Membrane Distillation Crystallization Applied to Brine Desalination

Jan 7, 2016 - Membrane distillation crystallization (MDC) is an attractive alternative process for the separation of brines into salts and purified wa...
0 downloads 0 Views 1MB Size
Article pubs.acs.org/IECR

Membrane Distillation Crystallization Applied to Brine Desalination: Additional Design Criteria Carlos E. Pantoja,* Yuri N. Nariyoshi, and Marcelo M. Seckler Department of Chemical Engineering, Polytechnic School, University of São Paulo, Avenida Prof. Luciano Gualberto, Travessa 3, número 380, 05508-900 São Paulo, São Paulo, Brazil ABSTRACT: A hierarchical design procedure for membrane distillation crystallization processes based on a rigorous mathematical model and presented in a previous article is extended to address important aspects of this technology not yet discussed. The possibility of pore blockage due to premature crystallization inside the membrane distillation module is analyzed, and additional design criteria based on solubility curve shapes and metastable zone limit widths are established. An alternative process configuration is proposed for the treatment of dilute feeds in order to take advantage of the higher transmembrane fluxes characteristic of low-concentration solutions.

1. INTRODUCTION

2. POSSIBILITY OF PORE BLOCKAGE BY CRYSTAL DEPOSITION

Membrane distillation crystallization (MDC) is an attractive alternative process for the separation of brines into salts and purified water.1 In a previous paper, the authors presented a comprehensive mathematical model and a hierarchical procedure to be used as tools in the design of MDC processes.2 However, important aspects of the MDC process need further consideration for the achievement of an effective design. One aspect concerns the possibility of pore blockage due to crystal deposition on the microporous membrane surface, which is regarded as the main factor to cause permeability deterioration in MDC applications.3 The simultaneous heat and mass transport phenomena that take place in membrane distillation processes give rise to concentration and temperature polarization effects that result in higher concentrations and lower temperatures at the retentate−membrane interfacial region, and considering the fact that in MDC processes the retentate inside the membrane distillation module is usually at saturated (and even supersaturated) conditions, the possibility of crystal nucleation and growth on the membrane surface is increased.4,5 Another important aspect, which ultimately is also related to pore blockage, concerns the behavior of the MDC process with salts that present different solubility curve shapes and metastable zone widths because brines originating from industrial wastewater treatment systems may present several saline species in their composition.6,7 The last aspect addressed in this paper regards the treatment of diluted solutions, where the MDC flowsheet originally proposed by Curcio, Criscuoli, and Drioli8 was modified in order to accommodate a preconcentration step that takes advantage of the higher transmembrane fluxes characteristic of low-concentration solutions (due to higher water activity and consequently lower vapor pressure depletion). Hence, the aim of this work is to present additional design criteria, developed with the aid of the mathematical model previously demonstrated,2 to allow consideration of the above-mentioned aspects in the design of MDC processes. © 2016 American Chemical Society

Figure 1 depicts a typical process flow diagram for an MDC process as proposed by Curcio, Criscuoli, and Drioli.8 In a possible operating mode, brine to be treated is fed to the retentate circulation loop where it is mixed with the saturated mother liquor, leaves the crystallizer and is heated in a heat exchanger to the desired temperature (normally between 70 and 90 °C) before entering the tube-side of a direct-contact membrane distillation (DCMD) module, where water evaporation takes place. Solvent removal and cooling of the retentate stream inside the module, due to solvent vaporization itself and to sensible heat exchange with the cold distillate stream flowing in the shell-side, causes the retentate to increase its concentration and eventually reach supersaturation at the module exit. The thus supersaturated stream is fed to the crystallizer where supersaturation is released by crystallization. The distillate stream in turn gets warmer as it flows through the shell side and is then cooled in the distillate circulation loop. Because of the temperature and concentration polarization effects mentioned earlier, concentration at the retentate− membrane interface region inside the membrane distillation module is higher than the bulk concentration, whereas temperature at that region is lower than the bulk temperature. This causes the supersaturation ratio at the retentate−membrane interface region to be significantly higher than the bulk supersaturation. Because the interface is situated near the membrane surface, polarization potentially stimulates scaling upon the membrane. To quantify the tendency to crystallize, it is convenient to use the concept of metastable limit. For a given solute−solvent system at a given temperature, crystallization starts only if the solution concentration exceeds the solubility xSAT by a given Received: Revised: Accepted: Published: 1004

