Research Interests

My research interests are in a broad area of biological engineering. Over the years, we have applied chemical engineering principles to the problems in the domain of bioprocesses. A brief glimpse of our work and important results are highlighted below.

Bioprocess involving microbial cells and enzymes are multiphase processes with biotic phase (microorganisms) and abiotic phase consisting of gas, liquid and solid. Our research in the area of modeling of bioprocesses has focused on understanding the interactions between multiphase transport phenomena and biochemical reaction kinetics. Several processes investigated include multiphase multienzyme oxidation of glucose, growth and metabolite production in hollow fiber bioreactors and bioremediation using immobilized cell systems.

Multiphase Bioreactors

Schematic representation of transport and reaction processes in a single fibre of the HFBR

Schematic representation of tray bioreactor for solid state fermentation process

We have developed a generalized model for flow fields in hollow fiber bioreactor in the presence of growth of the microorganisms. Our analysis of the model shows that the secondary flows arising in these reactors can change the nutrient and/or product distributions significantly , and thereby, alter the bioreactor performance. We have been able to bring out the complex interactions between multiphase transport phenomena and biochemical reaction kinetics . Several interesting criteria for assessing the limitations posed by the mass transfer processes on the performance of the bioreactor for processes have been proposed. Our work in the area of solid-state fermentation has clearly brought out several interesting observations for the first time, namely, (i) the occurrence of near anaerobic conditions at inner depth in biofilms in spite of the presence of oxygen in the voids of the bed, a phenomena which was subsequently verified experimentally and (ii) the complex transport and bioreaction interplay leads to transition from mass transport to energy transport regimes. We have shown for the first time the spatial and temporal variations in the growth processes and its implications for monitoring and controlling “difficult to control” solid state fermentation bioreactors. The growth of cells around single particle and resulting expansion of biofilm around the particle is analyzed for simplified zero and first order oxygen consumption kinetics. Under conditions of zero order kinetics, the model predicts upper limit on cell density. The model simulations for packed bed of solid particles in tray bioreactor show distinct limitations on growth due to simultaneous heat and mass transport phenomena accompanying solid state fermentation process. The extent of limitation due to heat and/or mass transport phenomena is analyzed during different stages of fermentation. It is expected that the model will lead to better understanding of the transport processes in SSF, and therefore, will assist in optimal design of bioreactors for SSF.

Metabolic Control Analysis

General reaction mechanism of the pyruvate dehydrogenase multienzyme (PDH) com

Growth and metabolite production by cells involves complex network of reactions at two levels: metabolite and gene. We have developed a model for pyruvate decarboxylation for multienzyme pyruvate dehydrogenase complex. Metabolic control analysis (MCA) of pyruvate dehydrogenase multienzyme (PDH) complex of eucaryotic cells has been carried out using both in vitro and in vivo mechanistic models. Flux control coefficients (FCC) for the sensitivity of pyruvate decarboxylation rate to activities of various PDH complex reactions are determined. At high pyruvate concentrations, the control is shared by all of the components, with E1 having a negative influence while the other three components, E2, X, and K, exert a positive control over the pyruvate decarboxylation rate. An unusual behavior of deactivation of the E1 component leading to higher net PDH activity is shown to be linked to the combined effect of protein X acylation and E1 deactivation. The steady-state analysis of the in vivo model reveals multiple steady state behavior of pyruvate metabolism with two stable and one unstable steady-states branches. FCCs also display multiplicity, showing completely different control distribution exerted by pyruvate and PDH components on three branches. At low pyruvate concentrations, pyruvate supply dominates the decarboxylation rate and PDH components do not exert any significant control. Reverse control distribution is observed at high pyruvate concentration. The effect of dilution due to cell growth on pyruvate metabolism is investigated in detail. While pyruvate dilution effects are shown to be negligible under all conditions, significant PDH complex dilution effects are observed under certain conditions. Comparison of in vitro and in vivo models shows that PDH components exert different degrees of control outside and inside the cells. At high pyruvate levels, PDH components are shown to exert a higher degree of control when reactions are taking place inside the cells as compared to the in vitro situation.It is expected that such an understanding would lead to better genetic engineering of the cells.

Pichia Pastoris Fermentaion

Schematic representation of Pichia pastoris fermentation

Modeling hCG Fermentation

Pichia pastoris, rapidly growing yeast, has gained a widespread acceptance as a host for production of a number of recombinant proteins. Recent exciting developments in glycosylation engineering have further propelled Pichia pastoris as an expression host for the manufacture of various vaccines and biopharmaceuticals, including biogenerics. While significant advances have been made in genetically engineering the cells, engineering of fermentation systems to develop scalable and cost-effective human like therapeutic protein manufacturing systems is a challenging task. The critical design considerations to achieve high cell densities, and thereby high protein yields, are nutrient(s) supply rate, oxygen transfer and temperature control.Our research group is involved in developing model based strategies for high cell density Pichia pastoris fermentations for production of glycoprotein hormones such as human chorionic gonadotropin (hCG), follicle stimulating hormone (FSH) and luteinizing hormone (LH). A simple unstructured semi-stoichiometric model for fed-batch process is developed using experimental data gathered from a preliminary set of experiments. The model incorporates several important aspects of Pichia pastoris fermentation: glycerol inhibition on cell growth, methanol inhibition on cell growth and methanol inhibition on the production rate. One of the key observations from experimental study is the delayed protein expression after the methanol feed is initiated. This phenomenon of delayed protein expression is captured by introducing shock-recovery term in the model which includes AOX promoter repression by glycerol and induction by methanol. As a result, the comprehensive model developed in this work is capable of emulating all the features of the fermentation with the same set of kinetic parameters unlike some reported work wherein different models are proposed for different phases of fermentation.

