Greg Foley

Effect of cell morphology on dead-end filtration of the dimorphic yeast Kluyveromyces marxianus Var. marxianus NRRLy2415

The dead‐end filtration characteristics of the dimorphic yeast Kluyveromyces marxianus var. marxianus (formerly fragilis) NRRLy2415 were investigated for a range of mean cell morphologies, ranging from predominantly yeast‐like to predominantly filamentous. Semiautomated image analysis was used to measure the mean cell specific surface area, Sv, and the mean ratio of cell length to equivalent cylindrical diameter, Ldm, in each broth. The method of Ju and Ho (Biotechnol. Bioeng. 1988, 32, 95–99) was used to show that for broths with Ldm values between 1.72 and 10.03, the voidage of cell pellets formed by centrifugation increased with increasing Ldm. In the pressure range 30–180 kPa, the specific filter cake resistance, α, was found to be related to pressure, ΔP, through the equation α = α0(1 + kcΔP) . The dependence of α0/Sv2 on Ldm was found to be qualitatively consistent with the pellet voidage data and the Carman−Kozeny equation. Considerably better agreement with the experimental data was obtained when the Kozeny constant, K, was treated as variable and related to Ldm through the equation K = 4.83 + 7.08 log10 Ldm. The cake compressibility constant, kc, was found to increase with increasing Ldm, a phenomenon consistent with the wide range of voidages that can be displayed by beds of long cylinders.

Experimental techniques for quantifying the cake mass, the cake and membrane resistances and the specific cake resistance during crossflow filtration of microbial suspensions

Abstract Simple techniques are reported for (i) measuring the cake mass formed during crossflow filtration of microbial cells, (ii) recovering and measuring the cake mass after crossflow filtration, and (iii) measuring the specific cake resistance during crossflow filtration. The first technique is based on measuring the decline in cell concentration in the feed reservoir during total recycle operation. Cell concentrations are measured on a dry basis and converted to wet weight concentrations using the previously measured wet-to-dry weight ratio of the cells. The second technique involves a series of flushing steps at zero transmembrane pressure followed by wet weight analysis of the suspension formed by the recovered cake. In the third technique, two types of specific cake resistance are defined; an apparent value, α app , and the true value, α true . These are related by the expression α app = α true +( R m − R m0 )/ m where R m − R m0 is the flow resistance due to membrane fouling (as distinct from cake formation), and m is the cake mass per unit area. A methodology for decoupling the effects of cake formation and membrane fouling during crossflow filtration is presented, thus, providing a way of monitoring the time evolution of α true . The techniques described were used to investigate the crossflow filtration of the polymorphic microorganism Kluyveromyces marxianus var. marxianus NRRLy2415 in tubular ceramic membranes. This microorganism displays a range of morphologies ranging from simple ovoid yeast to branched filamentous forms. The approach of measuring the cake and membrane resistances, as well as the specific cake resistance, provides a wealth of information, which greatly adds to our understanding of the mechanism of flux decline in complex biological suspensions.

Minimisation of process time in ultrafiltration and continuous diafiltration: the effect of incomplete macrosolute rejection

Abstract The total process time for the concentration of a macrosolute solution by ultrafiltration and the removal of a microsolute by diafiltration, is calculated given that the limiting flux, J lim , is related to the macrosolute bulk concentration, C , by J lim =a ln (C lim −C(1−σ)/Cσ) where a is a constant, C lim is an apparent limiting ‘wall’ concentration of the macrosolute and σ is the macrosolute rejection coefficient. It is shown that the optimum macrosolute concentration for commencing the diafiltration increases as the rejection coefficient decreases. Previous attempts to model the effect of rejection coefficient on the optimum concentration are shown to be derived in an inconsistent manner and under-predict the optimum concentration. It is also shown that the optimum macrosolute concentration increases as more microsolute is removed in the diafiltration step.

Characterising the packing and dead-end filter cake compressibility of the polymorphic yeast Kluyveromyces marxianus var. marxianus NRRLy2415

Suspensions of the polymorphic yeast Kluyveromyces marxianus var. marxianus NRRLy2415 were grown in batch and continuous culture, producing broths with a range of mean cell morphologies. The mean cell aspect ratio, Ldm, defined as the mean ratio of total cell length to equivalent cylindrical diameter, was measured for each broth by image analysis and ranged from 1.7 to 51. The porosity of the cell pellet formed in a bench-top centrifuge was measured for each broth and in all cases, the pellet porosity decreased linearly with increasing centrifuge speed. This decrease became more pronounced as Ldm increased. The extrapolated zero-speed pellet porosity, e0, was correlated with Ldm by the expression e0=1−1/(1.24+0.062Ldm). The dead-end specific cake resistance, α, was measured for each broth in the pressure range 30–180 kPa. The cake compressibility constant, kc, defined by the expression α=α0(1+kcΔP) was found to increase with increasing values of Ldm. The increase was most rapid in the region, Ldm<10. Microscopic observations indicate that cell deformation at branching points may contribute to cake compressibility.