October 11, 2015 December 27, 2015 January 7, 2016 January 7, 2016 DOI: 10.1021/acs.iecr.5b03807 Ind. Eng. Chem. Res. 2016, 55, 1004−1012

Article

Industrial & Engineering Chemistry Research

Figure 1. Membrane distillation crystallization process flowsheet.

maximum allowable value Δxmax, i.e., if the solution concentration xMSZL assumes the value x MSZL = x SAT + Δxmax

Table 1. Maximum Allowable Subcooling for Aqueous Saline Solutions at 25 °C Measured in the Presence of Crystals under Low Cooling Rate (∼5 °C/h) and Mild Agitation9

(1)

where the superscript MSZL stands for metastable zone limit. Similarly, for a given solution concentration, there is a maximum allowable undercooling ΔTmax. Because values of ΔTmax are available for a number of solute−solvent systems, it is convenient to obtain Δxmax from the following expression:9 Δxmax =

dx SAT ΔTmax dT

salt

maximum allowable subcooling (°C)

sodium chloride (NaCl) potassium nitrate (KNO3) potassium sulfate (K2SO4)

1.0 0.4 6.0

In the analysis discussed in the next section, only clean solutions of the above-mentioned salts are considered. No sparingly soluble contaminants such as CaSO4 or CaCO3 are considered in the feed solutions. 2.1. Salts with a Flat Solubility Curve. An example of the monitoring of the concentration profile along the membrane module considering the retentate−membrane interface region as described above is illustrated in Figure 2, regarding a DCMD

(2)

The derivative in the equation above is the slope of the solubility curve. Δxmax is expressed in terms of mass fraction (kg of solute per kg of solution). Using eqs 1 and 2, the metastable limit xMSZL may be derived from the solubility curve and the allowable undercooling for a given system. A mathematical model presented previously2 already considers such polarization effects by means of the temperature polarization coefficient (TPC) and the concentration polarization coefficient (CPC), allowing the estimation of local temperature and concentration at the retentate−membrane interface along the module length. The model has been extended to calculate the metastable limit. The ability to predict the possibility of premature nucleation in the interfacial region inside a DCMD module−introduced in the mathematical model as described above−is useful in the design of MDC processes, and such analysis is explored to salts with different behaviors regarding the shape of the solubility curve and the metastable zone width. For that matter, three salts were selected: (a) sodium chloride, which presents a flat temperature dependency of solubility (dx/dT ranging from 0.0001 to 0.0005 °C−1 between 0 and 100 °C) and a narrow metastable zone width (1 °C of maximum undercooling, Table 1); (b) potassium nitrate, which presents a large variation of solubility with temperature (dx/dT ranging from 0.003 to 0.007 °C−1 between 0 and 100 °C) and a narrow metastable zone width (narrower than that of NaCl, 0.4 °C of maximum undercooling; Table 1), and (c) potassium sulfate, which has a large metastable zone width (6 °C of maximum undercooling, Table 1) but a mildly temperature-dependent solubility (dx/dT ranging from 0.0008 to 0.0017 °C−1 between 0 and 100 °C).

Figure 2. Concentration profiles along a DCMD module in a MDC process applied to water recovery of a 5% NaCl brine. Retentate and distillate inlet temperatures 70 and 30 °C, respectively. Operation without additional cooling in the crystallizer. Polypropylene membrane with 100 μm thickness and 0.2 μm pore diameter.

module in a MDC process designed according to the hierarchical procedure presented earlier2 applied to the water recovery of a 1005

