The honest answer for a plant with motor load and an existing capacitor bank: your measured power factor will get worse before it gets better, and the harmonic problem that appears afterwards was usually already there.
A factory engineer looking at a solar proposal usually has two questions that never appear in the proposal. The tariff has a power factor penalty clause, and there is an automatic capacitor bank sitting on the LT board that somebody set up years ago. What does putting a megawatt of inverters on the roof do to either of them?
This is the answer that surprises people, and it is arithmetic rather than opinion.
A grid-tied solar inverter, in its default mode, supplies almost pure active power at close to unity power factor. It gives you kilowatts. It does not give your motors the reactive power they need, and your motors go on needing exactly as much of it as they did before.
So at the utility meter, the kilowatts drawn from the grid fall while the reactive demand stays where it was. Power factor is the ratio of active power to apparent power, and you have just reduced the numerator without touching the other term. A plant that comfortably held its contracted power factor before solar can drop below the threshold at midday, in bright sun, while generating most of its own energy, and attract a penalty for the privilege.
The solar is not misbehaving. The correction equipment is now sized and controlled for a load profile that no longer exists.
An automatic power factor correction panel will chase the new condition if it is allowed to, and two things stop it.
The first is the current transformer feeding the APFC controller. It must see the same current the utility meter sees, which means it belongs at the incomer, upstream of the point where the solar ties in. If the solar connection lands between the CT and the load, the controller is measuring a current that no longer represents what you are being billed on, and it will correct to the wrong target all day. Moving that CT is the single most common commissioning correction when solar is added to an existing factory board, and it costs almost nothing if it is identified before the tie-in rather than after the first bill.
The second is step resolution. With midday active power much lower, the smallest capacitor step may now be too coarse for the correction the controller is trying to make, so it switches in, overshoots, switches out, and hunts. Hunting wears contactors, cooks capacitors and produces switching transients that annoy everything else on the board. Review the step sizes against the new midday profile, not the old one.
Fixed, unswitched capacitors deserve a separate look. At light load with solar running, they can push you into leading power factor, which some tariffs also penalise and which is harder on equipment than a mild lag.
Grid inverters can supply reactive power themselves, through a fixed power factor setting, a set point or a voltage-dependent curve, and it is a useful trim. Two limits keep it from being a solution. It only works while the plant is generating, so it does nothing at night or through a wet monsoon week. And reactive output consumes apparent power capacity, so a standing power factor setting costs you active energy at full sun. Also check the utility approval, because the operating mode may be specified in it and changing it unilaterally is a compliance issue rather than a settings tweak.
Grid connection standards require inverters to keep total harmonic current distortion low at rated output, with limits on each individual order, and a decent modern inverter sits comfortably inside them. Two real issues nevertheless come up.
The first is a measurement artefact. At low output, on a cloudy morning, the harmonic current is a much larger percentage of a much smaller fundamental, so a meter reports an alarming distortion figure. The absolute harmonic amps are tiny. Always ask what the distortion figure was measured against, and look at harmonic current in amperes rather than as a percentage when the plant is lightly loaded.
The second is real, and it is not the inverter's fault. A capacitor bank and the supply transformer's inductance form a parallel resonant circuit at some frequency. If that frequency sits near a harmonic order your variable frequency drives are already producing, and on six-pulse drives the fifth and seventh are the usual candidates, the bank amplifies it. The symptoms are capacitors running hot, fuses failing, contactors welding and unexplained tripping. Adding solar changes which capacitor steps are in service at which times of day, and that can move the resonance onto a harmonic that was previously missed. The problem then appears the month after commissioning and gets blamed on the solar, when what actually happened is that a pre-existing drive and capacitor interaction was uncovered.
The fix is detuned reactors on the capacitor steps, which shift the resonance below the fifth harmonic, or an active filter where the drive load is large. On any board with significant drive load these should have been fitted in the first place.
Log the incomer for a full week before the plant is built: active and reactive power, power factor profile, voltage and current distortion, individual harmonic orders, and which capacitor steps are in service through the day and night. One instrument, one week, and a report filed with the project documents.
Then log it again after commissioning and compare like with like. Every later dispute about whether the solar caused a power quality problem becomes a two-minute conversation with two graphs, instead of a season of assertions. It is the cheapest insurance in the project.