/GJa

Low pressure steam (160 8C) Medium pressure steam (184 8C) Electricity Cooling waterb Refrigeration (5 8C)

3.17 3.66 16.80 0.16 20.00

a b

Basis 1996. Based on 10 8C change in cooling water temperature.

process [8], ROSE [4,9], DEMEX [4] and Solvahl /Asvahl [5] processes is widely known and has been in commercial use for decades. Supercritical processes such as decaffeination of coffee and tea, spice and hop flavour extraction and polymer processing have also been commercialised during the last couple of decades. For the processing of paraffinic material we have considered CO2, ethane or propane as supercritical solvents. The higher density of CO2 may lead to a smaller diameter extraction column, compared to ethane at the same flow rate. CO2 is also non-flammable, whilst ethane is flammable. For this separation, ethane is preferred as the solvent due to the following reasons: . CO2 has lower solvent power than ethane at the same pressure. A higher extraction pressure is required to extract the same amount of wax with CO2, compared to ethane. . The higher density of the solvent phase results in a smaller density difference between the wax

Fig. 1. Composition of the n -paraffin wax feed used in the case studies.

J.C. Crause, I. Nieuwoudt / J. of Supercritical Fluids 27 (2003) 39 /54

and solvent phases, which can lead to premature flooding in the extraction column. . There is a possibility of phase inversion in CO2 /n -alkane systems at high pressures due to the high CO2 density [10,11]. . Ethane is normally available in a Fischer / Tropsch plant. 3.1. Modelling of supercritical extraction Phase equilibrium in the supercritical fluid extraction column is calculated using the Patel / Teja equation of state (PT EOS) [12]: P

RT a(T)  v  b v(v  b)  c(v  b)

(1)

Parameters a, b and c are evaluated as follows: R2 Tc2 a(T)aVa Pc

(2)

RTc Pc

(3)

b Vb c Vc

RTc Pc

The separator used to separate the solvent and dissolved wax in the extract is modelled using the simplified perturbed hard chain equation of state (SPHC EOS) [15]. The compressibility factor of the SPHC EOS is given by: z1czrep zattr

with c being Prigogine’s parameter [16] which represents one third of the total number of density dependent degrees of freedom. The repulsive term was given by Carnahan and Starling [17]: zrep 

4h  2h2

The packing fraction h is calculated as h

tv v+

(15)

Kim et al. [15] used a simple expression for the attractive part of the equation, based on the local composition model of Lee et al. [18]. For mixtures the attractive term is written as:

where Zm is the maximum co-ordination number, equal to 36. The mixing rules for the properties are given below: X v+  xi v+i (17) i X c xi ci (18)

The original correlations for m and zc were used for ethane: (9) (10)

The m and zc values for long chain n -alkanes have been fitted to experimental saturated vapour pressure and liquid volume data. These parameters were correlated as follows [14]: m  0:630:8695v

ŽcvY  ŽcvY  v c

(5) (6)

The Soave [13] form of the a function is used: pffiffiffiffiffi a (1m(1 Tr ))2 (8)

m 0:4524131:30982v0:295937v zc 0:3290320:076799v0:0211947v2

(14)

(1  h)3

zattr Zm

(7)

2

(13)

(4)

Vb is the smallest positive root of the following equation: V3b (23zc )V2b 3z2c Vb z3c 0

(12)

zc 0:2550:12exp(2v)

with Vc 13zc Va 3z2c 3(12zc )Vb V2b 13zc

41

(11)

i

ŽcvY  

XX i

xi xj ci v+ji

j

NA ss3ij ffiffiffi v+ij  p 2 Tij+  sij 

o ij qi ci k

si  sj

2 pffiffiffiffiffiffiffiffi o ij (1kij ) o ii o jj

(16)



  o ij qi 1 exp 2ci kT 

(19) (20) (21) (22) (23)

The values of o and s were evaluated from the

42

J.C. Crause, I. Nieuwoudt / J. of Supercritical Fluids 27 (2003) 39 /54

T *, v* and c parameters for n -alkanes by Kim et al. [15] as follows: o 62:5 K k NA ss3 pffiffiffi 8:667106 m3 =mol 2

(24) (25)

Pure component parameters reported by Kim et al. were used for ethane and CO2. Generalised parameter correlations for v * and c for n-alkanes, developed by Gasem and Robinson [19], were used for the heavy n -alkanes. A simple exponential function in terms of carbon number was used to correlate Gasem and Robinson’s T * values, since their correlation appears to be inconsistent with the values they reported: T + 215:4 81:46 exp(0:349799(CN1)0:7462 )

(26)

The interaction parameter kij have been fitted to published equilibrium data for CO2 /n -alkane systems and correlated as follows [14]: kij 0:545105 Mr 0:08463 ln(T)

