1176
Ind. Eng. Chem. Res. 1997, 36, 1176-1180
Heat Removal from Gas-Phase Polyethylene Reactors in the Supercondensed Mode Yan Jiang, Kim B. McAuley,* and James C. C. Hsu Department of Chemical Engineering, Queen’s University, Kingston, Ontario, Canada K7L 3N6
For gas-phase fluidized bed polyethylene reactors, heat removal is one of the controlling factors that limits the production rate. Since the invention of condensed mode operation, several process improvements for enhancing heat transfer have been discovered. In the current article, a nonequilibrium model for multicomponent condensation in vertical heat exchangers described by Jiang et al. has been used to simulate supercondensed mode operation and to test other new designs for heat removal from polyethylene reactors. It appears that a larger heat exchanger or a significant change in the inlet conditions of cooling water is not required for supercondensed operation, in which substantially higher amounts of heat are removed and substantially higher production rates are achieved. It is found that operating at higher reactor temperatures can lead to a 1.1%/°C increase in both heat removal and production rates. Decreasing the operating pressure of the heat exchanger inhibits heat transfer. Makeup ethylene should be added after the heat exchanger, and makeup hexene should be added before the exchanger to maximize heat removal. Introduction The discovery of the gas-phase fluidized bed polymerization process in the late 1970s lead to a dramatic reduction in capital investment for polyethylene plants and to new possibilities for polymer properties. Since that time, industrial research has continued toward further process improvements, including increases in production rates. The production rate in commercial gas-phase polyethylene reactors is restricted by the maximum rate at which heat can be removed from the reactor. The preferred method for removing the heat of polymerization is through removing the unreacted gas from the reactor and cooling it by passing it through an external heat exchanger, before recycling it to the bed at a temperature lower than the desired polymerization temperature (see Figure 1). Obviously, the rate of heat removal can be increased by reducing the temperature of the gas recycled to the reactor. However, it had long been assumed (Jenkins et al., 1986) that the temperature of the recycle gas could not be lowered any further than its dew point, because the introduction of liquid into a gas-phase fluidized bed reactor would inevitably result in plugging of the recycle lines or gas distributor plate or accumulation of liquid at the bottom of the reactor, which would interfere with continuous operation or result in complete reactor shutdown. Contrary to this widely held belief, Jenkins et al. (1985, 1986) discovered that the recycle gas can be cooled below its dew point and the resulting liquid/vapor mixture returned to the reactor without plugging or other problems. The introduction of this “condensed mode” operation can lead to substantial increases in production rate without enlarging the reactor. In condensed mode operation, the extent to which the recycle gas can be cooled is limited by the temperature of the industrial cooling water available on site. This constraint can be alleviated by adding condensable material to raise the dew point of the recycle steam and * Author to whom correspondence should be sent. E-mail:
[email protected]. Fax: 613-545-6637. Telephone: 613-545-2768. S0888-5885(96)00485-X CCC: $14.00
Figure 1. Industrial gas-phase polyethylene reactor system scheme.
encourage greater levels of condensation, as suggested by Jenkins et al. (1986). To satisfy the requirement that the gas-to-liquid ratio in the recycle stream be maintained at a level sufficient to keep the liquid phase of the two-phase flow entrained in the vapor, Jenkins et al. (1986) recommended that the level of liquid in the recycle stream be in the range between 2 and 12 wt % and not exceed 20 wt %. Recently, DeChellis and Griffin (1994) demonstrated that gas-phase polyethylene reactors can be safely operated at even greater levels of condensation, thereby leading to even greater increases in the production rates. However, if too much condensable inert is added to the reaction gas, an undesirable fluidization regime can be established in the reactor. DeChellis and Griffin (1994) recommend using pressure measurements to monitor the fluidized bulk density in the bed to ensure that the condensation rates are kept low enough to prevent this problem. In this improved process, the level of liquid in the recycle stream can be greater than © 1997 American Chemical Society
Ind. Eng. Chem. Res., Vol. 36, No. 4, 1997 1177 Table 1. Model Equations differential material balance for each species of the mixture dVi
dLi i ) 1, 2, ..., n dA molar flux (i.e., multicomponent mass-transfer) equations dA
) -Ni ) -
(N) ) [kV][Ξ](yV - yI) + Nt(yV)
(1)
(2)
liquid phase assumed completely mixed with regard to composition xiI ) xiL ) Li
/∑
Lj
(3)
assumed equilibrium between the liquid and vapor phases at the interface yiI ) KixiI
i ) 1, 2, ..., n
(4)
energy balance for the vapor phase
Figure 2. Separator introduced into the gas-phase polyethylene reactor system to split the recycle stream into a liquid stream and a vapor stream.
