Hydrodynamic studies of innovative membrane reactor for enzymatic hydrolysis of lignocellulosic waste

This paper presents the study concerning the impact of the basic operational parameters on the performance of an innovative microfiltration membrane reactor applied for enzymatic hydrolysis of lignocellulosic biomass. The concept and basic hydrodynamics of the reactor with tubular ceramic membranes and a propeller agitator were shown. Besides, the efficiency of enzymatic hydrolysis of corn straw was studied to check reactor functionality. It has been proven that the proposed reactor construction can improve the microfiltration of lignocellulosic suspension by reducing the cake layer on the membrane surface. Increasing the rotational speed of the propeller agitator also improved the filtration efficiency. The permeate flux during the microfiltration experiments was lower for smaller lignocellulose biomass fraction (D < 425 μm) when compared to the less fragmented corn straw (425 < D < 900 μm). For larger solid fractions, a stirring speed increase enhanced the separation efficiency regardless of the differences in biomass concentration. In contrast, this trend for the finer biomass fraction was only noticeable for the highest used biomass concentration (C = 2.0%). Considering the enzymatic hydrolysis of corn straw, membrane separation of reaction products positively influenced the process yield, and the results depended on the applied operational parameters.

its digestibility.In practice, various pretreatment methods, including physical, chemical, biological, and combined techniques, are used, [9][10][11] which, on the one hand, improve the hydrolysis efficiency while generating energy and reagents costs.It has been proven in previous studies that the efficiency of both pretreatment and enzymatic hydrolysis was enhanced as the biomass particle size decreased. [12]This improvement resulted from enhanced specific surface area and reduced crystallinity index of lignocellulosic material, leading to a more effective mass and heat transfer and better substrate accessibility for enzymes.Apart from the influence on lignocellulose conversion, biomass particle size is also a crucial factor in transportation logistics; therefore, its reduction may improve the overall economics of the biorefinery.
Another relevant obstacle in the processing of lignocellulosic biomass is the high cost of enzymes used as catalysts for cellulose and hemicellulose hydrolysis to fermentable monosaccharides and the inhibition of the enzymes by reaction products. [13][18] The main advantage of using membrane reactors is the ability to recover and reuse enzymes and receive pure monosaccharides in the permeate.Configurations of membrane reactor systems used for different lignocellulosic substrate hydrolysis proposed so far have usually involved the application of ultrafiltrationbased separation of low-molecular reaction products. [14,19,20]In this case, unreacted biomass particles and cellulolytic enzymes (molecular weight of 23-150 kDa [21] ) were retained in the reaction mixture due to the higher sizes than pores of the used membrane.At the same time, monosaccharides (mainly glucose and xylose) passed through to the permeate site and thus were removed from the hydrolysis solution.
Among studies conducted so far on enzymatic hydrolysis of cellulosic raw materials in membrane reactors, most literature reports concern the application of an integrated reaction vessel with an ultrafiltration membrane (dead-end filtration system) [16,22] or reactor equipped with an external membrane module, usually with hollow fiber membrane working in a cross-flow mode. [17,23]It should be noted that since lignocellulose is insoluble, pumping high-solids concentration feed streams into the outside module from a reactor is problematic due to the high solution viscosity.Other solutions were used only in a few cases.For example, Al-Zuhair et al. [19] developed a jacketed cylindrical continuous stirred tank membrane reactor with vertical sides made of ultrafiltration membrane.A similar concept was applied by Al-Mardeai et al. [24] In both cases, the reactor construction ensured a radial circulation direction of the feed.In contrast, Belafi-Bako et al. [25] used a tubular membrane reactor with a porous stainless steel tube placed in a cylindrical vessel without mixing.
