@drunkill I had no idea that decking could be that expensive. It's also the sort of thing that seems hard to believe if what makes it that expensive isn't explained.
It's not that it's that expensive; it's just more than the land is worth.
The typical cost of land around Carnegie, for example, is around $1 to $1.5 million per house block (other areas would be cheaper). It's not worth building a deck if the marginal cost of adding the deck is more than the value of the land created.
And $1 to $1.5 million per house block is not all that much money. It has to cover the costs of:
* the deck itself
* any additional excavation (the rail line needs to stay down; it can't immediately begin to return to the surface)
* the additional cost to the sidewalls and foundations to support the deck and whatever you build on it
* the additional construction time (e.g. bus substitution, slower construction due to working next to an operational railway)
* (if long sections, which are closer to tunnels) fire control and emergency services
* the geotechnical investigations, design and supervision work
The land becomes more valuable if you can use it for multistory buildings. This is why the Ormond redevelopment was designed for a multistory building, and why the Glen Waverley line proposal was the same. This additional value means that the cost of the decking more viable. But it means the deck, walls, and foundations become more expensive, so it's an optimisation. Ormond clearly demonstrated that the cost optimisation point in inner Melbourne for decking was a multistory building. But this means the locals are more likely to protest against the use - this is what's stalled the Ormond redevelopment. It's also what's stalled the Burke Rd, Camberwell, proposal.
On the other hand, a very light building on top means a very simple deck and foundations. That can be worthwhile if the land values around are high enough. The single story redevelopment over the Caulfield lines at Chapel St, South Yarra, is a good example.
Finally, if you don't redevelop it and use it as a 'park' or a bike track, it would almost certainly be more cost effective to buy the adjacent land instead.