Automatic transfer switches and the transfer sequence

A generator that starts but is not connected to the load does nothing. The equipment that decides when the utility has failed, starts the generators, connects them to the load, and later returns the load to the utility, is the transfer system. In smaller buildings this is an automatic transfer switch (ATS). In large data centers it is often a set of motor-operated breakers in paralleling switchgear controlled by a programmable controller, but the logic is the same. This lesson walks through the transfer sequence second by second, explains open and closed transitions and retransfer, and shows what the technician sees on the screens and in the logs.

The automatic transfer switch

An ATS has two inputs, normal (the utility) and emergency (the generator), and one output to the load. It contains a switching mechanism that connects the output to one input or the other, and a controller that monitors both sources. When the normal source fails or goes outside its voltage and frequency limits for longer than a set delay, the controller signals the generator to start. When the generator's voltage and frequency are within limits, the controller transfers the load to the emergency source. When the normal source returns and stays stable for a set time, the controller transfers the load back and, after a cool-down period, stops the generator.

At Stonebridge, the Hall A block uses ATS units for its mechanical and house loads and motor-operated breakers in the generator switchboard for the UPS inputs. Halls B, C and D use medium-voltage paralleling switchgear: all the generators of a block are synchronized onto a common generator bus, and motor-operated breakers transfer each unit substation from utility to generator bus. Both arrangements follow the sequence below.

The sequence after a utility failure

The load is applied to the generators in steps, not all at once. A diesel generator's frequency dips when a large load is suddenly added, and too large a step can cause a dip big enough to trip loads or stall the engine. UPS rectifiers therefore walk in their load over several seconds, and the mechanical plant's chillers, pumps and fans are restarted by the building management system in a sequence planned during commissioning. The cooling restart is critical: the IT load keeps producing heat throughout, and halls without cooling warm up quickly, especially the dense AI racks in Hall D, which is why Hall D's coolant pumps and CDUs are on UPS.

# Simulate a transfer sequence and check the total time to load (illustrative timings, seconds)
events = [
    ("utility lost", 0.0),
    ("start delay ends, start signal", 2.0),
    ("engines firing", 2.0 + 1.6),
    ("first generator at voltage and frequency", 2.0 + 6.4),
    ("paralleled sets on bus", 2.0 + 7.2),
    ("transfer to generator", 2.0 + 7.6),
]
ups_runtime_s = 23.0 * 60
for name, t in events:
    print(f"{t:6.1f} s  {name}")
t_load = events[-1][1]
print(f"Load on generator after {t_load:.1f} s; UPS battery reserve at present load {ups_runtime_s / 60:.1f} min")
print(f"Battery reserve used: {t_load / ups_runtime_s * 100:.1f} %")

   0.0 s  utility lost
   2.0 s  start delay ends, start signal
   3.6 s  engines firing
   8.4 s  first generator at voltage and frequency
   9.2 s  paralleled sets on bus
   9.6 s  transfer to generator
Load on generator after 9.6 s; UPS battery reserve at present load 23.0 min
Battery reserve used: 0.7 %

Under ten seconds of the roughly twenty-three-minute battery reserve are used in a normal transfer. The reserve is there for when the sequence does not go normally: a generator fails to start, a breaker fails to close, or the paralleling controller needs a second attempt.

Open and closed transition

When an ATS transfers from one live source to another, it can do so in two ways. In an open transition, also called break-before-make, the switch disconnects from the first source before connecting to the second, leaving the load without power for a fraction of a second. For loads on UPS, this short break is covered by the batteries. In a closed transition, or make-before-break, the switch connects to the second source before disconnecting from the first, so the two sources are briefly paralleled. Closed transition needs the two sources to be synchronized and is limited to a fraction of a second by utility rules, because the generator is briefly connected to the utility network.

After a utility failure, the first transfer is always an open transition, because the utility is dead. Closed transition matters on the way back: when the utility returns, a closed-transition retransfer moves the load back to the utility without the second interruption an open transition would cause. Stonebridge's transfer equipment uses closed-transition retransfer, approved by Dominion Energy under the interconnection agreement. Some ATS units offer delayed transition, which pauses in a neutral position to let motor loads slow down before reconnecting, protecting motors from out-of-phase reconnection.