Compared to microfiltration, the disadvantage of ultrafiltration of lignocellulosic suspension is primarily the more quickly growing resistance of hydrolysate flow through membranes.It is related to concentration polarization, fouling of the membrane, and cake layer formation, which lead to low permeate yields. [14,26]The membrane fouling mechanism based on filter cake formation is the most serious concern, especially at high insoluble solids in the feed.These phenomena limit the application of ultrafiltration of lignocellulosic hydrolysates in industrial conditions.The overall mass transport can be improved by reducing the boundary layer around the biomass particles through the intensive turbulences of the feed.However, it should be aware that too intensive shear forces may decrease enzyme activity. [27]Although the filter cake formed on the membrane reduces the permeate flow, the enzymatic hydrolysis process occurs there with the highest intensity. [28]Therefore propeller and its rotations must be carefully matched to the process.Besides, the enzyme protein fouling on the surface or in the pores of ultrafiltration membranes may occur, decreasing permeate flux. [29]It may be alleviated by using hydrophilic membranes with a rough surface.It was often confirmed that the membrane reactor configuration significantly influences the enzymatic reaction and the separation process. [30]Thus, both membrane type, module characteristics, and operating parameters must be appropriately considered during process design to maximize the membrane reactor performance.
Applying two membrane separation units (as shown in Figure S1), including microfiltration as a first step to retain the unreacted lignocellulosic substrate and ultrafiltration (second step) to recover enzymes, could enhance the permeate flux and ensure a long-term process.
In this case, the risk of the ultrafiltration membrane's permeability quickly decreasing during its use is reduced.The enzymes can be constantly recycled into the reaction space, whereas the hydrolysates are received in the permeate.The residue after the reaction, containing exhausted lignocellulose (mainly composed of lignin), can be removed continuously or periodically in a cyclic mode of operation (see Figure S1).A scarcity of reports on the application of microfiltration units for enzymatic hydrolysis of lignocellulosic substrates has been published, e.g., ref. [31, 32] The area in Figure S1 marked by the dotted line refers to the microfiltration membrane reactor used and described further in this work.Constant recovery and return of enzymes to the reactor space improve this process; however, it wasn't considered in our research presented in this paper.Our previous investigations proved that ultrafiltration membrane enabled partial cellulolytic enzyme recovery from the permeate obtained after hydrolysis in a microfiltration membrane reactor. [33]e efficiency of obtaining the final products depends on the permeate flux and the reaction rate (kinetics) simultaneously.For this reason, the impact of operational parameters such as pressure, temperature, concentrations, and agitator rotations on the final effects cannot be predicted without experimental studies.
The scope of the present work covers the impact of the basic operational parameters on the performance of the new microfiltration membrane reactor for enzymatic hydrolysis of lignocellulosic biomass.The research aimed to investigate the basic hydrodynamics of the system.Tubular ceramic microfiltration membranes, made of abrasion-resistant material, were used for separating the hydrolysate (C5 and C6 sugars solution) from the lignocellulosic waste.Circulation of the biomass particle suspension inside the reactor was forced by a propeller, further increasing the turbulence and, thus, the pro-cess intensity.A guiding tube installed between the propeller and membranes ensured an advantageous axial circulation of the reaction mixture.The proposed reactor construction leads to the scouring effect, sometimes achieved in membrane modules using bentonite or diatomite to reduce the concentration polarization. [34]The membrane scouring effect by the particulate matter of lignocellulosic biomass was expected to maintain the high efficiency of the filtration process.
Besides, the course and efficiency of enzymatic hydrolysis of corn straw in the proposed membrane reactor, compared to the batch mode, were tested.

Feedstock, chemicals, and enzymes
Corn straw, provided by Bioagra S.A. (Poland), was used in our investigation as the lignocellulosic raw material.Its main composition, determined by the National Renewable Energy Laboratory (NREL) procedure, [35]  The shredded raw material was screened into two fractions of the different size ranges (425 < D < 900 μm and D < 425 μm) using a vibrating screen.Before that, corn straw biomass was pretreated using 2% sodium hydroxide, according to the procedure applied in our previous investigations, [33] to increase its susceptibility to enzymatic digestion.
The carbohydrate composition of the pretreated biomass, determined by the NREL procedure, was 54.

