PFC that stopped working, at a Much Asphalt plant.

By Oliver Nesbitt, Sales & Client Relations, Augos Energy

Seventeen good months, then a slip

This Much Asphalt plant has two metered incomers, one for the office and one for the plant. In late 2024 the site's power factor at maximum demand was 0.80 to 0.82, and the avoidable part of the demand charge was R22,410 and R24,230 a month.

In December 2024 that changed. Power factor went to 0.999, and from then to April 2026 it stayed between 0.998 and 1.000 in all 16 months with data. Avoidable demand cost fell to under R200 a month.

Then in May 2026 it dropped to 0.86. It came back to 1.00 in June, and from July it sat at 0.87 to 0.88. Nothing tripped and production carried on. The only sign was in the demand charge.

How we saw it

Our PFC analyst reporting works on 30-minute cost signals. It takes the plant's measured kVA and kVAr, runs them through the site's own tariff, and works out what the demand charge would be at a power factor of 1.0. When the gap passes a threshold, it raises an alert.

The alert landed on 1 September 2026: power factor 0.86 at maximum demand, and correcting to 1.0 could bring maximum demand down from 540.5 kVA to 469.8 kVA, worth an estimated R11,264 a month. The platform's measured loss for July, August and September was R12,581, R12,468 and R11,636, so the alert was in the right range.

Month	Power factor (site)	Avoidable demand cost
April 2026	0.999	R93
May 2026	0.858	R13,199
June 2026	1.000	R12
July 2026	0.876	R12,581
August 2026	0.870	R12,468
September 2026	0.875	R11,636

Where on site was the problem?

The office supply stayed close to 1.0 throughout. The plant incomer is where the reactive power appeared: about 280 to 300 kVAr lagging at peak from July to September, where for most of the previous year it had read slightly leading. On this plant's City of Ekurhuleni tariff no reactive energy charge appears, so every rand of loss came through kVA demand. From May to September the avoidable demand cost came to R49,896.

The June reading is worth noting. Power factor recovered for one month and then dropped again, which pointed to something intermittent, such as a contactor or a fuse, rather than only an aged capacitor.

What the survey found

The annual PFC survey was completed in early October. The fuses and contactors had failed, and the plant's PFC panel wasn't running at full capacity. The repair was approved on 19 October and is booked for early November.

What October shows so far

From 1 October the plant incomer reads close to 1.0 again, and the avoidable demand cost for October to the 26th is R221. That happened before the repair, just as it did in June, which fits an intermittent fault rather than a fix. So once the repair is done, we'll compare the first full month after it, on the same tariff, against July to September before we call it fixed and put a payback on it.

What I'd tell any plant engineer

Installing PFC isn't the end of the job. Capacitors, fuses and contactors wear out, and the panel won't tell you. Watch your power factor every month, not once a year, and when it moves, get the panel surveyed before the next bill.

Figures from the Augos Power Factor and Demand report for the site and its two incomers, October 2024 to October 2026, as calculated on 27 October 2026. June 2025 is missing from the report. Avoidable demand cost is the demand charge at the actual power factor minus the demand charge at 1.0.