A comparison of N-glycosylation pathway in Human and P. pastoris

In vivo analysis

Pichia can perform post-translational modifications and hence has been used for expression of glycosylated proteins. However, the N-glycosyalation in yeast is of the high mannose type and varies greatly from the hybrid type of glycoproteins synthesized by the cells of the higher eukaryotic systems. The high mannose structures attached to the Pichia expressed proteins can affect its proper folding and activity. Also the non-human nature of these glycans restrict its therapeutic use, as the recombinant proteins are cleared from the system by the mannose receptors of the endothelial cells and macrophages and hence have a reduced serum half-life. We have re-engineered the Pichia N-glycosylation pathway to mimic the human type of N-glycosylation. This begins with the disruption of the inherent reactions in yeast leading to the formation of high mannose structures. A OCH1 strain of P. pastoris lacking the enzyme och1p transferase, which catalyzes the first step in hypermannosylation was used as a starter strain. Further, to replicate the reactions occurring in higher eukaryotes, we introduced enzymes from various eukaryotic sources into the system. By the use of combinatorial libraries, where the catalytic domains of the enzymes are fused to yeast derived signal sequences, proper targeting of these enzymes to the intra-cellular organelles was achieved. These modifications result in the conversion of the yeast Man9-20GlcNAc2 glycan structure to a more human like GlcNAc2Man3GlcNAc2 form. In order to demonstrate the application of these strains as efficient protein expression hosts, the glycoengineerd Pichia was used for large scale expression of the glycoprotein hormones, Follicle Stimulating Hormone (FSH) and human Chorionic Gonadotropin (hCG). The hormone sub-units were cloned, transformed and expressed in the engineeredPichia pastoris. Upon validating the ability of the engineered Pichia to produce functional hormone dimers, the proteins were expressed and puried in large scale bioreactors. l hormones. The glycan profile of these hormones showed the presence of human like complex and hybrid glycans. These recombinant hormones were also able to elicit responses in animal models, demonstrating novel therapeutic potential of the Pichia expressed glycoproteins. Thus, we report the generation of a glycoengineered Pichia pastoris, which can be considered as a serious contender for the expression of glycosylated proteins of therapeutic importance.

Computational Fluid Dynamics Modeling of Bioreactors

Computation fluid dynamics has emerged as an important tool for understanding of mass-energy-momentum interactions in multiphase reactors. One of the research area of interest to us is extending the application of CFD to enhancing the understanding of biotic-abiotic interactions in bioreactor. We are investigating several systems such as anaerobic wastewater treatment processes and photo-bioreactors for algal growth.

Anaerobic Bioreactor for Coffee Plantation Effluent

Contours of liquid velocity in waste water treatment bioreactor

Contours of X-direction mean velocity (a) Experimental and (b ) Simulation

A large amount of effluent is produced in the coffee processing industry. This effluent generated during the pulping process has a high organic content (BOD 10 g/l to 15 g/l) and is highly acidic, about 3.8 pH. The presence high organic matter may lead to excessive depletion of dissolved oxygen, which along with the conditions of low pH causes immensedamage to aquatic life. An anaerobic bioreactor was designed and successfully deployed (more than 90% BOD removal) based on the preliminary results in the laboratory. Study of the interaction between fluid flow and biological reaction kinetics in ananaerobic bioreactor is essential to improve its performance. CFD modeling approach allows us to combine these effects. Basic objective of this project is to integrate the fluid flow with the reaction kinetics model so as to predict the actual conditions and hence to obtain a better reactor design if possible. fluent. A two phase packed bed model for anaerobic bioreactor the Solid biomass is developed and a complete 2D CFD analysis was done for the system. m. The CFD results shows the presence of circulations in the liquidregion and a negligible fluid velocity in more than half of bed. The short circuiting of the fluid was also quite evident. These results were validated with an experimental setup made of glass and perplex sheet. PIV technique was used for visualizing the flow profile near inlet and the CFD results were found to be inreasonable agreement with the experimental value. The effect of baffle and its position is studied and an enhancement in conversion was seen for baffled reactor width on flow patterns was studied. A system with three equispaced baffle was found to be the most effective. A comparaive study of 3D simulation of existing reactor and the best baffle configuration was done and in the later case a better distribution of fluid velocity in the bed region is seen. An hybrid CFD based multizonal approach was used to simulate the transient profiles of concentration inside the reactor. While the complete CFD solution would have taken computational resources, this hybrid approach is quite fast. An algorithm for mapping flow data from CFD to zones was developed. A structured model for anaerobic digestion taking into account complex set of biological reactions is solved. Initial stage rapid build up of acids can lead the system to a low pH condition which is undesirable. The model explicates the importance of pH in anaerobic digestion process. Simulations results suggests reactor influent to be pretreated with suitable alkaline medium so as avoid low pH conditions which can lead to reactor failure.

