Field reference for grid scale battery storage technicians. 11 entries, taken from the StoreWatt app.
These notes are the web copy of the StoreWatt reference, built for grid scale battery energy storage technicians. The same library is free inside the app, with no time limit and no account, and it works with no signal.
Reference notes from StoreWatt, the offline field toolkit for grid scale battery energy storage technicians. It works with no cell signal, because the sites do not have any.
Get StoreWatt on the App StoreThe power conversion system converts the DC bus to grid AC and back. Its protection stack typically includes DC and AC over and under voltage, overcurrent, short circuit, ground fault, over temperature, anti islanding, and an emergency shutdown chain. Communication paths are CAN, RS 485, Ethernet, and Modbus in both RTU and TCP flavors. Most PCS faults come with an event code and a timestamped log; pull the log before cycling anything, because the first fault in the sequence is the story and the rest are consequences.
Pull the event log before cycling anything, because the first fault in the sequence is the story and everything after it is consequence; note the timestamp and check whether other units logged the same thing at the same moment, since that separates a grid event from an equipment fault immediately; identify which protection function tripped, because DC, AC, thermal and ground faults lead to completely different investigations; check the communication path is healthy before trusting any reported value
The DC bus left the PCS operating window. Under voltage usually means the battery is empty, a rack dropped off and sagged the bus, or contactors opened upstream. Over voltage points at charge control, a regen transient, or a voltage measurement fault. Cross check the PCS DC reading against the BMS bus voltage. If they disagree by more than a few volts, chase the measurement before the battery.
Confirm which side actually moved before resetting, because a real bus excursion is a battery or control problem that a PCS reset only hides; cross check the PCS DC reading against the BMS bus voltage, and if they disagree by more than a few volts chase the measurement before the battery; for under voltage, check whether the battery is simply empty, a rack dropped off and sagged the bus, or contactors opened upstream; for over voltage, check charge control, a regen transient, and the voltage measurement in that order; check the BMS for rack level alarms in the same window, since a rack dropping out explains a bus sag without any PCS fault at all
Grid side voltage or frequency left the ride through window, so the PCS tripped or rode through per its IEEE 1547 settings. On a healthy plant these events cluster across every PCS at the same timestamp, which tells you it was the grid, not the equipment. One unit tripping alone on grid limits points at its own measurement, its settings file, or its AC connection.
Correlate the timestamp across every PCS on the plant first, because that single check separates a grid event from an equipment fault; if every unit tripped together, treat it as the grid and check with the utility rather than investigating the equipment; if one unit tripped alone, look at that unit's measurement, its settings file and its AC connection; check the IEEE 1547 ride through settings match the interconnection agreement, and do not widen them to stop the alarms; check AC terminations on the affected unit for a high resistance joint, which raises local voltage under load
Current exceeded the PCS limit on the AC or DC side. Occasional trips during aggressive setpoint steps or grid transients happen; repeated trips at the same operating point are a real signal. Causes include an undersized setpoint ramp, a failing current sensor, an AC side fault, or a DC bus problem making the PCS work harder for the same power. Compare the trip current and side against the event log history before touching hardware.
Do not raise current limits to clear repeated trips; find what changed instead; compare the trip current and which side it occurred on against the event log history before touching hardware; check whether trips cluster at the same operating point, because occasional trips on aggressive setpoint steps or grid transients are normal and repeated trips at one point are a real signal; check the setpoint ramp rate, since an undersized ramp produces exactly this; check the DC bus, because a bus problem makes the PCS work harder for the same power; check the current sensor against another unit under the same duty before replacing anything
The PCS detected a short on its output or across its bridge. This is a protective trip against a real fault current event, not a nuisance alarm. Possible causes: a cable or busbar fault, a failed power module, or switchgear closing into a fault. A DC side short on a battery plant is especially serious because the battery can feed enormous current with no zero crossing to help clear it.
Stop work. Do not reset and retry, because this is a protective trip against real fault current rather than a nuisance alarm; treat a DC side short as especially serious on a battery plant, since the battery feeds enormous current with no zero crossing to help clear it; isolate under the qualified person's direction before any inspection; inspect the fault path for cable and busbar damage, then the power modules, then the switchgear that may have closed into the fault; test insulation before any re energisation; establish what caused it, because re energising into an uncleared fault repeats the event with more damage
Current found a path to ground on the AC or DC side. On the DC side this overlaps with the BMS insulation monitor; on the AC side it is residual or zero sequence current. Either way, some conductor or component is leaking where it should not, and grounded metal can be energized. Moisture, damaged cable, coolant intrusion, and failed surge devices are the usual suspects.
