This paper is a review of the challenges faced by designers and implementers of software and hardware to aid in the design of solutions to serious environmental problems. The complexity of the problems usually requires that the software simulate the underlying process and present the results visually. The challenges include the domain size, both physical and temporal, and the complexity of the model. The authors use the example of designing deep geological repositories for the disposal of high-level nuclear waste. Proposed configurations for such repositories include multiple barriers to control the migration of radionuclides. The domain for this problem is physically large, but it requires detailed resolution since the thermal and hydraulic gradients across the relatively thin barriers must be correctly captured. Also, the timescale must be long since the materials will be subject to mineral alteration, and boundary values will be affected by global changes. However, because of the gradients at the barriers, the resolution in time cannot be coarse. Moreover, the authors say that for these problems, common simplifying assumptions with regard to geometry may lead to “features of the system’s behavior being neglected or significant errors.”

The complexity of the models is affected by the number of processes and variables, as well as the coupling between variables and the nonlinearity of the system. Like the domain issues, the complexity increases the dimensionality of the numerical problem. As a result, a numerical solution to these problems will use a combination of discretization methods, and the computational demands will be dominated by the solution of large sets of simultaneous equations. The authors refer to a software package, COMPASS, for a class of problems including the disposal example. The numerical portion is written in FORTRAN 90 with a graphical user interface written in Visual Basic.

The challenges described here are similar to those encountered in other areas, such as weather prediction. The difference here is that we are designing a process, and we have some control.