Poor power factor value through energy meter on maximum solar power exported to grid

We installed a couple of 100kW grid tied rooftop solar PV plants in factories which worked without flaw for quite sometime thereby contributing to electricity consumption and reducing the overall utility bills. Noted power factor was always in the range of .95 to .98 lagging, energy meters are bi-directional.

For reasons beyond their control, some of these factories had to shut down for a good part of the month. At this time solar plants were fully operational so majority of kWh was exported and only about 10-15% of total load (lighting, air-conditioning, fans, etc.) was catered to during this period.

To the horror of factory owners, electricity bill for this period carried around 70% penalty for low overall pf, around 0.25 lagging. No change in method of pf calculation, which is, accumulated kWh divided by accumulated kVAh for the duration. Capacitor banks were intact and operational.

Analysis showed that if the solar plants were shut down, all other conditions remaining same, the power factor improved to .91 lagging. Why? To the best of my knowledge, solar power is supplied at unity power factor and GT inverters are not power factor controllers. So it boiled down to 'some' issue with the bi-dir energy meter. But this happened at 3-4 factories so not all meters were defective.

Utility answered that this problem will get resolved on installation of Netmeter. What's the difference - bi-dir vs. Netmeter? How will it resolve the pf issue? Answers are welcome.

I'd always assumed that "bi-directional" and "net metering" were the same thing, so I can't make sense of that bit.

Could you post some of the measured accumulated kWh and accumulated kVAh numbers with time periods for both the "normal" and "bad power factor" time periods please?

However, what I suspect is happening is one of two things:

1) the power company's power factor measuring system is single-directional and treats reverse flow as a 180 degree out-of-phase power factor, calling it "bad".

2) the meter is single-directional and only calculated power factor when the panels were not exporting. Since electricity consumption was down to a few items, the power factor was much worse - but calculated over a much smaller amount of electricity.

• I am sharing below data of 'good (May)' and 'bad (Oct.)' period. Note the dramatic improvement in pf on 10.10.2016 when the solar power was switched off. !Good period !Bad period Oct 30, 2016 at 18:35
• Meters are confirmed bi-directional, but your suspect (1) made me think. Take a look at the crude representation below of what may actually be happening. While the utility takes into account S & P for calculating pf, what about P1, it is available on the meter, why is it being ignored? Probably that is where the Net meter will be corrective, presenting S for a resultant (P1-P). Just a thought. !Thinking pf Oct 30, 2016 at 18:36
• There's a third possibility, that one of the few things still running has terrible PF, and normally its hidden by the rest of the plant averaging it up. Oct 30, 2016 at 19:58
• Not to forget that when solar plant is shut down, the pf improves to an above 0.90 value... so third possibility is remote. Nov 2, 2016 at 15:37
• Update: Net meters solved the problem! Nov 11, 2016 at 18:41

Having read carefully the problem described above, my suggestions are as follows:
1- Inverters are designed at a PF above 0.9 - so inverter PF is assumed to be more than 0.9.
2- As per usual practice, Utility supply is connected on one side of the Main CB (the incoming side) while Inverter supply and Electrical load connected on other side of CB - hopefully correct.
3- Now care to be taken is keeping PF of utility supply healthy because inverter supply PF shall always be 0.9.
4- PFI relay is installed to keep PF healthy, connect it as follows:
4.1- Auxiliary supply (220V/440V) to PFI relay is given from Utility supply i.e. incoming side of CB.
4.2- Install CT on different phase than those used at 4.1 and connect on PFI.
4.3- Carefully calculate C/K Ratio and set it on PFI relay.
5- Hopefully above mentioned problem shall be solved.

• See the last comment to pjc50's answer. The problem was solved by "Net meters" - whatever that means. OP should have posted an answer and accepted it to mark the question solved. Welcome to EE.SE. Jan 6, 2021 at 6:58