Treat everything as live and stop work, because grounded metal can be energised; do not reset to see whether it clears; establish whether it is the AC or DC side, since DC overlaps with the BMS insulation monitor and AC is residual or zero sequence current; check the usual paths in order: moisture, damaged cable, coolant intrusion, then failed surge devices; check whether other units or the BMS insulation monitor are reporting at the same time, which localises it to a shared cause such as weather; locate the fault under the qualified person's direction and prove insulation before returning to service
Heatsink, coolant, or cabinet temperature crossed the limit and the unit derated or tripped. Check the boring things first: blocked or dirty filters, failed fans or pumps, low coolant, radiator fouling, doors left open in summer, and control room HVAC failures for the electronics. A unit that derates every afternoon at peak sun and load is a cooling capacity problem, not a defect that a reset fixes.
Restore cooling before restoring load, because repeated thermal trips age power electronics fast; check the boring things first, since they are usually it: blocked or dirty filters, failed fans or pumps, low coolant, fouled radiator, doors left open; check control room HVAC for the electronics, which is an easy one to miss; check whether it derates at the same time each day, because a unit that derates every afternoon at peak is a cooling capacity problem rather than a defect; check the temperature sensor against another unit before assuming the reading is real; do not raise the trip threshold to keep the unit online
The PCS believed it was energizing an island and disconnected, as designed. Real islanding events are rare on transmission connected plants; most anti islanding trips trace to grid switching operations, weak grid conditions, or oversensitive settings interacting with the plant controller. Cluster analysis helps: all units together says the grid did something; one unit alone says look at that unit's settings and measurements.
Run the cluster analysis first, because it answers the question: all units together says the grid did something, one unit alone says look at that unit; check for grid switching operations at that timestamp, since most anti islanding trips trace to switching rather than genuine islanding; check the settings against the interconnection agreement if one unit is oversensitive relative to its peers; check how the plant controller interacts with the anti islanding settings, because weak grid conditions plus aggressive settings produce nuisance trips; treat genuine islanding as rare on a transmission connected plant, and do not weaken the protection to stop nuisance trips
The hardwired emergency shutdown loop is open: a pressed button, a fire panel contact, a door interlock, a broken wire, or a stuck relay anywhere in the chain stops the PCS. The log usually says only that the chain opened, not where. Walk the loop with the drawings. Fire and gas panel contacts into the ESD chain are common on BESS sites and trip the whole block by design.
Find out why the chain opened before resetting it, and if a fire or gas signal opened it treat that as real until proven otherwise; walk the loop with the drawings, because the log usually says only that the chain opened and not where; check the fire and gas panel contacts first on a BESS site, since those are wired into the chain by design and trip the whole block; check for a pressed button and a door interlock next; check for a broken wire or a stuck relay anywhere in the chain, which is what remains once the deliberate causes are ruled out; reset only once the cause is understood and the site is confirmed safe
The PCS lost its link to the EMS, plant controller, or BMS. Depending on configuration it holds last setpoint, ramps to zero, or trips. Modbus timeouts, duplicate IP addresses after a controller swap, failed switches, and broken RS 485 segments are the workhorse causes. A PCS running normally at zero setpoint after a comm loss is doing its job; the fault is in the network, not the power stage.
Recognise that a PCS sitting at zero setpoint after a comm loss is doing its job, so the fault is in the network rather than the power stage; check what the configuration does on comm loss, since holding last setpoint, ramping to zero and tripping are all normal designs and change the urgency; check for duplicate IP addresses after any controller swap, which is a workhorse cause; check Modbus timeouts and whether the master is polling faster than the device answers; check the switch and the physical path, and for RS 485 check the segment for a break, termination and bias; restore the link before investigating any electrical symptom, because stale setpoints look like power faults
PCS cabinets carry their own auxiliary loads: control power supplies, fans, pumps, and heaters. A failed auxiliary breaker or supply can present as anything from a dead HMI to random restarts. Aux power quality problems, like a sagging control transformer, cause faults that follow plant load. When a unit logs scattered unrelated faults, check its control power first.
Check control power first whenever a unit logs scattered unrelated faults, because that is the usual explanation; check the auxiliary breaker and the control power supply; check aux power quality, since a sagging control transformer causes faults that follow plant load and look mysterious; check fans, pumps and heaters individually, because a failed auxiliary can present as anything from a dead HMI to random restarts; check whether the faults correlate with plant load or time of day, which points straight at aux power quality
Reference notes from StoreWatt, the offline field toolkit for grid scale battery energy storage technicians. It works with no cell signal, because the sites do not have any.
Get StoreWatt on the App StoreThese notes are a field aid, not a substitute for the governing codes, the stamped drawings, the authority having jurisdiction, or manufacturer manuals. Verify against the current documentation for your installed equipment.