Algal Growth in Raceway Ponds

Contours of liquid velocity in original raceway pond

Raceway pond with 2 delta wings and deflectors.

Distribution of light intensity along the depth of the raceway pond.

Distribution algal biomass in the raceway pond.

The urgent need for substantive reduction in CO2 emissions to the atmosphere can be potentially addressed via biological CO2 sequestration, coupled with its conversion to value added food, feed and fuel-grade products. Microalgae have recently attracted a great deal of attention for CO2 fixation and utilisation of biomass for food, feed and fuel production because they can convert CO2 into biomass via photosynthesis at much higher rates than conventional land-based crops. We have undertaken modeling CO2 sequestration and conversion to value added products (food, feed and fuel) in algal photobioreactors to develop a functional relationship between input and output process parameters of algal photobioreactors. The input parameters will include light intensity, temperature (diurnal fluctuation of solar radiation and temperature in case of open raceway), pond depth and flow rate (in case of open raceway), CO2 concentration, reactor hydrodynamics, mass transfer; kinetics; while the output parameters will include CO2 sequestration (algal growth), value added product yield and productivity. The major emphasis has been on Computational Fluid Dynamics (CFD) modelling to optimise the process and structure parameters of open and closed photobioreactors.

In raceway ponds uniform circulation of nutrients and CO2 and exposure of sunlight for the algal cells is required. However higher mixing rates damage algae cell membrane. Flow profile of the Raceway ond is obtained by using CFD revealed presence of large patches of dead zones (regions with : velocity < 0.1 m/s) and high circulations (non-uniform distribution of flow) in the raceway pond which are highly undesired. In order to improve the mixing. We tested design of delta wings in the raceway pond. CFD simulations of improved design showed a) complete elimination of dead zones in the curvature sections, b) more uniform distribution of flow velocities in straighter channels. For qualitative analysis, mixing studies are performed by solving tracer transport equation for both the reactor designs which were operated equal power input. Results reveal that the presence of two delta wings in the domain has significant localized impact but not much to the overall mixing in the reactor. Further quantification can be performed by analyzing the influence of these mixing characteristics on the growth of microalgae in the two reactors. As the time scale required for fluid flow in the reactor (raceway pond) to attain a quasi-steady state (i.e, physics of the domain) IS around few minutes whereas the growth of algae takes place over period of 2-3 months. So, it is nearly impossible to simulate the CFD code at shorter time steps with the existing computational resources. This lead us to follow a two stage simulation developed, in which initially physics of the system is allowed to converge with shorter time steps and then kinetic equations for the growth of algae is solved with larger time steps.

Optimal design and control of fermentation processes is a challenging dynamic control problem. We have carried out pioneering work in the area of optimization fed-batch bio-reactor system. Our research in this area focuses on developing computationally efficient algorithms, which combine the rigour of optimal control theory and advanced search techniques such as genetic algorithm.

Understanding optimal feeding policies for fed-batch fermentation.

Optimal Feed Rate Profiles

Optimisation of Penicillin Fermentation.

General characteristics of the optimal feed rate profiles have been deduced for various fed-batch fermentation processes by analyzing singular controls and singular arcs. The optimal control sequences depend on the shapes of the specific growth and product formation rates μ and π , and the initial conditions. For fed-batch processes described by four mass balance equations, the most general optimal control sequence consists of a period of maxmum feed rate, a period of minimum feed rate (a batch period), a period of singular feed rate (variable and intermediate), and a batch period. Degenerate sequences in which one or more periods are missing can result with a particular set of initial conditions. If the fermentation time is not critical, the singular control maximizes the net yield of product and only when the time is also important, it balances a trade off between the yield of product and the specific growth rate which dictates the fermentation time. With the sequence of optimal control known, the optimal feed rate profile determination is reduced to a problem of determining switching times. Based upon the general characteristics of the optimal feed rate profiles, efficient computational algorithms have been developed for fedbatch fermentation processes described by four or less mass balance equations. These algorithms make computations of optimal substrate feed rate profiles straight forward and simple for various fed-batch cultures for such products as antibiotics, amino acids, enzymes, alcohols, and cell mass.

