An uninterruptible power supply (UPS) provides instantaneous backup power to connected equipment when the mains supply fails or degrades. It bridges the gap between loss of mains and either restoration of supply or a controlled shutdown — and on sites with standby generation, it covers the seconds a generator needs to start and take load.
On an industrial site the load is rarely just IT equipment. Fire and gas panels, emergency lighting inverters, PAGA amplifiers, control room workstations, network switches and security systems all commonly sit behind a UPS, and each has different tolerance for voltage and frequency variation.
The three topologies, and when each is appropriate
- Standby (offline) — the load runs on mains and transfers to inverter when mains fails. Cheapest, with a short transfer time and no conditioning of incoming power. Suitable for individual workstations, not for critical plant.
- Line-interactive — adds automatic voltage regulation, correcting sags and surges without drawing on the battery. A good balance of cost and protection for network cabinets and small distributed loads.
- Online double-conversion — incoming AC is rectified to DC and re-inverted to AC continuously, so the load is permanently isolated from mains disturbance with effectively zero transfer time. The default choice for control rooms, safety systems and anything that must not see a break.
The temptation is to specify online double-conversion everywhere. It is the safest default, but it is also the most expensive and least efficient, and on a large distributed site the cumulative energy cost is real. Match topology to load criticality rather than applying one answer across the board.
Sizing: two numbers, not one
Sizing a UPS means answering two separate questions: how much load must it carry, and for how long. These are independent, and conflating them is the most common specification error.
- Capacity is limited by both apparent power (kVA) and real power (kW). Modern UPS units are commonly rated at unity power factor, but older equipment is not — check both figures against the actual load.
- Autonomy is set by the battery, and it is not linear. Doubling runtime requires substantially more than double the battery capacity, because discharging a battery faster reduces its usable capacity.
- Allow headroom for growth. Most sites add load over time, and a UPS running near its limit has no margin for a failed module in a redundant configuration.
- Account for inrush. Motor loads and some power supplies draw far more at start-up than in steady state.
Where a standby generator is present, autonomy usually only needs to cover start-up and load transfer plus a safety margin. Where there is no generator, autonomy must cover either the expected outage duration or the time needed for a controlled, safe shutdown of the process.
VRLA or lithium-ion?
Valve-regulated lead-acid (VRLA) batteries remain the most common choice on cost grounds, with a typical design life of five to ten years — though real service life is often shorter, because life falls sharply with elevated temperature. A VRLA battery operated consistently ten degrees above its rated temperature can lose around half its expected life.
Lithium-ion has become a credible alternative. It offers longer service life, higher energy density, much better tolerance of elevated ambient temperature, and faster recharge — at a higher upfront cost, and with different fire safety and transport considerations that must be addressed at design stage. Over a ten to fifteen year horizon the total cost of ownership frequently favours lithium, particularly where replacing batteries means shutting down a critical system.
Most UPS failures are battery failures
The electronics in a modern UPS are generally reliable. The battery is a consumable that degrades continuously from the day it is installed, and it fails silently — a string can appear perfectly healthy on a front panel and still collapse under load. The only way to know is to test it.
- Monitor battery impedance, not just voltage. Rising internal impedance is the earliest reliable indicator of a failing cell.
- Control the battery room temperature. Temperature is the single biggest determinant of battery life.
- Carry out periodic discharge testing under real load, on a planned basis rather than discovering capacity during an outage.
- Replace whole strings rather than individual cells. A new cell in an aged string is dragged down to the condition of its neighbours.
- Keep a documented installation date and a planned replacement date for every string on site.
Monitoring and integration
A UPS that cannot report its status is a single point of failure nobody is watching. Network management cards allow UPS units to report load, battery condition, runtime remaining and alarms into a building management system or SCADA, and to trigger graceful shutdown of servers before autonomy expires. On distributed sites this is what turns a collection of individual units into something maintainable.
How Leap Networks Global approaches it
We specify, install, commission and maintain UPS systems as part of the wider critical infrastructure they protect — control rooms, data centres, security systems, and fire, gas and emergency alerting equipment. That includes load surveys and autonomy calculations, battery replacement programmes, and integrating UPS monitoring into the systems your team already watches.