"Numerical Recipes" is probably what you need the most. It contains tons of useful scientific/numerical computation codes. It's a book, but I believe it was also available online.
Apart from that, you probably need some more theoretical background, and some programming experience.
"Linear Algebra" lectures would help a lot with solving sets of equations. Ocw.mit.edu has open courses available, for example.
Fortran, C/C++, Python and Matlab would all help you model and solve the equations for you. And there are several good textbooks and open source materials on the Internet.
Matlab takes the most investment (unless it's already licensed at your institution). It has many solvers already built in, and it handles matrices extremely well. Plus it has built in visualization capability. (Though it takes just a few lines of code in the other languages to come up with some graphs, or you could write your data to a file and use another application for plotting).
If you'd rather learn how to do the computations yourself, Fortran and C/C++ would be the best teaching way to go.