Dali Blackout Timeline: Building A Technical Incident Sequence

Applicability depends on the incident consequence, flag and class reporting requirements, whether VDR or equivalent data exists, whether personal data is involved, whether insurers or authorities may need evidence, and whether technical systems such as OT, bridge, AV/IT, SatCom, CCTV, or automation were involved.

This article uses a large commercial casualty to teach a yacht investigation method. The scale is different, but the evidence problem is familiar: a technical event happens fast, several systems record only part of the truth, crew recollection is under pressure, and the first repair action can destroy useful evidence.

On 26 March 2024, the Singapore-flagged containership Dali departed Baltimore Harbor. At about 0125, while approaching the Francis Scott Key Bridge, the ship suffered an electrical blackout. The event led to loss of lighting and machinery, including systems needed for propulsion and steering. About four minutes later, at about 0129, Dali struck Pier 17 of the bridge. A substantial portion of the bridge collapsed. Six highway workers died.

For YachtByte, the article is not about bridge design or cargo-ship regulation. It is about timeline discipline.

Dali shows why a technical incident timeline must include more than the first visible failure. The meaningful sequence started before the casualty voyage, with in-port blackouts and electrical configuration changes the previous day. It continued through departure, tug release, breaker trips, VDR data interruptions, steering and propulsion loss, emergency generator behavior, bridge orders, radio calls, anchor orders, external traffic closure, impact, post-impact calls, alarm handling, evidence downloads, and later technical analysis.

Most yacht incidents are smaller, but the investigation problem is the same. If a yacht loses shore power, suffers a PMS fault, has a generator trip during manoeuvring, experiences a GNSS conflict, or has a vendor change go wrong, the timeline is what separates evidence from opinion.

Dali was a 984-foot containership outbound from Baltimore. The vessel had a single main propulsion engine connected to a single propeller and was operating in the Fort McHenry Federal Channel with pilots onboard. According to the NTSB preliminary report, at about 0045 the senior pilot ordered dead slow ahead. At about 0107, the tugs were let go in normal practice. At about 0109, the main engine speed was increased to slow ahead and a course of 141 degrees was ordered for the transit under the Key Bridge.

At about 0125, the vessel was about 0.6 miles from the bridge, roughly three ship lengths. Breakers feeding most equipment and lighting opened unexpectedly. The NTSB preliminary report states that this caused the first blackout to lighting and most equipment, including main engine cooling-water pumps and steering gear pumps. The main propulsion engine then shut down because the pumps required for its operation had lost power. The loss of power also stopped all three steering pumps, so the rudder could not be moved at that point.

The bridge team and pilots acted quickly. Power was restored to some systems, emergency equipment became available, helm and anchor orders were given, tug assist was called, and external parties were warned. But the ship was already close to the bridge, moving at speed, without effective propulsion and steering recovery in time.

The NTSB's later board summary identified the probable cause as an electrical blackout due to a loose signal wire connection to a terminal block, stemming from improper wire-label banding. That loose connection led to loss of electrical power, and then loss of propulsion and steering close to the bridge.

A weak investigation timeline starts at the first alarm. A better timeline asks what changed before the alarm.

In Dali, the previous day mattered. The NTSB preliminary report says the vessel experienced two in-port blackouts on 25 March, about 10 hours before departing Baltimore. The first involved maintenance on the exhaust scrubber system for the diesel engine driving the only online generator. A crewmember mistakenly closed an inline exhaust damper, causing the generator engine to stall. Power was restored after another generator started and crewmembers manually closed breakers.

Then a second in-port blackout occurred after generator no. 3 continued to run for a short time but suffered insufficient fuel pressure. During recovery, the crew switched the bus configuration so that TR1 and its associated breakers were in use when the vessel departed on 26 March.

The board summary later connected the second in-port blackout and the second underway blackout to the same operational pattern: the use of a flushing pump as the service pump for online diesel generators. The pump did not provide the redundancy and automatic restart behavior expected of the normal fuel-service arrangement.

