Towards web based Monitoring and Optimization of Microbial Fermentations – A Homemade Prototype. Gueguim Kana E B, Oloke , J. k. Zebaze Kana and Lateef A. Presented by Gueguim Kana E. B. Biotechnology Centre Ladoke Akintola University of Technology Nigeria ICTP, Trieste ITALY April 2007
Some Biotechnology products obtained through fermentation ICTP, Trieste ITALY April 2007
Complexity of Microbial Fermentations Unlike chemical processes bioprocesses are more complex due to : The optimal setpoints of a given process change with time. Some kinds of “randomly moving targets”(Gueguim kana et al. 2003 a). The same microbe can produce highly divergent products under a slightly modified environment. These processes generate large volume of data, which are poorly gathered and managed, but may contain useful hidden process information. ICTP, Trieste ITALY April 2007
Bioreactors &Bioprocesses Bioreactors offer a possibility to provide an optimally controlled environment for fermentation processes, a condition required for potential higher yield (Williams, 2002 Some controllable variables affecting the process are: • p. H • Temperature Agitation rate • Aeration • Substrate flow rate ICTP, Trieste ITALY April 2007
(The bioreactors) The capacity varies between 1 - 400, 000 litres) ICTP, Trieste ITALY April 2007
Aims/Objectives (3) 1 - Interface the various sensors and actuators of home implemented bioreactor for monitoring and control of fermentations. 2 -Design and implement a monitoring , control & optimization software bioreactor operation which functions in realtime and pseudo multitasking. 3 - Remotely access process data in realtime for client monitoring. ICTP, Trieste ITALY April 2007
4, 10 and 17 -litre implemented bioreactors The Control modules considered were: • Temperature control module • p. H control module • Substrate flow module • Agitation system • Aeration system ICTP, Trieste ITALY April 2007
ICTP, Trieste ITALY April 2007
PCL 818 L (Advantech USA. ) ICTP, Trieste ITALY April 2007
The p. H and DO sensors (Hanna Instruments) Donated by TWAS, trieste. Italy 12 volts Dynodrive motor from Junk yard Master Flex peristaltic pump (Cole Parmer) ICTP, Trieste ITALY April 2007
Signal amplification & actuator relay board ICTP, Trieste ITALY April 2007
The big control loop ICTP, Trieste ITALY April 2007
The feedback p. H control loop ICTP, Trieste ITALY April 2007
2. The monitoring and control software. Biopro_Optimizer The Computer software named Biopro_optimizer was developed to provide a real time control, monitoring and reproducibility required for research and optimization of fermentation processes. It incorporates control loop modules for p. H, Temperature, Aeration, Feed flow rate and dissolved oxygen. ICTP, Trieste ITALY April 2007
The control Software (main panel) ICTP, Trieste ITALY April 2007
The control software (Control panel) ICTP, Trieste ITALY April 2007
Process monitoring panel ICTP, Trieste ITALY April 2007
System experimentation on Baker’s Yeast Production Nine process batches of Saccharomyces cerevisae fermentation were run on the above machine using different feeding trajectories. A glucose medium was fed exponentially to the reactor. The control software was programmed to implement the exponential feeding equation below on the fermentation process. FFR(0)= X. V. µ/y. s. The computer automatically updated the control setpoints every 5 mins for 24 hours: All 5 mins Repeat FFR(t)=FFR(0). exp(µ. t) End. ICTP, Trieste ITALY April 2007
Let’s view the running process ICTP, Trieste ITALY April 2007
Process yield with the experimented feeding trajectory ICTP, Trieste ITALY April 2007
3. The Optimization module Genetic algorithm(GA) and Artificial Neural Network(ANN). The first submodule generates a set of potential optimal profiles which constitute the search space. 0. 001% of these are evaluated experimentally. The obtained output and input data are used to train and validate the ANN submodule which serves as evaluation function for the GA. The GA submodule randomly select few of the initial profiles and evaluates using ANN, performs genetic operations on the best profiles to produce the next generation. Generations evolve by iteration after genetic operations until an optimal profile emerges. A new feeding profile enhancing the yield value from 0. 99 to 1. 52 was generated. ICTP, Trieste ITALY April 2007
Genetic Algorithm optimization process ICTP, Trieste ITALY April 2007
ANN structure The ANN learns a nonlinear relationship on a process when the Input/ output process data are presented to it repeatedly, it does this by modifying the synaptic weights towards minimizing the error on the output to 0. ICTP, Trieste ITALY April 2007
GA configuration Panel ICTP, Trieste ITALY April 2007
Profile performance evolution ICTP, Trieste ITALY April 2007
Structure of the Remote Monitoring Sub. System The Control software populate the My. SQL Database module every minute on the stand alone computer. This machine runs an Apache server and a PHP interpreter. Client machines on the LAN use their browser to access the php scripts on the server end to fetch process data from the database server. ICTP, Trieste ITALY April 2007
Conclusion & prospects • The concept of Grid Technology applied to the Biotech industry , where process experts across the globe can share process resources will certainly enhance production process while reducing production cost. • Large volume of sampled data which were originally ignored may now reveal some useful transient phenomena across process time. ICTP, Trieste ITALY April 2007
Our web based bioprocess optimization service • www. pro-optimizer. net. • With this web service, researchers across the globe in different laboratories can share different optimization tasks for the same process using the Optimization Search Engine of the service. • Thus optimizing the same process at the same time but in different labs. ICTP, Trieste ITALY April 2007
Published papers on the present work : 1. Gueguim kana , E. B. , Oloke, J. K. and Lateef, 1 A. (2003 a). Construction of rugged 4. 5 -litre Bioreactor for the fermentation of Actinomycetes. African Scientist. 4(1): 1 -5. 2. Gueguim kana , E. B. , Oloke, J. K. , Lateef, A and Zebaze kana, M. G. (2003). Constructional features of 15 -litre homemade bioreactor fedbatch fermentations. African Journal of Biotechnology (AJB). 2(8): 233 -236 www. academicjournals. org/AJB 3 -Gueguim kana , E. B. , Oloke, J. K. , Zebaze kana, M. G. and Lateef, A (2004). Computer Software for Real time Monitoring and Control of Fermentations. Asian Journal of Microbiology, Biotechnology and Environment Sciences. ICTP, Trieste ITALY April 2007
I am highly grateful to: • Prof Carlos Cavka, Universidad nacional de san luis, Argentina. (ICTP associate). Prof. Induruwa A. S, University of Kent, U. K. (ICTP associate) • Prof B. O. Solomon, OAU , Nigeria • Prof J. K Oloke , Lautech , Nigeria • The ICTP Administration, trieste Italy. ICTP, Trieste ITALY April 2007
ICTP, Trieste ITALY April 2007
• ICTP, Trieste ITALY April 2007
Скачать презентацию Towards web based Monitoring and Optimization of Microbial
b13276fb8af92bd3a299ad3a1b74bc96.ppt