Genetic Algorithm and Neural Network Based Optimal Control

Population chromosome for fed-batch fermentation

Evolution of feeing pattern

The rigorous application of the optimal control theory for complex fermentation processes represented by more than four mass balances is a challenging task. An optimisation procedure based on genetic algorithms is developed for the determination of substrate feeding policies for such complex processes. The proposed algorithm combines genetic algorithm with certain knowledge generated through the rigorous applications of the optimal control theory. According to the optimal control theory, the optimal feeding sequence consists of feed at the maximum, minimum or intermediate (during the singular interval) levels. Furthermore, the substrate concentration during the singular interval is maintained constant for a limited class of problems only, and in general, the substrate concentration varies during the singular interval. This problem specific knowledge (domain knowledge) is incorporated in formulating the decision variables and developing a novel hybrid genetic algorithm for optimisation of fed-batch bioreactors. The feed rate pattern is represented by an integer set {1,2, 3} and the determination of feed rate sequence is thus formulated as a pattern identification (sequences such as 2321, etc.) problem. The switching times between different patterns are also treated as decision variables. The feed rate in the singular interval has been expressed as a nonlincar function of state variables and few (two or three) unknown parameters. Thus, the assumption of piece-wise constant or linear function for feed rate is completely avoided. Con sequently, the number of search variables is significantly reduced. The optimal evolution of the unknown parameters enable us to determine the singular interval feed rate a nonlinear feedback control law expressed in terms of state variables alone. Since the decision variables involved are both integers (pattern identification) and real numbers (switching times and feed rate correction), a specialized mixed discrete-continuous type chromosome has been used. This mixed representation requires combination of different sets of crossover and mutation operators. An order maintain ing mechanism for switching times is also included. As : result of all these features, the proposed algorithm yields the optimal feed rate as a continuous function of ime. All the key features of the optimal feeding policy, such as, switching structure and characteristics of singular interval, which result from the rigorous application of the optimal control theory are retained. This is possible because of the proper choice of decision variables that we use in our GA-based search method, namely, he pattern of feeding (switching structure), the switching times, and the singular interval feed rate as a feedback control law. In this respect, the proposed method is different from other search based methods including those employing genetic algorithms.

Multiobjective optimisation of Pichia pastoris

Pareto optimal from for hCG production

Many optimal control problems are characterized by their multiple performance measures that are often noncommensurable and competing with each other. The presence of multiple objectives in a problem usually give rise to a set of optimal solutions, largely known as Pareto-optimal solutions. Evolutionary algorithms have been recognized to be well suited for multi-objective optimization because of their capability evolve a set of nondominated solutions distributed along the Pareto front. This has led to the development of many evolutionary multi-objective optimization algorithms among which Nondominated Sorting Genetic Algorithm (NSGA and its enhanced version NSGA-I) has been found effective solving wide variety of problems. We have developed a genetic algorithm based technique for solving dynamic single-objective optimization problems, with single as well as multiple control variables, that appear in fed-batch bioreactor applications. The purpose of this study is to extend this methodology for solution of multi-objective optimal control problems under the framework of NSGA-1I. The applicability of the technique is illustrated by solving two optimal control problems, taken from literature, which have usually been solved by several methods as single-objective dynamic optimization problems. The results the proposed algorithm can evolve optimal control profiles which result in an acceptable compromise between various (and possibly conflicting) objectives. Further, the algorithm is also capable ‹ quickly identifying and sustaining the entire Pareto domain which provides the decision maker with more immediate information about the optimal operating regime of the process. Since different Pareto-optima! solutions have different properties, the inspection of the entire Pareto front helps the decision maker in deciding which specifications of the performance objective can be achieved and which cannot, and how the specifications could be traded off in order to find a suitable operating condition for the process. Although the proposed method has been demonstrated for optimal control problems with only objective functions, it can be used directly for problems with more than two objective functions. We have experimentally demonstrated the benefits of multiobjective optimisation for hCG production by Pichia pastoris.

A three-layer feed-forward ANN architecture

Schematic representation of fed-batch process in discrete time domain

The approaches mentioned above works well when the fermentation model is available. An algorithm using feedforward neural network model for determining optimal substrate feeding policies or fed-batch fermentation process is developed. The algorithm involves developing the neural network model of the process using the sampled data. The trained neural network model in turn is used for optimization purposes. The advantages of this technique is that optimization can achieved without detailed kinetic model of the process and the computation of gradient of objective function with respect to control variable is straightforward. We have extended this approach for solving nonlinear optimal control problems, The universal function approximation capability of a three-layer feedforward neural network has been combined with a simulated annealing algorithm to develop a simple yet elicient hybrid optimisation algorithm determine optimal control profiles. The ability, feedforward neural network with a single hidden layer to approximate any arbitrary function has been exploited and the optimal control problem transformed into a nonlincar programming problem where the decision variables are the weights and biases of the networks. Simulation studies show that not many neurons in the hidden layer are necessary for a reasonably good approximation of the control profiles. The applicability of the proposed ANNSA method has been demonstrated by solving four well-known challenging optimal control problems from the literature. The results obtained for the case studies considered are in excellent agreement with the published results. The computational difficulties associated with the solution of singular optimal control problems can be very easily avoided as the ANNSA method can clearly identify the singular ares in the optimal control problems. Also, the efficiency of the method is not dependent on the choice of initial guesses of the deci sion variables. The present approach is very flexible and can easily solve both multicontrol and free final time problems.