For a yacht team, this is the first lesson. A blackout timeline should not begin with "blackout at 0125." It should begin with the earlier defects, configuration changes, maintenance work, alarm resets, shore-power transfers, generator selection, breaker modes, fuel-pump arrangements, and handover notes that shaped the casualty.

The most useful timeline window in Dali is the short period between the first underway blackout and the bridge contact.

At about 0125, the vessel was 0.6 miles from the bridge. Breakers HR1 and LR1 opened unexpectedly, causing the first blackout to most shipboard lighting and equipment. Bridge equipment lost power and the VDR lost vessel system data feeds, although bridge audio continued to be recorded.

At 0124:59.2, the VDR audio transcript records the sound of multiple alarms activating at once. Those alarms continued through the bridge contact and collapse, with changes in tone noted by the transcript.

At 0126:02, VDR system data resumed. The preliminary report records the vessel's heading, course over ground and speed over ground at that point. At 0126:13, the senior pilot ordered 20 degrees of port rudder.

At 0126:39, the pilots called for tug assist. The tug Eric McAllister was about 3 miles away and headed back, but did not reach Dali before the impact. Around this same minute, the pilot's call to shoreside command was captured in both the vessel VDR and MDTA command-center audio. The NTSB VDR report used that shared audio event to correlate the recordings and align the MDTA command-center timestamp with the Dali VDR time.

At 0127:01, the senior pilot ordered an anchor dropped. The pilot dispatcher called the MDTA Police duty officer and relayed that the ship had lost power. The Coast Guard was also notified of the loss of power.

At 0127:23, the pilot ordered hard port rudder. At 0127:25, a VHF warning was made to waterborne traffic.

The vessel then suffered a second electrical blackout when generator breakers opened, causing loss of both high-voltage and low-voltage buses. The emergency generator, by then connected to the emergency bus, supplied emergency equipment through the second blackout. Generator no. 2 later restored power to the high-voltage bus.

At 0128:56.5, the pilot ordered full bow thruster to port. Seconds later, the bridge team confirmed it was not working. At 0128:59.9, the VDR transcript records a sound similar to the anchor chain running.

At 0129:15.2, the VDR transcript records a sound similar to the bridge collapse. At 0129:37.4, the training pilot called the Coast Guard on VHF and reported that the Key Bridge was down.

That is a technical timeline. It links alarms, power loss, breaker behavior, propulsion and steering consequence, VDR gaps, human orders, external communications, anchor action, and the moment of impact. It does not treat "blackout" as one line in the log.

The Dali timeline revealed several different types of causation.

The initiating technical failure was small: a signal wire connection to a terminal block became electrically disconnected. The board summary explained that wire-label banding covered the ferrule's insulated collar, increasing its circumference and preventing full insertion into the terminal block spring-clamp gate. That left the wire vulnerable to disconnecting.

The system consequence was large. When the wire disconnected, a high-voltage breaker opened, interrupting power to a step-down transformer and the low-voltage bus. The low-voltage bus supplied lighting and equipment including steering gear pumps, the fuel oil flushing pump, and main engine cooling-water pumps.

The operational consequence was severe. The loss of power to cooling-water pumps contributed to the main engine shutdown, and the loss of steering pumps removed normal rudder control. The ship was close to the bridge with limited time.

The recovery problem was layered. The board summary identified several safety concerns related to preventing or recovering from loss of propulsion and steering after a blackout: main engine shutdown due to low cooling-water pressure, use of the flushing pump as a fuel service pump, breaker control modes being Manual rather than Automatic, and emergency generator radiator damper behavior.

The evidence problem also mattered. The board summary noted VDR deficiencies affecting practical investigation and organizational safety review. The VDR audio report described proprietary formats, playback limitations, difficulty isolating channels, and the time required to convert large amounts of recorded data.

For yacht teams, this is the point. A technical timeline does not only answer "what failed?" It shows where the failure crossed boundaries: electrical, machinery, steering, bridge operations, communications, emergency response, and investigation evidence.

