Simulations were invented hundreds of years ago to test models. That has remained their purpose in science ever since.
Science education is another thing entirely. Simulations were not invented for education but adapted to it. The earliest science education simulations that I saw were for projectile motion, hardly too expensive or dangerous. It was merely convenient for the purposes of visualizing trajectories under different circumstances. I have reproduced in real life many trajectories without great cost and with no danger.
Some point to the Manhattan Project and the space program as examples of using simulations.
The purposes of the Manhattan Project and the space program are quite different than those in a science classroom. This analogy fails completely. Space engineers are not attempting to visualize something for the purpose of learning a new concept. They're designing equipment. Ultimately, they do as much real-life testing as possible before committing their devices to space.
Here's my approach to when it's too expensive or dangerous to do the real thing. Have someone else carefully plan and record those experiments, just as I have. Do the recordings in such a manner that students can take data from them interactively. Don't bother to figure out the model because the real world IS your model in this case. That's the best possible model for science.
You don't have to worry whether the model on which your images are based is correct or not because the real world is always correct. When you drop an object, it accelerates at a rate determined by mass, shape, size, air resistance, gravity, wind, and any other parameters that may be involved without writing one line of code or using a single equation.
Simulations are great tools for visualizing the unseeable. They can also be an adjunct to or replacement for videos of difficult-to-understand concepts. They are not a replacement for good science lab investigation experiences. Many people think that they have to be that because they're the only alternative to hands-on. BUT they're NOT the only alternative.
One, very limited, alternative is remote robotic labs. They don't have interactive data collection and work only on a limited range of experiments. But, they're real and are online; they do provide access to expensive equipment.
My alternative is to record the experiments ahead of time. Theoretically, you can record every parameter combination likely to be chosen by students. Although you've time-warped the experiment, it makes no difference to the students. They cannot tell that the experiment was performed in the last minute or month or year. Once those bits get onto the Internet, they become virtual and can be stored for recall at any time.
However, just watching the experiment is not enough. With simulations, that's all you get, and you get it in an unreal world that I'd even call fake or cartoon. The real experiment begs you to take the data yourself. The data are not known beforehand as with a simulation. Your taking of data has real meaning to you. It's your data, not the data some programmer set up with an algorithm (possibly flawed and definitely imperfect as a representation of the real world).
So, you get to choose. Would you have your students investigating an algorithm non-interactively or investigating the real world interactively? In both cases, they do it online with all of the benefits that flow from that medium.
© 2012 by Smart Science Education Inc., U.S.A. www.smartscience.net