Bjarne A. Foss^#, Tor I. Eikaas^*, and Morten Hovd^#

#Department of Engineering Cybernetics, Norwegian University of Science and Technology, N-7491 Trondheim, Norway.

Key words: Remote laboratories, education, control engineering

This paper proposes to improve access to laboratory experiments in control education, by making remote experimentation accessible through the use of the Internet. It is proposed to form a network of providers of remotely operable laboratory experiments, and to delegate many of the technical, managerial and financial tasks involved in operating such a network to a commercial organizing company.

Further, the paper discusses the motivation for running remotely operated experiments as a part of an engineering education, and describes some typical control experiments which can be made available over the Web. Issues related to the technical implementation and quality assurance of experiments are also addressed.

Learning an engineering discipline is accomplished through a diverse set of activities, such as lectures, tutorial exercises, simulation, and pilot plant experiments. It is widely recognized that efficient learning requires a mixture of theoretical and practical exercises. One of the main features which sets engineering apart form the pure natural sciences, is that engineering is more or less exclusively focused on solving problems in the ‘real’ world. This real-world focus of the engineering disciplines makes practical exercises particularly important, in order to develop an understanding of how to apply theoretical knowledge to real-world problems; [1], [2].

During the last decades there has been a definite trend towards increased use of simulation in engineering education, coupled with a decline in the use of physical experiments. There are several reasons for this. One obvious reason is the fact that physical experiments are costly both to build and maintain. Another reason has been a strong belief that simulators can replace physical experiments. The ability to simulate a process is clearly very helpful in finding solution to many real-world problems. Knowledge of modeling and simulation are therefore important skills for an engineer. Nevertheless, it is the strong belief of the authors that physical experiments constitute an important ingredient in order to achieve efficient learning. Real-life processes, even at laboratory scale, display features that are not captured by simulators. Such features include:

• They are physical entities, which can be seen, and often also heard and touched. The typical student therefore finds it motivating to work with laboratory experiments. A successful laboratory experiment is some sense proof that the student has been able to perform a task which is of relevance to the real world.

The discussion above points to the need for making physical experiments more available to engineering students, and more affordable for educational institutions. This paper will report on one initiative which aims to achieve this, with the use of modern communication technology.

The Cyberlab initiative is based on the idea that although a physical process cannot easily be moved, it can be made available for on-line remote experimentation via the Internet. The Cyberlab initiative originates in part from professor emeritus J. G. Balchen in Trondheim, and is also inspired by related work in Switzerland and Germany, see [3], [4]. The providers of laboratory facilities will be paid for the use of their facilities, while the customers are relieved of the costs involved in building and maintaining the laboratory equipment. Both therefore stand to gain. This also enables a much more efficient use of laboratory facilities, since much laboratory equipment in universities is at present used for only a fraction of the academic year. A key aspect of the Cyberlab initiative is the realisation that a commercial company is needed for handling many important tasks:

• All aspects regarding booking and reservation of experiment time.
• Handling of invoicing and payments.
• Quality assurance of the experiments.
• Provide functional specifications and software components for remote experimentation.
• Marketing of the experiments that meet the defined quality standards and functional specifications.
• Provide the software necessary for bringing an experiment on-line.

The quality standards that the organizing company will have to ensure, should cover both the quality of the physical equipment and its instrumentation, the quality of the available interface for remote operation, as well as the educational value of the experiments that are available on the specific laboratory equipment. The concept is illustrated in Fig. 3.

The Cyberlab initiative has received a positive response from key control groups on both sides of the Atlantic [5]. The aim is to establish a network of providers of physical experiments. Each Provider may concentrate their effort on maintaining a few high-quality experiments for remote on-line experimentation, while having access to many other remote laboratories from other Providers. Further, other users, without their own laboratory facilities, may purchase access to remote laboratories. This latter category of users will include

• universities that have given up on practical experiments due to high costs,
• open universities like FernUniversität Hagen in Germany and Open University in the UK,
• commercial organizations offering continuing education courses, and
• in-house training departments in engineering and production companies.

There are a number of technological and practical issues that need to be resolved in order to achieve close to the same learning from running a remote experiment as from running the same experiment locally. Some of these issues are discussed below.

One issue that needs to be addressed is how to give the user the necessary feeling of interacting with real, physical equipment. The ability to manipulate the experiment remotely, and seeing the results of such manipulations in real time, will contribute to this aim. The feeling of being present at the experimental site can be further enhanced by the transmission of sound and video images from the experiment, and by the use of virtual-reality techniques.

On the other hand, the fact that the experiment is located far away from the experimenter can also be regarded as an educational asset. There are two reasons for this:

• Plants are usually operated ‘remotely’, by operators located in a control room some distance from the physical plant. Hence, running remote experiments actually reflects the real working conditions.
• The trend towards virtual organizations where an operating organization is located in different physical locations is quite apparent in many organizations. This trend can also be anticipated for the most highly skilled technical support staff in many industries. This means that the use of remotely operated experiments is likely to be quite representative of how many control engineers will work in the future.

