Kinetics and boundary flux optimization of integrated photocatalysis and ultrafiltration process for two-phase vegetation and olive washing wastewaters treatment

Chemical Engineering Journal 279 (2015) 387–395

Contents lists available at ScienceDirect

Chemical Engineering Journal journal homepage: www.elsevier.com/locate/cej

Kinetics and boundary flux optimization of integrated photocatalysis and ultrafiltration process for two-phase vegetation and olive washing wastewaters treatment J.M. Ochando-Pulido a,⇑, M. Stoller b a b

Chemical Engineering Department, University of Granada, 18071 Granada, Spain Department of Chemical Engineering, University of Rome ‘‘La Sapienza’’, Via Eudossiana, 18-00184 Rome, Italy

h i g h l i g h t s  Simultaneous treatment of effluents of olive mills operating with two-phase technology.  pH-T flocculation + UV/TiO2 photocatalysis + UF yields effluent apt for irrigation.  Significant and stable flux (26.5% increment) and minor fouling (28.6% reduction).  Proposed boundary model fits accurately the UF membrane performance.

a r t i c l e

i n f o

Article history: Received 5 March 2015 Received in revised form 6 May 2015 Accepted 14 May 2015 Available online 21 May 2015 Keywords: Olive mill wastewater Modeling Fouling Ultrafiltration Photocatalysis Wastewater reclamation

a b s t r a c t In many of the studies available on the treatment or fractionation by membrane technology of the effluents by-produced by olive oil factories, the problem of fouling is not correctly approached or not even addressed. In the present study, the operating framework of a spiral wound polymeric ultrafiltration (UF) membrane module was optimized by the boundary flux theory, which merges both the critical and threshold flux theories for simplification purpose and was formerly validated by the Authors. The raw wastewater, a mixture of olive washing and olive vegetation wastewaters, was pretreated by two processes developed in prior research: pH-temperature flocculation (pH-T F) and photocatalysis with lab-made ferromagnetic-core titanium dioxide nanoparticles under ultraviolet light (UV/TiO2 PC). The organic matter removal during UV/TiO2 PC fitted accurately a two-step first-order kinetic model. Also, the proposed boundary model fits the membrane experimental data with accuracy. Higher boundary flux values were confirmed for batch UF when the feedstream is further pretreated by UV/TiO2 PC (23.3–23.6% increment), and also slightly higher feed recovery and significant minor sub-boundary fouling index a. Moreover, the higher rejection of the organic pollutants (53.3%) permits achieving the standard limits to reutilize the purified effluent for irrigation purposes. Ó 2015 Elsevier B.V. All rights reserved.

1. Introduction As a sample of the popularity of ultrafiltration (UF) membranes for effluents tertiary treatments, a plethora of scientific papers have been published in international scientific journals in the last decade [1]. Currently, UF membranes have replaced many conventional separation operations in wastewater treatment processes, provided their lower specific energy consumption, minor investment and maintenance costs as well as higher efficiency. UF membranes have already been employed in the decontamination of ⇑ Corresponding author. Tel.: +34 958241581; fax: +34 958248992. E-mail address: [email protected] (J.M. Ochando-Pulido). http://dx.doi.org/10.1016/j.cej.2015.05.050 1385-8947/Ó 2015 Elsevier B.V. All rights reserved.

wastewater from very different sources, including metalworking industry [2], oil field wastewater [3], refinery wastewater [4], pulp and paper [5], textile wastewater [6], dairy effluents [7], protein production [8], olive mills wastewater [9–11], restaurant wastewater [12] and municipal sewage [13], among others. Much effort has been invested to attain novel membranes capable of offering higher technical and economical performances since the development and commercialization of the first cellulose acetate asymmetric membranes. However, there is an important lack of knowledge concerning membrane fouling still to be fulfilled. Fouling is a complex phenomenon which involves different mechanisms comprising pore blocking, plugging or constriction, cake, gel and biofilm formation as well as cake-enhanced

388

J.M. Ochando-Pulido, M. Stoller / Chemical Engineering Journal 279 (2015) 387–395

concentration polarization [14–16]. Moreover, several variables influence to different extent the type and rate of fouling formation on UF membranes, related to the membrane (layer and support chemical nature, roughness and mean porosity), the hydrodynamic conditions inside the membrane module (net operating pressure, turbulence and temperature) and also the characteristics of the feedstream (physico-chemical composition, concentration, particle size distribution, pH, ionic strength and divalent ions concentration). In the present work, UF treatment of the effluents by-produced by olive oil industries (OME), in particular olive washing (OWW) and olive vegetation wastewaters (OVW), is discussed. In most of the studies available on the treatment of these effluents by membranes, and also in others in which the fractionation of OME has been studied [17,18], the problem of fouling is not correctly approached or not even addressed. Fouling is always present in the treatment of wastewater streams by UF membranes and its control is imperative to assure the appropriate operation and design of the plant. During operation, fouling soars on one hand the energy costs to maintain the target permeate production, and on the other the operating costs associated to frequent plant shut-downs for in situ membrane cleaning procedures. Furthermore, the longevity of the membranes might be irretrievably shortened if irreversible fouling arises. In this regard, OME contain high concentrations of a wide range of solutes in the form of suspended solids and colloidal particles which are all very prone to cause membrane fouling, such as organic pollutants, as well as inorganic matter, which may also lead to deleterious scaling problems. Concerning this, the Authors have observed in previous works that an optimized membrane plant design requires a properly-tailored pretreatment process in order to avoid high fouling rates, which would rapidly lead to zero flux conditions if no pretreatment is conducted on the raw effluent upstream the UF operation [19,20]. In particular, it is key to shift the mean particle size distribution of the foulants in the feedstream away from the average pore diameter of the selected membrane to avoid constriction, blocking and plugging of the pores, which often cause irreversible fouling. On the other hand, it is essential to operate the plant under the appropriate operating framework. Bacchin et al. introduced in 1996 an important concept that has served as a very reliable tool to define the operating framework of membrane operations, the critical flux [21]. They observed for microfiltration (MF) membranes that there is a permeate flux below which fouling is not promptly attained, but above which fouling becomes critically triggered. This pattern was also confirmed in UF membranes afterwards [22,23], as well as for nanofiltration (NF) [24,25]. Lately, a new concept was introduced, the threshold flux. In this case, the concept evaluates the maximum permeate flux that can be yielded by a membrane upon a low constant fouling rate regime [26,27]. This concept is a novel practical tool for membrane process designers, more than the critical flux, especially in those cases where the presence of fouling is unavoidable even below the critical conditions, such as in wastewater treatments by membranes [16,26–32]. Subsequently, both concepts were merged by Stoller and Ochando in a recent paper for ease of application [33]. In the present manuscript, the operating framework of a spiral wound (SW) polymeric UF membrane module was optimized by means of the boundary flux theory. Membrane fouling control was addressed by this model, and also for this purpose the raw wastewater stream was pretreated by two processes developed in previous work by the Authors, that is, pH-temperature flocculation (pH-T F) and photocatalysis with lab-made ferromagnetic-core titanium dioxide nanoparticles under ultraviolet light (UV/TiO2 PC) [12,34], for which the kinetics was also studied. Finally, the parametric