DOI: 10.1021/acs.iecr.5b03807 Ind. Eng. Chem. Res. 2016, 55, 1004−1012

Article

Industrial & Engineering Chemistry Research

temperature kept at 30 °C, i.e., 25 °C below the solution exit temperature from the DCMD module of 55 °C, which caused the bulk concentration to remain below saturation along the entire membrane module. It is noteworthy, however, that because of this strategy saturation in the interfacial region is achieved only at the very end of the module length but at the cost of a 160% increase in the overall energy consumption of the MDC process when compared to the “isothermal” operation depicted in Figure 2 (no cooling in the crystallizer). Finally, it must be noted in Figures 2 and 3 that the retentate concentrations at the module inlet are slightly different because the supersaturation ratio at the module exit is different in each scenario (design variable). Nonetheless, the resulting average transmembrane flux for both isothermal and cooling modes were very close in each case (19.9 kcal/(h·m2) in the isothermal mode and 20.1 kcal/(h·m2) in the cooling mode); hence, the required membrane area in both cases were practically the same. The resulting retentate concentrations inside the membrane module correspond to very similar water activities, and because the fluid dynamic conditions were the same in both cases, the resulting fluxes were very close to each other. 2.2. Other Salts. Figure 4 illustrates KNO3 and K2SO4 solubility curves in comparison to that of NaCl, and Table 1 shows the metastable zone widths of those salts (representing the maximum allowable subcooling of their aqueous solutions). The mathematical model and hierarchical design procedure developed by the authors and presented in a previous paper2 were applied to the KNO3 and K2SO4 brines, with the same operating conditions and membrane module characteristics used previously for the NaCl brine so that the simulation results could be compared to the NaCl results discussed in the previous section. Physical properties and Pitzer parameters used in the KNO3 and K2SO4 simulations are shown in Table 2, whereas operating conditions, membrane module characteristics, and design variables used in the simulations are shown in Table 3. Simulation results for the KNO3 and K2SO4 brines are shown in Table 4, where the results for the NaCl brine discussed in section 2 are also displayed for comparison purposes. Also, Table 4 results consider the MDC process in isothermal mode (no cooling in the crystallizer) where a low supersaturation was imposed at the DCMD module exit in each case (retentate concentration above saturation but below the MSZL). The local transmembrane flux as a function of the dimensionless module length L/dT obtained in each case is illustrated in Figure 5, where it can be observed that the flux decrease rate along the membrane module was practically identical for all cases. This result was expected because the same operating and fluid dynamic conditions have been employed. However, the absolute fluxes for the KNO3 and K2SO4 brines were slightly higher than the flux obtained for the NaCl brine, and these results may be attributed to the following reasons: (i) differences in water activity in the retentate side for each case, because it is a characteristic of the dissolved ionic species in each solution; (ii) differences in the physical properties of each solution, which resulted in different Prandtl numbers and consequently different heat transfer coefficients for the retentate side, producing a different TPC for each brine; and (iii) each brine producing retentates with different CPCs because this is dependent on physical properties such as salt diffusivity and solution density. It is also noteworthy that each brine resulted in a different module length (and consequently that the retentate exit temperature was also different in each case) with this result being due to the different shapes of the solubility curve of each

5% NaCl containing brine. The graph in Figure 2 was generated by the model and shows that the retentate bulk concentration reaches saturation at a point close to the module exit, whereas the desired supersaturation ratio (design variable) is reached exactly at the module exit. (In this case no cooling in the crystallizer was considered; hence, supersaturation at the DCMD module exit is required in order for crystallization to occur.) It is noticeable in the MDC process that the retentate is already concentrated (and close to saturation) at the module inlet because saturated mother liquor is recirculated from the crystallizer and mixed with the brine feed. It is noteworthy that in the case of the NaCl brine although the bulk concentration reaches supersaturation only at the module exit (as designed) the concentration at the retentate− membrane interface is above the MSZL along the whole length of the DCMD module even in a well-designed process where the minimization of polarization effects was addressed which indicate a high possibility of nucleation in that region (where even primary homogeneous nucleation is likely to occur). The above-described behavior for the concentration at the retentate−membrane interface can be attributed to the almost invariable solubility with respect to temperature (“flat” solubility curve) and to the very narrow metastable zone width of sodium chloride aqueous solutions. Some authors recommend keeping the saline solution under subsaturated conditions inside the membrane module,1 in a modus operandi where supersaturation is achieved only inside the crystallizer by application of cooling. For salts with a flat solubility−temperature curve such as that exhibited by sodium chloride, such recommendation is not effective for two reasons: (a) This recommendation refers to the bulk conditions, and the analysis of Figure 2 indicates that even when the bulk concentration is below saturation the interface region is already supersaturated. (b) To ensure subsaturated conditions even in the interfacial region, the crystallizer temperature must be kept at such low temperatures (in relation to the retentate exit temperature from the DCMD module) that the process’ overall energy consumption increases dramatically. Figure 3 shows the concentration profile for the same case shown before in Figure 2, but this time with the crystallizer

Figure 3. Concentration profiles along a DCMD module in a MDC process applied to water recovery of a 5% NaCl brine. Retentate and distillate inlet temperatures 70 and 30 °C, respectively. Crystallizer temperature 30 °C. Polypropylene membrane with 100 μm thickness and 0.2 μm pore diameter. 1006