(27)

The supercritical extraction column is simulated as a multistage equilibrium process. It is assumed that the extraction column operates isothermally, since the column would normally be jacketed to prevent the wax from congealing. An in-house simulator was used to solve the coupled and mass balance and equilibrium equations [14,20]. 3.2. Process flow diagram and operating conditions Using ethane as solvent, the extraction column conditions were chosen as 70 8C and approximately 100 bar. At this temperature and pressure there is an acceptable balance between the capacity of the solvent and the density difference between the wax and solvent phases. Simulations were done with 25, 30, 35 and 40 theoretical stages in the extraction column. The pressure, solvent to feed ratio (S/F) and reflux ratio for a simulation were adjusted to attain the specified wax product purity and recovery. The extraction column pressure, S/F ratio and reflux ratio were selected to ensure a minimum

density difference of at least 120 kg/m3 between the solvent and wax phases through the extraction column. This was done to prevent premature flooding caused by too small a density difference in the extraction column. Brunner [21] recommends a value of 100 kg/m3 (based mostly on CO2 data), but a 20% safety factor is included to compensate for possible errors resulting from the use of the PT EOS for phase density calculations. To minimise solvent cost all efforts are made to recycle as much of the solvent as possible. The recycled solvent contains trace amounts of the light components in the wax feed, which may have some influence on the product composition. This is taken into account in the simulations. The solubility of wax in the solvent phase decreases with a decrease in separator pressure, but decreasing the separator pressure leads to an increase in energy required to recycle the solvent. Expanding to too low a pressure will also lead to congealing of the wax. A separator pressure of about 55 bar was chosen as a compromise between energy usage and residual wax solubility in the solvent. This also ensures that the temperature after the expansion is high enough to prevent congealing of the wax. The operating conditions were selected to minimise the solvent:feed (S/F) ratio, while conforming to the product recovery and purity specifications, and the minimum density difference requirement. This was done as follows: 1) Specify S/F ratio and feed location for fixed number of stages. 2) Adjust reflux ratio and extraction pressure to obtain required product purity and recovery. 3) If the minimum density difference is larger than specified decrease the S/F ratio, else increase the S/F ratio. Repeat step 2 until minimum density difference requirement is met. 4) Adjust feed location and repeat 2/3 until minimum S/F ratio is obtained. The following trends were observed during simulation (pressure was adjusted to maintain a fixed raffinate to feed fraction):

J.C. Crause, I. Nieuwoudt / J. of Supercritical Fluids 27 (2003) 39 /54

. Increasing the reflux ratio (at constant S/F) decreases light fraction in raffinate and decrease density difference between wax and solvent phases. . Increasing S/F (at constant reflux ratio) decreases light fraction in raffinate and increase density difference between wax and solvent phases. The sets of operating conditions, determined from simulations, that satisfy the required product specifications for the different number of equilibrium stages are shown in Table 2. Several possible flow schemes exist for the solvent recycle circuit, and three alternatives have been investigated. The first flow scheme, PFD1, is shown in Fig. 2. This flow diagram can be split in two conceptual circuits: a wax circuit and a solvent recycle circuit. The solvent circuit can be configured in many different ways, but the wax circuit is kept the same for all the different flow diagrams and will be discussed only for the PFD1 flow diagram. The feed wax is pumped into the extraction column through pump P1 and heated to the extraction temperature with heat exchanger HX1. Some of the wax extract from the separator, SEP1, is returned to the extraction column as reflux through pump P2 and heated to the extraction column temperature with heat exchanger HX2. The rest of the wax extract is collected in vessel V1. The wax residue from the extraction column is collected in vessel V2. The product vessels V1 and V2 have internal heating elements to heat the wax to 70 8C to liberate dissolved solvent from the wax. These heating elements mean that the liquid (and the gas liberated from the liquid) only is

43

heated to the higher temperature. The bulk of the gas that enters will thus not be heated. It can be modelled as an adiabatic flash on the whole stream followed by an isothermal (70 8C) flash on the liquid only. The gases from the adiabatic flash and the isothermal flash are mixed to determine the final temperature of the gas leaving the vessel. It is important to conserve energy in this solvent recovery step, since it amounts to a significant portion of the energy usage. Vessels V1 and V2 are operated at atmospheric pressure. Using the SPHC EOS, the residual ethane dissolved in the wax at these conditions is calculated to be about 0.2 wt.%. Solvent gas from V1 and V2 are compressed to 55 bar with multistage compressor CMP1, cooled to 55 8C with air cooler HX3 and returned to the solvent recycle circuit. Since all the temperatures in the supercritical unit are below 100 8C, significant saving may be realised by doing heating with low value waste heat such as from condensate streams. Due to the high temperatures required this is not an option in SPD units. 3.2.1. Description of PFD1 The solvent stream from the extraction column is first expanded over an expansion valve to the separator pressure, then heated to the separator temperature by heat exchanger HX4. Due to the lowered solvent power of the solvent at the separator conditions, the wax extract precipitates out of solution in the separator vessel. The solvent from the separator vessel is condensed in heat exchanger HX5, using cooling water. The exit temperature of the solvent from HX5 is about 28 8C. This liquid is pumped to the extraction column with positive displacement pump P3 and

