New load shedding stages for South Africa – the big changes, and the big problems

Energy regulator Nersa has approved and published the third edition of the NRS 048-9 Code of Practice, which governs load shedding in South Africa.

The latest edition of the code introduced some changes to the load shedding structure in the country, the most notable of which is the extension of load shedding stages to stage 16.

In previous editions of the code, load shedding was only structured to stage 4 in the first edition and then to stage 8 in the second, showing the exact schedule and time-periods attached to rolling blackouts, for 25% and then 50% of the total load, respectively.

In the third edition, the full load is now divided and scheduled to account for the unlikely event that outages move past stage 8.

According to independent energy analyst Pieter Jordaan, while some wording and the load shedding algorithms have changed, the actual scheduling of outages – the part that impacts end-users and customers the most – has not shifted significantly.

Under the new schedules, Eskom’s electricity grid is segmented into 16 blocks with similar load profiles.

Each block represents 6.25% (100/16) of total demand, and the load shedding algorithm cuts an additional block for each additional stage required.

However, 20% of each block’s demand is deemed critical, and thus only 5% of total demand per stage – known as the Manual Load Reduction (MLR) yield – can theoretically be shed.

“This was also the case with the algorithm in the second edition, which already referred to 5% load shed as 1 stage,” Jordaan said, adding that the only difference in the third edition is that this has been expanded to 16 stages.

What has actually changed?

One thing that has changed in the new edition, which may impact end-users and customers, is that the load shedding cycles have been altered.

While the third edition schedules still follow a 32-hour cycle (i.e., X hours are shed in a 32-hour cycle), this is only true for the first 16 calendar days.

Previously, the primary stages of load shedding (stage 1 to stage 4) followed a ‘retreating pattern’ from day 17 to 31.

With the third edition, the cycle resets, and days 17 to 31 now follow the same pattern as days 1 to 15.

Jordaan said the time slots remained the same, but the order of the stages was slightly different, which will have a negligible net effect.

When we move to the secondary stages of load shedding (stage 5 to stage 8), a bigger change kicks in.

In the second edition, these stages of load shedding followed the blocks of the primary stages – but in the third edition, these blocks now precede the primary stages.

For example, in Edition 2, at stage 1, a household experiences load shedding between 08h00 and 10h00, and then at stage 5, this becomes an outage between 08h00 and 12h00.

In Edition 3, this would change to an outage between 06h00 and 12h00.

While the outage time is the same, the shifting of the blocks makes it slightly more restrictive for customers at higher stages, Jordaan said, with the 32-hour cycle effectively becoming a 30-hour cycle.

Ultimately, the actual impact on customers will be negligible, Jordaan said, with the new structure giving the System Operator a small buffer when stages hit stage 5 and above.

Edition 2 load shedding stages (Block 1)

Edition 3 load shedding stages (Block 1)

The table below outlines how households will experience load shedding in terms of hours off. The blocks of time out increase by 2 hours until there are four 6-hour blocks at stage 12.

From stage 13, the blocks start merging, and customers would experience a full 14 hours without power (a 6-hour block immediately following an 8-hour block). By stage 15, this merges again to 30 hours off.

Stage	Blocks (32 hour cycle)	Hours off
Stage 1	2 x 2-hour blocks	2
Stage 2	3 x 2-hour blocks	4
Stage 3	4 x 2-hour blocks	6
Stage 4	4 x 4-hour blocks	8
Stage 5	1 x 4-hour block, 3 x 2-hour blocks	10
Stage 6	2 x 4-hour blocks, 2 x 2-hour blocks	12
Stage 7	3 x 4-hour blocks, 1 x 2-hour block	14
Stage 8	4 x 6-hour blocks	16
Stage 9	1 x 6-hour block, 3 x 4-hour blocks	18
Stage 10	2 x 6-hour blocks, 2 x 4-hour blocks	20
Stage 11	3 x 6-hour blocks, 1 x 4-hour block	22
Stage 12	2 x 14-hour blocks	24
Stage 13	1 x 14-hour block, 2 x 6-hour block	26
Stage 14	1 x 30-hour block	28
Stage 15	1 x 30 hour block	30
Stage 16	Power off	32

The big problem

Jordaan noted that the new edition of the code doesn’t address some problems that emerged during South Africa’s worst load-shedding year last year—the main one being that the ‘critical demand’ is far greater than the 20% assumed by the algorithm.

The analyst said that using the 80% theoretical load assumed by the code should have seen a blackout ratio of 9.3% in 2023—but in reality, the ratio was 19.9%, with the ‘effectiveness’ on load shedding only being about half the intended rate.

This means that the 5% each stage of load shedding is supposed to ‘recover’ from the grid is only yielding around 2.33%, he said.

This is because of two key factors: load slippage and load shifting, Jordaan said.

The first refers to more critical demand being present than the code assumes – where more customers are exempt from load shedding and load curtailment than the NRS factors in.

The second refers to the load shifting to subsequent time slots – i.e., customers using more power when the lights come back on. The load shed in time slot one is partially shifted to time slot 2, etc, negating the efficiency of the system

This is exacerbated by the spread of inverters and battery chargers, which are net power consumers.

“These two factors were not addressed in Edition 3 – and therefore, the effective MLR yield will, at best, remain around half of its potential 5% per stage.