Voltage Sags and Swells: Why Your Plant Trips

Part 19 of 23 in Cobler's Electricity Fundamentals series. New here? Start with the course map.

A fault on someone else's TNB feeder, a kilometre away, drops the voltage in your plant to 70 percent for five cycles. That is one tenth of a second. Upstairs the office lights barely flicker and nobody notices. Down on the floor a contactor lets go, a robot faults out, and the injection-moulding line stops cold. Restarting it, purging the scrap and re-qualifying the first good parts takes four hours. The event that caused it was over before you could blink.

That brief drop is a voltage sag, the single most common power-quality problem industrial sites face. It never trips your main breaker and never shows on your bill. This is why sags are so frustrating: the damage lasts four hours, the cause lasts a tenth of a second, and nothing in between records that it happened.

What is a voltage sag, and how is it different from an outage?

A voltage sag is a short drop in supply voltage that recovers on its own, without the power ever fully going off. An outage is different: the voltage collapses to nothing and stays there until someone or something restores it.

The IEEE 1159 standard draws the line with numbers. A sag (the American term; "dip" in IEC usage) is RMS voltage (the effective, working voltage) falling to between 0.1 and 0.9 per unit of nominal, that is, to between 10 and 90 percent of normal supply voltage, lasting anywhere from half a cycle to one minute (powerquality.blog). At Malaysia's 50 Hz, one cycle is 20 milliseconds, so half a cycle is 10 ms. A swell is the mirror image, a rise to between 1.1 and 1.8 per unit over the same range of durations. Drop below 0.1 per unit and it is no longer a sag but an interruption, which is the technical name for a genuine outage.

Sags themselves come in flavours by duration: instantaneous (half a cycle to 30 cycles), momentary (30 cycles to 3 seconds) and temporary (3 seconds to a minute). Almost every event that stops a plant is in the instantaneous band, gone in well under a second.

What causes voltage sags?

Three things cause most of them: a fault somewhere on the utility network, lightning, and large motors starting up. EPRI attributes roughly 92 percent of all transmission and distribution power-quality problems to voltage sags, and its US monitoring found a typical customer sees somewhere between 40 and 60 sag events a year (powerquality.blog).

The most common single cause is a fault on a nearby feeder. Most of these are single-line-to-ground faults: a tree branch, a failed cable joint, an animal across a busbar, or a contractor's excavator on the next lot. For the few cycles it takes the utility's protection to clear that fault, the voltage across a wide area sags, including yours. Someone else's excavator becomes your downtime, and the journey electricity takes from power station to your plug means you share a feeder with neighbours you will never meet.

Lightning matters more here than almost anywhere on Earth. The Klang Valley averages around 240 thunderstorm days a year, against roughly 10 for London, and Subang ranks among the most lightning-prone locations on the planet (Daily Express). A strike does not have to hit your building to hurt you. A strike onto the network kilometres away causes a fault, the fault causes a sag, and the sag reaches your contactors. Malaysia is not quite the world's lightning capital, that title usually goes to Bogor in Indonesia, but it sits firmly in the top tier, and any plant here plans for it.

The third cause is internal, and it is yours. When a large induction motor starts direct-on-line (switched straight onto full mains voltage, with no soft-starter), it draws six to ten times its running current for a second or two, and that inrush pulls your own bus voltage down. In a typical plant, internal causes like this account for 60 to 70 percent of the sags equipment actually sees. The chiller compressor that kicks in and browns out the PLC two panels over is a sag you created.

Why does a half-second dip stop a whole production line?

Because the cheapest component gives up first, and it takes everything downstream with it. Survival is decided at the weakest link, not the strongest.

The usual weak link is an ordinary AC contactor, the electromechanical relay that holds a motor's power on. Its coil relies on magnetic force to keep the contacts closed, and that force falls with voltage. Many contactors drop out when the voltage sags to somewhere between 70 and 85 percent of nominal, some as low as 57.5 to 65 percent (Lee Contracting). Once it opens, the motor stops, and no amount of recovered voltage a few cycles later closes it back by itself.