Retransfer and cool-down

When the utility comes back, the transfer controls do not return the load immediately. Utility supplies often come back and fail again several times while crews restore the network. The controller waits for the utility to be stable for a set time, the retransfer delay, before moving the load back. Common settings range from five to thirty minutes; Stonebridge uses thirty minutes after a storm-related outage and lets the operations team extend it if the utility advises that more switching is planned. After retransfer, the generators run unloaded for a cool-down period, typically five minutes, so that the engines cool evenly before stopping. The controller then stops the sets and returns them to automatic, ready for the next event.

Many sites choose not to retransfer automatically at all after a major event, but to keep the load on generator until the operations team decides the utility is reliable. The decision weighs the risk of another utility failure against the generator run hours, fuel and emissions. Whatever the policy, it is written into the site procedures so that nobody improvises during an event.

Controller settings a technician should know

The transfer controller holds a set of timers and limits. Technicians do not change them, but knowing them makes the logs readable. The values below are typical; each site's are recorded in its operating procedures.

Paralleling switchgear

In a paralleling plant, each generator has its own breaker onto a common generator bus. The first set to reach voltage and frequency closes onto the dead bus. Each following set is synchronized by its controller: the controller adjusts the engine speed until the set's frequency and phase angle match the bus, then closes its breaker at the right moment. Once enough sets are on the bus to carry the load, the transfer breakers close. If one set fails, the load is shared by those that are running, and if the remaining capacity is not enough, the controller sheds the lowest-priority loads, such as office lighting and some mechanical equipment, to protect the critical load.

# Paralleling plant: is there enough capacity after a failed start? Which loads are shed? (illustrative)
sets = {"GEN-C1": 3000, "GEN-C2": 3000, "GEN-C3": 3000, "GEN-C4": 3000, "GEN-C5": 3000}
failed = ["GEN-C4"]
loads = [  # (priority, load, kW) - priority 1 is most critical
    (1, "UPS inputs, Halls B and C", 7200), (1, "cooling for Halls B and C", 3300),
    (2, "office and support areas", 250), (3, "car park lighting and heaters", 220),
]
available = sum(kw for g, kw in sets.items() if g not in failed) * 0.9   # plan to 90 % of rating
running = 0
for prio, name, kw in sorted(loads):
    if running + kw <= available:
        running += kw
        print(f"keep   priority {prio}: {name:32s}{kw:6.0f} kW")
    else:
        print(f"SHED   priority {prio}: {name:32s}{kw:6.0f} kW")
print(f"Available {available:,.0f} kW, carried {running:,.0f} kW ({running / available * 100:.0f} %)")

keep   priority 1: UPS inputs, Halls B and C         7200 kW
keep   priority 1: cooling for Halls B and C         3300 kW
keep   priority 2: office and support areas           250 kW
SHED   priority 3: car park lighting and heaters      220 kW
Available 10,800 kW, carried 10,750 kW (100 %)

With one of the five sets failed, the four remaining can carry all the critical load and the support areas, so only the car park lighting is shed. This is the N+1 design doing its job. If a second set failed, more load would have to be shed, and the redundancy whiteboard would turn red.

Bypass-isolation ATS units

ATS units serving critical loads are often the bypass-isolation type. A second switch inside the same enclosure can carry the load while the main transfer switch is withdrawn for maintenance or repair, without interrupting power. Operating a bypass-isolation ATS is done by trained critical facilities staff with a procedure, much like a UPS maintenance bypass. The ATS front shows which source is connected, the position of the bypass switch, and any alarms.

What to watch and record during a transfer

• The time the utility was lost and the time the load was on generator, from the controller or monitoring system.
• Which generators started, how long each took to be ready, and any that failed.
• UPS battery time used, and the battery state of charge as recharging begins.
• Cooling restart: CRAH, chiller and CDU restart times and any hall temperature rise.
• Any load shedding and which loads were affected.
• When the utility returned, when retransfer happened and how long the generators ran.

The monitoring system records all of this, but during a real event the technician's written notes of what was seen and heard are often what makes the incident report clear. Lesson 12 follows a full event and the report that comes after it.