DOI: 10.1021/acs.iecr.5b03807 Ind. Eng. Chem. Res. 2016, 55, 1004−1012

Article

Industrial & Engineering Chemistry Research

Figure 4. Solubility of NaCl, KNO3, and K2SO4 in water as a function of temperature.9

Table 2. Physical Properties of Salts and Saturated Aqueous Solutions9−11 property

NaCl 1.59 × 10−9 58.44 2170 −665.70 1178 0.95 × 10−3 3404 Pitzer Parameters 0.0765 0.2664 0.00127 2.0

diffusivity in water at 25 °C (m2/s) molecular weight (g/mol) crystal density (kg/m3) enthalpy of crystallization (kJ/kg) solution density @ 70 °C (kg/m3) solution viscosity @ 70 °C (Pa·s) solution specific heat (J/(kg·°C)) β0 β1 CΦ α1

Table 3. Operating Conditions, Membrane Module Characteristics, and Design Variables Used in the Simulationsa brine feed flow rate to the MDC process (stream 1) brine feed concentration to the MDC process (stream 1) brine feed temperature to the MDC process (stream 1) retentate inlet temperature at the DCMD module (stream 3) distillate inlet temperature at the DCMD module (stream 10) retentate feed to the DCMD module Reynolds number in tubes fow velocity in tubes membrane material membrane thickness membrane pores diameter membrane porosity a

KNO3

K2SO4

2.35 × 10−9 101.10 2110 −827.61 1440 0.89 × 10−3 3266

2.38 × 10−9 174.26 2660 −504.21 1114 1.20 × 10−3 3289

−0.0816 0.0494 0.00660 2.0

0.0499 0.7793 0.0 2.0

case, and the characteristic supersaturation ratio of each solution at the module exit yielded different recirculation rates and total number of tubes. (A low supersaturation at the module exit was imposed as a design variable as mentioned earlier, but the different solubility curves and different MSZW produced different supersaturation ratios in each case.) However, the most striking difference in the design results observed for the KNO3 and K2SO4 brines, when compared to the NaCl results discussed in the previous section, concerns the concentration profiles in the retentate−membrane region. In the case of the KNO3 brine, the interfacial concentration remained below the saturation and MSZL curves along most part of the module length, intercepting both curves only at point very close to the exit, as shown in Figure 6 (a very different behavior than the one observed for the NaCl brine, where the interfacial concentration remained above the MSZL in the whole length of the DCMD module). This result can be explained by the steep inclination of the KNO3 solubility curve: The mixing of the brine feed stream with the saturated mother liquor recirculated from the crystallizer, resulting in a close-to-saturation solution as discussed previously, and its subsequent heating before entering the DCMD module caused a significant subsaturation, which even with the water removal and temperature drop inside the module kept the interfacial concentration below the MSZL in

1000 kg/h 5% (mass) 25 °C 70 °C 30 °C tubes side 8000 1.5 m/s polypropylene 100 μm 0.2 μm 0.8

Stream numbers refer to the flowsheet depicted in Figure 1.

solution as well as to their different MSZW (Table 5). Other differences in the resulting module geometries may be explained by the following facts: Because the retentate Reynolds number and flow velocity are design variables, the different physical properties of the brines yielded different tube diameters for each 1007

DOI: 10.1021/acs.iecr.5b03807 Ind. Eng. Chem. Res. 2016, 55, 1004−1012

Article

Industrial & Engineering Chemistry Research Table 4. MDC Process Design Results for NaCl, KNO3, and K2SO4 Brines result

NaCl

KNO3

K2SO4

retentate recirculation flow rate (kg/h) retentate exit temperature (°C) saturation ratio at module exit tubes internal diameter, dT (mm) number of tubes, nT tubes length, L (m) pressure drop tubes side, ΔPT (kPa) liquid entry pressure, LEP (kPa) membrane area, A (m2) average transmembrane flux, N̅ (kg/(h·m2)) water recovery (kg/h)

91350 55.0 1.002 1.95 4787 1.63 80 482 47.8 19.9 950

86550 54.7 1.005 3.31 1301 3.26 83 482 44.0 21.7 950

57800 49.5 1.031 5.55 397 6.90 73 482 47.9 19.8 950

Figure 5. Transmembrane flux as a function of the dimensionless length L/dT. Design results for NaCl, K2SO4, and KNO3 brines (design conditions according to Tables 2 and 3).