A Tiny Electrical Detail Can Become A Vessel-Control Event

The Dali case is a reminder that "minor" electrical details are not minor when they sit inside critical power distribution, PMS, steering, propulsion, or emergency-power paths.

On a yacht, comparable details include loose control wires, badly crimped ferrules, unsupported terminal wiring, terminal-label sleeves that interfere with insertion, undocumented bypasses, poor cable dressing, control-panel vibration, water ingress, and repeated opening of a panel during refit.

The investigation question should not be "could a loose wire really do this?" The better question is "what does this wire control, and what happens if it opens at the worst time?"

Previous Faults Belong In The Timeline

Dali had in-port blackouts before the casualty transit. Those events did not automatically mean the vessel would strike a bridge, but they were relevant to the later configuration and recovery sequence.

Yacht teams should treat earlier defects as part of the incident record. If a generator tripped during the previous charter, if a UPS alarm was silenced, if a shore-power converter fault was reset, or if a vendor changed a PMS setting during refit, it belongs in the timeline.

Recovery Time Matters As Much As Failure Cause

The NTSB board summary found that breaker Automatic mode would not have prevented the underway blackouts, but could have shortened the first blackout from 58 seconds to 10 seconds. That is exactly the kind of finding yacht teams often miss.

The question is not only "what failed?" It is also "how much time did recovery take, and did the yacht have that time?"

In close-quarters manoeuvring, a 40-second difference can be operationally significant. For yachts, this applies to harbour entries, stern-to berthing, narrow channels, anchorages, tender operations, DP-like holding, and guest embarkation with wind or current.

Evidence Has Its Own Failure Modes

Dali's VDR preserved critical information, but the investigation still had to deal with data-feed gaps, proprietary formats, audio limitations, and time-correlation work.

Yachts may have weaker evidence systems than a SOLAS cargo ship. CCTV may overwrite quickly. AV processors may have no useful logs. PLC/HMI clocks may be wrong. Firewalls may use UTC. PMS logs may be vendor-only. Bridge displays may not record source status. Owner dashboards may show a polished track without raw evidence.

If the yacht has not practised evidence capture, the best data may be gone before anyone starts investigating.

1. Define The Event Window

Do not start at the obvious failure. Start at the last known normal state and work forward.

For a yacht blackout or control incident, include:

• Previous alarms, resets, and known defects.
• Recent maintenance, refit, software updates, and vendor access.
• Generator, shore-power, battery, UPS, and bus configuration before the event.
• Watch handover, bridge orders, and engine-room status.
• First alarm or first abnormal reading.
• Crew response and recovery actions.
• Return-to-service testing.
• Corrective actions and verification.

2. Keep Separate Timeline Layers

Build the layers separately before merging them.

System layer:

• PMS and switchboard events.
• Generator alarms and breaker operations.
• Automation or ICMS events.
• UPS, battery, VFD, inverter, or shore-power logs.
• Steering, propulsion, stabilizer, or thruster status.

Bridge and operational layer:

• Bridge log.
• VDR/ECDIS/AIS/GNSS data where available.
• Helm, engine, anchor, tender, or radio orders.
• Pilot, marina, port, or VTS communication.
• Watchkeeper observations.

Human and procedural layer:

• Work permits.
• Toolbox talks.
• Defect reports.
• SMS entries.
• Crew interviews.
• Vendor tickets.
• Management decisions.

Physical evidence layer:

• Switch positions.
• Breaker states.
• Damaged cables, terminals, fuses, relays, connectors, batteries, chargers, or pumps.
• Photos before panels are disturbed.
• Thermal images if used.
• Labels and serial numbers.

3. Mark Evidence Quality

Every timeline entry should have a source and confidence level.

Use simple labels:

• Confirmed by system log.
• Confirmed by two independent sources.
• VDR or CCTV supported.
• Crew recollection only.
• Vendor statement only.
• Estimated time.
• Time mismatch unresolved.
• Evidence missing or overwritten.