CHARACTERISATION OF CAKE COMPRESSIBILITY IN DEAD-END MICROFILTRATION OF MICROBIAL SUSPENSIONS

Abstract Culture broths of Saccharomyces cerevisiae and Kluyveromyces marxianus var marxianus NRRLy2415, suspensions of rehydrated active dry bakers yeast and suspensions or calcium carbonate were filtered in dead-end mode at pressures below 200 kPa. In the case of all the microbial suspensions, the specific cake resistance was found to be a linear function of pressure. Cake compression was found to be reversible or weakly irreversible with respect to changes in pressure, i.e., the specific resistance measured at a given pressure was only weakly dependent on whether the filter cake had previously been exposed to higher pressures. The greatest irreversibility effects were obtained with unwashed active dry yeast suspensions, consistent with the breakdown of cell aggregates in these suspensions. The specific cake resistance of calcium carbonate suspensions was found to be a non-linear function of pressure. The compression of calcium carbonate cakes was irreversible, consistent with the breakdown of the large...

Ultrafiltration flux theory based on viscosity and osmotic effects: application to diafiltration optimisation

Abstract The optimisation of diafiltration is examined using the new theory of ultrafiltration based on osmotic pressure and the viscosity dependence of the mass transfer coefficient. It is shown that, if the permeate flux is maintained at the limiting value at each bulk macrosolute concentration, the diafiltration time always decreases with increasing concentration. However, the transmembrane pressure required to reach the limiting flux increases rapidly with concentration, thus putting a practical limit on the concentrations at which an ultrafiltration system can be operated under limiting flux conditions. In contrast, when the osmotic pressure is a polynomial function of the wall concentration and operation at a fixed transmembrane pressure is considered, a finite optimum concentration is found which minimises the diafiltration time. A methodology for computing this optimum concentration is developed. It is shown that the optimum concentration increases with increasing transmembrane pressure but is only weakly dependent on feed velocity and membrane resistance.

On the Relation between Filtrate Flux and Particle Concentration in Batch Cross flow Micro filtration

Abstract The relation between the filtrate flux and particle concentration in batch cross flow micro filtration is investigated using a model based on classical filtration theory and the Kern–-Seaton theory of surface fouling. The model, which includes the effects of cake compressibility but not of membrane fouling, is solved for both laminar and turbulent tangential flows. It is found that the sole effect of cake compressibility is to reduce the flux without altering the general shape of the flux versus concentration curve. Fluxes which increase with increasing concentration are shown to be a result of enhanced cake removal due to the increased wall shear stress brought about by increased suspension viscosity. A sigmoidal relation between flux and concentration is reproduced by the model only if there is a reduction in the cake removal rate as the tangential flow regime changes from turbulent to laminar.

Dimensional Analysis of Crossflow Microfiltration Data

Abstract A new approach to correlating crossflow microfiltration (CFMF) data based on dimensional analysis is presented. The steady state flux was assumed to be a function of the trans‐membrane pressure (ΔP), the crossflow velocity (u), the particle concentration (c), filtrate viscosity (μ), and membrane resistance (R m). Correlations of the form J/u=K(ΔP/cu 2) a (ΔP/μuR m) b were tested on three sets of published data: one for CFMF of dried yeast suspensions in a laminar flow hollow fiber module, one for dried yeast suspensions in a turbulent flow tubular module and one for suspensions of latex particles in a laminar flow flat sheet module. The R 2 values for the fits of the correlations to the data were 0.98, 0.94, and 0.91 respectively.

Time-optimal batch diafiltration*

This study concentrates on time-optimal operation of a general batch diafiltration process. The process model consists of a set of input affine differential equations. We apply Pontryagins minimum principle to formulate necessary conditions of optimality. These are then used to identify the shape of optimal control as well as singular surface in concentration space. Comparison is made between traditional control techniques and obtained minimum time control. It is shown that traditionally used operations are suboptimal for selected case studies.

MODELING PROCESS DYNAMICS USING A NOVEL NEURAL NETWORK ARCHITECTURE: APPLICATION TO STIRRED CELL MICROFILTRATION

A novel neural network architecture is presented for dynamic process modeling, using stirred cell microfiltration of bentonite suspensions as a model system. Unlike previous studies that include time explicitly as a network input and have a single output at that time, the network architecture presented contains the process variables as inputs and many outputs representing the output (filtrate flux in this case) at different selected times. The network is shown to represent the stirred cell microfiltration of bentonite suspensions over a range of pressures (0.2–1.5 bar), initial concentrations (0.5–2.0 g/L), stirrer tip speeds (0.04–0.17 m/s), membrane resistances (3.09 × 1010–6.85 × 1010 m−1), pH values (2.5–10.4), and temperatures (20°–24°C) with good accuracy (R2 = 0.91 on network test data). With this network architecture, it becomes easy to track the time dependence of the relative effect of the various process parameters on the system output. Thus, for example, the network weights show that the effect of stirring rate on flux increases as time progresses, while the opposite effect is seen for membrane resistance, as expected.