An interesting recent monograph about project planning, from home improvement up to large-scale projects: Bent Flyvbjerg and Dan Gardner How Big Things Get Done, Currency 2023.
Besides important recommendations about planning, especially getting the errors made and corrected on paper rather than during construction, is a list of project types ranked from historically worst results to best results.
"Bad" means over budget; with statistics for mean overrun, % of projects in tail of distribution, and mean overrun of projects in tail.
Going from worst to best in his list, with major omissions:
Nuclear waste storage, Olympic Games, Nuclear Power, Hydroelectric dams, IT, ... , Bus transit, Rail, Airports, ... , Fossil thermal power, Roads, Pipelines, Wind power, Energy transmission, Solar power.
See his discussion of the list: "solar power" is incredibly good, with a mean cost overrun of only 1%.
In his discussion, he points out that the worst stuff is always a one-off unique design (nuclear reactors are not in mass production, although future projects could change that); Olympic Games are the poster child for this.
The best stuff is inherently modular, such as solar panels made in factories and installed on site.
Even fossil thermal power is just bolting together catalog items from the boiler factories.
The transport projects are somewhere in the middle of the list, with distribution tails falling between his "fat tails" (Olympic Games) and "thin tails" (power plants and transmission).