The Provider will connect a laboratory experiment to the Internet through a Web service, and the user interacts with the experiment through a Web browser. The information received from the experiment will typically include time series plots (strip charts), and video and audio information. When the experiment is completed, it should also be possible to obtain comprehensive documentation of the experiment by downloading files consisting of time series of experimental results. Exactly which variables the user should be able to manipulate remotely, as well as when, how and how much the variables can be manipulated, will depend on the experimental design. Some typical examples will be described below.

Cyberlab.Org will specify key aspects of how the user should be able to interact with the process, and will assist the Provider in connecting the experiment to the Internet, if the Provider requires such assistance. However, all technical spec-ifications will be independent of the actual hardware and software the Provider uses. At present, there are at least two software packages that can be used for providing two-way interaction with an experiment over the Internet:

• A solution based on the APIS product suite from the Norwegian company Prediktor, and

• Labview from National Instruments.

There are several types of control experiments that can be run remotely. Some typical examples are:

• Experiments where the user acts as an operator for the laboratory process. Typically the user will change the setpoints for some loops and otherwise monitor the equipment. Further, the user may change some control parameter settings, such as gain and integral time for a PI controller. Such parameter changes may be based on intuition only (ad hoc tuning), or result from some sort of systematic tuning procedure (e.g. Ziegler-Nichols tuning). The experiment may be designed to introduce (modest or severe) disturbances, to make the operator function more challenging.
• The user may run identification experiments by perturbing the system and collecting process data. These may be used to identify an empirical dynamical model, to determine parameters in a 1^st principles model of the process, or to validate such a 1^st principles model. Both open loop and closed loop experiments may be run.
• The user may program a controller and download it to the control system, and then assess its performance by running experiments with this controller. This opens possibilities both for giving students more realistic experience with advanced control concepts, as well as to provide researchers with a testbed for new controller algorithms that is closer to industrial reality than what simulation alone can offer.

The different experiments support different concepts in control courses like controller tuning, model development, and the actual programming and implementation of a control algorithm.

In the following, we will illustrate some of the possibilities that exist for remote operation of laboratory equipment over the Internet, by using as an example a laboratory scale refrigeration process at the Engineering Cybernetics Department at the Norwegian University of Science and Technology. The refrigeration process consists of a compressor, condenser, expansion valve, and an evaporator, as shown schematically on the left hand side of Fig. 1. The process will cool down the liquid in a tank. The fluid in the tank is heated, which is the main disturbance.

The process is equipped with temperature, pressure, level and flow transmitters. This instrumentation is connected to a Simatic S7-400 PLC, which again is connected to an industry standard PC. The PC has software for communicating with the PLC, and for programming the PLC. In addition, Simatic WinCC is installed on the PC, and is used both for a local HMI and for communicating with a Web server through an OPC interface.

The Web server uses the APIS program suite from Prediktor to access process data from WinCC and make this data available over the Web. Over the Web it is at present possible to

• view a P&ID of the process,
• to show or hide specific types of transmitters to make it easier to focus on the most important information, and
• view time series of responses from selected transmitters in strip charts as shown in Fig. 2.

The use of Web technology also allows for a tight integration of tutorial material on the process or the control theory (or practice) which the experiments are designed to demonstrate. This is illustrated in Fig. 1, where we see a screen dump of a Web page showing the process layout. By pointing to key equipment in the process, a pressure-enthalpy diagram pops up, in which the refrigeration cycle is drawn in. It is also explained where the equipment pointed to is located in the pressure-enthalpy diagram.

The reader may wish to visit the web pages for this laboratory equipment. They are found at:

http://kybpc213.itk.ntnu.no/bisp/index.htm

Please note that most of the time there will not be any experiment running on the process, and this will obviously affect what data is available.

The company Cyberlab.Org has been founded as the organizing company for the Cyberlab initiative. The plans for the future include:

• develop the necessary functional specifications for remote experimentation,
• assemble a network of providers of high-quality control experiments, who make their experiments available for remote experimentation over the web, and
• establish Cyberlab.Org as the organizer of this network, taking care of the tasks that should be handled by a commercial company, as explained above.

The remote experimentation network shall be subject to a full-scale evaluation with respect to didactical, technological and commerical issues during 2000 and 2001 provided a EU-funded trial project including participants from Norway, Germany and Switzerland obtains the necessary funding. It should be noted that there is an intense interest from the EU towards innovative internet-based education concepts [6], [7], [8].

This paper argues that the Internet makes remote operation of laboratory equipment feasible, and that such remote operation can have high educational value if the experiments are properly designed. Remote experimentation has the potential for both making experiments on real physical equipment more affordable, and provide for more efficient use of laboratory facilities.

Fig. 1. Tutorial material on the process is integrated with drawings showing the process layout.

Fig. 2. Strip charts showing time series of process responses can be configured and displayed over the Web (apologies for the weak contrast in black and white).

Fig. 3. Network structure for remote online experimentation.