Again Cos Fi

Because it isn't worth the effort for the additional 0.03.

To expensive to do so with too little gain achieved.

This is one of those times when the Engineers and Accountants are in agreement.

Economics....

..."The demand charge is more complicated. It's typically based on the peak kW demand over a given 15-min, 30-min, or 60-min interval during the month, which is nominally multiplied by the demand charge rate ($/kW).

Penalties for low PF - typically anything less than .95 - have become increasingly common. Most utilities use one of the following two common formulas for determining the billed demand when PF is lower than 0.95 lagging:

KW_Billed = kW_Actual x (0.95/PF)

or

KW_Billed = kW_Actual x (1 + 0.95 - PF)

Both of these formulas are applied only when PF is less than .95 lagging. Otherwise, the billed demand is the same as the actual demand.

The difference between the amount paid for the billed demand and the amount that would be paid for the actual demand is often termed the PF penalty. This quantity, calculated using the following formula, is generally responsible for the bulk of the justification for PF correction:"....

..." Most of the economic justification for PF correction will come from eliminating penalty fees on your utility bills. Reducing energy losses can contribute to the justification, but contrary to manufacturer claims, it's generally insufficient by itself to justify the cost of installing correction. Some manufacturers boast their products can yield energy savings as high as 10%, but it's typically only possible to reduce average kW demand by .5% to 1.5%. "...

http://ecmweb.com/content/inside-pq-0

http://www.michaelsenergy.com/industrial-energy-briefs-2/miscellaneous/power-factor-correction/

PF correction yields $avings even if there are no penalties if you operate your own generators. Fuel savings.....

The concept is called the "Law of Diminishing Returns". It just means that to go from .8 PF to .95 PF is relatively inexpensive for the gains made in the big picture of things. But going from .95 to .99 becomes exponentially more expensive and the benefits of doing so are exponentially fewer at the same time.

Why "expensive" - capacitive load losses are not large indeed, aint they ?

Moreover, if you go "overboard" into a leading PF, some instabilities can occur.

BTW, the Greek letter is "phi", not "fi".

"Phi" is not 1 but 3 latin letters, neither Greek.

To be above 0.96 is also not allowed. Now it is 0.98, the target being 0.96. If I drive it with capacitors to 1 and then to the leading 0.96 what instabilities can occur ?

The suppliers put that out there is the minimum power factor for complex (combined resistive and reactive loading at an industrial facility), not the maximum. If a facility has enough capacitor banks to shift their local power factor back up to 0.99, then I am not sure the power supplier would mind at all. The power companies like to provide MW, not so much MVARS. MW are easier to transmit at lower loss factor (I suppose) than are MVARS at a low power factor. Usually the Pf limitation may only apply during peak demand time when HVAC loads are highest on the grid. I am sure the power suppliers have less problem with reactive loads being higher at times when the grid is off peak demand (typically at night).