Specifically, stacked against C3, Britecharts has a more 'D3-like' API, while C3 is a lot more declarative. This means that on C3, we will render charts by passing large configuration objects, while in Britecharts we strive to use a different API, chaining settings like we do on D3.
Also, C3 hasn't been updated to D3 v4, and we can say it is an inactive project.
One other big difference is that Britecharts is compatible with the latest version of D3. D3 V4 underwent a major refactor to increase modularity. NVD3 is actively looking for volunteers to help them get there though - https://github.com/novus/nvd3#do-we-support-d3-v4x.
Disclaimer: I work with the folks who develop Britecharts.
We've had it in production for a couple of years now. The best thing about it was it got us off the ground quickly but at this point I think we've likely sunk more hours into debugging issues and hacking features than it would have cost us to have built it with d3.