dTV ) -qV dA energy balance at the liquid/vapor interface mVcPV
ho(TI - Tc) ) qV +
20% by weight and the production rates are typically 85% greater than in the traditional condensed mode. The amount of liquid that can be condensed in the recycle stream and, consequently, the amount of heat that can be removed also depend on the concentrations of certain components in the recycle gas, such as hydrogen, nitrogen, and light comonomers, which are difficult to condense and can impede heat transfer (Jiang et al., 1996b). Gas-phase polymerization processes are now being operated with new metallocene catalysts that require significantly lower levels of hydrogen and comonomer in the gas phase than traditional ZieglerNatta catalysts to produce a polyethylene product with specified properties. When these new catalysts are used and heavy inert components are added to augment heat removal in the external exchanger, condensation levels of up to 50%, most preferably between 25 and 40 wt %, can be obtained (DeChellis et al., 1995). Operation at such high condensation levels has been called the “supercondensed mode” (Morgan-Grampian, 1995). Because the polymer particles produced using metallocene catalysts can have higher sticking temperatures than the polyethylene produced using ZieglerNatta catalysts, gas-phase reactors can be safely operated at slightly higher temperatures with metallocenes without risking particle agglomeration. As shown in Figure 1, an increase in the reactor temperature will naturally lead to a higher gas inlet temperature to the heat exchanger and to a greater driving force for heat transfer to the cooling water. This greater driving force should lead to greater heat-removal rates from the reactor system for a given cooling water inlet temperature and flow rate. An additional improvement that has been suggested for condensed mode operations is to divide the recycle stream into two or more separate streams (Jenkins et al., 1986; DeChellis and Griffin, 1994). For example, the recycle stream can be divided into a liquid and a gas stream which can then be separately introduced into the reactor (see Figure 2). A more recent patent (BP, 1994) fully focuses on this issue and gives a detailed description about how and where the separating and reintroducing could be done. By separating the liquid from the gas and then feeding the liquid directly to the fluidized bed at several locations, the total amount of liquid which can be reintroduced into the fluidized bed
∑N
cPiV(TV - TI) +
L i
∑N
(5)
L i
∆hi
(6)
energy balance at the wall hL(TI - TW) ) hWc(TW - Tc)
(7)
energy balance for the coolant mccPc
dTc ) -hWc(TW - Tc) dAouts
(8)
might be increased without introducing fluidization problems. Splitting the liquid and vapor stream also gives some flexibility in the location of the compressor which is used to pressurize the recycle gas before it is fed to the bottom of the reactor. The compressor could either be placed before the heat exchanger as in Figure 1 or on the vapor line following the heat exchanger as in Figure 2 if the liquid and vapor are separated. These two placements of the compressor result in different pressures in the heat exchanger and may result in different heat-removal rates. There is also considerable flexibility in where the makeup monomer and comonomer can be added to the system. It could be added to the recycle stream before the exchanger or after the exchanger to the liquid or gas feed stream. From the above overview, it can be seen that many improvements and new designs have been introduced for condensed mode operation of gas-phase polyethylene reactors. The objective of this article is to use the model developed by Jiang et al. (1996a,b) to simulate these proposed new designs and to determine which factors lead to greater improvements in heat-removal and, hence, production rates. For example, can a heat exchanger used for traditional condensed mode operation remove enough heat to operate at significantly higher rates in the supercondensed mode, how does the potential temperature increase in the reactor affect the amount of heat that can be removed, how does the operating pressure in the heat exchanger affect the heat-removal rate, and where should the makeup monomer be added to the system? Model Equations The complete model equations are summarized in Table 1. A detailed description of the model development can be found in Jiang et al. (1996a). Two equations in Table 1, eqs 3 and 4, are model assumptions.
