9:47 on a Saturday night, in the middle of a headline set. I was standing side-stage at an outdoor amphitheater — 30,000 seats, all of them sold — consulting on the venue’s backup power upgrade, when every stage light, every PA cabinet, and every concession sign died at once. One hard cut. No brownout, no warning.
Here’s what I remember most about that night: nobody panicked. Thirty thousand people in sudden darkness, and the crowd stayed calm. Why? Because about twelve seconds later the house emergency lights came up, the exit signs never blinked at all, and the announcer’s voice rolled back over the PA — thin and small off the emergency system, but there. People filed out in an orderly river toward the gates. The show was dead. The evacuation wasn’t. That night is the cleanest illustration I own of what a venue’s backup power is actually for — and it is not the show.
Every stadium, theme park, and arena generator serves two masters, and buyers keep designing for the wrong one first. Master one is life safety: egress lighting, exit signage, the fire alarm panel, smoke control, elevator recall, the PA channels used for announcements. Master two is everything else — stage lighting, the sound system, video walls, rides, concessions, HVAC. The industry’s unspoken ranking puts life safety first and the show second, and the codes in most jurisdictions say the same thing: emergency circuits typically have to be back within about ten seconds. Ten seconds. That’s the number that decides the whole design.
Look, nobody’s insurance company writes the check because the concert continued. They write it — or refuse to — based on whether fifty thousand people could find the exits. I tell every venue owner the same thing in the first meeting: your generator’s job on its worst night is to move people, not to entertain them. Once that lands, the rest of the design conversation gets a lot more honest.
Show power is a strange animal, and it’s worth understanding before you size anything. A modern concert rig is one of the ugliest electrical loads in the business. Moving-head fixtures and LED walls are switching loads that throw harmonics back at the source. Big audio amplifiers hit the system with sharp, fast current spikes — the kick drum alone can make a current waveform look like a mountain range. Variable-speed drives on arena HVAC add their own harmonic garbage to the mix. On utility power, none of this matters much. The grid is a huge, stiff source that swallows harmonics without flinching.
A generator is not the grid. It’s a small, relatively soft source, and dirty loads show up in its voltage waveform immediately. I’ve watched an undersized set feed a stage rig and produce power so distorted the lighting desk started doing things nobody asked it to do. That same harmonic mess is why I wrote a whole piece on generator THD and sensitive loads — if your venue runs big LED walls or VFD-driven equipment, read it before you buy. The fix is usually some combination of an oversized alternator, a properly matched AVR, and honest load math, not wishful thinking.
The practical rule I give production managers: the generator that runs the show wants to be sized for the show’s nastiest thirty seconds, not its average draw. Averages lie. Peaks trip breakers. One festival crew I worked with had a rig that cruised along fine at two-thirds of the set’s rating — then the headliner hit the first chorus with every fixture at full tilt, the load spiked, and the voltage dipped hard enough to reboot half the stage automation. Same generator, same show, two completely different nights depending on the setlist. Size for the chorus, not the soundcheck.
Theme parks are a different puzzle from stadiums, because the loads don’t just go dark — they strand people. A coaster stopped on a lift hill at 9 p.m. with a full train is not a power problem, it’s an evacuation problem. Dark-ride vehicles stall mid-scene with guests inside. That’s why parks design ride power and life-safety power as separate conversations, and why ride manufacturers publish restart procedures that assume you don’t just slam the power back on.
Here’s the part that surprises park operators: most modern rides run on variable-frequency drives, and VFDs are fussy about the power they accept. Reclose onto a generator whose voltage is still sagging from the initial load slam and the drives fault out — undervoltage trip, and now your ride is dark again even though the generator is running fine. I saw this exact sequence at a park I’d been servicing for about two years. Grid dropped, the set started beautifully, the ATS threw, and half the rides’ drives refused to come back. Not a generator problem. A sequencing problem. We fixed it by programming the ride loads to come on in steps, with the drives’ restart delays honored, instead of one big slam. Twenty minutes of programming saved them from the same embarrassment twice.
And rides mean loads that cycle violently — launch coasters, drop towers. Those are short, brutal current pulses that look a lot like motor starting, which is its own sizing discipline. My motor starting and generator sizing piece walks through the locked-rotor math, and every bit of it applies to a park with three coasters that can all dispatch at once.
This is the question venue managers actually argue about in my meetings, and both answers are defensible. One big set is simpler: one engine to maintain, one fuel system, one ATS scheme. But it has exactly one failure mode, and that failure mode arrives on the worst night of the year.
Multiple smaller sets — paralleled, N+1 – buy you redundancy and something people forget: efficiency. A stadium’s life-safety load alone might be a fraction of the full show load. Run that fraction on a megawatt machine and you’re loafing the engine at light load all night, which is how you earn yourself wet stacking. Run it on one of two smaller sets and both machines work in their happy zone. I’ve also seen parks that keep one permanent set for life safety and rent towable sets for big event days, which is honestly the smartest money move for a venue with a lumpy calendar. You don’t buy a generator for your busiest Saturday; you buy for your typical Tuesday and rent the difference.
Whatever you choose, the transfer scheme has to match the load priority. Life-safety loads go through their own transfer switch with the fast time, show loads through theirs with whatever delay the sequence needs. That separation — one switch, two brains – is the single most common thing I find missing in older venues. The ATS explainer I wrote covers the mechanics; the venue lesson is that the switch layout is the design, not an accessory to it.
Ask a venue manager about their generator and they’ll talk about kilowatts. Ask about fuel and the room goes quiet. A stadium that only needs life-safety power can sip from a sub-base tank for a full evacuation with fuel to spare. A park running show power through a blackout drinks like a small factory — and a Saturday blackout that stretches past midnight is exactly when the fuel truck isn’t coming.
My honest guidance: size the onboard fuel for the longest outage you refuse to be embarrassed by, then add a margin, because outages love to outlast plans. And don’t let the fuel sit. Venue generators are classic standby machines — months of idle, then everything depends on them. Stale fuel in the tank is the quietest way to fail the loudest night. I keep a fuel storage guide for exactly this audience: water, microbes, and the sampling habit that takes five minutes a year.
The amphitheater I opened with? Their monthly test ran the engine unloaded, every month, logbook all green. Sound familiar? It’s the most common lie in standby power. An unloaded test proves the engine starts. It proves nothing about the transfer switch actually transferring, nothing about the life-safety panel picking up, nothing about the rides’ drives accepting generator power. I’ve written about why weekly exercise doesn’t equal readiness, and venues are where that gap bites hardest, because the test that matters is the full-transfer test with the actual loads — and almost nobody runs it.
Once or twice a year, on a dark maintenance day, throw the transfer switches for real. Time the emergency lighting. Watch the egress path. Confirm the PA announcement channels come up. Confirm the rides’ restart sequence behaves. It takes an hour and a tired facilities team, and it is worth more than twelve months of green logbook entries. The night the grid actually dies, nobody in the crowd cares about your logbook. They care that the exit signs are on and somebody with a calm voice is telling them where to go.
That amphitheater show never restarted – you can’t un-ring a blackout for thirty thousand people. But the evacuation took eleven minutes, nobody was hurt, and the venue’s insurance renewal the next year had a very different tone. The owner told me later he’d finally understood what he’d bought. Not a show machine. A people-moving machine that happens to burn diesel. I’ve never heard a better one-sentence spec for venue backup power, so I’ll steal it and give it to you: buy the machine that moves the crowd. The show can wait for the grid.
Email：[email protected]
Work：Monday to Friday,9:30-18:30