Table 5. TPC, CPC, and Water Activity at the Module Inlet (Retentate Side)a CPC TPC water activity

NaCl

KNO3

K2SO4

1.0389 0.8821 0.7398

1.0399 0.8337 0.8263

1.0759 0.7228 0.9729

Figure 6. Retentate concentration profile (KNO3 brine) along the DCMD module length for the operating conditions described in this section (isothermal operation).

a

Design results for NaCl, K2SO4, and KNO3 brines (design conditions according to Tables 2 and 3).

most part of its length, indicating that the possibility of nucleation in the interfacial region in this case is likely to occur only at a very small section close to the module exit. In the case of K2SO4, the design results show that the interfacial concentration intercepted the solubility curve at a point around the middle of the module length, but because of the very large metastable zone width of this salt, the intersection with the MSZL curve occurred only at a point close to the module exit, as shown in Figure 7. This means that supersaturation at the membrane−retentate interface region in this case occurs in the second half of the module length but always below the MSZL and is less likely to promote crystallization when compared to the NaCl case discussed above, where the MSZL is exceeded along the whole length of the membrane module. The design results presented above for the KNO3 and K2SO4 brines considered the “isothermal” operation of the MDC process, meaning that the crystallizer was operated at the same temperature at which the retentate leaves the membrane module. Similar to what was done previously for the NaCl case, further simulations were performed for the KNO3 and K2SO4 brines

Figure 7. Retentate concentration profile (K2SO4 brine) along the DCMD module length for the operating conditions described in this section (isothermal operation).

where cooling was applied in the crystallizer in order to ensure subsaturated conditions inside the membrane module not only in the liquid bulk but also in the retentate−membrane interface 1008

DOI: 10.1021/acs.iecr.5b03807 Ind. Eng. Chem. Res. 2016, 55, 1004−1012

Article

Industrial & Engineering Chemistry Research region, with supersaturation achieved only inside the crystallizer as discusses above. On the basis of the same operating conditions applied to the “isothermal cases” simulations described in Table 3, the obtained results for the “cooling mode” operation showed that in the case of the KNO3 brine if the crystallizer is kept at 53.5 °C or just 1.2 °C below the retentate exit temperature from the membrane module then subsaturated conditions in the retentate−membrane region along the whole module length is achieved as illustrated in Figure 8. Unlike what

Figure 9. Retentate concentration profile (K2SO4 brine) along the DCMD module length for the operating conditions described in this section (cooling in the crystallizer).

distillation module no significant flux decrease was perceived.12 It is noteworthy that our research group is presently conducting experiments in MDC with the objective of investigating the membrane blockage problem. Preliminary results with NaCl solutions show that significant flux decrease was observed only after a few hours of operation even with laminar flow regime in the retentate side of the DCMD module. (These results will be reported in a future paper.) Even if the isothermal mode in such cases necessarily results in a flux decrease after a certain amount of hours of operation, the existence of cleaning cycles is not unusual in membrane separation processes. Two membrane modules could be installed in parallel if necessary: one inline with the process and one offline in cleaning mode. 2.2.2. Salts with Small, Positive dx/dT and Large MSZW. For salts with a flat solubility curve but a large MSZW (maximum undercooling > 4 °C), it was shown that only part of the membrane module is likely to present the possibility of nucleation. The cooling crystallization option in this case also penalizes the energy consumption of the process; hence, the isothermal operation is recommended. The fact that only part of the module length may present pore blockage and consequent flux decline after a certain time extent allows the inversion of the retentate and distillate inlet streams of the membrane module from time to time. Such flow reversal may cause the dissolution of crystals that occasionally start to grow on the membrane surface without the need of process interruptions or additional membrane modules. With a previous knowledge of the flux decline profiles, with time, it is even possible to automate the process with a timer controller acting on automatic valves in the retentate and distillate inlet streams. Flow reversal intervals could also be based on the previous knowledge of the induction times of the supersaturated species: By rinsing the part of the module where crystallization is most likely to occur with undersaturated solution at time intervals smaller than the induction time, completion of the nucleation process and consequent crystal growth may never occur.13 2.2.3. Salts with Large, Positive dx/dT. For salts with a steep solubility curve (dx/dT > 0.003 °C−1), supersaturated conditions are located at the very end of the module length as observed in the simulation results for the isothermal operation. Because of the