1178 Ind. Eng. Chem. Res., Vol. 36, No. 4, 1997 Table 2. Comparison of Simulation Results for a Traditional Condensed Mode Operation (Run 1) with or without the Impurities without impurities base case
pat.a
TinV, °C pressure, kPa g ethylene, mol % ethane, mol % iC5, mol % H2, mol % N2, mol % hexene, mol % hexane, mol % total mol fract, mol % Tdew, °C % condensed ToutV, °C prod rate, tons/h heat removed/tube, MJ/h Tinc, °C Toutc, °C water flow rate, kg/s recycle gas rate, tons/h
78.9 2110 32.5
simul
pat.a
simul
78.9 2110 32.5 0.65 1.0 11.8 48.6 4.3 1.15 100
1.0 11.8 48.6 4.3 98.2 64.3 11.6 40.4 26.1
with impurities
57.20 6.32 40.43
64.34 9.98 40.42
74.891 36.13 51.31 550.7 1100
90.922 36.21 52.80 550.7 1100
a Data in bold are from Table 1, run C-1, in DeChellis et al. (1995).
The model equations are completed by the following additional two assumptions: (i) The pressure is assumed to be constant throughout the heat exchanger. (ii) The bulk temperature in the liquid phase is approximated by the arithmetic mean of the interfacial temperature and the tube wall temperature: TL ) (TI + TW)/2. Numerical methods for solving the model equations are reported by Jiang et al. (1996a). Simulation of the Supercondensed Mode Operation DeChellis et al. (1995) employed several examples to provide a better understanding of the supercondensed
mode operation and its benefits. We have used the nonequilibrium model developed by Jiang et al. (1996a) to simulate several of the examples from DeChellis et al.’s patent (1995). Because the information about the heat-exchanger unit is not provided, a typical vertical heat exchanger (Jiang et al., 1996b) employed in a traditional condensed mode operation is used to see whether it is adequate for providing cooling under supercondensed mode operation. Run 1 in Table 2 was chosen as a base case, because it presents traditional condensed mode operation and is actual plant data. As described by DeChellis et al., the patent data of runs 2 and 3 of Table 3 were prepared by extrapolating information from actual operations using thermodynamic equations, well-known in the art, to project actual conditions and production rates in the supercondensed mode. For run 1, from the computation results shown in Table 2, it can be seen that the calculation results of the dew-point temperature are quite different between the patent data (64.3 °C) and this simulation (57.2 °C calculated by normalizing the mole fraction of the components), because the total mole fraction of the recycle gas mixture is only 98.2%. Impurities which are not shown in the patent were probably included in dewpoint calculations by DeChellis et al. (1995). Two possible impurities which are present in gas-phase polyethylene reactors but are not listed in the patent information are ethane and hexane. In the third column of Table 2 (and in Table 3), appropriate amounts of ethane and hexane were added to the recycle stream to match the dew point with that provided by DeChellis et al. The inlet cooling temperature, Tinc, and the coolant flow rate were not provided by DeChellis et al. (1995). The values shown in Table 2 were adjusted so that the simulation results can match the outlet vaporphase temperature, ToutV, for run 1. A typical value of the recycle gas flow rate was selected based on the gas velocity, the gas density provided in the patent, and the geometry of a typical industrial reactor. Using the same coolant flow rate as in the case without adding impurities, Tinc was adjusted so that the simulation results for
Table 3. Comparison of Supercondensed Mode Operation with Traditional Condensed Mode Operation (Assuming Ethane and Hexane as Impurities To Make Up the Total Composition of the Recycle Gas Mixture to 100 mol % and To Match Up the Dew Point in the Patent) run 1, base case (trad condensed mode) pat.a V,
Tin °C pressure, kPa g ethylene, mol % ethane, mol % iC5, mol % H2, mol % N2, mol % hexene, mol % hexane, mol % Tdew, °C % condensed ToutV, °C gas density, kg/m3 prod rate, tons/h heat removed/tube, MJ/h heat-removal rate/prod rate, (MJ/tube)/ton Tinc, °C Toutc, °C water flow rate, kg/s recycle gas rate, tons/h a
78.9 2110 32.5 0.65 1.0 11.8 48.6 4.3 1.15 64.3 11.6 40.4 21.1 26.1
simul
64.34 9.98 40.42 21.21 90.922 3.484 36.21 52.80 550.7 1100
Data in bold are from Table 1, runs C-1, -3, and -6 in DeChellis et al. (1995).