quality standards to reuse the purified effluent whether for irrigation purposes or for discharge into public waterways or in municipal sewers were checked.

2. Experimental 2.1. The feedstock to the UF membrane module As a result of the production of olive oil in olive oil factories working with the modern two-phase continuous centrifugation process, two main effluents are generated. Once received in the olive mills, olives are washed prior to their entrance in the olive oil production line, leading to the generation of olive washing wastewater (OWW). Afterwards, during the vertical centrifugation of the oil, olive vegetation wastewater (OVW) is by-produced. OWW is a moderately polluted effluent, presenting high amount of suspended solids but low concentration of dissolved organic matter. Concentration values depend mainly on the water flowrate employed in the olives washing machines during the cleaning procedure of the fruit, but normally stand below the limits for discharge on suitable superficial terrains (Guadalquivir Hydrographical Confederation, 2006: total suspended solids TSS < 500 mg L1 and chemical oxygen demand COD < 1000 mg L1). On the other hand, high organic pollutants load in the form of dissolved matter is confirmed in OVW, most of them phytotoxic and recalcitrant to biological degradation [12]. Samples of OWW and OVW were taken from olive oil mills located in Jaén and Granada (Spain) operating with the two-phase olive oil extraction process. Olive oil factories operating with the two-phase extraction technology generate on average a daily amount of more than 1 m3 of OWW per ton of processed olives and 10 m3 of OVW, respectively. Thereby, in order to treat OWW and OVW simultaneously, both effluents were mixed in 1:1 (v/v) proportion – hereafter labeled as olive mixed wastewater (OMW) – to stabilize the average organic matter concentration of the stream entering the treatment system and thus avoid sensible fluctuations in the COD parameter. Subsequently, the OMW stream was subjected to two different pretreatment processes studied and thoroughly described in previous work by the Authors [12,34]. Beforehand, gridding was carried out to remove the coarse particles present in OMW (cut-size equal to 300 lm), then pH-T flocculation (pH-T F) by adding HNO3 (70% w/w) followed by photocatalysis of the supernatant under ultraviolet irradiation (UV) with commercial Degussa P-25 or lab-made ferromagnetic-core TiO2 nanoparticles (UV/TiO2 PC). The feedstream at the end of this pretreatment process will be hereafter assigned as OMW-FPC. The alternative pretreatment process comprised only gridding followed by pH-T F, therefore the feedstream exiting this pretreatment line will be referred to as OMW-F. The pH-T F experiments were performed at ambient temperature (20 ± 0.5 °C) in a 20 L stirred batch reactor equipped with a turbine impeller stirrer providing short, initial high stirring rate mixing (90 s, 1000 rpm) followed by slow stirring for a longer period of time (20 min, 320 rpm). UV/TiO2 PC was carried out in an 8 L agitated batch reactor, provided with an UV lamp on top (nominal power 45 W, wavelength 365 nm), at ambient temperature (20 ± 0.5 °C) and medium agitation speed (500 rpm), lasting 2–4 h residence time. Two different nanometric TiO2 catalysts were used for the experiments: one commercial catalyst Degussa P-25 (mean particle size 40 nm, crystal phase composition 70% anatase and 30% rutile) and laboratory-made composite photocatalytic nanoparticles with a ferromagnetic core and two subsequent layers of silica and titania. Different catalysts dosages (0.5–9 g L1 for both catalysts, plus 20 g L1 for the commercial Degussa P-25) were tested and the

389

J.M. Ochando-Pulido, M. Stoller / Chemical Engineering Journal 279 (2015) 387–395

reduction of the COD values was followed during the course of the experiments. Finally, the achieved reduction of COD, total phenols concentration and suspended solids was measured after each pilot-scale pretreatment step. 2.2. The lab-made nanocatalyst production The lab-made photocatalyst was produced in three subsequent steps, following the procedure reported elsewhere [12,34]. Firstly, magnetite was produced by a sol–gel process continuously in a spinning disk reactor, in which a NH4OH aqueous solution, injected at 2 cm of distance from the center of the disk, was added to an aqueous solution of FeCl3, HCl and Na2SO3, injected at the center. This allowed producing magnetite with a modal particle size of 30 nm. Secondly, silica coating was performed by adding the dried magnetite particles to a TEOS–ethanol–NH3 solution. The coated particles were then recovered back by magnets, gently dried at 80 °C and calcined at 450 °C. Finally, TiO2 coating was carried out by pouring the silica-coated particles into a titanium tetraisoprop oxide–ethanol solution, then adding H2O2 dropwise to the solution under strong mixing conditions. Again, the recovered particles were dried at 80 °C and final calcination at 450 °C was performed. The particle size distribution of the obtained nanocatalyst (Fe2O3/SiO2/TiO2) was measured by means of a Plus90 nanosizer (Brookhaven) and showed a final modal particle size of 79 nm [34]. 2.3. The boundary flux model The boundary flux establishes the frontier between the nul or low fouling region from the high fouling operating framework of a membrane, which means that further increasing the applied pressure does not induce higher stable fluxes [26]. Below the boundary flux conditions, no or small but constant amount of fouling triggers, but above its value fouling starts to build up very quickly, leading to exponential permeate flux loss. The performance of a specific membrane according to the boundary flux model can therefore be described by the following equations:

dm=dt ¼ a; J p ðtÞ  Jb

ð1Þ

dm=dt ¼ a  b  ðJ p ðtÞ  J b Þ; J p ðtÞ > J b

ð2Þ

where dm/dt is the membrane permeability loss in time, Jp(t) is the permeate flux at any time t, and Jb is the boundary flux value, whereas a (L h2 m2 bar1) takes into account the constant permeability reduction rate suffered by the membrane in the sub-boundary operating conditions, a = [0, 1), thus hereafter called the sub-boundary fouling rate index, and the b parameter (h1 m2 bar1) represents the fouling behavior in the exponential fouling regime of the system, and will be hereafter called super-boundary fouling index. In case of a = 0 no fouling is triggered on the membrane, hence Eqs. (1) and (2) become reduced to the critical flux equations [21,26]. On the other hand, above the boundary flux conditions, fouling behaves by exponential permeability loss rate in time, given by the contribution of the b fouling index. Above the boundary flux conditions, the exponential part of Eq. (2) will thus quickly overwhelm the linear contribution of the a parameter [21,26]. The boundary flux value can be measured following critical flux measurement methods, but needs a different approach in order to estimate the value of a and the value of b, successively. Beside experimental data, the extended method requires the use of Eqs. (1) and (2) to separate both operating regimes. In this regard, the Authors suggest the use of the pressure-stepping method, extended from the one proposed by Espinasse et al. [23]. The

method has been proven to be quick and reliable. Fouling measurements by long-term experiments, which need to stop the plant and, as a consequence, the production, are not compatible with common industrial practices. The method basically consists in cycling the applied pressure up and down, by a constant PTM variation and equal to DPTM, and to check for the reproducibility of the membrane permeability when the same pressure level is again applied after one cycle (Fig. 1). Subsequently, the method requires the integration of Eq. (1) in time, as follows:

J b ðPTM ; tÞ  J b ðP TM ; t 0 Þ ¼ DJ b ¼ 

Z

t0

a  PTM  dt

ð3Þ

t

valid not only in the case the same PTM value is used at t and t0 , but also for different PTM values between t and t. As long as the adopted PTM values remain below the boundary one Eq. (3) will not be invalidated, such that no effect on changes of the permeability loss rate should be observed. As long as Eq. (3) holds, the boundary conditions are met. The lowest PTM value at which the difference between the experimentally measured permeate flux gap DJb and the theoretical DJ⁄b value evaluated from Eq. (3) in the same period of time dt becomes positive is the boundary pressure PTMb and the boundary flux value Jb is determined by taking into account the permeate flux value at the beginning of the correspondent pressure cycle. Subsequently, the boundary flux values were correlated as a function of a key parameter KP defining the fingerprint of the feedstream to the membrane module, which depends mainly on the organic pollutants load (CODFSi) and the particle size distribution (mp) of the feedstock solution [11,25]. The proposed fitting curve is based on the general relationship between permeate flux Jp and the applied PTM, that is:

J p ðKP;tÞ ¼ mðKP;tÞ  PTM ðKPÞ

ð4Þ

where m(KP,t) is the permeability and PTM(KP) the transmembrane pressure, as a function of the chosen key parameter KP at any given time t; m and PTM can be approximated by a linear function:

mðKP;tÞ ¼ m0 ðtÞ  m1  KP

ð5Þ

PTM ðKPÞ ¼ P  R  T  KP ¼ P  p1  KP

ð6Þ

where m0 is the pure water permeability of the membrane at t = 0 (see Table 1), P is the applied operating pressure, R is a constant, T is the temperature, and p1 and m1 are fitting parameters. The pure water permeability m0(t) is a function of time, since it depends on the amount of irreversible fouling formed over the membrane [33]. Substituting Eqs. (5) and (6) in Eq. (4) at the boundary conditions defined by Eq. (1) and re-arranging, the following second order polynomial equation is obtained:

J b ðKP;tÞ ¼ m0  Pb  a  t  Pb  ðm0  p1  a  p1  t þ m1  P b Þ  KP þ m1  p1  KP2

ð7Þ

valid in the physical range of KP = [0, +1) such that lim[KP?0] Jb(KP,t) = m0(t), that is, the pure water flux conditions. The fitting curve presents two roots, the first one hereafter called KP⁄ representing the upper limit of the key solute concentration in the feed solution to trigger almost instantaneously zero flux conditions, thus the validity of the fitting curve must be restricted to the range [0, KP⁄]. Otherwise, for [t ? 1], the minimum point tends to 1, which means that the relevant Jb(KP, t) values as a function of time becomes lower and lower, that is, the membrane is less productive. There is a time point t⁄ for a given KP where zero flux conditions are met and thus the module is completely dead, due to irreversible fouling build up and/or aging of the membrane.

390

J.M. Ochando-Pulido, M. Stoller / Chemical Engineering Journal 279 (2015) 387–395

Fig. 1. Boundary flux (Jb) determination by the pressure-stepping method: experimental (DJb) vs. theoretical (DJ⁄b) permeate flux gap as function of the applied pressure (PTM).