Table 2 Extraction column operating conditions for different number of equilibrium stages (at 70 8C) Number of stages

Feed stage

Solvent:feed ratio

Reflux ratio

Pressure (bar)

25 30 35 40

9 10 10 11

21.5 19.9 19.1 18.55

12 11 10.5 10.2

99.1 99.1 99.1 99.1

44

J.C. Crause, I. Nieuwoudt / J. of Supercritical Fluids 27 (2003) 39 /54

Fig. 2. Flow diagram PFD1, SCFE with liquid solvent recycle.

heated to the extraction temperature with heat exchanger HX6 using low pressure steam.

3.2.2. Description of PFD2 The energy demand of the PFD1 flow diagram is very high and can be decreased by heat integration. King et al. [22] discuss a flow diagram for near critical extraction where heat integration

is used. This concept is adapted to the SCFE processing of this work, as shown in Fig. 3. The solvent from the extraction column is expanded over expansion valve (EV1). The solvent stream from EV1 is heated to the separator temperature in heat exchanger HX4. The gaseous solvent from the separator SEP1 is compressed to the extraction pressure with vapour compressor CMP2. The hot vapour from the compressor is

Fig. 3. Flow diagram PFD2, SCFE with solvent expansion over valve and vapour compression.

J.C. Crause, I. Nieuwoudt / J. of Supercritical Fluids 27 (2003) 39 /54

cooled with the vapour stream from the expansion valve in HX4. The temperature of the compressed solvent stream from HX4 is cooled further to the extraction temperature with heat exchanger HX5 using cooling water. 3.2.3. Description of PFD3 It is possible not only to do heat integration, but also to integrate the work done by the compressor. In the third flow diagram, the expansion valve is replaced with an expander to recover mechanical work from the solvent expansion step, which is used to drive the compressor. This flow diagram is shown in Fig. 4. The pressure of the solvent stream from the extraction column is reduced to the separator pressure in two steps over expanders EX1 and EX2. This two-step pressure reduction with heating is used to prevent isentropic expansion temperatures below the congealing temperature of the extracted wax. The effect of different outlet pressures on the flow sheet calculations is shown in Table 3. The power requirement for the plant in flow sheet PFD3 increases as the outlet pressure of EX1 decreases. There is, however, an upper limit to the

45

outlet pressure of EX1 where the outlet temperature of EX2 enters the two-phase region. To protect the turbines and to prevent possible congealing of the extracted wax, the outlet pressure of EX1 is chosen as 90 bar. The solvent stream pressure is reduced to 55.5 bar and the temperature to 42.5 8C through expander EX2. The outlet from expander EX2 is heated to the separator temperature with heat exchanger HX5. Compressor CMP2 compresses the solvent vapour to the extraction column pressure. The hot outlet from CMP2 is cooled by heat exchanger HX6 using the separator outlet stream. The solvent vapour from HX6 is cooled down to 73 8C with heat exchanger HX7 using cooling water. The solvent stream from HX7 is used to heat the solvent stream going to the separator in HX5, and then goes to CMP2. The hot stream from CMP2 is cooled to the extraction column temperature in heat exchangers HX6 and HX4, by exchanging heat with colder streams. 3.3. Economic evaluation of SCFE The conceptual design of the SCFE unit was based on processing 100 000 t/a of wax. The costs

Fig. 4. Flow diagram PFD3, SCFE with solvent expansion through expanders, and vapour compression.

J.C. Crause, I. Nieuwoudt / J. of Supercritical Fluids 27 (2003) 39 /54

46

Table 3 Effect of EX1 outlet pressure on flow sheet EX1 Pout (bar)

EX2 outlet T (8C)

SEP1 T (8C)

HX7 outlet (8C)

Energy requireda kW

92 91 90 89

37.1b 42.2 42.5 42.9

54.8 54.6 55.3 56

72.6 73.3 74 74.7

670 692 697 702

a b

Difference between CMP1 power consumption and EX1/EX2 power output. Outlet stream in two-phase region.