Microbial contamination in water is a global issue which is escalating with severe health effects every year. Efforts made for safe drinking water access are not efficiently translating into reduction of the mortality rate worldwide. Therefore, considering the fact that poor countries like India, has to implement some cost and energy efficient measures for water treatment systems and improve their reach and awareness at every level. The statistic suggests that more than 1 billion people fetch their daily needs of water through unsafe water sources. In recent years photocatalysis has been used in wide variety of applications such as water purification, air purification, hydrogen generation, degradation of methanol, oxidation of H2S, oxidation of toluene and trichloroethylene, CO2 conversion, self-cleaning glass coating, paint coatings etc. Our studies focus on microbial inactivation using photocatalysis.

Inactivation Using TiO2

Schematic representation photo-bioreactor

Schematic of radical generations in a semiconductor.

Photocatalysis in presence of anatase TiO2 catalyst has shown to possess an enhanced capability to remove a wide range of contaminants. There are several methods to synthesize TiO2 however, solution combustion synthesis is a single step process to produce pure anatase phase TiO2. The catalyst produced by this method has been shown to be superior to the commercially available Degussa P-25 catalyst for the degradation of various chemical contaminants. The present investigation focuses on the use of combustion synthesized catalyst for the inactivation of microorganisms. The photocatalytic activity was compared with commercial Degussa P-25 catalyst. We have developed combustion synthesized TiO2 (CS-TiO2), 1% Ag substituted TiO2 (Ag/TiO2 (Sub)) and 1% Ag impregnated CS-TiO2 (Ag/TiO2 (Imp)). The results demonstrate higher photocatalytic activity of all the combustion synthesized catalysts than commercial Degussa P-25 catalyst. The optimum catalyst concentration was 0.25 g/L of Ag/TiO2 (Imp) catalyst. Rapid and complete inactivation of the microorganisms was observed at lower initial cell concentrations. A reduced photocatalytic inactivation was observed in presence of various anions and cations. Addition of H2O2 was observed to improve the photocatalytic inactivation. pH of the solution was observed to have no significant effect on the photocatalytic inactivation. The present study also demonstrates the potential use of catalyst immobilized thin films by LbL technique for the photocatalytic inactivation of E. coli in the presence of UV light. The effect of various operating parameters on inactivation has been investigated. The work also focused attention towards understanding the microorganism inactivation mechanism and kinetic aspects. The kinetics of inactivation was studied by various models available in literature. It was observed that power-law based kinetic model showed good agreement with the experimental data. A mechanistic Langmuir-Hinshelwood type model was also observed to model the inactivation reactions with certain assumptions.

Strategies for extended absorption for efficient designs of photocatalysts

Narrowing of band gap

. Schematics of possible photocatalytic mechanisms (a) type-1 (b) type-2 (c) Z-scheme (d) Z-scheme/type-2.

Commercially available catalysts such as TiO2, ZnO are wide band gap semiconductors and are highly efficient under UV irradiation. However, their wide band gap limits their viable usage in visible/ longer wavelength light. Due to limited access of UV wavelength from solar light (3-4 %), there is extensive need to design visible light responsive semiconductors or composites. Lowering the band gap, alteration in charge dynamics of the semiconductors and usage of novel lower band gap semiconductors are the ways to improve the photo response of the semiconductors. This study contains all the possible aspects of lowering of band gap by doping and charge dynamics alteration by band engineering with Type-1, Type-2 and Z-scheme. Metal substitution of Cu and interstitial doping of N in ZnO lattice and drastic improvement in optical properties were studied for inactivation of susceptible E. coli. Morphological aspect of the metal oxides is also a very crucial parameter in charge dynamics of the excitons, in this regard, ZnO/CdS/Ag nanorods and nanoparticles were synthesized to understand the photocatalytic mechanism involving surface plasmon effect due to the presence of Ag impregnation. Combined effect of doping and impregnation was also analysed by doping of Fe due effective coupling of orbitals due to its half-filled 3d orbitals. Ag is a widely used expensive antimicrobial agent. Therefore, CuO was introduced to increase the cost effectiveness with excellent photocatalytic properties. Leaching of metal ions from the semiconductor reduces the repeatability and stability of the catalysts. Therefore, metal free semiconductor C3N4 was coupled with CuWO4 in order to understand the Z-scheme mechanism of photocatalysis which is analogous to the photosynthesis. This Z-scheme composite was exploited for simultaneous inactivation of gram positive and gram-negative bacteria and extensive analysis of kinetics was studied for both the scenarios. These ways increase the charge separation, diffusion length and the carrier lifetime. In order to address the problem of recovery and the separation of the catalyst particles in slurry form immobilization of the catalyst on substrates such as FTO, ITO, glass slide, cellulose acetate and so on can be performed. Vertically aligned ZnO nanorods coupled with CuI were grown on a conducting substrate (FTO) to augment the photo response in visible light. These substrates with grown catalysts were used as working electrodes for photo-electro-catalysis. A model was derived considering all the possible aspects of interaction of catalysts with the pollutants for various cases such as electrolysis, photolysis, electrocatalysis, photocatalysis and photoelectrocatalysis. Kinetics and mechanistic aspects of all the processes were touched and explored for simultaneous inactivation of drug (chloramphenicol) and susceptible/ antibiotic resistant bacteria (E. coli).