Membrane reactor and experimental setup
The schematic construction of the new membrane reactor [36] proposed for the hydrolysis of lignocellulosic waste is shown in Figure 1A.An axial circulation of the suspension inside the reactor (marked with blue lines in Figure 1A) is generated by the propeller (2) with a guiding tube (7).This circulation plays an important but ambiguous role due to the complexity of the involved phenomena.The primary function of the agitator is to maintain the homogeneity of the reaction space in terms of temperature and concentration.But the mixing also controls the thickness of the filter cake and, thus, the permeate flow rate.The intensive tangential flow of the suspension to the membrane surface (cross-flow) reduces the thickness of the filter cake through turbulences so that the permeate flow can be high.
The hydrolyzed solution with enzymes flows through the microfiltration membrane as a permeate stream.The transmembrane pressure highly influences the permeate flux and the filter cake thickness.In turn, the permeate flow rate also ambiguously affects the amount of produced hydrolysis products due to its influence on residence time in the cake layer.Therefore, it is necessary to control the reactor hydrodynamics by properly synchronizing the permeate flow with the reaction kinetics.
The most crucial operating conditions influencing the performance of the membrane reactor used in this work are the rotation of the propeller, the transmembrane pressure, the lignocellulosic material concentration, and its particle size.They affect the membrane boundary layer thickness and permeability due to the influence on the hydrodynamics, which controls the turbulence, shear stresses, and scouring effects. [37]e scheme of the experimental setup used to study both the reactor hydrodynamic and enzymatic hydrolysis of lignocellulosic biomass is presented in Figure 2. The membrane reactor (1) was connected with an external feed tank (2) of 30 dm 3 attached to the nitrogen gas supply.
The experimental procedures are described in Section 2.3.2. and 2.3.3.

Membrane characteristics and cleaning
Before each experiment performed with different concentrations of biomass suspension in the reactor, membranes were characterized by determining the initial permeability (L p [dm 3 ⋅m -2 ⋅h -1 ⋅bar -1 ] ) according to the well-known equation. [27]The flow rates of pre-filtered tap water for three different transmembrane pressures of 0.15, 0.20, and 0.25 bar were measured using a cylinder and stopwatch to calculate the L p value.Average data obtained based on ten measurements for each pressure was used for calculations.
After each measuring series, the membrane washing operations were duly performed with tap water and alternately with 0.1 mol⋅dm -3 hydrochloric acid and then 0.1 mol⋅dm -3 sodium hydroxide solutions.
When stopping the installation, the 0.02% w/v sodium azide was used to protect against microbial growth.After membrane cleaning, the L p value was re-evaluated as described above.
The permeability loss (L p,loss ) was calculated after each membrane cleaning using the following formula (Equation 1): where L pi and L pf are the water permeability before and after both filtration and cleaning, respectively.

Study on reactor hydrodynamics
To Experiments were carried out in the system shown in Figure 2.
The membrane reactor (1) was entirely filled with the appropriate suspension before the measurements to keep a constant membrane separation area and the desired biomass concentration in the reactor during the microfiltration research.All assays were done at ambient temperature (22-23 • C).The suspension of lignocellulose was intensively mixed in the reactor before each measurement.The tap water was delivered to the reactor from the feed tank (2) under pressure gained using compressed nitrogen.The pressure in the reactor was set and controlled using a manometer (3, Figure 2).The pressurized gas forced water flow to the reactor (1, Figure 2), which, to prevent the backflow of the media, was installed below the level of the feed tank.
The suspension was stirred by a propeller agitator (4) powered by an induction motor (5).The retentate outlet (6) was closed, and the permeate was collected in the tank (7).Permeate flow was measured using a cylinder and stopwatch.After the whole set of microfiltration experiments for one of the studied biomass concentrations, the retentate was collected in the tank (8, Figure 2).The reactor was then cleaned following the procedure described in Section 2.3.1.Between each measuring series, the lignocellulose suspension was intensively mixed and left for sedimentation alternately to enable the filtration cake detachment from the membrane surface.For the initial 2 h, the hydrolysis was carried out in a batch mode, that is, without nitrogen supply to the reactor.After this time, the gas was provided into the feed tank to generate the overpressure in the whole system, forcing the permeate flow through the membrane.Due to the continuous supply of buffer solution containing enzymes from the feed tank in the reactor, the reaction volume was constant.The permeate flow measurements were conducted using a cylinder and a stopwatch.
During the process carried out at a pressure of 2.0 bars and propeller rotations of 240 min -1 (the highest among tested), the feed tank was emptied after ca.5.5 h of hydrolysis.Refilling the tank without putting the system into an unsteady state was impossible.Therefore, to reasonably compare the results obtained for all the process conditions tested, all the reactions were terminated 5.5 h after the hydrolysis was started.Samples were taken from the permeate stream and a collected volume of permeate during the process and from the retentate after reducing the pressure into atmospheric at the end of the reaction.The glucose and xylose concentrations in the samples were measured by the HPLC method, according to the procedure previously reported. [33]similar hydrolysis process was carried out in a batch mode with a propeller rotation of 240 min -1 to compare the reaction efficiency.
For this purpose, the same experimental setup was used, only without applying pressure inside the reactor, causing the permeate flow.
The instantaneous mass of released monosaccharides (summary glucose and xylose in time) was determined from the mass balance, according to Equation (2): where: The final (total) reaction yield was calculated using the following formula (Equation 3): where: content determined in the pretreated corn straw. [38] calculate the glucan and xylan saccharification yield, formulas similar to Equation (3) were involved, but the mass of glucose or xylose was used, respectively.

