Perhaps it was the flue gas duct of unit one at Kusile Power Station ripping away from the chimney that was the final straw for Rhulani Mathebula, who until Monday was acting Group Executive for Generation.
The first call made by the manager of Kusile would've been to Mathebula.
Not only does this 'structural failure' at Kusile mean this unit will be offline for a number of months - the ducts of units two and three are in the same chimney, so these units are offline too.
This translates into 2 400 megawatts (MW) of nameplate capacity lost.
More realistically, due to the numerous design defects at both Kusile and Medupi resulting in lower generation output, about 1 600MW is unavailable. This is nearly equivalent to Stage 2 load shedding.
Or maybe, for Mathebula, it was the fact that the percentage of Eskom's power stations actually producing power has plummeted to never-seen-before lows.
The energy availability factor (EAF) for its generation fleet is at 58,5 percent so far this year, a sizeable decline from the 62 percent achieved in the last financial year (to end-March).
Mathebula was surely a reluctant acting chief of the most troubled unit at Eskom.
He took the job in May after Phillip Dukashe, the then permanent head, resigned. The utility said at the time that 'in his resignation letter and discussions with Eskom executives, Mr Dukashe has cited the critical need to achieve a balance for the benefit of his health, family and work responsibilities'.
Fast forward six months and Eskom COO Jan Oberholzer on Tuesday told journalists: 'When (Mathebula) tabled his resignation, he said the demand of the job was untenable, and it impacted his health and family time. We need to understand the demand of this role, and the demand on the head of generation in turning this ship around.'
Mathebula's is the phone that rings when five units at different power stations trip overnight.
Or when there are trips in the very early hours of the morning necessitating Stage 6 load shedding, as was the case on September 20. Before him, Dukashe's rang.
Mathebula is no stranger to the role, having acted in it prior to Dukashe's appointment just 13 months before. Mathebula has been at the utility for 19 years. Dukashe had been there for 26. Both had been power station managers in the past and possessed vast experience.
He was not still supposed to be in this job - the recruitment process started in May.
Dream job that's a nightmare
This position - group executive for generation - ought to be the pinnacle of an Eskom manager's career.
It is arguably the most important executive role at Eskom and once the utility is finally split into three, whoever heads generation will run the biggest chunk of the business. But the current mess within this unit means that whoever takes the job will effectively just be doing crisis management.
Who would possibly want to willingly sign up for this? Were there even any applications from credible candidates?
Three white men . . .
And so, Eskom has turned to Thomas Conradie, the power station manager at Lethabo.
Conradie has been at the utility for 24 years, having previously run Kriel, Matla and Majuba. Lethabo is currently one of the power company's three best-performing power plants - Eskom describes it elsewhere in its presentation as a 'flagship station'.
This leaves Eskom with the politically, ahem, 'uncomfortable' situation where its three most senior executives are now (fairly old) white men.
Will Conradie be able to achieve any meaningful results?
Hiring and firing at Eskom is very hard work. For example, it's taken over a year to conclude the 'consequence management' process for the employees who blew the brand new unit four at Medupi to smithereens. Two employees were fired for their role in the incident, which will cost R2,5 billion to repair.
The plan is to somehow turn around the six worst-performing large power stations: Tutuka, Kendal, Duvha, Majuba, Kusile and Matla. These all have capacity of over 3 000MW and Eskom says they have been 'specifically selected as they are among the highest contributors to unplanned load losses'.
'Any improvement in these stations will result in massive gains in EAF for generation as a whole,' says Eskom.
By March 2024, in other words in its next financial year, the utility hopes to achieve an EAF of 65 percent. This is a mammoth ask, especially since Koeberg unit one will be offline for the replacement of its steam generators. It is however optimistic - it says it has turned around the performance of Medupi.
By 2024, things look better with the completion of Kusile and the return of Medupi unit four. And while additional renewables capacity is added to the grid by then, Eskom will ramp up the closure of old stations as per their dead-stop dates.
For now, load shedding will remain 'until there is adequate capacity in the country'.
Eskom's hopelessly out of touch 'base case' scenario for summer (September to March) sees just seven days of up to Stage 1 load shedding. A more realistic scenario, with an average of 14 500MW of breakdowns, sees 126 days of load shedding exceeding Stage 2. - Moneyweb