Table 1 Nominal characteristics of the selected UF membrane. Parameters

Parametric value

Model Surface, m2 Permeability (m0), L h1 m2 bar1 Configuration Chemical structure Chemical composition MWCO, kDa Average pore size, nm Maximum pressure, bar Maximum temperature, °C pH range

GM 2.5 5.2 ± 0.5 Spiral-wound Thin film composite (TFC) Polyamide/polysulfone 8 2 16 50 1–11

2.4. The UF pilot-scale membrane plant The membranes pilot plant used for the experiments is schematically shown in Fig. 2. It consisted of a 100 L feedstock tank (FT1) where the corresponding feedstock was loaded, two pumps – centrifugal booster (P1) and volumetric piston (P2), respectively – that could be selected to drive the feedstream to the UF membrane module inserted in housing M1, two regulation valves (V1 and V2) with a precision of 0.5 bar and 10 L h1, respectively, that served to set the desired operating pressure and crossflow over the membrane independently, also measured and displayed by analogue manometers and a turbine flow meter respectively, and two plate heat exchangers (E1 and E2) to maintain the temperature of the streams stable. During all experiments, both the temperature and the feed flow rate were controlled at fixed values, that is, ambient conditions (20 ± 0.5 °C) measured by a Pt100 sensor and turbulent tangential velocity over the membrane (550 L h1, to promote Reynolds number NRe > 4000), respectively. Otherwise, the permeate flux was steadily gauged during the operation time by a precision electronic mass balance (AX-120 Cobos, 0.1 mg accuracy). The membrane chosen for the experiments was an UF module in spiral wound (SW) configuration supplied by ‘GE Water and Process Technologies’ (model GM2540F). Prior to each UF assay, the membrane was equilibrated by filtering MilliQÒ water at a constant pressure and temperature until a stable flux was observed (approximately after 2 h time), and then the pure water permeability (Kw) was measured (Table 1). The active surface of the selected UF membrane was equal to 2.5 m2, and the rest of characteristics are hereafter reported in Table 1. During the measurement of the Jb values by the pressure-stepping method (recycling mode), both the permeate

Fig. 2. Flow diagram of the UF membrane pilot plant, FT1: feedstock tank, P1: booster pump, P2: volumetric pump, V1: bypass regulation valve, V2: concentrate regulation valve, E1 and E2: plate heat exchangers, M1: membrane housing provided with SW UF membrane.

and concentrate streams were cooled down to the temperature of the feedstock to mix and recycle them back into the raw wastewater tank to maintain the characteristics of the corresponding effluent constant. The pressure stepping method consists in cycling the applied pressure (P) up and down, by a constant P variation equal to DP, checking for the reproducibility of the permeate flux at the same pressure levels before and after the pressure changes [23]. Finally, the hydrodynamic behavior of the UF membrane was examined by conducting batch-run UF experiments upon the boundary conditions formerly estimated for the differently pretreated feedstocks, focusing on the steady-state performance of the measured permeate flux during operation time to check for the reliability of the method. At the end of each experiment the membrane was rinsed with tap water for 15 min and stored in fresh tap water if no longer necessary, after which chemical cleaning of the circuit with 1 N NaOH solution was performed in closed loop for 30 min.

J.M. Ochando-Pulido, M. Stoller / Chemical Engineering Journal 279 (2015) 387–395

2.5. Analytical methods The chemical oxygen demand (COD), total phenols (TPh), total suspended solids (TSS), electroconductivity (EC) and pH were measured following standard methods [35]. In addition, a Plus90 nanosizer supplied by Brookhaven served for the particle size distribution analysis of the suspended and colloidal matter. All the analytical methods were carried out in triplicate with analytical-grade reagents: 70% (w/w) HNO3, 98% (w/w) NaOH, 98% (w/w) Na2SO3, 30% (w/w) NH4OH, 37% (w/w) HCl and 30% (w/w) FeCl3, supplied by Panreac. 3. Results and discussion 3.1. Pretreatment of the feedstock The Authors observed in previous works that an optimized control of the membrane operation requires a good design of the pretreatment processes beforehand, called pretreatment tailoring, which may increase the boundary flux Jb accordingly. Hence, a detailed knowledge of how Jb varies as a function of the feedstock characteristics and a good pretreatment tailoring are essential [26,28–33]. Prior to the UF operation, in the present work OMW was whether pretreated by gridding followed by pH-T flocculation (pH-T F), thus called OMW-F feedstock, or further pretreated by photocatalysis under ultraviolet irradiation (UV) with lab-made ferromagnetic-core TiO2 (UV/TiO2 PC) nanoparticles (OMW-FPC) [13,34]. In recent years, an enormous interest has been devoted to heterogeneous photocatalysis using oxide semiconductors owing to its potential use in environmental applications. The key advantages of UV/TiO2 PC process rely on the very high surface area to volume ratio of the TiO2 nanoparticles that lead to the absence of mass transfer limitations, the possibility of being triggered at ambient operating conditions and to exploit sunlight instead of electric illumination in case of doped TiO2, as well as the fact that TiO2 is a highly active and non-toxic molecule capable of achieving complete mineralization of a wide range of organic pollutants or oxidizing them into harmless compounds [36,37]. A comparison of the COD reduction rates in OMW during UV/TiO2 PC operation time with both the commercial Degussa P-25 vs. the lab-made ferromagnetic-core nanocatalysts are reported in Fig. 3. The results of the evolution of the organic matter (COD) removal during the UV/TiO2 PC process were described by a two-step first-order kinetic model: 0

COD ¼ COD0  expðk1  t 1 Þ þ COD00  expðk1  t 01 Þ where COD is the chemical oxygen demand at a given time t, COD0 is the initial chemical oxygen demand, t is the irradiation time and k1 is the apparent first-order kinetic constant. The summarized results of the UV/TiO2 PC are reported in Table 2. The variation of organic matter in the PC reactor in time can be divided into two sequential stages: a first one, in which the organic matter concentration decreases sharply, followed by a second stage during which the COD decreases more smoothly until the end of the residence time. The variation of the COD values vs. time, for all experiments, can be explained by two sequential reactions responsible for OMW degradation. In the first stage, the OMW decomposition occurred quickly within approximately the first 10 min, whereas the second stage of reaction took place more slowly, supported by the much higher values of the reaction rate constant k1 than 0 k1. This model was already applied to describe effluents mineralization during wet oxidation and ozonation by Martins et al. [38]