This prevents a timeline from pretending that memory and logs are equally strong.

4. Correlate Clocks

Dali's VDR report shows why time correlation matters. Investigators aligned command-center audio to the Dali VDR by comparing an event captured by both recordings.

On yachts, do the same:

• Record the time source for each system.
• Note whether the system uses UTC, local time, or an incorrect clock.
• Use shared events to align logs, such as an alarm, radio call, CCTV movement, breaker trip, door access event, or phone call.
• Do not silently "correct" times without documenting the offset.

5. Preserve Before Resetting

Before repairs or resets, capture:

• HMI screens.
• Alarm history.
• Event logs.
• Switchboard and breaker positions.
• Photos of wiring, terminals, labels, and connectors.
• Failed parts.
• Vendor remote-access records.
• CCTV clips.
• ECDIS or bridge playback.
• Firewall and network logs if remote support or cloud systems were involved.

If immediate safety requires action first, record what was changed as soon as possible afterwards.

Technical Mitigation

• Add critical terminals and control wiring to inspection routines.
• Use thermal imaging where appropriate for switchboards and high-load electrical equipment.
• Check that wire labels, ferrules, sleeves, and cable dressing do not compromise terminal insertion.
• Review breaker control modes, PMS logic, blackout recovery, emergency generator start behavior, and UPS coverage.
• Test blackout recovery during controlled drills, not only during annual survey.
• Confirm what steering, thruster, propulsion, communication, and bridge equipment remain available during partial and total power loss.
• Verify whether emergency-power start times meet the yacht's operating needs, not only the written requirement.

Evidence Mitigation

• Make an incident evidence checklist part of the SMS.
• Know how to export logs from PMS, VMS, AMS, firewalls, CCTV, ECDIS, VDR, access control, AV/IT controllers, and SatCom systems.
• Keep admin access controlled but available for emergency evidence capture.
• Synchronize system clocks where practical.
• Preserve vendor support logs and tickets.
• Keep a controlled evidence folder with date, time, collector, and source notes.
• Train crew not to reset alarms or wipe logs before capture unless safety requires it.

Management Mitigation

• Treat previous blackouts and unexplained trips as management signals, not isolated engineering irritations.
• Require written closeout for repeated alarms and power-loss events.
• Ask whether a fix restored service or actually reduced recurrence risk.
• Review whether the yacht has enough redundancy for its intended operation.
• Make sure captain, chief engineer, ETO, management company, and DPA understand who controls incident evidence after a serious event.

For a yacht blackout, control loss, bridge data conflict, cyber event, or unexplained technical failure, preserve:

• PMS, AMS, VMS, and switchboard logs.
• VDR, ECDIS, radar, AIS, GNSS, and conning records where fitted.
• CCTV clips from bridge, engine room, technical spaces, passerelles, tender garages, and gangways if relevant and lawful.
• Firewall, VPN, remote-support, and cloud-controller logs.
• Photos of breaker positions, terminals, cables, damaged components, alarms, and temporary repairs.
• SMS defect reports, permits, toolbox talks, work orders, and vendor tickets.
• Crew statements taken as soon as practical.
• Time-source notes for every system.
• Any failed parts removed from service.

The timeline should say what evidence is missing. Missing evidence is not a reason to guess. It is a corrective action.

After building the timeline, close the loop with actions that map to the sequence.

Useful actions include:

• Correct the immediate fault and document the repair.
• Identify whether the failure could recur elsewhere.
• Review similar terminals, pumps, breaker modes, software settings, and bypasses.
• Test recovery under realistic operating conditions.
• Update the SMS, blackout checklist, bridge response procedure, or vendor-access rule.
• Train crew on the actual failure mode.
• Improve logging, clock synchronization, and evidence export.
• Verify the corrective action with a named person and date.

The final question is not "is the yacht working again?" The final question is "what evidence shows this sequence is less likely to happen again?"