Various types of biomass residues such as leafy waste, domestic solid waste, are potential source of energy if they can be efficiently converted to biogas. Unfavourable physical properties - biomass particles generally have a lower density than the digester liquid or acquire it as soon as biogas bubbles adhere to them, render conventional biogas reactors such as slurry reactors unsuitable for the treatment of biomass. Pre-treatment alternatives are costly. We, therefore, are focusing of using solid state fermentation processes and bioreactors for the treatment of biomass residues. The process of transformation of polysaccharide from biomass to biogas consists of two steps: acidogenesis, polysaccharide to acids and methanogenesis, acids to methane. Each step requires different type of bacteria.

Solid-state Stratified Bed (SSB) fermentation

Schematic representation of solid-state stratified bed biogas digestor

Long term gas production pattern in a SSB pilot plant.

Attempts at minimising digester liquid to remove the root cause of floating (firstly) and the anaerobic pre-treatment approach throw up key process questions, namely the role of digester liquid in a biomass-based biogas reactor. It is known that fermentation in landfills need very little water, and also with no water content there can be no stratification, scum or floating. A landfill type process thus escapes floating and scum formation with biomass feedstock. The major issue however is how to increase the slow reaction rates of a landfill i.e. bring down the solid residence time (SRT) from 5000d to about 35d. Our biogas fermentor process and design overcomes these above mentioned limitations in biogas fermentation of solid biomass residues. The floating nature of biomass feedstocks (low initial density adhering biogas bubbles induced), the tendency for organic acid flux during initial stages of fermentation and the need to maintain anoxic conditions while feeding and removing digested material are three key problems to be overcome during anaerobic digestion of solid biomass residues. Fresh biomass feed is added to the top of the decomposing biomass bed and digested biomass is withdrawn from the outlet below as a result partially digested biomass between 10-35d SRT becomes the methanogenic zone. Fermenting biomass as a vertically oriented moist but reasonably firm solid bed avoids floating problem posed by biomass feedstocks in slurry type fermentors. The constant weight of decomposing bed above each layer as well as softening of the plant tissue during decomposition expel biogas bubbles adhering and compacts the biomass. Feeding biomass feedstocks from the top and sprinkling the biomass bed with 3-5% volume of digester liquid initiates decomposition process. This facilitates a top fed decomposing biomass bed to be predominantly in an acidogenic stage at the top of the bed. Following the flux of organic acids there is a gradual methanogenic biofilm formation on older and partially digested biomass. As organic acids leach down to methanogen colonized biofilms below from the acidogenic zone above, they are converted to biogas. Venting the biogas produced from this zone itself to a gas storage device ensures that the gas quality is always ideal and is not diluted by air or nitrogen entering with feedstocks during the feeding operation. The biomass feedstocks undergo significant compaction and initial bulk densities of 60-200/m3 increases rapidly to reach up to 600 kg/m3. Digested biomass is removed from outlets placed below. This as well constant compaction of feed provides room for fresh material to be fed at the top of the bed. This allows quasi-continuous operation.

Plug Flow Biogas Digestor

Schematic representation of plug flow digestor

Long term gas production pattern plug flow biogas digester

Experimentation with anaerobic digestion of biomass under forced submergence revealed that most of the biomass feedstock suffered a rapid initial decomposition phase lasting 3-5 days. During this rapid fermentation stage over 30% of the organic matter (measured as volatile solids, VS) is converted to produce a collection of simple 2-6 carbon containing volatile fatty acids (VFA) intermediates. In these systems, concentrations of VFA at greater than 6 g/l inhibit their subsequent conversion to biogas. In case biomass feedstock was forcibly placed under digester liquid for this period, VFA over-production quickly diffused into the digester liquid surrounding it without seriously suppressing methanogen colonisation on this feedstock as well as achieving normal biogas production rates in the latter stages of decomposition. After this initial decomposition stage the biomass feedstocks acquire higher methanogenic rates that match or exceed acidogenic rates and therefore biogas is produced without serious impediments. This happens even in the absence of the feed stock being submerged in digester liquid. These observations gave the explanations necessary to design and operate larger fermentors designed to hold biomass feedstock submerged only for an initial period of 3-4 days after which feedstocks were free to move horizontally, in a partially floating state, towards an outlet placed at the opposite end. During this second phase, decomposition rates gradually fall while feedstocks acquired densities of up to 0.95 g/cc within a fermentation period of 30-35 days. However, it remained afloat throughout its useful stay in the digester. Studies of the biomass profile recorded during normal operation or while using biomass fed in marked bags bore out this flow pattern. This then necessitated designing wide outlets for manual removal of spent feedstock. Today many plants working on this principle have now been built and operated over a long period. The biomass is pushed through one end of the digestor wherein it remains submerged in liquid for about 3-5 days. Pushing the biomass in the digestor causes the horizontal movement of the feed. The buoyancy forces tend to lift the biomass which is opposed by the previously fed biomass which is compacted as a result of decomposition. As a result, the horizontal movement of biomass during daily feeding results into plug flow (pockets of biomass) type displacement of fluid. We have further adopted the plug flow bioreactor for waste water effluent treatment.