391

and Shende and Levec [39], as well as by Guedes et al. for Fenton oxidation of cork cooking wastewaters [40]. As it can be observed, the lab-made ferromagnetic-core photocatalyst provided not only slightly higher COD removal (DCODs) than the commercial one upon lower initial catalyst dosages, equal to 1 vs. 0.5 g L1, but was also kinetically advantageous since higher apparent first-order kinetic constant values (k1) were confirmed for the latter (Table 2). These facts are supported by the 100% anatase TiO2 phase of the lab-made nanopowder provided by the sol–gel process performed for the fabrication of the ferromagnetic core catalyst, which ensured the formation of homogeneous, pure and very uniform particles [13,34]. Among the three crystalline phases of TiO2, called anatase, rutile and brookite, the first is the most active under UV irradiation, also due to the particle size and habit. Mixed-phase submicron particles such as commercial Degussa P-25, consisting of 70% anatase and 30% rutile, are found to provide less effective catalytic action. Finally, the physicochemical characterization of the OMW stream exiting each pilot-scale pretreatment is in contrast with the composition of the raw OWW and OVW as well as the mixed effluent (OMW) are reported in Table 3. The main differences between OMW-F and OMW-FPC are related to the lower COD and TPh concentration observed in the latter, but also the minor number of particles (mp) in the near range of the mean pore diameter (Dp) of the selected UF membrane. The parameter mp represents the percentage of total number of interfering particles with respect to the membrane mean pore diameter (Dp). For this reason, particle size distributions were measured on the feedstocks to the UF membrane. According to the proposed pore blocking model, only particles of certain size can lead to fouling issues in short time, which are those having a size (dP) similar to that of the membrane DP, that is, blocking may be statistically significant when 0.1 < dP/DP < 10. The novel lab-made titania ferromagnetic-core photocatalyst is easily recoverable back from the wastewater stream by a magnetic trap, solving the problem of the recovery of the catalyst and thus enhancing the cost-effectiveness of the process. 3.2. Boundary flux modelization The different pretreated OMW streams, that is, OMW-F and OMW-FPC, were finally conducted to the pilot-scale UF operation. Previously, the appropriate operating conditions for the batch UF process were searched by means of the boundary flux theory formerly addressed by the Authors [33]. The results of the pressure-stepping method followed for the measurement of the boundary flux values on the UF membrane for both OMW-derived feedstocks are reported in Table 4, where the permeate flux values experimentally measured before and after one pressure cycle (DJp) are contrasted with the theoretical ones (DJ⁄p) calculated by means of Eq. (3) (see Section 3.2). As it can be noted, for each pretreated feedstock there is a different point at which the measured DJp values start to deviate from the theoretical DJ⁄p ones, which means that from that points thereon Eq. (1) does not hold anymore and thus the boundary conditions are surpassed. This occurs for a pressure (P) value equal to 11 bar (Pb OMW-F) for OMW-F, whereas for OMW-FPC the operating pressure can be further increased up to a value equal to 13 bar (Pb OMW-PC) without incurring in critical fouling build-up. Translated to permeate flux values, the Jb increases from 11.1 up to 15.1 L h1 m2, which significates a 26.5% increment. This can be more graphically understood by a guided focus on Fig. 4, where the pressure-stepping for each pretreated feedstock is reported for ease of comprehension. Once traced the boundary points (Jb  Pb) for each pretreated feedstock, which define the proper operating framework of the

392

J.M. Ochando-Pulido, M. Stoller / Chemical Engineering Journal 279 (2015) 387–395

(b)

1.1

1.1

1.0

1.0

0.9

0.9

-Ln (COD/COD0)

-Ln (COD/COD0)

(a)

0.8 0.7 0.6 0.5 0.4

0.8 0.7 0.6 0.5 0.4

0.3

0.3

0.2

0.2

0.1

0.1

0.0

0

50

100

150

200

250

0.0

300

0

50

100

150

200

250

300

t, min

t, min

Fig. 3. Organic matter concentration abatement (COD/COD0) in time during lab-scale UV/TiO2 photocatalysis on OMW. Comparison of commercial Degussa P-25:  = 3 g L1, = 9 g L1 (panel a) vs. lab-made ferromagnetic-core nanocatalyst: = 0.5 g L1, = 1 g L1 (panel b).

= 1 g L1,

Table 2 COD reduction and kinetics of UV/TiO2 photocatalysis in OMW. 0

Catalyst type

Catalyst dosage, g L1

CODfinal, g L1

DCODs, %

k1102, min1

R2

k1102, min1

R2

Degussa P-25

1 3 9

1.7 1.6 1.5

54.3 56.5 57.0

8.16 9.10 9.32

0.977 0.985 0.986

0.05 0.04 0.05

0.980 0.989 0.988

Ferromag. TiO2

0.5 1 3

1.5 1.6 1.7

58.4 55.7 52.7

9.33 8.46 7.91

0.982 0.992 0.991

0.010 0.006 0.008

0.987 0.979 0.989

Table 3 Physicochemical composition of raw OWW and OVW, as well as the 1:1 (v/v) mixture (OMW) and the effluent exiting each pilot-scale pretreatment.

pH EC, mS cm1 TSS, g L1 COD, g L1 mp, % TPh, mg L1

OWW

OVW

OMW

OMW-F

OMW-FPC

6.3–6.5 1.4–1.7 14.1–16.7 0.7–0.8 – 1.5–3.3

4.7–4.9 2.4–2.6 8.0–9.8 7.2–7.4 – 162.7–171.1

5.9–6.3 1.4–1.6 6.1–6.9 4.1–4.2 – 83.1–87.2

2.5 1.5 0.4 3.6 47.2 82.5

3.1 1.5 0.3 1.5 35.5 59.8

EC: conductivity; TSS: total suspended solids; COD: chemical oxygen demand; TPh: total phenols; mp: interfering particles.

