Tracking Unplanned Downtime Across Shifts, Sites and Languages

10/11/2026

A packaging line stops before shift handover and restarts after it. The controller records one interruption, both crews log it, and the regional report counts two breakdowns. Finance then has to price a loss that operations cannot reconcile. Siemens’ 2024 downtime study estimated that an hour of unplanned downtime at a large automotive plant can cost US$2.3 million. Use your own plant economics to value the loss, starting with a record of when production actually stopped. [1]

Track one event per physical interruption, allocate its minutes to the shifts affected, and use stable reason codes with local-language labels. For APAC manufacturing groups, this gives plant teams a record they can verify and regional leaders a consistent basis for deciding where to intervene. [2][3][4]

1. Agree on an event definition

For this rollout, define unplanned downtime as an unscheduled loss of the ability to produce during a window when the asset or line was expected to run. Agree on which production schedule supplies that window, what starts and ends an event, and how to classify planned changeovers, short stops, and slow running. Approve schedule corrections separately so the original production plan remains available for review. [2]

A machine state is evidence of what happened. PackML distinguishes machine states and includes signals for equipment blocked by downstream conditions or starved by upstream conditions. Use those signals to distinguish a machine fault from a line dependency. If an upstream feeder stops three downstream machines, report each affected asset where useful. Calculate line downtime from the line’s interruption interval so overlapping machine stops are counted once at line level. [2]

Define restart in terms operators can verify, such as the first valid production cycle after recovery. Keep the first alarm, maintenance completion, and production restart as separate timestamps when available. This lets supervisors see where the delay occurred. [2][3]

2. Create the minimum event record

Give each stop a stable event ID and capture the site, line, asset, source, start and end times, and reason status. Add the production order or product when it helps explain the stop. Keep links to source evidence and any maintenance work order. Leave ongoing events open and distinguish provisional records from approved ones. [2][3][5]

Store event times in UTC and assign reporting minutes using the site’s local time zone and effective shift calendar. Preserve the source timestamp where available and record receipt time separately. OPC UA specifies a UTC source timestamp associated with the value at its origin. This distinction matters when a gateway uploads buffered records after a network outage. [3]

For example, a stop from 21:55 to 22:15 across a 22:00 handover is one event with 20 elapsed minutes. Attribute five minutes to the outgoing shift and 15 to the incoming shift if both intervals were scheduled for production. Retain the event ID on both allocations and use distinct event IDs for the plant-wide count. [2][3]

3. Design reason codes for real use

Store a stable reason code and display a label in the operator’s language. An illustrative MATERIAL_STARVED code could appear as “Waiting for material” in English, “Chờ nguyên liệu” in Vietnamese, or “材料待ち” in Japanese. Have site teams validate the wording against actual scenarios. W3C’s SKOS model supports language-tagged labels for a shared concept. [4]

Keep the first choice short enough for an operator to use during a stop. Separate the observed reason, such as “drive tripped,” from the verified root cause entered later by maintenance. Preserve local notes in their original language. Provide a “pending review” option and assign a supervisor to review unresolved reasons at handover. Version changes to the catalogue and preserve the code used when each event was recorded. [2][4][5]

4. Reconcile automatic and manual records

Use TMA’s published architecture as a starting point. Its Machine Monitoring System (MMS) categorises stops and compares overall equipment effectiveness (OEE) across shifts, days, and machines. Its integration with a computerized maintenance management system (CMMS) connects detected issues with maintenance tasks. The following table proposes how to govern the records between those systems. [5]

Design areaDisconnected shift logsProposed regional event model
Stop identitySeparate entry from each crewOne event ID with shift allocations
TimingLocal entry or upload timeSource time, receipt time, and local shift calendar
ReasonFree text or site-specific labelsShared code, local label, and separate root cause
CorrectionsEdited spreadsheet cellsOriginal evidence and approved correction history
MaintenanceTicket outside the stop logLinked work order and verified return to production

Match an operator report to an automatic record by asset and overlapping interval, then apply a documented tolerance for clock differences. Send ambiguous matches to review and retain operator-only events for uninstrumented equipment. Treat lost telemetry as an evidence gap. When delayed signals arrive, match them to the original event and recalculate affected shift totals through an auditable correction. [2][3][5]