associated with utility consumption are calculated using the utilities costs listed in Table 1. The cost of make-up solvent is calculated assuming an average residual solvent level of 0.2 wt.% as calculated using the SPHC EOS. The free-onboard cost of ethane is quoted as approximately CAN$95/m3 [23]. Since the synthetic wax will always be within a Fischer /Tropsch complex where ethane will definitely be available as cracker feedstock, the price of ethane is taken as US$50/ ton. The capital cost estimations were done with the CapCost capital cost estimation program [24]. The CapCost values are given in 1996 US$. The chemical engineering journal’s plant cost index (CEPCI) is used to adjust costs to present values (CEPCI(1996) /386, CEPCI(2001) /396). Details of the equipment design can be found elsewhere [14]. The results of the cash flow analyses are shown in Table 4. In general the capital cost of a plant based on the PFD2 flow diagram is cheapest, closely followed by the cost of a plant based on PFD1. Plants based on PFD3 flow diagrams are about 10 /15% more expensive. Increasing the number of stages in the extraction column from 35 to 40 stages increases the capital cost by approximately 1%. The reduced solvent/feed ratio causes this small increase in capital cost required for the separation. Based on the IRR and payback period (PBP) values, the following plants appear to be the more economical alternatives: PFD2-30 and PFD3-30. It is interesting to note that the energy cost per ton of feed is about the same for PFD1 and PFD2, even though PFD2 is designed with heat integration. This is caused by the fact that most of the

energy supplied in PFD1 is in the form of steam, while only electricity is supplied to PFD2. Even though the energy requirements for PFD2 is lower than for PFD1, the higher cost of electricity (compared with steam) negates the cost effectiveness of heat integration. The capital cost of PFD2 is however less than PFD1. It is clear that plants based on both flow sheets PFD2 and PFD3 result in significant economic improvements compared with plants based on flow sheet PFD1. The number of theoretical separation stages leading to the best economic performance for flow sheets PFD2 and PFD3 are 30. Deciding between flow sheet PFD2 or PFD3 for a design will depend on several factors: . Available capital: If capital is in short supply, flow sheets based on PFD2 must be considered, since the required capital is lowest for this flow sheet. . Complexity: A plant based on PFD3 contains more equipment, will require more advanced control strategies, and may possibly be more expensive to maintain. . Energy expenditure: Plants based on flow sheet PFD3 will consume 2.5 times less energy than plants based on flow sheet PFD2. Lower energy consumption should also be seen as a reduced environmental impact since high energy consumption simply means that the environmental impact is shifted to another location.

4. Short path distillation SPD is commonly used to separate or recover low volatility or heat labile components from

J.C. Crause, I. Nieuwoudt / J. of Supercritical Fluids 27 (2003) 39 /54

47

Table 4 Cash flow analyses for different SCFE plants Number of stages

Capital cost (US$ million)

IRRa (%)

PBPb (years)

Energy cost US$/ton feed

PFD1 25 30 35 40

13.78 13.73 14.13 14.26

19.8 20.2 19.9 19.8

4.40 4.34 4.38 4.39

25.19 23.64 22.86 22.32

PFD2 25 30 35 40

13.57 13.49 13.71 13.99

21.7 22.0 21.8 21.5

4.15 4.11 4.13 4.17

18.46 17.41 16.88 16.52

PFD3 25 30 35 40

15.52 15.34 15.50 15.77

21.6 21.9 21.7 21.4

4.15 4.11 4.13 4.17

7.25 6.80 6.58 6.43

a b

Internal (or interest) rate of return. Payback period.

liquid mixtures. It is typically used to fractionate waxes since an SPD unit can operate at very low pressures (0.1 /10 Pa)*/much lower than is possible in standard vacuum distillation units. It is necessary to operate at these low pressures to prevent high temperatures that might lead to thermal degradation of the wax. 4.1. Simulation of SPD plant It was shown that a short path wax distillation unit can be accurately approximated as a single stage flash [14,25]. This approximation was used in the in-house simulator that was used to calculate the product compositions from wax SPD units. The operating pressure is kept constant at 40 Pa, and the operating temperature is adjusted to obtain the desired (bottoms) wax recovery. Lower pressures will lead to larger SPD units and a significant increase in the cost of the vacuum pumps. The wax product of a single stage SPD unit is shown in Fig. 5. From this figure it can be seen that the total fraction of the wax product below nC20 amounts to 5.9 wt.% of the total composition */well above the specified value of 0.5 wt.%. It is clear that one SPD unit cannot

produce a product that meets the required purity and recovery. Several SPD units in cascade are needed to achieve the required wax purity and recovery. This cascade of separation stages is simulated using a stage-by-stage flash algorithm with a constant molar vapour flow between stages. At each stage the required temperature was calculated to produce the specified vapour from the feed to the stage, with the total pressure kept constant at 40 Pa. This procedure is repeated until the calculated stage temperatures have converged. 4.2. Cost estimation of SPD plant The capital cost of the simulated SPD plants is based on an existing SPD plant [26]. The installed cost (IC) of the simulated SPD plants are calculated as follows:  0:7   C  (V =F ) Nunits IC ICref (28) Cref  (V =F )ref Nunits;ref C is the capacity of the simulated plant, and Nunits is the number of SPD units in the plant. IC is the installed cost of a plant. Subscript ref refers to the reference SPD units (Cref /30 000 ton/a, Nunits,ref /5, ICref /US$3/107, (V /F )ref /0.75,