Figure 8. Retentate concentration profile (KNO3 brine) along the DCMD module length for the operating conditions described in this section (cooling in the crystallizer).

was demonstrated for the NaCl brine, where the crystallizer cooling approach caused a high energy-consumption increase (a 160% increase when compared to the isothermal operation), for the KNO3 case the energy cost was small: The specific energy consumption increased from 1.30 kW·h/kg of pure water produced to 1.37 kW·h/kg, only a 5.4% increase. For the K2SO4 brine, fully subsaturated conditions inside the module were achieved when the crystallizer was maintained at 46 °C (3.5 °C lower than the retentate exit temperature from the membrane module), representing a 22.3% increase in the specific energy consumption of the MDC process (concentration profiles shown in Figure 9). On the basis of the above results, the following design criteria based on the solubility’s temperature dependence (dx/dT) and MSZW of a salt can be established. 2.2.1. Salts with Small, Positive dx/dT and Narrow MSZW. For salts with a flat solubility curve (dx/dT < 0.0005 °C1−) and narrow MSZW (maximum undercooling < 2 °C), the strategy of applying cooling crystallization to the retentate leaving the membrane module is not indicated because the energy penalty is excessive. In such cases, the isothermal mode is a better choice, even considering the possibility of nucleation inside the membrane module. Whether nucleation (which is very likely to occur) will result in crystal growth on the membrane surface, potentially blocking its pores, or if the nuclei thus formed will be dragged to the crystallizer is an open question. Several researchers have reported a transmembrane flux decrease in MDC experiments with NaCl solutions,1,5,8 but there are also works reporting that although nuclei formation have been observed inside the membrane 1009

DOI: 10.1021/acs.iecr.5b03807 Ind. Eng. Chem. Res. 2016, 55, 1004−1012

Article

Industrial & Engineering Chemistry Research

Table 6. Design Results for the Preconcentration Loopa

large dx/dT, a small decrease in the retentate temperature inside the crystallizer is enough to ensure overall subsaturation inside the module at no significant increase in energy consumption, as shown in the simulation results for the cooling operation. Hence, in such cases cooling operation is recommended.

3. FEED PRECONCENTRATION In the MDC process flowsheet illustrated in Figure 1, the feed (stream 1) is mixed with the saturated mother liquor recirculated from the crystallizer (stream 6) prior to being heated up and fed to the membrane module (stream 3). Because normally in MDC processes the recirculation stream 6 has a flow rate several times higher than that of feed stream 1, the concentration of stream 3 (membrane module feed) is always close to saturation. It was discussed during the DCMD model development2 that the presence of the dissolved ions causes the water activity in the solution to be lower than unity provoking a reduction in the solution’s vapor pressure when compared to that of pure water at the same temperature. This reduction, in turn, has a negative effect in the transmembrane flux because it decreases the vapor pressure difference between the retentate and the distillate streams that flow in each side of the microporous hydrophobic membrane, ultimately the driving force of the membrane distillation process. To take advantage of the potentially higher transmembrane fluxes of diluted feeds, a modified MDC process was conceived, where a preconcentration section is introduced. In this process, the flowsheet of which is depicted in Figure 11, the diluted solution to be treated by the MDC process is fed to the preconcentration loop instead of being fed directly to the crystallization loop. The preconcentration loop process design is similar to what was previously discussed for the original MDC process. In that case, it was shown that once the supersaturation ratio at the module exit is selected by the designer (stream 3 in Figure 1) the recirculation flow rate (stream 6 in Figure 1) is determined solely by the mass and energy balance of the process. In the case of the preconcentration loop, a similar procedure is adopted: Both temperature and concentration at the module exit are defined by the designer (design variables), with the recirculation stream flow rate being determined via mass and energy balance. The lower the exit temperature, the longer the module will be, but with a fewer number of tubes. In contrast, longer modules imply reduced average fluxes due to retentate cooling and distillate warming along the module length (with consequent loss of “driving force”). It is required for the designer, hence, to conduct a previous analysis when selecting the exit temperature, where the investment cost is weighed against the operating cost. To illustrate this process, Table 6 shows results for three different designs intended for the concentration of a 1000 kg/h stream of a 5% NaCl brine up to 10%. Table 6 indicate that modules with smaller L/dT ratios result in smaller membrane areas (because the module average flux is higher) but higher recirculation flow rates (which implies larger pumps and piping and also higher pumping energy consumption). Therefore, on the basis of design results calculated by the mathematical model, the designer can verify which configuration is the most cost-effective. Furthermore, to illustrate the effectiveness of the modified MDC process with a preconcentration loop, Table 7 presents design results for the modified process in comparison to those of the “regular” process (without preconcentration), for the same set of operating conditions. Material balance results for both designs are depicted in Figure 10 (regular process) and

