Published by Terry Chandler, Director of Engineering, Power Quality Thailand LTD/Power Quality Inc., USA.
E-mail: terryc@powerquality.org, terryc@powerquality.co.th
FYI, PQView calculates these automatically from the PQView database of all instruments. For more information on PQView, you can visit the blog post PQView & PQView DE
System Average RMS (Variation) Frequency 
represents the average number of specified rms variation measurement events that occurred over the assessment period per customer served, where the specified disturbances are those with a magnitude less than x for sags or a magnitude greater than x for swells:
where
x = percentage of nominal rms voltage threshold; possible values – 140, 120, 110, 90, 80, 70, 50, and 10
= number of customers experiencing short-duration voltage deviations with magnitudes above X percent for X > 100 or below X percent for X percent for X < 100 due to measurement event i
= total number of customers served from the section of the system to be assessed
Notice that SARFI is defined with respect to the voltage threshold x. For example, if a utility has customers that are only susceptible to sags below 70 percent of nominal voltage, this disturbance group can be assessed using 
. The eight defined threshold values for the index are not arbitrary. They are chosen to coincide with the following:
140,120, and 110. Overvoltage segments of the ITI curve.
90, 80, and 70. Undervoltage segments of ITI curve.
50. Typical break point for assessing motor contactors.
10. IEEE Standard 1159 definition of an interruption.
An increasing popular use of SARFI is to define the threshold as a curve. For example, 
would represent the frequency of rms variation events outside the ITI curve voltage tolerance envelope. Three such curve indices are commonly computed:
This group of indices is similar to the System Average Interruption Frequency Index (SAIFI) value that man utilities have calculated for years. , however, assesses more than just interruptions. The frequency of occurrences of rms variations of varying magnitudes can be assessed using . Note that is defined for short-duration variations as defined by IEEE Standard 1159.
There are three additional indices that are subsets of . These indices assess variations of a specific IEEE Standard 1159 duration category:
SARFI for the EPRI DPQ project
Table 8.1 shows that statistics for various forms of SARFI computed for the measurements taken by the EPRI DPQ project. These particular values are rms variation frequencies for substation sites in number of events per 365 days. One-minute temporal aggregation was used, and the data were treated using sampling weights. This can serve as a reference benchmark for distribution systems in the United States.
Example index computation procedure
This example is based on actual data recorded on one of the feeders monitored during the EPRI DPQ project. This illustrates some of the practical issues involved in computing the indices.
Reference
Electrical Power Systems Quality, Second Edition. Chapter 8 Power Quality Benchmarking.
Power Quality Blog 