run 2, removing most of H2 pat.a 78.9 2110 39.8 0.15 8.5 0.010 50.1 0.6 0.84 46.9 4.70 40.6 25.1 22.7
simul
46.90 3.73 41.02 24.33 62.356 2.747 36.21 50.28 550.7 1100
run 3, replacing H2 and N2 by iC5 pat.a 78.9 2110 39.8 0.886 21.1 0.010 37.6 0.6 0.004 73.4 30.59 40.6 30.9 59.7
simul
73.82 29.20 41.71 29.84 170.179 2.851 36.21 58.89 550.7 1100
Ind. Eng. Chem. Res., Vol. 36, No. 4, 1997 1179 Table 4. Effects of Operating Conditions on the Cooling Capacity base case (traditional condensed mode) pat.a V,
Tii °C pressure, kPa g ethylene, mol % ethane, mol % iC5, mol % H2, mol % N2, mol % hexene, mol % hexane, mol % % condensed ToutV, °C Tdew, °C gas density, kg/m3 heat removed/tube, MJ/h Tinc, °C Toutc, °C a
78.9 2110 32.5 0.65 1.0 11.8 48.6 4.3 1.15 11.6 40.4 64.3 21.1
simul
9.98 40.42 64.34 21.21 90.922 36.21 52.80
increase and decrease in TinV by
increase and decrease in pressure by
-1 °C
+1 °C
-5%
+5%
77.9
79.9
78.9 2004.5
78.9 2215.5
10.03 40.29 64.34 21.28 89.942 36.21 52.52
9.95 40.52 64.34 21.14 91.926 36.21 53.08
9.58 40.50 63.06
10.35 40.33 65.56 22.30 92.434 36.21 52.92
89.179 36.21 52.65
adding 30 tons/h C2 at 20 °C before/after heat exchanger
adding 0.9 tons/h C6 at 20 °C before/after heat exchanger
before
after
before
after
77.70 2110 34.320 0.632 0.973 11.482 47.290 4.184 1.119 9.45 40.53 63.40 21.30 89.636 36.21 52.68
78.9 2110 32.5 0.65 1.0 11.8 48.6 4.3 1.15 9.98 40.42 63.34 21.21 91.732 36.21 52.80
78.86 2110 32.491 0.650 1.000 11.797 48.586 4.326 1.150 10.07 40.40 64.48 21.22 91.278 36.21 52.82
78.9 2110 32.5 0.65 1.0 11.8 48.6 4.3 1.15 9.98 40.42 64.34 21.21 91.145 36.21 52.80
All data in bold are from Table 1 in Dechellis et al. (1995).
