i am using ricardo wave to model the exhaust for brunels 2008 car. i was hoping someone could recommend a book that would help me find a way to find a starting point for theoretical values of the header dimensions that i can improve upon using wave. thanks.

Theoretical dimensions?

You could start with i and work your way up to 2i.

You can improve on just about every useful dimension you choose in wave, so long as you properly model the lengths, diameters, and volumes.

For example, you'd model a two-valve head as two ducts from the valves connecting to a volume(the junction) that exits as a duct of the proper diameter to exit the head.

Careful measuring of the actual engine will let you approximate these dimensions - properly dimensioning your model is CRITICAL to obtaining useful results.

After that, model a current engine - preferably one for which you have a lot of test power curves. WAVE data without calibration is about as useful as drawing a power curve in MS paint. Try to model this engine as exactly as possible in WAVE - and tweak values as appropriately to get the power curves to look similar in as many cases as possible. If you can actually make dyno runs side by side to correlate your model, so much the better.

After this is done, you can use WAVE to give you good indications on which direction you should go with your design - to be followed with TESTING! In my experience, it is hard to get WAVE to give you accurate empirical numbers - what it can easily show you when properly set up is the shape of your power curve - where your choices of intake and exhaust lengths will give you more and less power, and in relative amounts.

If you're looking for a good starting point for where to start changing things - try the stock bike dimensions. Tweak in each direction, observe the result, determine the pattern. Change the next variable, determine pattern. Once you know what the effect of changing these variables is, see how they affect each other. Are they transitive or commutative?

Then keep going until you have a power curve that you want, and then build and test according to your model to see if you are right.

I agree with Wesley, it can be quite hard to get an accurate wave model fast. i sudgest that you concentrate on the volumetric efficency and try to get the smoothest graph from that, but you'll have to play around with the inlet runner lengths to to find the best curve.
A cool trick is to set up a multipul variable so you can produce a 3d graph of vol.eff vs. rpm vs. primary lengths.

Oh by the way MSpaint saved my ass on a wave assignment i had to do last year, i had a great torque curve in 30sec!