Variable-frequency drives fail differently but just as readily. A VFD rectifies AC into a DC bus, and it watches that bus for undervoltage to protect its own components. When the sag pulls the bus down, commonly around 65 to 70 percent of nominal, the drive trips itself off on an undervoltage fault (Voltage Disturbance). It is doing exactly what it was designed to do, and the result is a stopped conveyor.

Then there are the controllers. PLCs and their 24 V control supplies, robots and vision systems simply reset if their supply is not backed up. Often the controller blinks out before the motor even notices, so the brain stops before the muscle does. Put these together and a tenth-of-a-second event becomes a full line stop, hours of scrap, and a re-qualification run before the first sellable part comes off.

Deep or long: which decides whether equipment survives a sag?

Both, together, not voltage alone. A sag that goes very deep but lasts only a couple of cycles can be harmless, while a shallower sag that drags on for half a second will trip things the deep-but-brief one did not.

This is the idea behind the ITIC curve (formerly CBEMA), a voltage-versus-duration envelope first drawn for computer equipment and standardised in its current form in 1996 (Voltage Disturbance). You do not need the chart to use the thinking. Every piece of equipment has a boundary: for a very short event it can tolerate a very deep dip, but the longer the dip lasts the less depth it will survive. A sag to 70 percent for two cycles sits inside almost anything's tolerance; the same 70 percent held for 40 cycles, about 0.8 seconds, falls outside the ITIC envelope, which only tolerates 70 percent for around half a second. One caveat: the ITIC curve is a 60 Hz American reference, so treat it as a way to reason about depth and duration together, not as a local compliance limit.

How do you ride through a voltage sag?

You do not try to hold up the whole plant. You protect the parts whose loss cascades, and you buy equipment that can take a punch.

- Put a UPS on the controls, not the whole line. A small UPS holding up your PLCs, HMIs, relays and vision systems means the brain never resets. The motors may coast for a moment, but if the controls stay alive the line recovers instead of faulting. Backing up every motor is expensive and usually unnecessary; backing up the 24 V control bus is cheap and decisive.
- Fit contactor ride-through or coil-hold modules. These keep a contactor energised through a short sag so it does not chatter or drop out, closing the most common single point of failure.
- Use dip-proofing inverters or a dynamic voltage restorer. For a critical sub-process, a device that injects the missing voltage for the duration of a sag can hold the load up through events that would otherwise trip it.
- Specify sag tolerance when you buy. For semiconductor and similar tools, the SEMI F47 standard defines how deep and long a sag a machine must ride through (Voltage Disturbance). Ask for immunity ratings before you sign the purchase order, because the cheapest way to survive sags is to not buy equipment that folds at the first one.

Any of these is far cheaper than the downtime it prevents. ABB puts industrial downtime at up to 500,000 US dollars an hour (ABB). A four-hour restart after a one-tenth-of-a-second dip is not a rare disaster, it is a Tuesday.

Why does a voltage sag leave no trace on your bill?

Because your energy meter measures kWh over 30-minute windows, and a sag is over in milliseconds. It draws almost no extra energy, so it never registers. The line stopped, the shift lost four hours, and the meter shows nothing unusual. When you call TNB to argue the fault was on their side, you have no timestamp, no depth, no duration, nothing to put on the table.

This is the operator's real problem, and it is a measurement problem before it is an electrical one. Diagnosing sags, distinguishing a utility fault from your own motor inrush, and settling disputes all need timestamped event records that a billing meter will never give you. CobiNeural monitors power quality separately from your billing meter and logs sag and swell events with a timestamp, and its Alerts module pushes those notifications straight to your team by WhatsApp and email, so the record exists before anyone starts guessing. It runs today in hospital deployments, where a dip that resets a PLC is not an academic exercise, and where knowing exactly when and how deep the event was is the difference between a fix and a shrug.

This is Part 19 of 23 in Cobler's Electricity Fundamentals series. Previous: Electrical Harmonics: How VFDs Distort Your Power. Next: Earthing and RCDs: Why Birds Don't Get Shocked.

Sags cost you hours and leave no evidence. See what a timestamped power-quality record looks like for your site: talk to us about CobiNeural.