the base case with impurities can match ToutV as well. It can be seen from Table 2 that after adding ethane and hexane into the recycle stream, a better match with the amount of condensate reported in the patent was obtained. Using almost the same conditions of the inlet cooling water, a significantly higher heat-removal rate was obtained because of the higher dew-point temperature of the recycle vapor and the resulting larger amount of liquid condensed. The level of liquid condensed obtained by this simulation, 9.98 wt %, is still less than the value reported in the patent. Most likely, the value in the patent was obtained by a flash calculation assuming that the vapor and liquid are in equilibrium at the exit conditions of the exchanger, whereas the simulated values reported in Table 2 were determined using the more rigorous nonequilibrium technique described by Jiang et al. (1996a). Using the new metallocene system described by DeChellis et al. (1995), significantly less hydrogen and comonomer are required in the reactor to produce polyethylene with a specified melt index and density or, alternatively, with a specified weight average molecular weight and comonomer content. For run 2, in Table 3, it can be seen that by replacing most of the hydrogen and hexene (which decreased from 11.8% and 4.3% to 0.01% and 0.6%, respectively) by isopentane and ethylene, the dew point of the recycle stream is decreased. Hence, the heat-removal rate is reduced, as is the production rate, compared with run 1. If some of the noncondensable nitrogen is replaced by condensable isopentane (run 3), i.e., the system is operated in supercondensed mode, much more liquid is condensed in the recycle stream, and substantially higher heat-removal rates are obtained. If we look at the ratio of the simulated heat-removal rate per tube to the reported production rate for runs 1-3, it can be seen that these three values are of similar magnitude for the same recycle gas flow rate under the same cooling conditions (the same inlet cooling water temperature and the same water flow rate). If it is possible to remove enough heat to operate the reactor system at 22.7 tons/h using the conditions in run 2, then it will most certainly be possible to operate at 59.7 tons/h using the recycle gas composition suggested in run 3. It appears that a larger exchanger or access to colder cooling water is not required for supercondensed operation.
Effects of Operating Conditions on Heat-Removal Rates As mentioned in the Introduction, both DeChellis et al. (1995) and the BP patent (1994) suggested some new process operating conditions and designs that could affect the ability of the external heat exchanger to provide cooling for the reactor system. For example, higher reactor and heat-exchanger inlet temperatures can be tolerated using new metallocene catalysts. These higher temperatures could enhance cooling and heat removal. Also, as shown in Figures 1 and 2, whether the compressor is placed before the heat exchanger or following the heat exchanger on the vapor line will affect the actual operating pressure in the exchanger and will result in different heat-removal rates. Where the makeup monomers are added to the system may also result in different heat-removal rates. How these new designs affect the heat-removal rate from the reactor will be investigated in this section. Run 1 in Table 2 is still used as a base case for examining these potential improvements. As shown in Table 4, for a 1 °C disturbance in the reactor temperature with the cooling water inlet conditions held constant, the exchanger will automatically remove 1.1% more heat if the reactor temperature is increased or 1.1% less heat if the reactor temperature is decreased. As a result, for every 1 °C increase in reactor operating temperature that becomes possible due to higher polymer sticking temperatures, the production rate can be increased by approximately 1.1%. In Table 4, it can also be seen that the cooling capability of the recycle stream increases as the pressure is raised and decreases as the pressure is reduced. Compared to the base case, there will be a 1.7% increase in the heat-removal rate if the pressure increases by 5% and a 1.9% decrease in the heat-removal rate if the pressure decreases by 5%. This result shows that it is better to compress the recycle vapor before rather than after the exchanger in order to maximize the heatremoval and production rates. It can be seen that adding 30 tons/h of ethylene at 20 °C after the heat exchanger (normal situation) is better than adding the same ethylene before the heat exchanger (which reduces TinV to 77.7 °C), because introducing more ethylene reduces the dew point of the recycle stream. There is about a 2.3% decrease in heat removal obtained by
