To supplement Halos's answer.
The 66 kV network adjacent the Lonsdale substation is very strong, four feeds into the adjacent 66kV substation. This area used to supply the Pt Stanvac Refinery and the Lonsdale Foundry. The 66 kV feeders originate from either the Happy Valley or Morphett Vale Zone substations.
There are two 66 kV feeders supplying the Lonsdale traction substation which for some reason is on s cramped site. Gas insulated switchgear is used to save space and supposedly enhance reliability. There are separate traction transformers either of which, I understand, can supply the entire system. Although all switching equipment is in the one building there is adequate physical separation to prevent a fault in one section tacking out the entire system.
What happened when the system failed is as follows.
A fault occurred in one module of the incoming circuit breaker at such a location that the switchgear itself could not instantaneously detect it, an inherent blind spot. This meant the fault was sustained longer than necessary doing significant damage until an external circuit breaker could isolate the system. The fault energy was sufficient to blow out a blast cover on the switchgear and blow open the door of the switchroom. The healthy part of the system was still live.
First responders were sent to the site to determine why one part of the system had failed and on arrival found the switchroom door open and smoke in the air. Their immediate assumption was some kind of foul play and the entire system was shut down. At this stage nobody was aware what had happened, remember the switchgear could not itself detect that a fault had occurred because it occurred in a blind spot in the system.
It took a while for engineers to determine the failure events, this involved liaison between SAPN and PTS staff. Then confidence could be gained to know what equipment would still be serviceable such that power could be restored.
Meanwhile overhead line staff had to ensure the network was capable of being reenergised.
Understandably this took some hours before trains could run.
I have not been privy to why the module in the incoming circuit breaker failed, suffice to such that such an event is extremely rare. There may have been some shortcomings in the commissioning which meant that systems detecting faults in the blind spot were slow to act or failed to act.