48

J.C. Crause, I. Nieuwoudt / J. of Supercritical Fluids 27 (2003) 39 /54

Fig. 5. Simulated product composition for a single stage SPD separation. Pressure/40 Pa, temperature/137.5 8C.

basis /1998 [26]). The size of a SPD unit is determined by the required evaporation rate. This is accounted for by the V /F term; with V being the average vapour flow per SPD unit and F the feed rate to the plant. The heating duty of the SPD plant consists of two contributions: energy required for evaporating wax and energy required for heating up wax feed into a unit. The energy used to evaporate wax per stage is calculated as the vapour flow rate from that stage, multiplied by the heat of vaporisation. An average value of 370 kJ/kg wax is used for the heat of vaporisation. It is further assumed that the distillate of an SPD unit is subcooled by 40 8C to ensure good separation efficiency and high capacity, since high condenser temperatures will lead to re-evaporation of the distillate. When the distillate from a SPD unit flows to the next SPD unit in the cascade, this distillate must then be heated to the operating temperature again. An average Cp value of 2.4 kJ/kg K for the liquid wax is used. The SPD units are heated with recirulating heating oil, which in turn is heated using steam or electricity. The highest process side temperature is about 156 8C, therefore medium pressure steam (184 8C) is used. The cooling duty is assumed to be the same as the heating duty.

4.3. Economic evaluation of SPD plants To determine the number of SPD units, feed stage and reflux ratio that would lead to the most economical plant design, the following strategy was used: 1) A fixed number of SPD units are selected. 2) A feed stage is selected. 3) The reflux ratio is adjusted to match the required purity and recovery specifications. 4) Repeat steps 2 and 3 until the optimum feed stage is located for the chosen number of SPD units. 5) Repeat steps 1/4 until the optimum total number of units is determined. The optimum plant design is chosen as the design leading to the highest IRR and the shortest PBP. The results for different number of SPD units are shown in Table 5. From Table 5 it is clear that a 5 unit SPD plant will result in the highest return. The simulated product composition of the 5 unit SPD plant is shown in Fig. 6. It is interesting to note that the capital cost for a 5 unit SPD plant is less than a 4 unit SPD plant. When fewer SPD units are used, more reflux is required to obtain the same wax

J.C. Crause, I. Nieuwoudt / J. of Supercritical Fluids 27 (2003) 39 /54

49

Table 5 Cash flow analysis for different SPD plant designs Number of stages (feed)

Reflux ratio

Capital cost US$ million

IRRa (%)

PBPb (years)

Energy cost US$/ton feed

4 (2) 5 (2) 6 (1)

1.4 0.23 0

17.1 13.4 14.1

23.8 25.4 20.0

3.9 3.7 4.3

5.84 3.74 3.75

a b

Internal (or interest) rate of return. Payback period.

product. More reflux implies a higher evaporation rate per unit, which increases the required surface area of an SPD unit.

5. Static crystallisation In an effort to replace solvent based wax deoiling processes, Sulzer Chemtech and Schuemann Sasol jointly developed a static crystallisation deoiling process [27 /29]. The advantages of wax deoiling by static crystallisation, compared to wax sweating and solvent based deoiling processes are: . Low energy consumption . High yield . No residual solvent in product

. Preferential removal of iso-paraffins and aromatics Due to these advantages of static crystallisation over processes such as solvent recrystallisation and wax sweating, static crystallisation is chosen as the preferred state-of-the-art modern crystallisation process. 5.1. Simulation of static crystallisation plant As a first approximation the separation efficiency of a static crystallisation unit will be simulated as an equilibrium stage. Although the crystalliser units are designed to operate as differential crystallisers which would result in separation efficiency slightly better than an equilibrium crystallisation, practical problems such as inclu-

Fig. 6. Simulated product composition of run SPD 5.