Membrane permeability
The surface and cross-section pictures of used membranes, made using a scanning electron microscope (SEM) (Figure S2), show a rough and porous membrane surface.The results of determining membrane permeabilities before (initial) and after measurement series, conducted with different corn straw concentrations and membrane cleaning, are presented in Table 1.The increase in average hydraulic resistance can be observed due to the irreversible membrane fouling caused by corn straw biomass.However, the permeability loss was the highest after the first measuring series.

Influence of process parameters on permeate flux
In Figure 3 It can be observed that permeate fluxes depended on the process conditions and were even 20% higher for a fraction of larger particles.In both cases, the rotations of the agitator (Figure 3A,B) increased the permeate flow linearly.At lower concentrations of the suspension (C = 0.1%), this increase is the most apparent, whereas, for higher concentrations (C = 0.5% and 2.0%), the rotations of the agitator had a much lower effect on the permeate flux (the circles and the triangles in Figure 3A,B).This result is related to reducing the filter cake thickness on the membrane surface (i.e., surface renewal effect) and implies that shear stress on the membrane surface, which is a function of agitator rotations, affected membrane fouling behavior.These results agree with previous investigations, [39] proving that the weaker membrane fouling occurs for higher shear rates.
On the other hand, the suspension particles are deposited during filtration on the membrane, which increases the resistance of the filter cake and decreases the permeate flux.It is an accumulation effect and must be counterbalanced with the effect of surface renewal if the flux is to be constant (i.e., steady-state process).
Figure 3C,D show that the impact of suspended solid concentration on the permeate flux reduction is very strong regardless of the other process conditions.The results are consistent with previous studies proving that increasing the feed concentrations makes the membrane fouling more severe due to forming a thicker filtration cake layer. [40]sides, lower values of permeate flux were found for particles with D < 425 μm, indicating that the smaller particles form a less porous, thus less permeable cake layer on the membrane surface. [39]