Dissolution of metals from ores, removal of metal ions from effluents and generation of acidic waters at mine sites are some of the processes which result from mineral-microbe interactions. In the area of mineral biotechnology, we have investigated several processes which result from mineral-microbe interactions, namely, microbial ecology of gold ore and bauxite deposits, dissolution of metals from minerals and ores, removal of metal ions from effluents and generation of acidic waters at mine sites. Our research focuses on understanding the mineral-microbe interactions and developing suitable mathematical models for quantification of these processes.

Bioleaching of minerals and ores

Thiobaccilus ferroxidans attached to sulphide mineral

Leaching rate as a function of particle size

We have clearly brought out the importance of direct microbial attachment to sulphide minerals. Increase in leaching rates with particle size is one of the most unexpected and hitherto unreported finding of our work. Our investigations have clearly brought out the dynamic nature of mineral-microbe interactions, that is, both mineral surface and microorganism properties such as hydrophobic, surface charge and composition undergo a change when they come in contact with each other. These changes are shown to result in enhanced attachment of bacteria to the mineral surface, higher metal tolerance and hence higher leaching rates. One of the impediments in commercial utilization of biomineral technology is the slow kinetics of the processes. We have investigated various strategies for improving the kinetics of the bioleaching process. Serial subculturing is shown to increase the tolerance of the microorganisms to high dissolved metal ion concentrations as well as high solid loading. However, such an acquired tolerance is shown to be stress sensitive. In the absence of stress (high metal concentration), the organisms lose their tolerance. Immobilization of bacteria in polyurethane foam is shown to be a simple technique for enhancing the bioleaching kinetics. Apart from bioleaching, we have also investigated removal of calcium from bauxite ores. Our results have showed that both direct attachment of bacteria as well as leaching by metabolites is responsible for leaching of calcium from bauxite. An equilibrium model is developed for predicting the solubility of calcium in the metabolite solution. Ocean nodules are a rich source of cobalt and nickel and processing of these nodules has been studied by various routes. But none have been successful. We have isolated a microorganism from the nodules and used the organism and its byproducts for leaching the metals from the nodule. The organism has been characterized to be a halophile , which grows rapidly in artificial seawater nutrient broth at room temperature. In a single stage leaching process using the cell free supernatant, we are able to leach about 40% of cobalt and about 15-20% of copper and nickel within 4 hours. When the residue of the first stage leaching was leached with a fresh metabolite, similar leaching efficiencies were obtained. The cumulative leaching efficiencies are far superior compared to conventional mineral acid leaching. Most importantly, the leaching with the microorganism is carried out in the neutral pH range (pH 7.5), which makes it very attractive when compared to acid leaching. The metabolite produces a reducing environment and also contains siderophore like phenolic compound/compounds with an attached carboxyl group that might form soluble organic complexes with the metals. In the presence of organic reducing agents such as ascorbic acid and salicylic acid along with the metabolite, the leaching of the metal values could be enhanced to more than 60%.

Biosorption of metal ions

Modeling framework for biosorption

Biological removal of toxic metal ions is another area in which microbe-mineral interactions are gainfully utilized. We have investigated removal of several metal ions such as nickel, iron, calcium, chromium etc. using waste fungal biomass from fermentation industry. Various important parameters which influence the biosorption have been identified and mechanisms for biosorption proposed. We have shown that biosorption with living cells can completely remove metal ions from solution. However, processes with living cells are time consuming and costly, and therefore, biosorption with nonliving cells have been investigated in detail. For successful scale up of the process, modeling is an essential component. We have developed a framework for mathematical modeling for such processes by combining statistical thermodynamic principles of adsorption with double layer theories of electrostatic interactions. We are able to capture some of the inherent characteristics of biosorption such as effect of pH and metal ion concentration using such a fundamental approach

Demonstration Bioreactor Plant for Hutti Gold Mines Limited, Hutti, Karnataka

Continuous bioreactor assembly at Hutti Gold Mines, Karnataka

Gold, a precious noble metal, occurs in nature in its native form and usually associated with quartz. Such ores, referred to as free milling ores, are treated in a conventional process with cyanide to solubilize gold-cyanide complex. The gold is then recovered from the solution by using either zinc (old technology) or more recent, carbon-in-pulp technology. With extensive mining activities over centuries, the reserves of such free milling ores are fast depleting worldwide. Increasingly, the gold ores are found to be refractory in nature, that is, gold is associated with sulphides such as pyrite and arsenopyrite. In early nineties, The Hutti Gold Mines Company Limited (HGML) faced with dwindling grade free milling ores and ex-cavations of sulphidic zones, recognized the need for exploring new technology for gold processing. Initial collaborative research project sponsored by HGML dealt with exploring the feasibility of biotechnological route for gold processing. These projects were essentially a laboratory scale study of investigating microbial ecology of Hutti Gold Mines and establishing a process for biotreatment of sulphidic ores. Having successfully demonstrated the technology at laboratory level, a need was felt to scale up this process and demonstrate it on a larger scale. In collaboration with Department of Materials Engineering-IISc, We have successfully setup a demonstration plant to process about 100 kg/day of sulphide gold bearing concentrate using Thiobacillus ferrooxidans. Without biooxidation the gold recovery was about 40%. But after biooxidation, the gold recovery increased to 90% while silver recovery increased to 95%. This technology has potential for increasing the production of gold by using some of the newly explored reserves in Karnataka.