50

J.C. Crause, I. Nieuwoudt / J. of Supercritical Fluids 27 (2003) 39 /54

sion of residual oil in the solid wax should decrease the actual separation efficiency achieved with a unit. In this work it is assumed that the separation achieved with one crystalliser unit can be modelled as an equilibrium stage. The equilibrium ratio of component i between the liquid and solid phase can be calculated from the following equation [30]. This was implemented in the in-house simulator [14]. x KiS  i si 

     DHim T DHit T 1 1 exp   Tim Tit gLi RT RT  m  m  DCpm Ti T i 1ln i (29)  R T T gSi

where the following pure component properties m are required: Tm i , melting temperature (K); DHi , m heat of fusion (J/mol); DCpi , heat capacity change upon fusion (J/mol K). The activity coefficient of a component in the solid phase (gSi ) is calculated using the Wilson local composition model as modified by Coutinho and Stenby [31]. The liquid phase activity coefficient (gLi ) is calculated using the Flory free volume liquid activity model [32]. (Correlations for the pure component properties required in Eq. (29) are given elsewhere [14].) The composition of the wax product from a single stage crystalliser is calculated using a flash algorithm, with the temperature adjusted to obtain the desired wax fraction. The wax product of this single stage crystallisation is shown in Fig. 7. The material lighter than n -C20 amounts to 10 wt.% of the total composition, well above the specified value of 0.5 wt.%. Since a single stage crystalliser cannot achieve the product specification, a multistage static crystallisation plant will be necessary to achieve the recovery and purity specifications. This is similar to the Schuemann /Sasol deoiling plant, where five static crystallisers in series are used, together with internal reflux, to achieve both high product purity and high recovery [29]. This cascade of separation stages are simulated (Fig. 8) using a stage by stage flash algorithm with

a constant light oil fraction recovered per stage (analogous to a constant molar vapour flow in distillation). At each stage the required temperature was calculated to produce the specified light oil fraction from the feed to the stage. This procedure is repeated until the calculated stage temperatures have converged. 5.2. Cost estimation of crystalliser plant The capital costs of the simulated crystalliser plants are based on an existing static wax crystalliser plant of Schuemann /Sasol, built by Sulzer Chemtech AG [27,28]. This wax deoiling plant has a capacity of 100 000 ton/year, with 5 crystalliser units in series to obtain good separation efficiency and high recovery. The IC of the simulated crystalliser plants are calculated as follows:     C 0:7 Nunits IC (30) ICref Cref Nunits;ref C is the capacity of the simulated plant, and Nunits is the number of crystalliser units in the plant. An equipment factored capital cost estimate was done for the Schuemann /Sasol reference plant (Cref /100 000 ton/a, Nunits,ref /5, ICref / US$3 /107, 1998 basis). The internal flow through a stage is affected by the amount of reflux used, which would affect the size of the crystallisers. Since details of the flowrates in the Schuemann / Sasol plant is not available, the effect of reflux on the capital cost cannot be accurately quantified. The cooling duty required is calculated as the energy needed to crystallise a certain fraction of wax per stage, while the heating duty is calculated as the energy needed to melt the crystallised wax again. An average heat of fusion of 236 kJ/kg wax was used for these calculations. The stage temperatures are between 20 and 40 8C, so low pressure steam will be used for heating. It is further assumed that the molten wax is heated to 5 8C above the operating temperature of the stage to prevent wax precipitation during transfer between stages. An average Cp value of 2.4 kJ/kg K for the liquid wax is used. The pumping cost for transferring the wax from stage to stage is assumed to be negligible.

J.C. Crause, I. Nieuwoudt / J. of Supercritical Fluids 27 (2003) 39 /54

51

Fig. 7. Simulated wax product composition for a single stage crystallisation. Dotted line indicates cut-off carbon number below which not more than 0.5 wt.% wax is desired. Simulated temperature /30.9 8C.

5.3. Economic evaluation of static crystalliser plants To determine the number of crystalliser stages, feed stage and reflux ratio that would lead to the most economical plant design, the following strategy was used: A fixed number of crystalliser stages are selected. 1) The feed stage is selected. 2) The reflux ratio is adjusted to match the required purity and recovery specifications. 3) Repeat steps 2 and 3 until the optimum feed stage is located for the chosen number of stages. 4) Repeat the preceding steps until the optimum total number of stages is determined.

Fig. 8. Schematic diagram for modelling of crystallisers in series.