1180 Ind. Eng. Chem. Res., Vol. 36, No. 4, 1997
adding the makeup ethylene monomer before the heat exchanger compared to adding the makeup monomer after the exchanger. From the final column of Table 4, it appears, however, that is is better to add makeup hexene comonomer before the heat exchanger to enhance condensation and heat removal. Conclusions Using the nonequilibrium model for multicomponent condensation described by Jiang et al. (1996), the supercondensed mode operation introduced by DeChellis et al. (1995) has been simulated. The effects of changing the operating conditions on the cooling capacity of the recycle gas for polyethylene reactor systems has been investigated. The following conclusions can be drawn: With little change in the cooling water temperature or flow rates, a typical vertical heat exchanger used in traditional condensed mode operation can be employed in supercondensed mode operation, thereby removing the extra heat generated at the resulting higher production rates. The possible production and heat-removal benefits obtainable from operating with metallocene catalysts at higher reactor temperatures without sticking are approximately 1.1%/°C. Decreasing the operating pressure of the heat exchanger inhibits heat transfer. If reactor systems are designed with the compressor after the heat exchanger, a reduction of approximately 1.9% in heat removal and production rates will result from each 5% reduction in exchanger pressure. Makeup ethylene should be added after the heat exchanger and makeup hexene should be added before the heat exchanger to maximize heat removal. Acknowledgment We thank Queen’s University and the Natural Science and Engineering Research Council of Canada for support of this research. Nomenclature A ) interfacial area, m2 Aouts ) surface area of the tube outside, m2 cP ) molar heat capacity, J kmol-1 K-1 h ) heat-transfer coefficient, W m-2 K-1 hWc ) heat-transfer coefficient accounting for the heattransfer resistances in the tube wall and in the coolant, W m-2 K-1 ∆hi ) latent heat of vaporization of component i, J/kmol [kV] ) matrix of zero-flux mass-transfer coefficients, m s-1 Ki ) vapor/liquid equilibrium constants
Li ) molar flow rate of component i in the liquid phase, kmol s-1 mV ) total molar flow rate of the vapor mixture, kmol s-1 mc ) coolant mass flow rate, kg s-1 Ni ) molar flux of component i, kmol m-2 s-1 P ) operating pressure in the exchanger, kPa qV ) conductive heat flux out of the bulk vapor phase, W m-2 T ) temperature, K Vi ) molar flow rate of component i in the vapor, kmol s-1 x ) mole fraction in the liquid phase y ) mole fraction in the vapor phase [Ξ] ) matrix of high flux correction factors Subscripts and Superscripts c ) coolant i, j ) component indexes or section numbers I ) interface in ) inlet condition L ) liquid phase o ) referring to overall parameter out ) outlet condition V ) vapor phase W ) tube wall
Literature Cited BP. Polymerization Process. BP Patent, International Patent Classification 5, C08F 2/34, 10/00, B01J 8/24, Dec 8, 1994. DeChellis, M. L.; Griffin, J. R. Process for polymerizing monomers in fluidized beds. U.S. Patent 5,352,749, Oct 4, 1994. DeChellis, M. L.; Griffin, J. R.; Muhle, M. E. Process for polymerizing monomers in fluidized beds. U.S. Patent 5,405,922, April 11, 1995. Jenkins, J. M., III; Jones, R. L.; Jones, T. M. Fluidized bed reaction systems. U.S. Patent 4,543,399, Sept 24, 1985. Jenkins, J. M., III; Jones, R. L.; Jones, T. M.; Beret, S. Method for fluidized bed polymerization. U.S. Patent 4,588,790, May 13, 1986. Jiang, Y.; McAuley, K. B.; Hsu, C. C. Nonequilibrium Modelling of Condensed Mode Cooling of Polyethylene Reactors. AIChE J. 1997a, 43 (1), 13. Jiang, Y.; McAuley, K. B.; Hsu, C. C. Sensitivity Analysis of Model Predictions to Multicomponent Condensation Process in an Industrial Polyethylene Reactor System. AIChE J. 1997b, submitted. Morgan-Grampian. Supercondensing Trebles Output. Process Eng. 1995, 76, 23.
Received for review August 6, 1996 Revised manuscript received October 31, 1996 Accepted November 2, 1996X IE960485T
X Abstract published in Advance ACS Abstracts, February 15, 1997.