Formation of the filtration cake and its characteristics
In Figure 4A-F, the effect of a flux decline, that is, the typical changes (reductions) of the permeate stream during the microfiltration, [37,41] for different process parameters, is presented.It should be assumed that all initial fluxes correspond to streams of pure water because, at the beginning of the process, there is no filter cake yet on the membrane.After a prolonged period, the stream stabilized under the influence of the opposing factors, that is, the accumulation and surface renewal.This time-dependency has been recorded under different In Figure 4A the effect of instability of the film formed from small particles (D1) under low pressure (P1), lower concentration (C1) and intense rotations (N3) can be observed (black circle, P1C1N3D1 on the legend).The oscillating change of the permeate stream with time testifies to the alternating creation and demolition of the filter cake on the membrane.This phenomenon was observed only for the lowest studied concentrations and smallest suspension particles.After a longer time (>50 min), the filter cake was stabilized for all process conditions.For both fractions with the highest concentrations of sus-pensions C = 2.0%, (see Figure 4E,F), the film stabilized very quickly (after ca. 10 min).This behavior can be explained by the threshold flux concept, [42] whereby a low and almost constant fouling rate is observed below a certain permeate flux.In this case, membranes fouled rapidly, then the fouling rate decreased, and the flux profiles flattened.
Many advantages of applying such a system for realizing enzymatic hydrolysis are due to the possibility of using large fragmentation of the lignocellulosic raw material.The small size of lignocellulosic particles significantly enhances the hydrolysis process.Thus, it improves the efficiency and kinetics of the process because it increases the specific surface of solid material.This also facilitates the access of the enzyme to the hydrolyzed substance (cellulose, hemicelluloses), which is to be hydrolyzed independently of the structure of plant tissues.However, not all optimistic assumptions of such an innovative reactor can be ity.It should also be noted that the shredding raises the process costs to a very significant extent.Moreover, the finer the biomass particles in the suspension, the smaller the permeability of the filter cake at the membrane surface, which reduces the entire process performance.By ensuring rapid mixing, membrane fouling due to the deposition of the particles on the membrane surface can be minimized.for example, [17,22,33] In general, it was recognized that hydrolysis efficiency in membrane reactors was improved due to the removal of inhibitory products.Additionally, it may be explained by the fact that the biomass concentration at the membrane surface is much larger than that in the core region of the reactor, which accelerates the reaction rate.However, it is worth adding that the observed improvement in saccharification yield may be attributed not solely to the use of a membrane reactor but also to the continuous supply of fresh buffer with enzymes into the reactor.Therefore, additional experiments should be performed to study possible enzyme activity loss, for example, due to the shear forces and/or unproductive adsorption on lignin.

The course and efficiency of enzymatic hydrolysis
The results showing the total concentrations of released monosaccharides in the collected permeate as a function of time are presented in Figure 5C,D.Since during the first 2 h the reaction was carried out in a batch mode, the contents of glucose and xylose in the samples taken from the permeate tank are presented as recently as the first permeate stream appeared after generating the overpressure in the system.
The concentrations of hydrolysis products changed with time, but their course depended on the applied operational parameters.In general, a decrease in monosaccharide concentrations over time was observed, which is consistent with previous studies on enzymatic hydrolysis of lignocellulosic feedstock in a membrane reactor. [22,33]ly in the case of reaction at a P = 0.3 bar and N = 240 min -1 , the increase of the monosaccharide concentrations in the permeate stream during the process was found.
Maintaining constant flow conditions until the whole substrate is converted into products, the mass of the products per volume of permeate is increasingly lower, which is a significant drawback of using membrane reactors.Therefore, process optimization, including the pressure and agitation conditions for various feedstock used concentrations, is needed to maximize the C5 and C6-sugar content in the obtained permeate.Nevertheless, the results clearly showed that the operational parameters are key factors on which the course and efficiency of the enzymatic hydrolysis process depend.Properly selecting the process parameters makes it possible to maximize process efficiency and productivity.