The optimum plant design is chosen as the design leading to the highest IRR and the shortest PBP. The results for different number of crystalliser stages are shown below in Table 6. From Table 6 it is clear that a static crystalliser plant with 5 crystalliser units will result in the highest return. This is also the minimum number of stages that can obtain the desired product specifications. If the number of crystallisation

J.C. Crause, I. Nieuwoudt / J. of Supercritical Fluids 27 (2003) 39 /54

52

Table 6 Cash flow analysis for different crystalliser plant designs Number of stages (feed)

Reflux ratio

Capital cost US$ million

IRRa (%)

PBPb (years)

Energy cost US$ /ton feed

5 (3) 6 (3) 7 (4)

4.2 1.5 1.1

22.0 26.4 30.8

15 12.4 10.1

5.2 6 7

7.79 5.71 5.43

a b

Internal (or interest) rate of return. Payback period.

units is increased, the required reflux ratio drops, but the total energy usage increases. This increase in energy usage is due to the fact that the total energy usage is proportional to the product of the energy used per stage and the total number of stages. It must be noted that the liquid loading for the plant with five stages is double that of the plant with six stages. It should be borne in mind that the effect of reflux (internal liquid loading) on the cost of the crystalliser units is not accurately reflected in the present analysis, since not enough detail on the internal liquid loadings of the reference plant is known. The simulated composition of the wax product of the five stage crystalliser plant is shown in Fig. 9.

6. Discussion An economic comparison for the more economical plants based on wax recrystallisation, SPD and SCFE is shown in Table 7. Wax crystallisation is an expensive technology with the highest capital cost. However, when the aim is the preferential removal of iso-paraffins or aromatics it is by far the best technology */ yielding a low oil content and a good wax recovery. It is only when deoiling entails the removal of low molecular weight n-paraffins that SPD and SCFE can compete. Even though SPD appears to have a better economic potential, it must be seen in context. SPD should be limited to the processing of short

Fig. 9. Simulated product composition of five stage static crystalliser plant.

J.C. Crause, I. Nieuwoudt / J. of Supercritical Fluids 27 (2003) 39 /54

53

Table 7 Economic comparison of wax recrystallisation, SPD and SCFE plants Type of plant

No. units or stages

Capital cost (US$ million)

IRR %

Energy cost (US$/ton wax)

Wax cryst. SPD SPD (electr)a SCFE, PFD1 SCFE, PDF2 SCFE, PFD3

5 5 5 30 30 30

22 13.4 13.4 13.8 13.5 15.4

15 25.4 22.4 20.2 22 21.9

7.8 3.8 16.5 23.6 17.4 6.8

a

SPD unit with electrical heating.

chain length n -paraffin waxes only. The distillation temperature rises steeply as the wax chain length increases, e.g. the boiling point at 20 Pa of n -C20 and n-C40 are 125 and 250 8C, respectively. At temperatures higher than about 200 8C electricity or an auxiliary furnace has to be used to heat the heat transfer fluid. This is expensive compared to steam */electricity is about four times as expensive as steam (Table 1). When the wax distillate’s molecular weight becomes too high, normal cooling water will cause congealing of the distillate. To prevent this a tempered cooling water system should be used, which is an added expense. It is thus not only the stability of the wax compounds but also the economics that points towards the use of supercritical fluid processing. It is believed that the processing of waxes with a significant portion of the feed heavier than n-C45 is best done with SCFE rather than with SPD. An SCFE wax processing plant can be operated just above the melting point of the waxes (which should be several degrees centigrade above the critical temperature of the solvent). This means that an SCFE wax processing plant will typically be operated between 60 and 130 8C. It is thus possible to use the lowest pressure steam in the factory. It is also conceivable that waste heat, possibly in the form of steam condensate, could be used to supply some of the heating duty. This will dramatically reduce the operating costs which may mean that the flowsheets with the lower capital costs may be selected. It also means that there will be less environmental impact at the source where fossil fuel is burnt to generate steam, electricity or heat. With an increase in the molecular weight of the feed, the temperature does not need to be

raised. The solubility can be regulated by manipulating the extraction pressure. Considering capital and operating costs and taking a holistic view of environmental impact, it can be concluded that for waxes with a significant amount of material heavier than n -C45 the most fractionation method with the lowest environmental impact and which is the most profitable is supercritical fluid extraction.

References [1] R.E. Irwin, A. Aufhauser, Methods and apparatus for wax de-oiling. Patent Number US3926776, 1975. [2] T.H. West, Modified deoiling-dewaxing process. Patent Number US4541917, 1985. [3] Y. Katayama, Producing method of fractionated wax products having different molecular weights from solid wax. Patent Number US5223122, 1993. [4] A. Sequeira, Lubricant Base Oil and Wax Processing, Marcel Decker Inc, New York, 1994. [5] I. Nieuwoudt, The fractionation of high molecular weight alkane mixtures with supercritical fluids. PhD dissertation, University of Stellenbosch, Stellenbosch 1994. [6] Pope, Pope wiped-film molecular stills and evaporators. Available from http://www.popeinc.com/wiped.htm 2000. [7] A.B. de Haan, Supercritical fluid extraction of liquid hydrocarbon mixtures. PhD dissertation, Faculty of Chemical Technology and Materials Science, Delft University of Technology, Delft 1991. [8] R.E. Wilson, P.C. Keith, R.E. Haylett, Liquid propane: use in dewaxing, deasphalting and refining heavy oils, Industrial Engineering Chemistry 28 (1936) 1065 /1078. [9] S.R. Nelson, R.G. Roodman, ROSE: the energy efficient bottom of the barrel alternative, Chemical Engineering Progress 81 (1985) 63 /68. [10] T. Charoensombut-Amon, R.J. Martin, R. Kobyashi, Application of a generalized multiproperty apparatus to measure phase equilibrium and vapor phase densities of

