Eating capacity
Disclaimer: to answer this question, you really have to go looking through the timetables or speaking to planners to get a good idea, but there are some common sense ones as well.
There are quite a few key junctions on the network that are such constrained pinch points that they by definition set the timetable for an entire route. Woking junction is by far the most famous, as the flat junction there is the starting point for writing the the timetable for almost the whole LSWR route. You'd solve that with a flyover and a bit of remodelling, but ultimately it wouldn't cost the world.
Similar junctions include previously mentioned Castlefield, Newbridge and Haymarket South near Edinburgh, and the complex at Windmill Bridge. Windmill Bridge was planned, now on hold, and doing anything with Castlefield would be very pricey indeed. Off the top of my head, some others that fill the same criteria include;
Colwich
Slade Lane
Trent Valley Junction
Heaton Norrish
Law Junction
The entirety of Crewe station
Swinton
Syston
The problem is, the capacity loss at these junctions isn't because they're slow, it's because they're badly designed flat junctions in key places for a modern day railway.
Haughley junction in Suffolk fits this category; it limits quite severely the amount of trains that can go north out of Felixstowe, and it's been on the cards for ages. You'd fix it with a double lead junction, but that's ten million or so for the resignalling and remodelling, and that's a lot of nurses and doctors.
These junctions, and they are almost universally single lead where they should be double lead, or flat where they should be grade separated, eat capacity on the modern railway. Defensive driving and signal protections mean that you have to approach slowly, and wait your turn. The moment something runs late, you're holding up everything else waiting for your slot.
But, and there is always a bit, it's never that simple. Take for example the thirty-odd mile section between Carstairs and Glasgow Central: there are no less than 8 flat junctions along this stretch. Solving Carstairs is sorely needed, eliminates a dodgy PSR, and speeds up three sides of a triangle, but equally, it isn't going to give you any capacity. Instead, any conflicts are shuffled around, and reappear somewhere else.
It's the same with Newark: speeding up the junction does nothing, because the conflicts are still there. You need grade separation to make a difference, and suddenly we are talking about budgets in the high tens of millions at least.
That is the sad reality of our network. If we were rebuilding it from scratch today, we wouldn't build it as the Victorians did. We'd have much more grade separation, and more segregation of fast and stopping and freight services for a start.