CONCLUSIONS
This paper presents the new membrane reactor sion (C = 0.1%), whereas, for higher concentrations (C = 0.5% and 2.0%), the rotations of the agitator much less influenced the permeate stream.These results can be used to optimize the working conditions of the reactor in the future, as membrane fouling is more likely to develop for small particle sizes.
In addition, it was found that the efficiency of enzymatic hydrolysis performed in a membrane reactor was remarkably higher than in a batch mode, and the results depended on the used operational parameters.However, further experiments are needed to explain if the observed improvement in saccharification yield may be attributed solely to using a membrane reactor and not to the continuous supply of fresh buffer with enzymes.Analyzing the system hydrodynamics without hydrolysis and the influence of the operational parameters on the enzymatic reaction is essential to describe the reactor performance completely.
provided by the manufacturer, Accellerase 1500 shows the highest activity at 50-60 • C and pH in the range of 4.6-5.0.Supplementing this enzymatic mixture with Accellerase XC and Accellerase XY is desirable to increase the efficiency of Accellerase 1500 action.Accellerase XC, containing mainly xylanases and additional cellulases, was derived from the culture of Trichoderma reesei.This preparation has the highest activity at a temperature of 45-65 • C and pH of 3.5-6.5.Accelerase XY is a complex of hemicellulolytic enzymes (secreted by T. reesei), that positively influences glucan and xylan hydrolysis.This mixture's optimum temperature is 50-75 • C and a pH between 4.5 and 7.0.The recommended doses of the above enzyme complexes are as follows: 0.05 to 0.25 cm 3 per gram of biomass (depending on biomass compo-sition) for Accellerase 1500, 0.005 to 0.05 cm 3 per gram of biomass for Accellerase XY and 0.0125 to 0.125 cm 3 per gram of biomass for Accellerase XC.
The reactor (volume: V = 14 dm3 ) is integrated with membranes, where in one housing(1) are the propeller mixer (2) and 16 tubular ceramic microfiltration membranes (3) (inner diameter: d in = 10 mm, length: l = 200 mm) with a pore size of ∼ 1 μm, made from ceramic material Al 2 O 3 -α (Atech innovations GmbH, Germany).One membrane area was 0.011 m 2 ; thus, for all sixteen membranes total area equaled 0.176 m 2 .Membranes were concentrically placed around the propeller stirrer (2) (diameter, d = 11 cm).A guiding tube (d = 13 cm) is located between the membranes and the propeller(7).The suspension of the feedstock (i.e., grounded lignocellulosic waste) and enzymes in the buffer flow into the reactor by the inlet (4).After passing through the membranes, the hydrolysate solution and enzymes outflow by the permeate outlet(5).The suspension of heavily concentrated particles of lignocellulosic material forms a layer on a membrane surface (i.e.filter cake), through which enzymes flow.The layer has a dynamic character because it is permanently renewed by turbulence due to the intense operation of the propeller.The retentate can outflow from the reactor by an outlet (6) continuously or periodically.Tubular ceramic membranes (2) are mounted upright on the plate manifold(10).The lower ends of membranes are closed with the caps (9), allowing the effective isolation of the retentate space (R), and the permeate space (P), as indicated in Figure1A.The real photos of the reactor's main components are presented in Figure1B-D.

2. 3 . 3
Study on biomass hydrolysisThe second part of the present work investigated the feasibility of using the proposed membrane reactor for the enzymatic hydrolysis of lignocellulosic biomass.Studies included the performance of experiments, which were conducted for different transmembrane pressures (P [bar]: 0.3; 1.0 and 2.0) and propeller rotations (N [min -1 ]: 80; 160 and 240) under flow conditions.The pretreated corn straw biomass with a particle size of 495 < D < 900 μm was used as a substrate in each experiment.The reactor (1, Figure2) was filled with 200 g of dry matter of pretreated corn straw in 14 dm 3 of citric buffer (0.05 mol⋅dm -3 , pH 5.0) containing a dissolved mixture of Accelerase enzymatic preparations (274 cm 3 of Accelerase 1500, 55 cm3 of Accelerase XY and 137 cm 3 of Accelerase XC).The reactor content was thermostated at 48 • C using the electric heat jacket.The ratios of enzymes were selected based on the recommendation of the enzyme producer; however, doses of enzymes per g of the dry mass of biomass were higher than recommended to increase the reaction rate.The same solution of enzymatic preparations in a citric buffer (volume of 30 dm 3 ) was placed in a feed tank (2, Figure2).Since the reactor was equipped with microfiltration membranes, during the hydrolysis at flow conditions, low-molecular-weight reaction products (mainly glucose and xylose) and enzymes passed through the membrane pores.At the same time, unreacted substrate components (cellulose, hemicellulose, lignin, insoluble oligosaccharides) remained in the feed.The collected permeate was clear, meaning it contained no insoluble substances.The reactor was continuously replenished with a buffer containing fresh enzymes from the feed tank to maintain a constant concentration of enzymes in the reaction mixture.
gk total (t)-the total instantaneous mass of glucose and xylose produced in the reaction, [g]; C gk R (t)-the concentration of glucose and xylose in the reactor, [g⋅dm -3 ].It was determined based on the measurements of glucose and xylose concentrations in the samples taken from the permeate stream, assuming the ideal mixing inside the reactor; C gk P (t)-the concentration of glucose and xylose in the collected volume of permeate, [g⋅dm -3 ]; V R , V P (t)-the volume of the reaction mixture and collected permeate, respectively [dm 3 ]; final -the final reaction yield calculated based on the amount of produced monosaccharides after 5.5 h of reactions, [%]; m gk total,5.5h-the total mass of glucose and xylose produced in the reaction during 5.5 h, [g]; m gk max -the maximum mass of glucose and xylose that may be released from the used corn straw biomass in the case of the complete hydrolysis of glucan and xylan, [g].Following our previous report, it was calculated based on the glucan and xylan , a comparison of the permeate fluxes for the various process parameters, such as stirrer rotations N [min -1 ], concentrations of biomass in suspension C [%], and transmembrane pressures P [bar], are shown.Figures 3A,C present the results obtained for the suspensions of finer fractions of the ground biomass (D < 425 μm), whereas Figure 3B,D show the results for the thicker biomass fraction (425 < D < 900 μm).

