My usual structure:
You may do this project at almost any point during the semester. Here's the procedure...
Here are some guidelines for picking the problem...
There are some common problems which arise when learning mathematical writing. I recognize that you have almost certainly not done this before, and that this is probably all new to you.
(1) Technology -- Both the rough draft and the final copy must be computer-processed, including any pictures or graphs needed. I should see nothing handwritten on your paper. Here are fixes for some standard difficulties...
(2) Format -- One way to think about the assignment is to pretend you're writing a short paper, and the subject is the problem at hand. There should be an introduction, which often consists of a statement of the problem, a body, which often consists of a discussion and solution of the problem, and a concluding statement, which usually places the solution in context. (Examples of common conclusions: "This amount is realistic because..." or "...which is what we expected because...") There should not be sections labelled as corrresponding to parts (a), (b), etc. of a problem, if indeed the problem has such parts.
(3) Writing style -- While the style of your writeup should be similar to any regular old group homework problem, this assignment requires a bit more formality. (In my dreams, writing this will help you in writing up your regular homework...) Your reader will not have a textbook handy when perusing your paper, so don't refer to the text in such a way that you confuse the reader. For example, restate the problem, but do not quote it directly. ("Find the area under the curve" becomes "Suppose we wanted to find the area under the curve...") Similarly, if your chosen problem has several parts, don't list them at the beginning of the paper. Bring them up as they occur. And, because you're controlling the flow of the writing, you can discuss the parts in a different order than they're presented in the problem if that's more convenient!
Another example, taken from an email sent on this topic:
Instead of writing
"(a) Find the miniflarg of the flozzlenuck."
write something like
"Now let's find the miniflarg of the flozzlenuck."
"Suppose we wanted to find the miniflarg of the flozzlenuck..."
"What about the miniflarg of the flozzlenuck? To find..."
That way you're introducing what you're trying to do without actually posing it as a command.
You don't need to have intro and conclusion paragraphs, but you
should have a couple of sentences on each end that serve as a frame
for the problem. (You don't want people to be confused at the
beginning or wonder what happened to the punch line of the problem.)
One good way to end is to make a statement that ties together a few
parts of the problem.
When I critique your rough draft, I will not point out every flaw. In an English-style writing assignment, a faculty member generally doesn't make detailed grammatical comments and such. Because you don't have much experience with mathematical writing, I usually do some editing -- I'll correct grammar errors and usage errors as I see them... after all, mathematical grammar is slightly different from English grammar. So, when you're revising your paper, don't forget to read it over a few times to catch flaws that I didn't mark. And just as should happen with any other paper, the exposition should improve somewhat between the rough draft and the final copy.