Computing SAIDI, SAIFI, CAIDI & MAIFI

9 min · difficulty 5/10

This month’s log shows three sustained outages and one momentary interruption. The utility serves 10,000 customers. Event one dropped 500 customers for 90 minutes, event two dropped 1,200 for 45 minutes, event three dropped 300 for 120 minutes, and a recloser blinked 800 customers for a few seconds. Compute SAIDI, SAIFI, CAIDI, and MAIFI before reading on.

Start with the two indices you build directly from the log. SAIFI is total customer interruptions over total customers served (each event contributes its customer count, so a customer hit twice counts twice): (500 + 1,200 + 300) / 10,000 = 0.20 interruptions per customer. SAIDI is total customer-minutes over customers served: (500x90 + 1,200x45 + 300x120) / 10,000 = 135,000 / 10,000 = 13.5 minutes. SAIDI is always reported in minutes, so label the unit every time the number appears.

CAIDI is the one that catches people. You do not measure it from scratch. CAIDI equals SAIDI divided by SAIFI, so here it is 13.5 minutes / 0.20 = 67.5 minutes, the average restoration time for a customer who was actually interrupted. Sample the log directly and you get the same 67.5: total customer-minutes 135,000 over total customer interruptions 2,000. It falls out of the other two by algebra, which is exactly why a CAIDI that moves without SAIDI or SAIFI moving is a data error, not a reliability change.

From outage log to SAIFI, SAIDI, CAIDI and MAIFI A left column lists an outage log for 10,000 customers served: three sustained events (500 customers for 90 minutes, 1,200 for 45 minutes, 300 for 120 minutes) plus a dashed orange momentary recloser blink of 800 customers. Arrows flow from the sustained events into two right-column boxes, SAIFI equals 0.20 and SAIDI equals 13.5 minutes. A gold dashed derivation arrow combines those two into a third box, CAIDI equals 67.5 minutes, labeled SAIDI divided by SAIFI. A separate dashed orange arrow runs only from the momentary event into a fourth box, MAIFI equals 0.08, kept apart from the sustained math. Outage log → reliability indices (10,000 customers served) Sustained events Event 1 (sustained) 500 cust × 90 min Event 2 (sustained) 1,200 cust × 45 min Event 3 (sustained) 300 cust × 120 min Recloser blink (momentary) 800 cust, < 5 min SAIFI = 0.20 interruptions / cust SAIDI = 13.5 min customer-minutes / cust CAIDI = 67.5 min SAIDI ÷ SAIFI MAIFI = 0.08 momentary / cust CAIDI is derived from SAIDI ÷ SAIFI. MAIFI stays on its own line, separate from the sustained math.
CAIDI is derived (SAIDI divided by SAIFI), and momentary interruptions stay out of the sustained-outage math entirely.

The recloser blink is a different animal. MAIFI is a separate momentary counter that stays out of the sustained-outage math: momentary customers interrupted over customers served, 800 / 10,000 = 0.08 momentary interruptions per customer. Momentaries are counted apart because reclosers in fuse-saving mode clear temporary faults with a fast open-and-reclose before a fuse blows, so a healthy feeder can log many brief blinks that never became real outages. Add them into SAIDI and they would bury the sustained numbers that matter.

Whether you are a buyer comparing vendors on quoted reliability, an operator closing out a month, or an engineer defending a rate case, report SAIDI in minutes, treat CAIDI as SAIDI over SAIFI, and keep MAIFI on its own line. Those three habits are the difference between numbers a regulator trusts and numbers that quietly contradict each other.

SAIDI and SAIFI targets drive capital planning, and capital planning drives what gets ordered and when. The Feeder tracks the reliability investment and regulatory signals behind those orders, free, every week.

Question 1 of 3

The utility serves 10,000 customers. Sustained events: 500 out for 90 min, 1,200 out for 45 min, 300 out for 120 min. Compute SAIDI in minutes.

Educational material only. This is not engineering, safety, or procurement advice. Confirm any value against manufacturer documentation and a licensed professional before specifying equipment.