Every plant manager knows what a full power outage costs. Siemens' 2024 True Cost of Downtime report estimates that unplanned downtime costs the world's 500 largest manufacturers $1.4 trillion annually, about 11% of revenue.
Yet, in my conversations with facility operators, a different problem comes up more often. Most facilities aren't dealing with daily, plant-wide outages. They're dealing with smaller, harder-to-detect power events that trip a single drive, reset a PLC or fault a CNC machine mid-run. The line stops and recovery takes hours. Then it happens again the following week.
These events are voltage sags: brief dips in supply voltage, often lasting less than a second. The source is usually external to the facility. A fault on a nearby feeder, a large motor starting somewhere on the grid, a switching event at a substation miles away. The disturbance reaches a facility's electrical panel without warning.
The office building next door might not notice a voltage dip. Their lights stay on and their computers keep running. But on a manufacturing floor, the same dip is enough to trip a drive or reset a controller mid-cycle. The sag lasts 100 milliseconds. Restarting the line takes hours.
Why this is getting worse
At Torus, I spend a lot of time at the intersection of grid infrastructure and facility operations, and right now several forces are compounding on each other.
Grid infrastructure is aging. Seventy percent of U.S. transmission lines and power transformers are more than 25 years old, well beyond their original design life. That increases the frequency of faults and switching events, the primary drivers of voltage disturbances.
Demand is rising dramatically. NERC's 2026 Long-Term Reliability Assessment projects peak demand growth approaching 5% annually in some regions, driven by data centers, electrification and industrial expansion. As feeders operate closer to capacity, disturbances become more frequent and harder to absorb at the facility level.
And manufacturing equipment is less tolerant than it used to be. As facilities adopt more automation and digitally controlled processes, the voltage threshold for a disruptive event drops. Equipment that handled a 15% dip a decade ago now trips at 10%.
The grid is getting less reliable at the same time equipment is getting more sensitive.
The costs that don't show up
The most visible cost is lost production. When a drive trips or a PLC resets, the line stops. Recovery involves recalibration, clearing material, restarting processes and running quality checks. In food and pharmaceutical manufacturing, an interruption can mean scrapping an entire batch.
What gets overlooked more often is equipment wear. Repeated sags stress components that don't immediately fail: motors operating below proper voltage, devices cycling unexpectedly, capacitors absorbing transients. This shows up as increased maintenance and premature failures, without a clear root cause.
Then there's the cost that rarely gets measured. Facilities experiencing frequent nuisance trips adapt by running more conservatively, avoiding simultaneous startups, staying below capacity. These decisions protect uptime but quietly limit throughput. I've talked to operators who have been doing this for years without thinking of it as a cost. It is.
What actually works
Backup generators aren't designed for this problem. They take 10 to 30 seconds to start. A voltage sag is over in milliseconds.
Addressing the issue starts with visibility. Power quality monitoring at a facility's main electrical panels quantifies how often sag events occur, how severe they are and which equipment gets hit hardest. Without that data, the problem remains invisible.
From there, manufacturers evaluate mitigation options. UPS systems, dynamic voltage restorers and capacitor banks are effective at the machine level, but they protect only the circuits they're installed on.
A system-level approach installs energy storage where utility power enters the facility. It injects or absorbs power in real time, stabilizing voltage across the full electrical network before disturbances reach equipment. Paired with software, it also provides a continuous record of events and system response.
For many facilities, the economics extend beyond power quality. The same infrastructure also reduces peak demand, which accounts for 30 to 50% of electricity costs at most mid-to-large manufacturing operations.
A problem worth measuring
The most disruptive power events in manufacturing aren't always the ones that shut everything down. They're the ones that interrupt a process, delay a line and erode performance over time while the lights stay on.
Manufacturers that start measuring these events and treating them as an operational variable are in a better position to control both reliability and throughput.