Table 4 Pressure-stepping method for the measurement of the boundary flux values (marked in bold) on the UF membrane for both OMW-derived feedstocks. P, bar

2 3 4 5 6 7 8 9 10 11 12 13 14 15

OMW-F

OMW-FPC

DJp

DJ⁄p

DJp

DJ⁄p

0.01 0.03 0.03 0.05 0.04 0.06 0.04 0.04 0.07 0.30 0.42 0.54 – –

0.02 0.03 0.04 0.05 0.06 0.07 0.08 0.09 0.10 0.11 0.12 0.13 – –

0.01 0.02 0.02 0.03 0.04 0.04 0.03 0.03 0.03 0.06 0.06 0.40 0.60 0.73

0.01 0.02 0.03 0.03 0.04 0.05 0.05 0.06 0.06 0.07 0.08 0.08 0.09 0.10

selected UF membrane for each feedstream, the sub-boundary fouling index (a) was calculated in each particular case by integrating Eq. (3). The summarized results are given in Table 5. As it can be observed, lower a fouling index can be ensured on the UF membrane by previously subjecting the raw feedstock to the whole pretreatment process, that is, 28.6% reduction of the fouling build-up. Additionally to the higher permeate production (Jb), this would have important implications on the steady-state performance of the UF membrane module during the treatment of this highly polluted OMW stream when transferred to an industrial scale, enabling longer operation times without having to stop the plant to apply cleaning protocols, and would also safe the UF membrane from long-term and irreversible fouling. The higher Jb  Pb observed for the feedstock further pretreated by the UV/TiO2 PC process is justified not only by the lower COD but also by the lower number of particles present in the effluent (mp) exiting the photocatalysis process having a modal size (dp) in a close range to the mean pore diameter (Dp) of the UF membrane (10 < dp/Dp < 10 range) (Table 5). Once the data from the pressure-stepping experiments were gathered, they were used for the subsequent batch operation of the UF membrane. The performance of the selected UF membrane module was compared for both differently pretreated feedstocks. In case of batch membrane processes, it is advisable to lower the operating conditions in order to have enough safe operating space to not trigger fouling at high recovery values. For this reason, a safety margin (dp) for the operating pressure equal to 5–10% below the boundary pressure point (Pb) should be chosen to operate the UF membrane plant, justified on the basis that the feedstock becomes increasingly concentrated during the batch-run membrane operation. The results of the permeate flux evolution experimentally measured during the batch UF operation upon the boundary conditions

393

J.M. Ochando-Pulido, M. Stoller / Chemical Engineering Journal 279 (2015) 387–395

(a) 1.0

16 14 12

PTM, bar

ΔJp, L h-1m-2

0.8

0.6

Jb

0.4

10

Pb

8 6 4

0.2

2 0.0

0

100

200

300

0

400

0

100

t, min

200

300

400

t, min

(b) 1.0

16 14 12

PTM, bar

ΔJp, L h-1m-2

0.8

0.6

Jb 0.4

Pb

10 8 6 4

0.2

2 0.0

0

100

200

300

400

500

0

0

t, min

100

200

300

400

t, min

Fig. 4. Boundary flux (Jb) determination (left figures) by the pressure-stepping method (right figures) for OMW-F (a panels) and OMW-FPC (b panels): experimental DJb (d) vs. theoretical -DJ⁄b (o). Corresponding Jb  Pb values highlighted with arrows.

Table 5 Boundary flux values measured by the pressure-stepping method on the UF membrane. Membrane feedstock ⁄

1

KP , mg L J⁄b, L h1 m2 rs (J⁄b), % P⁄,b bar pf, bar P⁄TMb, bar a, L h2 m2 bar1

OMW-F

OMW-FPC

3600 11.1 0.16 11 0.9 10.1 0.007

1500 15.1 0.11 13 0.6 12.4 0.005

mp: percentage of interfering particles, with a mean diameter dp in the range of the membrane average pore size Dp (10 < dp/Dp < 10); Pb: measured boundary pressure; Jb: measured boundary permeate flux; pf: osmotic pressure of the corresponding feedstock; CODf: chemical oxygen demand of the corresponding feedstream; rs: standard deviation (%).

are reported in Fig. 5. The behavior of the permeate flux during operation time was also modelized with the proposed second order boundary flux polynomial equation, that is, Eq. (7). For both feedstreams, OMW-F and OMW-FPC, rapid achievement of a steady-state performance of the permeate flux is attained by adopting the boundary operating conditions formerly estimated. After the set-up of the initial concentration polarization and fouling due to the rapid accumulation of solutes within the membrane surface in the first moments of operation, the permeate flux reaches a plateau for both feedstocks (Fig. 5).

Furthermore, the proposed boundary model fits the experimental data accurately, such that the boundary flux values (Jb) estimated by the model correspond to those experimentally observed (Jss) (Table 6). Finally, the summarized results of the batch UF operation with both feedstocks are reported in Table 6, where the achieved feed recovery factor (Y,%) and the physico-chemical analyses of the final permeate streams – comprising pH, EC, TSS, COD and TPh – are given. For both feedstreams, total rejection of the suspended solids (TSS) is achieved, but it is important to highlight that higher rejection of the organic matter (RCOD) is ensured for the feedstream previously subjected to the whole pretreatment process including UV/TiO2 PC, that is, 53.3% for OMW-FPC in contrast with 42.0% for OMW-F. This substantial difference permits the stream exiting the UF of OMW-FPC to achieve the standard limits for reuse the purified effluent for irrigation purposes (COD values below 1000 mg L1), also ensuring slightly higher feed recovery (90.6%). 4. Conclusions In the present study, the operating framework of a spiral wound (SW) polymeric UF membrane module was optimized by means of the boundary flux theory, formerly validated by the Authors in previous work, for the simultaneous treatment of the two main effluents by-produced by olive oil factories (OME): olive washing (OWW) and olive vegetation wastewater (OVW).

