Tiina M. Komulainen

How extreme environments can impact the training of industrial operators

Extreme environments, either natural or artificial, can greatly affect the effectiveness and safety of operators. Scope of the paper is to provide a rather detailed list of extreme working environments and the effects produced on the workers in terms of disturbances, efforts, and misrepresentation of real operating conditions. These elements can influence the operators’ awareness and therefore call for the definition of dedicated training methods. The concluding part of the article discusses the recommended features that a training simulator for extreme environments should implement to improve the situational awareness and safety of operators.

Integrating Commercial Process Simulators into Engineering Courses

Abstract The aim of this article is to encourage university lecturers to utilize commercial process simulation software in engineering courses. This article provides examples on four different simulation modules which are implemented in chemical and electrical engineering courses at Oslo and Akershus University College. The students have given positive evaluation on the learning through the simulation exercises and considered the simulation tool easy to use. Together with the exam results the student feedback indicates enhanced learning outcome.This work describes a Simulink lab practice using Arduino as low cost hardware. A Shell must be developed in order to adapt Arduino signals to the real plant, consist of a DC motor. With Arduino architecture and with open hardware a cheap Data Acquisition card has been build. Several tests have been done to validate de full system and a frequency study has been completed in order to know the possibilities of the proposed architecture in the control of new other plants. Finally, a lab exercise using the Simulink and Arduino system is proposed to the student of engineering bachelors. The exercise consisting of apply the empirical method of Ziegler-Nichols for adjust a PID controller, testing this controller over the experimental plant.

Implementation of performance indicators for automatic assessment

Abstract Simulator training is widely used in different industries and in academia as a teaching tool. An expert instructor, who guides the users through the learning process with the simulator by giving verbal feedback, pausing the scenarios if necessary, and sometimes by developing certain evaluations, leads the simulation sessions. However, the availability of instructors is not always enough for the training demand, which can slow down the training process, as the trainees often need to wait for the instructor’s feedback before they can proceed with their tasks. This research evaluates the feasibility of combining the guidance of the instructor with an automatic live assessment tool, based on performance indicators, to improve simulator training. Two simulator training sessions were evaluated. The strategy used for the simulation sessions consists of three stages: briefing, simulation and debriefing. During both sessions, it was observed that the assessment tool makes the users aware of the state of the process and in most of the cases helps them to find faster the source of an abnormality in the system. However, further development of the automatic assessment tool is needed in order to require less instructor guidance during the simulator training.

Experimental design and probabilistic modeling of E. coli fermentation

Abstract In this paper the experimental design and modeling of E.coli ER2566 bacteria growth in a bioreactor are presented. The goal of the study is to explore the possibilities to use probabilistic models for bioreactor modeling. Flask tests were performed in conditions corresponding to a full 33 experimental design. Based on the flask test results, a 1/3 of the design matrix was chosen for the bioreactor experiments. The bacterial growth data was used to compare the proposed Markov chain model with the modified Grompertz model. The proposed modeling method combines biologically inspired Markov chain with probabilistic growth/no-growth functions. The performance of the Markov chain model was similar to the performance of the Grompertz model, giving a good basis for further modeling studies.