result

design 1

design 2

design 3

retentate recirculation flow rate (kg/h) retentate exit temperature (°C) tubes internal diameter, dT (mm) number of tubes, nT tubes length, L (m) membrane superficial area, A (m2) L/dT ratio average flux, N̅ (kg/(h·m2)) energy consumption (kW·h/kgDEST.) water recovered (kg/h)

342000 68.5 1.95 17.950 0.10 10.8 50 46.5 1.27 500

123600 65.7 1.95 6.504 0.29 11.7 150 42.8 1.22 500

25800 48.5 1.95 1.352 2.93 24.3 1500 20.6 1.20 500

a

Feed of a 5% NaCl brine 1000 kg/h. Retentate and distillate inlet temperatures 70 and 30 °C, respectively. Retentate side Reynolds number 8000. Polypropylene membrane 100 μm thickness, 0.2 μm pores, and 0.8 porosity. Exit concentration 10%.

Table 7. Design Results for the MDC Process Considering the Regular Configuration (Without Preconcentration) and the Modified Configuration (With Preconcentration)a result

regular process

Preconcentration DCMD Module retentate recirculation flow rate (kg/h) retentate exit temperature (°C) retentate concentration at exit (% mass) tubes internal diameter, dT (mm) number of tubes, nT tubes length, L (m) membrane superficial area, A (m2) L/dT ratio average flux, N̅ (kg/(h·m2)) energy consumption (kW·h/kgDEST.) water recovered (kg/h) Crystallization DCMD Module retentate recirculation flow rate (kg/h) 90350 retentate exit temperature (°C) 55.0 retentate concentration at exit (% mass) 1.0015 tubes internal diameter, dT (mm) 1.954 number of tubes, nT 4788 tubes length, L (m) 1.63 membrane superficial area, A (m2) 47.8 L/dT ratio 832 average flux, N̅ (kg/(h·m2)) 19.9 energy consumption (kW·h/kgDEST.) 1.41 water recovered (kg/h) 950 Totals average flux, N̅ (kg/(h·m2)) 19.9 energy consumption (kW·h/kgDEST.) 1.41 membrane superficial area, A (m2) 47.8 water recovered (kg/h) 950

modified process 31700 53.2 10.0% 1.954 1714 1.79 18.8 915 26.6 1.20 500 89424 63.1 1.0015 1.954 4712 0.56 16.0 284 27.9 1.34 450 27.3 1.27 34.8 950

a

Feed of a 5% NaCl brine 1000 kg/h. Retentate and distillate inlet temperatures 70 and 30 °C, respectively. Retentate side Reynolds number 8000. Polypropylene membrane 100 μm thickness, 0.2 μm pores, and 0.8 porosity. Isothermal operation (no cooling in the crystallizer).

Figure 11 (modified process). On the basis of those results, it is possible to verify that with the process with preconcentration of the feed from 5 to 10% that it was possible to reduce the total required membrane area from 47.8 m2 (regular process) to 1010

DOI: 10.1021/acs.iecr.5b03807 Ind. Eng. Chem. Res. 2016, 55, 1004−1012

Article

Industrial & Engineering Chemistry Research

Figure 10. Mass and energy balance results. MDC process design without preconcentration of the feed. Operating conditions described in Table 7.

Figure 11. Mass and energy balance results. MDC process design with preconcentration of the feed. Operating conditions described in Table 7.

