Reliability — Field Failure Cohort

// A field-failure cohort for one hardware fleet: units shipped, then
// registered, then followed for twenty-four months, with every unit that
// left the cohort attributed to a named failure mode. The FMEA scores
// those same modes, so the ranking and the flow are one document.

fmea inverter {
  title: "String inverter — field returns"
  system: "Rooftop PV inverter, 5 kW class"
  team: "Reliability engineering"
  revision: "rev 7"
}

failure_mode dc_arc {
  component: "DC input stage"
  mode: "Arc fault at the field-wired terminal block"
  cause: "Lug torque drifts below spec after thermal cycling"
  effect: "Unit trips and locks out; installer callout to reset"
  severity: 8
  occurrence: 4
  detection: 5
}

failure_mode fan_seize {
  component: "Heatsink fan"
  mode: "Bearing seizes, unit derates"
  cause: "Dust ingress past an IP54 gasket in dry climates"
  effect: "Yield loss through summer, then thermal shutdown"
  severity: 5
  occurrence: 6
  detection: 3
}

failure_mode capacitor_dryout {
  component: "DC-link capacitor bank"
  mode: "Electrolytic dry-out, ripple current rises"
  cause: "Sustained operation above the 65 C ambient rating"
  effect: "Silent efficiency loss, then hard failure near month 30"
  severity: 7
  occurrence: 5
  detection: 6
}

failure_mode firmware_hang {
  component: "Grid-interface controller"
  mode: "Controller hangs after a grid-frequency transient"
  cause: "Watchdog not armed during the reconnection routine"
  effect: "Unit stays offline until a manual power cycle"
  severity: 6
  occurrence: 3
  detection: 7
}

mitigation torque_check {
  mode: dc_arc
  action: "Spring-washer lug kit plus a torque-audit step in commissioning"
  owner: "Hardware engineering"
  target_severity: 8
  target_occurrence: 2
  target_detection: 3
}

mitigation fan_spec {
  mode: fan_seize
  action: "IP66 sealed-bearing fan, qualified to 40 000 h in dust chamber"
  owner: "Supplier quality"
  target_severity: 5
  target_occurrence: 2
  target_detection: 3
}

mitigation cap_derate {
  mode: capacitor_dryout
  action: "Derate the bank by 20 percent and report ripple over telemetry"
  owner: "Power electronics"
  target_severity: 7
  target_occurrence: 2
  target_detection: 2
}

mitigation watchdog_fix {
  mode: firmware_hang
  action: "Arm the watchdog across reconnection; ship in firmware 4.3"
  owner: "Embedded software"
  target_severity: 6
  target_occurrence: 1
  target_detection: 2
}

// ── The fleet, followed. 18 400 shipped, 16 455 still in service at
//    month 24, and every unit that left the cohort has a reason.
cohort shipped {
  n: 18400 source "ERP despatch ledger, FY24"
}

cohort registered {
  n: 17120
  from: shipped
  excluded: {
    never_commissioned: 812
    registration_not_returned: 468
  }
}

cohort in_service_12m {
  n: 16903
  from: registered
  excluded: {
    site_decommissioned: 141
    transferred_out_of_region: 76
  }
}

cohort in_service_24m {
  n: 16455
  from: in_service_12m
  excluded: {
    arc_fault_return: 96
    fan_failure_return: 173
    capacitor_return: 118
    firmware_return: 61
  }
}

view rpn: fmea_table
view fleet: consort(shipped)