Tero Eerikäinen

On-line biomass measurements in bioreactor cultivations: comparison study of two on-line probes.

Two on-line probes for biomass measurement in bioreactor cultivations were evaluated. One probe is based on near infrared (NIR) light absorption and the other on dielectric spectroscopy. The probes were used to monitor biomass production in cultivations of several different microorganisms. Differences in NIR probe response compared to off-line measurement methods revealed that the most significant factor affecting the response was cell shape. The NIR light absorption method is more developed and reliable for on-line in situ biomass estimation than dielectric spectroscopy. The NIR light absorption method is, however, of no significant use, when the cultivation medium is not clear, and especially in processes using adsorbents or solid matrix for the microorganism to grow on. The possibilities offered by dielectric spectroscopy are impressive, but the on-line probe technology needs to be improved.

Neural networks in extrusion process identification and control

Abstract Although neural networks have become one of the key research objects within artificial intelligence, relatively little information is available on neural networks related to food process control. The interest in such areas as dynamic modelling of food processes has increased, not least due to dramatic improvement and availability of the calculation methods and hardware. In the present case, flat bread extrusion was used as an example food process. Dynamic changes of torque, specific mechanical energy (SME) and pressure were identified (modelled) and controlled using two independently taught feed-forward artificial neural networks (ANN). SME, torque and pressure are system parameters which can be controlled with process parameters, such as feed moisture, mass feed rate and screw speed. Target parameters, such as product expansion index, bulk density, etc. are normally difficult to measure on-line, but can be estimated as functions of the system parameters. For the modelling of the whole flat bread extrusion cooking process a MIMO (multi input and multi output) approach was necessary. The neural network topology for the process model was 21-9-3 and for the controller 18-20-2. The process model was taught with 629 real data samples and the controller with 115 synthetic samples created with the process model. When testing the MIMO controller, the SME and pressure set points were quite well reached. One of the clear advantages of neural networks in the controller design is the ease of constructing a complex MIMO controller.

Cross-linked glucose isomerase crystals as a liquid chromatographic separation material

Cross-linked crystals of glucose isomerase (CLGI) were characterized as a liquid chromatographic separation material. The experiments were done with crystals having an average diameter of 83 µm. Porosity (epsilon(p)) and pore size distribution of the CLGI crystals were measured with size exclusion chromatography using D(2)O and polyethylene glycols as probes. CLGI material was capable of separating <1000 g/mol polyethylene glycols. Fifty two percent (epsilon(p) = 0.47) of the total crystal volume was in pores. Pore size measurement showed that CLGI crystals were microporous material, having an average apparent pore diameter of 29 +/- 0.08 Å. CLGI material separated n-alcohols C(1) to C(8) based on the hydrophobic interaction between the protein material and the carbon chain of the alcohols. Height equivalent to a theoretical plate (HETP, in millimeters) ranged from 1.6 to 0.89 for the C(1) to C(7) n-alcohol series. Despite the large crystal size, CLGI as a chirally active phase effectively separated D- and L-arabitol (R(s) = 0.58) and showed potential for chiral separation of amino acids.

Optimization of Streptomyces peucetius var. caesius N47 cultivation and ε-rhodomycinone production using experimental designs and response surface methods

Streptomyces peucetius var. caesius is an aerobic bacterium that produces doxorubicin as a secondary metabolite. A mixture design was applied for the screening of suitable complex medium components in the cultivation of S. peucetius var. caesius N47, which is an ε-rhodomycinone-accumulating mutant strain. ε-Rhodomycinone is a non-glycosylated precursor of doxorubicin. Best growth results were obtained with soy peptone and beef extract. A central composite face-centered (CCF) experimental design was constructed for the investigation of pH, temperature and dissolved oxygen (DO) effects on the cultivation growth phase. Another CCF was applied to the production phase to investigate the effects of aeration, pH, temperature and stirring rate on ε-rhodomycinone production. An increase in cultivation temperature increased both cell growth and glucose consumption rate. Best ε-rhodomycinone productivities were obtained in temperatures around 30°C. DO control increased all growth phase responses, but aeration in the production phase coupled with pH decrease resulted in rapid ε-rhodomycinone decay in the medium. In non-aerated production phases a pH change resulted in better productivity than in experiments without pH change. A pH increase with a temperature decrease seemed most beneficial for productivity. This implies that dynamic control strategies in batch production of ε-rhodomycinone could increase the overall process productivity.

Continuous propionic acid production with Propionibacterium acidipropionici immobilized in a novel xylan hydrogel matrix.

The cell immobilization potential of a novel xylan based disulfide-crosslinked hydrogel matrix reinforced with cellulose nanocrystals was studied with continuous cultivation of Propionibacterium acidipropionici using various dilution rates. The cells were immobilized to hydrogel beads suspended freely in the fermentation broth or else packed into a column connected to a stirred tank reactor. The maximum propionic acid productivity for the combined stirred tank and column was 0.88gL(-1)h(-1) and the maximum productivity for the column was determined to be 1.39gL(-1)h(-1). The maximum propionic acid titer for the combined system was 13.9gL(-1) with a dilution rate of 0.06h(-1). Dry cell density of 99.7gL(-1) was obtained within the column packed with hydrogel beads and productivity of 1.02gL(-1)h(-1) was maintained in the column even with the high circulation rate of 3.37h(-1).