394

J.M. Ochando-Pulido, M. Stoller / Chemical Engineering Journal 279 (2015) 387–395

(a)

(b) 18

Permeate flux, L m-2h-1

Permeate flux, L h-1m-2

13.5

13.0

12.5

0

50

100

150

200

16

14

PUF = 15 bar > Pb UF PUF = Pb UF - δp= 12 bar

12

10

0

50

100

150

200

t, min

t, min

Fig. 5. Permeate flux evolution during batch UF upon boundary conditions (panel a): experimental flux data for OMW-F (d) and OMW-FPC (o) vs. modelized data: ( supercritical ( ) vs. boundary flux (o) conditions (panel b).

);

Acknowledgments

Table 6 Results of the batch UF operation and final permeate streams. UF feedstock

OMW-F

OMW-FPC

PTM, bar Jss exp., L h1 m2 Jb model, L h1 m2 a, L h2 m2 bar1 COD, mg L1 EC, mS cm1 TSS, mg L1 TPh, g L1 pH Y, %

10 9.7 9.9 0.007 2100 1.1 0 40.2 2.6 89.4

12 12.7 12.9 0.005 700 1.0 0 27.5 3.2 90.6

PTM: transmembrane pressure; Jss: experimental steady-state permeate flux; Jth: modelized boundary permeate flux; Y: feed recovery; EC: conductivity; TSS: total suspended solids; COD: chemical oxygen demand; TPh: total phenols.

In many of the studies available on the treatment or fractionation by membrane technology of OME, the problem of fouling is not correctly approached or not even addressed. The organic matter (COD) removal during the UV/TiO2 photocatalysis (PC) process was described by a two-step first-order kinetic model, and the lab-made ferromagnetic-core photocatalyst was found to provide slightly higher removal than the commercial Degussa P-25 upon lower initial catalyst dosages (1 vs. 0.5 g L1) and was also kinetically advantageous. The proposed boundary model fits the experimental data accurately, such that the estimated boundary flux values correspond to those experimentally observed. Moreover, higher boundary flux values were confirmed for the batch UF operation when the feedstream to the membrane module is further pretreated by the proposed UV/TiO2 PC process (23.3–23.6% increment), and also slightly higher feed recovery and significant minor sub-boundary fouling index a. Especially this latter fact will have important implications on the steady-state performance of UF at industrial scale, enabling longer operation times without having to stop the plant to apply cleaning protocols, and safe the UF membrane from long-term and irreversible fouling. Finally, the higher rejection of the organic matter ensured for the feedstream previously subjected to the whole pretreatment process including UV/TiO2 PC (53.3%) permits achieving the standard limits to reuse the purified effluent for irrigation purposes.

The membrane pilot plant was constructed in the framework of the European project PHOTOMEM (contract no. FP7-SME-2011, Grant 262470) and revamped under the European project ETOILE (contract no. FP7-SME-2007-1, Grant 222331). Funding by the EC is gratefully acknowledged. The Spanish Ministry of Science and Innovation is also gratefully acknowledged for having funded the projects CTQ2007-66178 and CTQ2010-21411, as well as the University of Granada. References [1] Data Taken from Scopus, Aug 2014, . [2] A. Muric´, I. Petrinic´, M.L. Christensen, Comparison of ceramic and polymeric ultrafiltration membranes for treating wastewater from metalworking industry, Chem. Eng. J. 255 (1) (2014) 403–410. [3] X.S. Yi, S.L. Yu, W.X. Shi, S. Wang, N. Sun, L.M. Jin, X. Wang, L.P. Sun, Hydrodynamics behaviour of oil field wastewater advanced treatment by ultrafiltration process, Desalination 305 (2012) 12–16. [4] E. Yuliwati, A.F. Ismail, Effect of additives concentration on the surface properties and performance of PVDF ultrafiltration membranes for refinery produced wastewater treatment, Desalination 273 (1) (2011) 226–234. [5] M. Pizzichini, C. Russo, C. Di Meo, Purification of pulp and paper wastewater with membrane technology for water reuse in a closed loop, Desalination 178 (2005) 351–359. [6] B. Lee, K. Choo, D. Chang, S. Choi, Optimizing the coagulant dose to control membrane fouling in combined coagulation/ultrafiltration systems for textile wastewater reclamation, Chem. Eng. J. 155 (1–2) (2009) 101–107. [7] J. Luo, L. Ding, B. Qi, M.Y. Jaffrin, Y. Wan, A two-stage ultrafiltration and nanofiltration process for recycling dairy wastewater, Bioresour. Technol. 102 (16) (2011) 7437–7442. [8] A.S. Cassini, I.C. Tessaro, L.D.F. Marczak, C. Pertile, Ultrafiltration of wastewater from isolated soy protein production: a comparison of three UF membranes, J. Clean. Prod. 18 (3) (2010) 260–265. [9] S. Zulaikha, W.J. Lau, A.F. Ismail, J. Jaafar, Treatment of restaurant wastewater using ultrafiltration and nanofiltration membranes, J. Water Proc. Eng. 2 (2014) 58–62. [10] E.O. Akdemir, A. Ozer, Investigation of two ultrafiltration membranes for treatment of olive oil mill wastewater, Desalination 249 (2009) 660–666. [11] M. Stoller, M. Bravi, Critical flux analyses on differently pretreated olive vegetation wastewater streams: some case studies, Desalination 250 (2010) 578–582. [12] J.M. Ochando-Pulido, M. Stoller, M. Bravi, A. Martinez-Ferez, A. Chianese, Batch membrane treatment of olive vegetation wastewater from two-phase olive oil production process by threshold flux based methods, Sep. Purif. Technol. 101 (2012) 34–41. [13] S. Delgado Diaz, L. Vera Peña, E. González Cabrera, M. Martínez Soto, L.M. Vera Cabezas, L.R. Bravo Sánchez, Effect of previous coagulation in direct ultrafiltration of primary settled municipal wastewater, Desalination 304 (15) (2012) 41–48. [14] E.M.V. Hoek, M. Elimelech, Cake-enhanced concentration polarization: a new fouling mechanism for salt-rejecting membranes, Environ. Sci. Technol. 37 (2003) 5581–5588.

