Bioreactor control systems
During the existence of the BCL several applications of bioreactor control systems have been realised and implemented:
- CHP 3000 control system based on the SAPI 1 platform with an Intel 8080 microprocessor,
- application software of the CHEPOS 150 industrial control system,
- BIOGENES – knowledge-based control system (implementation on the Genesis software platform - Iconics),
- BIOGENES – knowledge-based control system (implementation on the InTouch software platform - Wonderware),
- BIOGENES II – multiagent knowledge-based control system.
The first knowledge-based control system BIOGENES© was a result of the
Knowledge-Based Control and Operation of Industrial Productive Bioprocesses international research project within the EU COPERNICUS programme. It was developed for the control of batch and fed-batch bioprocesses in bioreactors, both for laboratory and in industrial scale.
The system consists of two hierarchically arranged levels. The base level (classical control) realises all standard control tasks from data acquisition through process visualisation and direct digital control to data logging. The superior level (expert supervisory control) realises the sophisticated tasks of metabolic state classification and supervisory process control – inferring advice messages on possible corrective actions from the actual process control goal and state of the process.
The Knowledge-based supervisory control level is composed of expert system, which allows to use also the so-called heuristic (difficult to describe mathematically) knowledge of microbiologists, biotechnologists, and operators for control of a particular process. It is based on the concept of physiological modelling and control that assumes that the microbial culture can occur in different metabolic states and so the cultivation duration can be separated into several operational phases. The selection of an appropriate control strategy or its modification (a change of the corresponding setpoint or manipulated variable values or profiles) depends on the immediate assessment of the metabolic state, current process phase and the available set of process measurements. Physiological control allows a comprehensive assessment of the instantaneous characteristics of the microbial culture and the inference of decisions about the most appropriate control actions according to this information.
The BIOGENES system was implemented in the Genesis (Iconics, USA) process control software development environment and it was used for a research of control methods of following processes:
- baker's yeast production (Saccharomyces cerevisiae) on molasses wort,
- biosynthesis of ergosterol (provitamin D2) using Saccharomyces cerevisiae,
- xylitol preparation (non-caloric sweetener) using Debaromyces hansenii.
For a more detailed insight into the features and capabilities of the original system see operator communication with BIOGENES© system
This control system was also presented at the ICT booth at the CHEMTEC Prague '98 exhibition.
Subsequently the system had been entirely reprogrammed and transferred to Wonderware InTouch platform in the present multiagent version BIOGENES II, which consists of a distributed implementation in a computer network.
The basic level of real-time control including the tasks of process data measurement, conversion to physical units and regulation is implemented in a Compact programmable logic controller. Process visualisation is based on the InTouch environment with ActiveFactory application and for process data storage a real-time database (Wonderware Historian) is used, serving as a communication blackboard for individual agents running on separate computers. The agents include an expert system (Bioclips), on-line Matlab-based metabolic state estimator or an agent for processing of HPLC chromatograms. The agent layer represents the supervisory knowledge-based control level of the system.
BIOGENES II system allowed the research of the following processes:
- extracellular biopolymer xanthan gum production using the Xanthomonas campestris microorganisms,
- intracellular mcl-PHA biopolymers production using the Pseudomonas putida microorganisms.
Introduction of automation in sugar refineries
Our cooperation with the sugar industry dates back to the 80's of the last century. Starting from scratch, we have designed and built a process control microcomputer (hardware and software) for the control of the production line in the Lovosice sugar refinery, which was at the time the first successful implementation of its kind in Czechoslovakia.
Since the early nineties commercial control systems have been already available and thus it was possible to implement industrial applications on a professional level. We participated in a whole range of implementations, e.g.:
- sugar crystallisation control in boiling house of sugar refinery in Lovosice and
- sewage water treatment plant in sugar refinery in Uničov.
Mathematical models of technological processes
Work in this area is focused on creating mathematical models that are simple, but precise enough to be suitable for testing of different technological options and to predict the behaviour of processes and to design control strategies.
As an example, baker's yeast production model can be named. This is a fed-batch cultivation of Saccharomyces cerevisiae, where the feeding of substrate and the cooling of the bioreactor are controlled. The aim of simulation is to monitor the yield of the process and the duration of the process when different ways of process technological runs are applied.
We have been working on this issue for several years. One of the outcomes of our research is an electronic textbook - Modelling of Bioprocesses, that describes the basic principles of mathematical models and simulation of processes in biotechnology, engineering microbiology, ecology, and pharmacokinetics.
In the area of data processing a tool Graphical User Interface for Bioprocess States Visual Analysis was created. This tool serves for identification and visualisation of a multivariable biochemical processes. The application combines visualisation of process variables in time domain with 3D visualisation of a process trajectory obtained from the Principal Component Analysis (PCA) of selected physiological process variables. The application also can project the process classifications into the plot and it can be used for classification by selected centre points in a process trajectory space too.