Constraint-based genome-scale metabolic modeling of Clostridium acetobutylicum behavior in an immobilized column

In this study a step-wise optimization procedure was developed to predict solvent production using continuous ABE fermentation with immobilized cells. The modeling approach presented here utilizes previously published constraint-based metabolic model for Clostridium acetobutylicum without direct flux constraints. A recently developed flux ratio constraint method was adopted for the model. An experimental data set consisting of 25 experiments using different sugar mixtures as substrates and differing dilution rates was simulated successfully with the modeling approach. Converted to end product concentrations the mean absolute error for acetone was 0.31 g/l, for butanol 0.49 g/l, and for ethanol 0.17 g/l. The modeling approach was validated with another data set from similar experimental setup. The model errors for the validation data set was 0.24 g/l, 0.60 g/l, and 0.17 g/l for acetone, butanol, and ethanol, respectively.

The effect of temperature and pH gradients on Lactobacillus rhamnosus gene expression of stress-related genes

In this study, Lactobacillus rhamnosus, a renowned probiotic, was cultivated in fluctuating environment. Base gradients caused by a pH control in an industrial process and temperature gradients caused by uneven heating were simulated with a scale-down method. A pH gradient was created in a plug flow reactor (PFR). Expression of pH stress-related genes (atpA, aldB, cfa, groEL, hrcA and pstS) were studied as a relative gene expression study using ldhD as a reference gene. Expression measurements were carried out with the TRAC method. The responses of groEL, hrcA and atpA genes to temperature and pH changes were observed. The expression of phosphate uptake system-related pstS gene was induced almost linearly in the chemostat cultivation experiments when the base gradient in the PFR was increased. Correlations between the results from gene expression studies and freeze stability or acid stress survival were studied. However, by measuring the expression of these genes, we were not able to predict eventual freeze stability or survival from the acid stress test.

Carbon 13-Metabolic Flux Analysis derived constraint-based metabolic modelling of Clostridium acetobutylicum in stressed chemostat conditions

The metabolism of butanol producing bacteria Clostridium acetobutylicum was studied in chemostat with glucose limited conditions, butanol stimulus, and as a reference cultivation. COnstraint-Based Reconstruction and Analysis (COBRA) was applied using additional constraints from (13)C Metabolic Flux Analysis ((13)C-MFA) and experimental measurement results. A model consisting of 451 metabolites and 604 reactions was utilized in flux balance analysis (FBA). The stringency of the flux spaces considering different optimization objectives, i.e. growth rate maximization, ATP maintenance, and NADH/NADPH formation, for flux variance analysis (FVA) was studied in the different modelled conditions. Also a previously uncharacterized exopolysaccharide (EPS) produced by C. acetobutylicum was characterized on monosaccharide level. The major monosaccharide components of the EPS were 40n-% rhamnose, 34n-% glucose, 13n-% mannose, 10n-% galactose, and 2n-% arabinose. The EPS was studied to have butanol adsorbing property, 70(butanol)mg(EPS)g(-1) at 37°C.

The effects of pH oscillation on Lactobacillus rhamnosus batch cultivation

Inhomogeneous mixing in industrial-sized fermentation processes causes oscillations in process parameters such as temperature or pH value in the cultivation medium, which causes stress to the bacteria being cultivated. In this work, the impact of extracellular pH oscillations on the production of Lactobacillus rhamnosus, a well-studied probiotic bacteria, were investigated by means of a scale-down batch process, simulating inhomogeneous pH values by controlling the pH value of the medium on sinusoidal trajectories. Effects of pH stimulation on the bacteria were assessed by testing storage and freeze-drying stability of harvested cells, two factors relevant for the industrial process. Furthermore, gene expressions of six selected genes, i.e. atpA, fat, cfa, groEL, hrcA, and pstS, known to be related to stress response were monitored. Although storage stability is only slightly negatively affected by pH stimulation of the bacteria, gene expression of four of the studied genes, i.e. fat, hrcA, groEL, and pstS show to correlate with amplitude and frequency of the oscillation.

Developing a multipoint titration method with a variable dose implementation for anaerobic digestion monitoring

Determination of metabolites from an anaerobic digester with an acid base titration is considered as superior method for many reasons. This paper describes a practical at line compatible multipoint titration method. The titration procedure was improved by speed and data quality. A simple and novel control algorithm for estimating a variable titrant dose was derived for this purpose. This non-linear PI-controller like algorithm does not require any preliminary information from sample. Performance of this controller is superior compared to traditional linear PI-controllers. In addition, simplification for presenting polyprotic acids as a sum of multiple monoprotic acids is introduced along with a mathematical error examination. A method for inclusion of the ionic strength effect with stepwise iteration is shown. The titration model is presented with matrix notations enabling simple computation of all concentration estimates. All methods and algorithms are illustrated in the experimental part. A linear correlation better than 0.999 was obtained for both acetate and phosphate used as model compounds with slopes of 0.98 and 1.00 and average standard deviations of 0.6% and 0.8%, respectively. Furthermore, insensitivity of the presented method for overlapping buffer capacity curves was shown.