High Throughput Waste Water Treatment Plant

Coffee plantation effluent treatment plant at Chikmagaluru, Karnataka

Coffee is a major commercial crop of the country. The processing of coffee fruits to remove bean results in an effluent, which is high in organic load. If untreated and discharged into streams, this causes many a health and environmental hazard. A technology is developed to treat these effluents as a part of collaborative research with Center for Sustainable Technology, IISc. The coffee pulping is done only in winter months from November to February. Constructing an effluent plant that only serves during pulping season is not likely to be acceptable, and therefore, a multipurpose biogas plant is designed. During the pulping season, system functions with coffee plantation effluent at high throughput while low-level operation using biomass feedstocks such as coffee skin, weeds, and fallen leafy litter is carried out in the remainder of the year. We have shown that the high degree of pollution caused to Karnataka's rivers by coffee processing wastewater could instead become a resource (energy /fuel) that may be recovered for local and environmental benefits. This process of resource recovery also brought down the pollution levels. Bioreactors are also low cost options. Bioreactors substituted the large and unwieldy anaerobic lagoons that spread over acres of prime estate land. These bioreactors cost about 10% of conventional anaerobic-aerobic lagoon costs on typical coffee plantations. It provided up to 80 m3 biogas (about 3 LPG cylinders) for every ton clean coffee produced. It removed > 85% pollution (COD/BOD) in a single step and this facility could be used throughout the year (with other solid biomass feedstocks when coffee processing season concluded). Under ideal conditions, it showed a potential to recover the cost within 2 years. This successful demonstration resulted in many plantations coming forward to use the bioreactors without subsidy or incentives. Currently, there are more than 20 such bioreactors in operation at different coffee plantations in Chickmangalore district of Karnataka. It was envisaged that these multi-feed bioreactors could also be utilized for processing segregated urban solid wastes (USW) with minor design modifications to reap similar benefits. Several organizations including Municipal Corporations and Industrial Units have shown interest in utilizing these reactors.

Online Optimisation of Rifamycin Fermentation

Schematic of an exemplary on-line optimization of fermenter unit.e

Though different approaches/algorithms are reported for optimizing the substrate feed rate in fed-batch fermentation processes, the methods do not address the requirements for on-line optimization of an industrial fed batch fermentation unit. The optimization schemes often used simplified fermenter models and have not addressed the problem of the time varying nature of the model parameters adequately, particularly during deployment of the methods on-line in industrial environment. The best way to address all these issues is to use a model that adequately represents the phenomena occurring in the fermenter and use non-linear optimization techniques to estimate the model parameters and calculation of the optimal feed rate of the substrate to maximize the product yield. This scheme of parameter estimation and optimization is carried out periodically on-line based on the plant measurements and laboratory analysis results. This ensures that the model used in the optimization calculations is close to the behavior of the real fermentation unit. Factors such as variations in the quality of raw materials, characteristics of the initial charge media and disturbances in process conditions lead to mismatch between the model and the actual plant, adversely effecting the performance of the fermenter optimization system. The best way to address this issue is to use non-linear optimization techniques for updating the model on-line and optimization of the Substrate feeding profile to maximize the product yield. A method for on-line optimization of a fed-batch fermentation unit comprising: on-line measurement of plant parameters such as agitator speed, airflow rate, level measurement, Sugar feed rate, percentage of carbon dioxide and oxygen in the vent gas and dissolved oxygen in the broth; storing of the on-line measurements/plant data as well as laboratory analysis results in a computer connected to the plant control system; fermenter model parameter re-estimation based on past and present plant data so as to reduce the mismatch between the plant data and the model calculation; on-line calculation of optimum Sugar feed rate based on the current plant data and prediction of fermenter's future behavior so as to maximize the product yield. We have successfully demonstrated this approach for industrial rifamycin fermentation process. Based on the data available from the industry, the variables included in the model are: (i) state variables: concentrations of cells, dextrose, rifamycin, dissolved oxygen, fermenter broth volume, off-gas oxygen and off-gas carbon dioxide, (ii) manipulated variables: dextrose feed rate, sterile water addition, broth withdrawal (spillage), ammonia addition rate and air flow rate, (iii) controlled variables: temperature and pH. A detailed model is subjected to off-line open loop optimization strategy. The implementation of the optimal feeding policy is coupled with online parameter estimation is achieved with the sampling data window, process re-optimizied and moving this window throughout fermentation. 10-15% improvement in product yield has been demonstrated.