Full Paper

Kate Patrick and Sue Johnston

EPI Group, RMIT, Melbourne

The project which we describe set out to provide a computer-based learning package which would help students achieve a better understanding of heat transfer processes. Our initial intention was to produce a conventional CAL package, using simulated laboratory experiments with preprogrammed models of different heat exchangers, offering students multiple choices about how to grasp key concepts. While it was hoped that this package would be widely useful in courses where convective heat transfer is significant, the project was specifically intended to help engineering students develop a systems approach to convective heat transfer.

Serendipitously, however, the project has involved representatives from several disciplines, across the VET and Higher Education sectors. In the process of developing a shared vision of what we want the project to achieve, we have re-framed our view of how the package would work.

What we have ended up with is very different from a conventional CAL package. It is not structured as a directly instructional program; in fact it does not include any overt guidance or questions to the user. It is a game-like, manipulable set of simulated laboratory models which can be investigated and explored by a range of users, with or without external notes to prompt them. These models are directly presented: they are open and interesting, along the lines of StatPlay and SimCity, without the explanatory apparatus of a simulated setting replete with guides and conversations. They offer students a resource rather than an instructional experience. A working version of one of the models is available for anyone interested to explore at the Conference.

How did we manage to achieve agreement that this was what we wanted? What makes us think that it will help students to achieve a better understanding of heat transfer processes?

The project was initiated by an engineer and a multimedia courseware developer, who agreed that computer-assisted learning offered an improved way of teaching and learning about convective heating and cooling liquids and gases. Their proposal was to develop a module which

Key features in this original version of the project were:

• project to focus on key theoretical concepts/principles involved in convective heat transfer
• hypotheses to be presented to students
• students to be able to test these hypotheses interactively by changing parameters of the computer model
• visual material to consist of working animated diagrammatic models
• immediate and comprehensive feedback to be provided to students

The development process involved discussion and reconsideration of each of these features of the project.

The project's remit was expanded when funding was allocated to it, to ensure that the product would be used as widely as possible across departments, faculties, and both sectors of the university. The production team eventually included a mechanical engineer, a chemical and metallurgical engineer, a food scientist, a refrigeration teacher, three educational technology staff, and an educational adviser. We now had a mixture of disciplinary concerns and a range of students from VET through different levels of higher education.

Hence the first issue which faced the group was identifying the learning needs to which our package would respond. We had agreed that we would aim to develop a package which would meet the diverse needs of students in all of our subject areas. But what did these students need to know about convective heat transfer? What, indeed, did they know already? What ideas do students bring to the study of heat transfer?

First, we tried collecting questions which students were expected to answer.

These questions were extremely diverse. They did not share a clear focus, but they led us to discuss what students needed to know.

One of the group explored the constructivist literature; she found relatively few studies of students' conceptions of heat, and none involving tertiary students. We considered asking students to try qualitative problems and report their learning difficulties, but we decided that a full-scale qualitative research project would be needed if we were to identify students' underlying beliefs.

After six weeks of discussion around questions and problems, the notion of "key ideas" proved to be a breakthrough. Within a fortnight the subject experts had pooled their ideas and agreed on a list of key ideas about heat and heating processes. Two examples of these key ideas:

• Resistance to heat flow involves conduction and convection (forced and free) and the area across the heat flow.
• To model heat transfer, an engineer must be able to combine heat transfer coefficients, conduction, surface area and temperature differences.

Our eleven key ideas, a little reformulated over time, have formed the theoretical frame of reference for the project.

The initial concept of the project was instructional, in that hypotheses were to be presented for students to investigate. At the same time, students were to be active agents in the task.

Very early in the development process, the proposed instructional approach was questioned. Instead, we thought students should be involved in simulated practical tasks, using different tools and components to build virtual working models of increasing complexity. The aim would be for them to work towards an understanding of the principles involved.

The subject experts began to develop sketches of suitable models for the students to work on, and the team as a whole began to look for other computer simulations based on the idea of exploration by the user. This was the point when the group encountered StatPlay and SimCity, which we thought offered a useful approach.

The features of these programs which we thought were particularly relevant were that:

• the environments were enjoyable. They encouraged play and exploration
• students could investigate different hypotheses which they posed for themselves
• feedback was intrinsically related to the student's actions
• the features of the simulated environment were devised so as to enable key ideas to be explored

After considerable discussion, we decided that our project would incorporate these features, which we regarded as likely to be very supportive of students' learning (cf Laurillard 1988, 1993, Ramsden 1992). In terms of the approach to learning adopted by the project, this was a critical decision.

Initially we thought of constructing a series of virtual environments, each with models to be constructed, consisting of a laboratory, a factory, and a home. In practice, we have focused our attention on the laboratory models, and we have not attempted to simulate the laboratory environment. Implicitly, we have abandoned the idea of constructing artificial contextual settings which students enter and explore. These details are seen as irrelevant and likely to become annoying. Rather, we give students direct entry to a screen where they can work with a particular device.

We agreed that the laboratory models would be constructed with parameters which related to the key ideas, so that students would be encouraged to develop and test hypotheses relevant to them. Each academic member of the team took responsibility for proposing and developing a model which would relate to at least some of the key ideas. We arrived at a list of seven models, each of which addressed at least two of the key ideas.

Examples:

• a conduction bench
• a can in a retort
• a double pipe heat exchanger (the "penguins' feet" model)

We have gone on to construct four of the simulated models on computer. The most complete model so far is the conduction bench. The user can place heaters, coolers, and slabs of various materials (brick, steel, polystyrene... ) in a vice on the bench, and use temperature and flux probes to discover how temperature and heat flux change over time and over distance. This is the model available at this Conference.

