[justinph]

I do front end dev work, and mostly it comes down to experience. I usually try to work out a small test of some of the more complicated problems beforehand, so I might have a sense of how much time they'll take. Then I break the project down into parts and assign hours to it

You can bill your client for actual hours, or do a flat rate based off the hours you think it'll take. I tend to do a hybrid approach where I bill them based on a flat rate with a defined scope, then can charge more if they change the scope or something takes significantly longer than expected. Most of the time I come in under my hours, but not always. It's worked well for me.