J.M. Ochando-Pulido, M. Stoller / Chemical Engineering Journal 279 (2015) 387–395 [15] A. Seidel, M. Elimelech, Coupling between chemical and physical interactions in natural organic matter (NOM) fouling of nanofiltration membranes: implications for fouling control, J. Membr. Sci. 203 (2002) 245–255. [16] P. Le-Clech, V. Chen, T.A.G. Fane, Fouling in membrane bioreactors used in wastewater treatment, J. Membr. Sci. 284 (1–2) (2006) 17–53. [17] C.A. Paraskeva, V.G. Papadakis, E. Tsarouchi, D.G. Kanellopoulou, P.G. Koutsoukos, Membrane processing for olive mill wastewater fractionation, Desalination 213 (2007) 218–229. [18] C. Russo, A new membrane process for the selective fractionation and total recovery of polyphenols, water and organic substances from vegetation waters (VW), J. Membr. Sci. 288 (2007) 239–246. [19] E. Turano, S. Curcio, M.G. De Paola, V. Calabrò, G. Iorio, An integrated centrifugation–ultrafiltration system in the treatment of olive mill wastewater, J. Membr. Sci. 206 (2002) 519–531. [20] J.M. Ochando-Pulido, A. Martinez-Ferez, A focus on pressure-driven membrane technology in olive mill wastewater reclamation: state of the art, Water Sci. Technol. 66 (12) (2012) 2505–2516, http://dx.doi.org/10.2166/wst.2012.506. [21] P. Bacchin, P. Aimar, V. Sanchez, Influence of surface interaction on transfer during colloid ultrafiltration, J. Membr. Sci. 115 (1996) 49–63. [22] R.W. Field, P. Aimar, Ideal limiting fluxes in ultrafiltration: comparison of various theoretical relationships, J. Membr. Sci. 80 (1993) 107–115. [23] B. Espinasse, P. Bacchin, P. Aimar, On an experimental method to measure critical flux in ultrafiltration, Desalination 146 (2002) 91–96. [24] M. Mänttäri, M. Nystörm, Critical flux in NF of high molar mass polysaccharides and effluents from the paper industry, J. Membr. Sci. 170 (2000) 257–273. [25] M. Stoller, A. Chianese, Optimization of membrane batch processes by means of the critical flux, Desalination 191 (2006) 62–70. [26] R.W. Field, G.K. Pearce, Critical, sustainable and threshold fluxes for membrane filtration with water industry applications, Adv. Colloid Interface Sci. 164 (2011) 38–44. [27] J. Luo, L. Ding, Y. Wan, M.Y. Jaffrin, Threshold flux for shear-enhanced nanofiltration: experimental observation in dairy wastewater treatment, J. Membr. Sci. 409 (2012) 276–284. [28] M. Stoller, J.M. Ochando-Pulido, Going from a critical flux concept to a threshold flux concept on membrane processes treating olive mill wastewater streams, Proc. Eng. 44 (2012) 607–608.

395

[29] Stoller, B. De Caprariis, A. Cicci, N. Verdone, M. Bravi, A. Chianese, About proper membrane process design affected by fouling by means of the analysis of measured threshold flux data, Sep. Purif. Technol. 114 (2013) 83–89. [30] J.M. Ochando-Pulido, G. Hodaifa, M.D. Victor-Ortega, S. Rodriguez-Vives, A. Martinez-Ferez, Effective treatment of olive mill effluents from two-phase and three-phase extraction processes by batch membranes in series operation upon threshold conditions, J. Hazard. Mater. 263 (1) (2013) 168–176. [31] J.M. Ochando-Pulido, G. Hodaifa, M.D. Victor-Ortega, A. Martínez-Ferez, Fouling control by threshold flux measurements in the treatment of different olive mill wastewater streams by membranes-in-series process, Desalination 343 (2014) 162–168. [32] J.M. Ochando-Pulido, G. Hodaifa, A. Martinez-Ferez, Threshold flux measurement of an ultrafiltration membrane module in the treatment of two-phase olive mill wastewater, Chem. Eng. Res. Des. (2013). doi.org/ 10.1016/j.cherd.2013.11.021. [33] M. Stoller, J.M. Ochando-Pulido, About merging threshold and critical flux concepts into a single one: the boundary flux, Sci. World J. (2014), http:// dx.doi.org/10.1155/2014/656101. article ID 656101. [34] B. De Caprariis, M. Di Rita, M. Stoller, N. Verdone, A. Chianese, Reactionprecipitation by a spinning disc reactor: influence of hydrodynamics on nanoparticles production, Chem. Eng. Sci. 76 (2012) 73–80. [35] A.E. Greenberg, L.S. Clesceri, A.D. Eaton, Standard Methods for the Examination of Water and Wastewater, 16th ed., APHA/AWWA/WEF, Washington DC, 1992. Cabs. [36] D. Chatterjee, S. Dasgupta, Visible light induced photocatalytic degradation of organic pollutants, J. Photochem. Photobiol. 6 (2005) 186–205. [37] E. Chatzisymeon, N.P. Xekoukoulotakis, D. Mantzavinos, Determination of key operating conditions for the photocatalytic treatment of olive mill wastewaters, Catal. Today 144 (2009) 143–148. [38] R.C. Martins, R.J.G. Lopes, R.M. Quinta-Ferreira, Lumped kinetic models for single ozonation of phenolic effluents, Chem. Eng. J. 165 (2010) 678–685. [39] R.V. Shende, J. Levec, Subcritical aqueous-phase oxidation kinetics of acrylic, maleic, fumaric and muconic acids, Ind. Eng. Chem. Res. 39 (2000) 40–47. [40] A.M.F.M. Guedes, L.M.P. Madeira, R.A.R. Boaventura, C.A.V. Costa, Fenton oxidation of cork cooking wastewater-overall kinetic analysis, Water Res. 37 (2003) 3061–3069.