An important feature of our approach to developing laboratory models has been the sharing of responsibility between subject experts. This not only shared ownership of the project; it meant that each of the academics could design in detail a model which they felt confident would enable students to work with key ideas relevant to their field of study. In deciding on the details of each model, we constantly referred to the relevant key ideas, so that each model incorporated appropriate choices and feedback. This gave us more confidence about the likely effectiveness of the models in helping students understand the principles at stake.

We still saw it as important for students to be able to test and get feedback from hypotheses - the difference being that the student generated the hypothesis, rather than being presented with one.

When we felt we had a reasonable paper draft of the conduction bench, we trialled the paper model with a pair of chemical and metallurgical engineering students. We wanted to find out if the interface was intuitively intelligible, and particularly to find out whether the students could see how they could use it to answer questions about heat transfer.

The students were very positive about the ability to explore which the model offered them, and they liked being able to "play" with it; they also made some suggestions which helped us improve the interface. Their exploration of it suggested that it would help them tackle the questions we posed.

Since then, we have tested a working version of the conduction bench model with a series of students, working in pairs. We gave them brief written instructions about the capabilities of the interface; they played with the possibilities for about twenty minutes, and then they were given an experimental task and asked for their predictions and later their comments. An extract from the videotape of these trials is available.

First, the interface was intuitive and effective. All the students operated it with considerable success and were able to use it for the task without hesitation. They enjoyed using it and all of them had questions they were still investigating at the end of an hour.

As we hoped, the students generated and explored questions of their own, and also encountered puzzling results which they began to try and explore. Importantly, their discussion and experience of these issues disclosed uncertainties and misconceptions about heat transfer which their work with the simulation helped them clarify. Examples:

• A pair of students who seemed quite comfortable with the language used to describe heat transfer were nevertheless convinced that heat travels one way and cold the other way, so that there is a turning point within a block of material where the temperature gradient changes. They set up an experiment which they hoped would show this happening.
• Another pair of students became intrigued with the difference between the electric heater and the steam heater. They were convinced that the electric heater would heat the material more quickly, but couldn't work out how to get the initial temperatures to match. When their time was up they were still trying to set an appropriate initial wattage on the electric heater.
• A student, using the flux probe to discern constant flux across the material, quite suddenly connected the readings he was getting with the familiar "energy in = energy out" equivalence.

We did not explicitly plan this model to address students' misconceptions, because we were uncertain what they were. It looks as though watching students use the model may help us identify and understand them.

We concluded that the model was likely to be useful to students with a range of preconceptions and approaches. We expect its flexibility will enable us to adapt it to students in a range of disciplines. Critical to this use will be the development of appropriate tasks by the various subject experts.

We have changed our approach to the graphic interface and developed a simulation with a much more real-world feel. This has entailed a series of decisions about programming and development specifications, and it has also led to the inclusion of a graphic designer in the team.

Initially one of the team members was familiar with using Excel to provide students with diagrammatic models. As our proposals for a virtual environment developed, we decided to move further towards simulation. We reviewed a number of authoring software packages, and decided that Macromedia Director was the most suitable, because it:

• supported the necessary graphing mathematics; and
• provided powerful animated visuals which would effectively simulate reality

In working on his conduction bench model, one of the academic team members decided to do the Lingo programming which would hook the graphic interface to the graphs. Other academic members of the team are currently honing their Lingo skills for work on their respective models.

Our rather lengthy development process has allowed us to refine the basic framework for the models, so that specifications are written to a common template and graphic elements of the interface are repeated in the design of each screen.

We have also refined our process for reviewing specifications. We now work through each draft specification as it is tabled, examining its relevance to the key ideas matrix and its consistency with the intuitive interface style already developed. A draft interface design is mocked up and its operation discussed before the graphic designer makes too much progress with the details.

We have continued to see this feature of the project as essential to its success. As described above, our most recent evaluation of the virtual conduction bench using our working computer model has shown that the model can in fact provide extremely useful feedback to students.

Our development process has been quite lengthy, but we can now look back over our time working together and identify three key steps in our progress.

• We identified a set of key ideas for students to engage with in the package;
• We agreed on the concept of designing the multi-media element of the project as a resource rather than as a self-contained instructional package;
• We agreed that this resource would take the form of a set of "models", simulated laboratory set-ups, each of which became the responsibility of a particular academic within the team, and each of which was seen as contributing to the project by helping students to achieve designated "key ideas".

Individually, members of the team have been able to clarify their ideas about the requirements of an interactive resource to support learning in a technical field such as this. Our questions have changed: from quantitative to qualitative to relational questions. Our collaboration has resulted in a resource which should enable students to make satisfying connections between theory, computation and intuitive understanding.

This project, funded by RMIT as part of its LearnT program, has involved John Ball, Andrew Halmos, Peter Johnson, Sue Johnston, Andrew Pang, Kate Patrick, David Samulenok, and Chris van der Craats. We thank the students who have helped us evaluate the project as it developed.

© Kate Patrick and Sue Johnston

The author(s) assign to ASCILITE and educational and non-profit institutions a non-exclusive licence to use this document for personal use and in courses of instruction provided that the article is used in full and this copyright statement is reproduced. The author(s) also grant a non-exclusive licence to ASCILITE to publish this document in full on the World Wide Web and on CD-ROM and in printed form with the ASCILITE 97 conference papers, and for the documents to be published on mirrors on the World Wide Web. Any other usage is prohibited without the express permission of the authors.