TA B L E 1 9 F I G U R E 3
Comparison of initial water permeability and permeabilities determined after measurement and cleaning series.Permeability (L p ) [dm 3 ⋅m -2 ⋅h -1 ⋅bar -1 ] Permeability loss (L p,loss ) The effect of process parameters on the permeate flux in a studied membrane reactor.(A,B) Influence of the rotational speed of the propeller (N) for different pressure (P) and biomass concentration in a suspension (C).(C,D) Influence of the suspension concentration (C) for different rotational speeds of the propeller (N) and pressure (P).The suspension contained milled and sifted corn straw particles of sizes in the range of D < 425 μm (A,C) and 425 < D < 900 μm (B,D).The curves shown are representative of three replicates at each process condition.conditions of the pressure, concentration, and rotations rates of the agitator, separately for two different fractions of biomass, that is, smaller particles with D < 425 μm (Figure4A,C,E) and larger particles of suspension 425 < D < 900 μm (Figure4B,D,F).For better visualization, the research results of these two fractions are placed side by side, and the symbols in the legends correspond to the values of the process parameters, as described in the caption for Figure4.

Figure
Figure 5A,B show the mass changes of glucose and xylose (m gk total ) released during the hydrolysis as a function of time for all studied operational parameters at the flow conditions in the proposed mem- construction and hydrodynamic characteristics developed to improve the efficiency of enzymatic hydrolysis of lignocellulosic biomass.It has been proven that the proposed reactor with tubular ceramic membranes and a propeller agitator can be used to improve the microfiltration of lignocellulosic suspension by reducing the filtration cake on the membrane surface and to increase the lignocellulose hydrolysis yield compared to the batch process.Increasing the rotational speed of the propeller agitator has been shown to improve filtration efficiency by extending the time required to reduce the permeate flux partially.It was visible that the permeate flux during the microfiltration experiments was lower for smaller lignocellulose biomass fraction (D < 425 μm) when compared to larger particles (425 < D < 900 μm) in each corresponding pair of measurements.For larger solid fractions, a stirring speed increase seemed to enhance the separation efficiency, but the effect depended on biomass concentration.It was observed at lower concentrations of the suspen-

Process parameters Reaction yield, Y final [%]
process performed in a membrane reactor yielded a higher mass of glucose and xylose than in a batch reactor using the same enzyme and substrate dosage and process conditions.The values of final hydrolysis yields (Y final ) determined using Equation (3) are reported in Table2.The results indicate that the Y final values for the hydrolysis performed in a membrane reactor are ca.6-21% higher than in a batch mode, depending on the used transmembrane pressure.It is both in terms of the total reaction yield and the yield of glucan and xylan hydrolysis.The observed enhancement in the yield of monosaccharides release confirms the results found in many previous investigations,