Technical deployment pipeline

  • Define the rules. Approve event boundaries, planned production windows, and short-stop treatment. [2]
  • Map the signals. Verify controller and historian states against observed stops and restarts. [2][3]
  • Configure local operations. Load shift calendars and test translated labels with operators. [3][4]
  • Replay exceptions. Test cross-shift stops, duplicates, missing telemetry, and late uploads. [2][3][5]
  • Run in parallel. Compare event counts and minutes with independently reviewed shift records. [3][5]
  • Release to operations. Route approved events to action owners and review unresolved records at handover. [5]

5. Turn the event log into action

Rank approved downtime by lost minutes and recurrence. Compare sites using unplanned stop minutes divided by planned production minutes, with the same scope, exclusions, and short-stop rules. Review incomplete coverage and unresolved reasons alongside the rate. Keep OEE separate, since an availability loss is only one part of that measure. [2][5]

Lessons Learned from the Field

At a Singapore nutritional ingredient manufacturer, TMA found that shortages of critical spares could prolong shutdowns. Its implementation combined a hierarchical asset register, maintenance workflows, and low-stock alerts. The documented case identifies parts availability as a constraint on recovery. It reports qualitative benefits, so set downtime targets from your own baseline. [6]

Use that scenario to compare TMA’s engineering scope with a low-cost offshore development centre focused on dashboards. Ask each bidder to demonstrate a cross-shift stop that creates a work order, waits for a part, and closes after production resumes. The demonstration should show who owns the wait and how the plant verifies recovery. [5][6]

TMA reports 29 years of engineering experience, 4,000 engineers serving clients in 30 countries, and mission-critical programmes involving hundreds of engineers. Its telecom R&D centres provide a precedent for long-term partnerships. For an APAC rollout, contract for named ownership of machine interfaces, taxonomy governance, site onboarding, and support. TMA also lists ISO/IEC 27001 and reports GDPR checks in software quality assurance. Verify certification scope and agree on access, retention, and regional sharing for records that identify personnel. [7][8][9]

FAQ

How do you count downtime that crosses a shift?

Keep one event ID. Allocate minutes by intersecting the stop interval with each shift’s planned production window. Use distinct event IDs for the plant-wide event count. [2][3]

How do sites use different languages without splitting the reports?

Store a shared reason code and display an approved label in each site’s language. Keep operator notes separate and report against the code. [4]

What if the operator and machine timestamps disagree?

Keep both records and their sources. Apply the approved matching tolerance, route ambiguous cases to review, and record the evidence behind any correction. [3][5]

Which KPI should regional managers compare?

Compare unplanned stop minutes as a share of planned production minutes under common rules. Show unresolved reasons and data gaps alongside the rate. [2][5]

How should buyers evaluate a downtime tracking vendor?

Request relevant plant references and a replay of a cross-shift stop with delayed telemetry, a reason correction, and work-order closure. Confirm who owns integration, reconciliation, and support. [3][5][6]

Start with one line and a real shift handover

Bring a machine-state export, two adjacent shift logs, and a recent maintenance work order to a TMA Innovation downtime assessment. Use them to agree on event rules, test reconciliation, and define what the next site must verify before rollout. [5][6]

References

[1] Siemens — The True Cost of Downtime 2024.

[2] OPC Foundation — OPC UA for PackML Data Representation Model.

[3] OPC Foundation — OPC UA Part 4, SourceTimestamp.

[4] W3C — SKOS Reference, Labeling and Language Tags.

[5] TMA Solutions — Machine Monitoring and CMMS Integration.

[6] TMA Innovation — Smart CMMS and Singapore Production Maintenance Case Study.

[7] TMA Solutions — Company Overview and Engineering Capabilities.

[8] TMA Solutions — Telecommunications Engineering and R&D Centres.

[9] TMA Solutions — CSR Report 2024, Security and Compliance.

TMA Innovation
Author: TMA Innovation
Table Of Content
1. Agree on an event definition
2. Create the minimum event record
3. Design reason codes for real use
4. Reconcile automatic and manual records
Technical deployment pipeline
5. Turn the event log into action
Lessons Learned from the Field
FAQ
How do you count downtime that crosses a shift?
How do sites use different languages without splitting the reports?
What if the operator and machine timestamps disagree?
Which KPI should regional managers compare?
How should buyers evaluate a downtime tracking vendor?
Start with one line and a real shift handover
References
Start your projectRight now!
Contact Us
Share: