Just before the Easter long weekend, Eskom ramped up the amount of generation capacity taken offline for planned maintenance by 40%. This was a substantial move (and a very belated positive one). In practical terms, planned maintenance had been hovering around the 4GW and 4,5GW mark and this was increased to over 6GW on the Thursday before Easter.

This was absolutely the right thing to do, given that electricity demand would drop significantly over the long weekend as large industrial users shut down. (Bear in mind that there’s still the 900MW from Koeberg offline till the end of May, meaning that even with an increase to 6GW, the real maintenance happening on Eskom’s legacy coal fleet is closer to 5GW — or at 5,5GW, closer to 4,5GW.)

The problem, however, is that since Easter, Eskom has battled to get planned maintenance back down to what had become “normal” levels (in the 4GW range). Planned maintenance was at 5,5GW on last Thursday (9 April), at 5,4GW this Monday (13 April) and 5,5GW as at today. Eskom does not have the headroom to do this amount of maintenance without load shedding or without unplanned outages dropping significantly (and remaining there).

Now this could mean a number of things. One: Eskom has realised that it hasn’t been doing enough maintenance during summer, and that maintenance would have to be kept at these elevated levels for the foreseeable future. Two: Eskom hasn’t been able to restore generation capacity that has been undergoing planned maintenance as quickly as it had planned. In other words, whatever it has been fixing has taken longer to fix than it originally thought. Three: Eskom misread the reliability and performance of the rest of its fleet and took the chance to ramp up maintenance.

In all honesty, the most likely answer is a combination of all three.

The quandary for Eskom all along is that it has to do proper maintenance when it takes any plant out of commission. There are (more than) some hints that it has continued to rush maintenance and force plant back online to make up for a shortfall due to unplanned outages.

Then there’s the curious situation over Easter, where Eskom originally forecast it would have over 34GW of capacity on 2, 3, 4 and 5 April — this despite the elevated levels of planned maintenance. Days later, it was revised lower to anywhere from 31,5GW to 33,5GW. In the original forecast, it expected to have 34,9GW available on the Thursday and Friday. (We don’t know what capacity it actually had, because Eskom didn’t publish a status bulletin on Monday (given that it was a public holiday), but this is entirely academic given the far lower than average demand.)

But the last and only other time that Eskom forecast it would have anything close to 35GW of capacity was on 21 and 22 February. The original forecast was revised lower to about 27,5GW for that weekend, and demand was forced lower (it did not use open cycle gas turbines). It’s difficult to figure out what happened in both these cases. How did Eskom arrive at the over-generous supply forecasts in the first place? Based on that, did Eskom look at these and decide to use the window to do additional planned maintenance?

What happened this week, where load shedding was in place from 6am on Tuesday, Wednesday and Thursday (and from 4pm on Monday and 10am on Sunday), was downright predictable. Planned maintenance is still at stubbornly high levels. (This is good news, especially for winter.) The fact that Eskom was able to make the call and announce load shedding from 6am on the night before each of these days is telling. But, if it was forced into stage two from as early as 6am, the supply situation was bad. (Thankfully, it ran stage one from 6am on Thursday, meaning it had found another gigawatt, or some diesel.)

However, the situation on Wednesday was downright bizarre. In the morning, it had managed to restore 1,2GW of capacity (2x600MW) from the 9,5GW unplanned outages. It confirmed then that planned maintenance stood at around 5GW. Yet, by the afternoon it was forced into stage three load shedding (where it removes up to 4GW of demand).

Do the maths. There were 9,5GW of unplanned outages plus 5GW of planned maintenance as of Wednesday morning. Stage two load shedding meant a foreseeable shortfall (at peak) of 2GW.

But 1,2GW of the unplanned outages were restored during the course of the morning. This left a new theoretical shortfall of 800MW (call it 1GW), but load shedding wasn’t downgraded to stage one. Instead, load shedding was escalated to stage three by 4pm.

This meant Eskom lost the equivalent of all of the capacity it had managed to restore (1,2GW) plus another gigawatt (or two).

In its status bulletin published yesterday, Eskom says it there was peak demand (“reduced down”) of 27,5GW with available capacity of 29,3GW on Wednesday.

jerm16042015 — TechCentral cartoonist Jerm’s take on developments

Something, somewhere does not quite add up. How did capacity disappear on Wednesday?

Business Day says it was able to “build a picture of what transpired” and that Eskom reached the “tail-end of… [the] diesel supplies” for its open cycle gas turbines. Compounding this was the depletion of “pumped storage capacity” earlier in the week. This is (somewhat) good news. It means that Eskom didn’t lose any more capacity from its coal fleet.

In fact, Thursday’s system status bulletin further corroborates this, except its not evident Eskom even had the luxury of the “tail-end of diesel supplies”. In the bulletin, Eskom says available capacity on Monday, Tuesday and Wednesday this week was met “considering primary energy constraints”. This is code for ‘we did not run the open cycle gas turbines’. Curiously, it now references a demand picture “excluding load shedding” which is not really a demand picture at all – more on this next week.

Referring to the investigation underway while four executives are suspended, public enterprises minister Lynne Brown said in her press briefing on Wednesday that she is ”digging deep into Eskom for three months to find factual information about what is happening at Eskom”.

“I want a deep dive into Eskom. Don’t you? Don’t you want to know what’s inside Eskom. It’s one of the [three] largest utilities of its kind in the world. That’s why I am only giving them three months. At least I want to know its liquidity. I want to know how the maintenance is going.”

So, the minister and department don’t know “how the maintenance is going”? And the war room? Do they know? Does Eskom itself know? This is the hauntingly worrying part of the minister’s pronouncements.

Brown also said on Wednesday: “I think we need to stick to maintenance — even in winter.”

Hallelujah! Back in February, I called for Eskom to load shed daily until May. This was pooh-poohed (indirectly, of course). It seems the reality of the situation is finally obvious to government.

I’ll repeat what I said earlier: “Get the ramp up of Medupi unit six right, and get some of that planned maintenance done (properly!), and we’ll be alright for winter. Don’t, and watch the unplanned outages and general unpredictability rocket upwards because Eskom simply hasn’t had the headroom to get plant fixed. The vicious circle it’s been battling against all summer will continue.”

At this point, the sane path is to continue doing maintenance during winter. Eskom did so last year. Let’s see which route is chosen.