34.8 m2 (modified process, considering the areas of both modules) resulting in a 27.1% reduction. Likewise, the specific

energy consumption of the modified process was 1.27 kW·h/kg, a 10.6% reduction from that of the regular process. 1011

DOI: 10.1021/acs.iecr.5b03807 Ind. Eng. Chem. Res. 2016, 55, 1004−1012

Article

Industrial & Engineering Chemistry Research

(4) Gryta, M. Fouling in Direct Contact Membrane Distillation Process. J. Membr. Sci. 2008, 325, 383. (5) Yun, Y.; et al. Direct Contact Membrane Distillation Mechanism for High Concentration NaCl Solutions. Desalination 2006, 188, 251. (6) Li, J.; et al. Treatment of High Salinity Brines by Direct Contact Membrane Distillation: Effect of Membrane Characteristics and Salinity. Chemosphere 2015, 140, 143. (7) Gryta, M.; Tomaszewska, M.; Karakulski, K. Wastewater Treatment by Membrane Distillation. Desalination 2006, 198, 67. (8) Curcio, E.; Criscuoli, A.; Drioli, E. Membrane Crystallizers. Ind. Eng. Chem. Res. 2001, 40, 2679. (9) Mullin, J. W. Crystallization; Butterworths-Heinemann: Boston, MA, 2001. (10) Perry, R. H.; Green, D. W. Perry’s Chemical Engineers’ Handbook; McGraw-Hill: New York, 2008. (11) Zemaitis, J. F.; et al. Handbook of Aqueous Electrolyte Thermodynamics; AIChE: New York, 1986. (12) Ji, X.; et al. Membrane Distillation-Crystallization of Seawater Reverse Osmosis Brines. Sep. Purif. Technol. 2010, 71, 76. (13) Pomerantz, N.; et al. Prevention of Scaling of Reverse Osmosis Membranes by “Zeroing” the Elapsed Nucleation Time. Part I: Calcium Sulfate. Ind. Eng. Chem. Res. 2006, 45, 2008.

4. CONCLUSIONS Additional design criteria for design of membrane distillation crystallization (MDC) processes were developed in order to address important aspects of this promising technology not yet discussed in the hierarchical design procedure presented previously.2 The behavior of the MDC process with solutions with different characteristics regarding solubility dependence with temperature and metastable zone width (MSZW) was tested with the aid of the mathematical model also presented in a previous paper,2 aiming the analysis at the possibility of nucleation inside the membrane distillation module and consequent flux decrease due to pore blockage by crystals. Simulation results showed that for solutions of salts whose solubility varies greatly with temperature cooling in the crystallizer may ensure subsaturated conditions inside the module without increasing significantly the energy consumption of the process. It was also shown that for solutions of salts that present a large MSZW it is viable to minimize the possibility of pore blockage by the reversal of the inlet streams (retentate and distillate) in the module from time to time with no need for cooling in the crystallizer which in such cases would increase significantly the energy consumption of the process. In contrast, for solutions of salts that present a flat solubility curve it was demonstrated that in order to obtain subsaturated conditions inside the module (and in the retentate−membrane interface region) by cooling the effluent solution in the crystallizer the required temperature reduction is so high that it provokes an unacceptable increase in the overall energy consumption in such cases being more adequate to operate the process isothermally with no cooling at all in the crystallizer and to coexist with the possibility of pore blockage which can be minimized by a cyclic operation with two or more modules in parallel so that when a flux decline is observed the process flow can be directed to a “fresh” module and the previous one can be cleaned. For the treatment of dilute solutions by the MDC process, a modified flowsheet was proposed where a preconcentration section is included, allowing the maximization of the overall transmembrane flux by taking advantage of the higher water activity characteristic of dilute solutions. Simulations showed not only that the overall required membrane superficial area can be reduced but also that energy consumption can be decreased in such cases with the modified MDC process.



AUTHOR INFORMATION

Corresponding Author

*E-mail: [email protected]. Notes

The authors declare no competing financial interest.



ACKNOWLEDGMENTS We wish to acknowledge Brazil’s Coordination for the Improvement of Higher Education Personnel (CAPES) for the financial support.



REFERENCES

(1) Creusen, R. J. M.; et al. Brine Treatment by Membrane Distillationcrystallization (MDC) Process. Procedia Eng. 2012, 44, 1756. (2) Pantoja, C. E.; Nariyoshi, Y. N.; Seckler, M. M. Membrane Distillation Crystallization Applied to Brine Desalination: A Hierarchical Design Procedure. Ind. Eng. Chem. Res. 2015, 54, 2776. (3) Edwie, F.; Chung, T. Development of Simultaneous Membrane Distillation-Crystallization (SMDC) Technology for Treatment of Saturated Brine. Chem. Eng. Sci. 2013, 98, 160. 1012

DOI: 10.1021/acs.iecr.5b03807 Ind. Eng. Chem. Res. 2016, 55, 1004−1012