54

[11] [12]

[13]

[14]

[15]

[16] [17]

[18]

[19]

[20]

[21]

J.C. Crause, I. Nieuwoudt / J. of Supercritical Fluids 27 (2003) 39 /54 supercritical carbon dioxide in n -hexadecane systems up to 26 MPa, Fluid Phase Equilibria 31 (1990) 89 /104. M. du Rand, High pressure fluid phase equilibria. Masters thesis, University of Stellenbosch, Stellenbosch 2000. N.C. Patel, A.S. Teja, A new cubic equation of state for fluids and fluid mixtures, Chemical Engineering Science 37 (1982) 463 /473. G. Soave, Equilibrium constants from a modified RedlichKwong equation of state, Chemical Engineering Science 27 (1972) 1197 /1203. J.C. Crause, Supercritical fluid extraction of paraffin wax. PhD dissertation, University of Stellenbosch, Stellenbosch 2001. C.-H. Kim, P. Vimalchand, M.D. Donohue, S.I. Sandler, Local composition model for chainlike molecules: a new simplified version of the perturbed hard chain theory, American Chemical Engineering Journal 32 (1986) 1726 / 1734. I. Prigogine, Molecular theory of solutions. North-Holland Publishing Company, Amsterdam 1957. N.F. Carnahan, K.E. Starling, Equation of state for nonattracting rigid spheres, Journal of Chemistry Physics 51 (1969) 635. K.H. Lee, M. Lombardo, S.I. Sandler, The generalized van der Waals partition function. II. Application to square well fluids, Fluid Phase Equilibria 21 (1985) 177. K.A.M. Gasem, R.L. Robinson, Evaluation of the simplified perturbed hard chain theory (SPHCT) for predictions of phase behaviour of n -paraffins and mixtures of n paraffins with ethane, Fluid Phase Equilibria 58 (1990) 13 /33. J.C. Crause, I. Nieuwoudt, The fractionation of paraffin wax mixtures, Industrial and Engineering Chemistry Research 39 (12) (2000) 4871 /4876. G. Brunner, Industrial process development of countercurrent multistage gas extraction (SFE) processes, Journal of Supercritical Fluids 13 (1998) 283 /301.

[22] M.B. King, O.J. Catchpole, T.R. Bott, Energy and economic assessment of near-critical extraction processes, in Extraction 90, IChemE. Symposium Series 119, 1990. [23] B. Ashton, A. Ashton, J. Jenkins, R. Mann, Table 1, Ashton Jenkins Mann, Price Forecast 2001 04, Base Case Forecast Effective April 1, 2001., available from http:// www.ajma.net/pricetable.htm 2001. [24] R. Turton, R.C. Bailie, W.B. Whiting, J.A. Shaeiwitz, Analysis, Synthesis, and Design of Chemical Processes, Prentice Hall, Upper Saddle River, NJ, 1998. [25] I. Nieuwoudt, C. Crause, M. Du Rand, Oligomer fractionation with supercritical fluids, Journal of Supercritical Fluids 24 (2002) 47 /55. [26] UIC */Personal communication at Achema 2000, Frankfurt, Germany. [27] H. Engstler, M. Stepanski, B.J. Jans, F. Lippuner, G. Hildebrand, F. Richter, N. Matzat, Process for obtaining paraffin or paraffin fractions. Patent Number US6074548. [28] B. Jans, M. Stepanski, Solvent-free Deoiling of Paraffin., in Sulzer Tecnical Review. 1999, Sulzer Chemtech AG: Buchs. [29] G. Meyer, G. Hildebrand, Replacement of solvent deoiling of slack waxes by an environmentally friendly process (in German), Erdo¨l Erdgas Kohle 116 (2000) 192 /195. [30] J.M. Smith, H.C. Van Ness, Chemical Engineering Thermodynamics, sixth ed., McGraw-Hill, 2001. [31] J.A.P. Coutinho, E.H. Stenby, Predictive local composition models for solid/liquid equilibrium in n -alkane systems: Wilson equation for multicomponent systems, Industrial and Engineering Chemistry Research 35 (1996) 918 /925. [32] H.S. Elbro, A. Fredenslund, P. Rasmussen, A new simple equation for the prediction of solvent activities in polymer solutions, Macromolecules 23 (1990) 4707 /4714.