Timestamps, Missing Data and Trustworthy Industrial Reports

Use the working record

Download the editable worksheet. Record the actual evidence and responsible reviewer against each relevant row. The empty results are inputs for your project, not completed tests. Add the equipment identity, operating period and reference method before assigning an outcome. Retain unsuccessful checks and the evidence from any repeat after correction.

Name the time represented by every field

A field observation may have an acquisition time, a device timestamp, a gateway receipt time and an application arrival time. Determine which of these the architecture actually supplies. The report should not label an arrival timestamp as the measurement time simply because it is the easiest field to obtain.

Record the clock source and any known quality limitation. Some devices maintain their own time, while others depend on a receiving system. The accepted method should describe the resulting evidence accurately rather than assume that every instrument is precisely synchronised.

Preserve a consistent time-zone convention

Specify the stored time basis and the display convention used by operators. A report combining sites or applications needs an unambiguous representation of the relevant offset or zone. Keep the original source time where it helps investigate a conversion problem.

Test date boundaries and any relevant clock-setting change. An event close to midnight can fall into a different operating day if one system uses local time and another uses a different basis. The reporting owner should approve how shifts and business periods are assigned from the accepted timestamps.

Distinguish late arrival from a new event

Local buffering can deliver historical records after communication returns. Preserve their original acquisition order and identify their arrival time separately. A delayed observation should not become a current equipment event merely because it has just reached the application.

Define how late records affect already issued reports. The system may need to reopen a provisional interval or create a controlled revision. Keep the previous result and the reason for the change so the operating team can explain why a historical total or timeline changed.

Use stable record identities for recovery

Agree how a source record is identified across retries and retransmission. A timestamp alone may be insufficient when several observations share the same time resolution. The supplied architecture should provide an appropriate combination of identity, sequence or other documented fields.

Test duplicates, out-of-order arrival and missing sequence elements. Deduplication should preserve legitimate distinct events while preventing a retried record from being counted twice. Retain the rule and relevant diagnostic evidence so a later discrepancy can be investigated.

Keep interpolation distinct from measurement

An interpolated value is a calculation between available observations. Record the method, source points and interval. It can be useful for an approved reporting purpose without becoming evidence that the physical quantity was measured at every displayed time.

Decide which reports permit estimated values and which require actual observations. A smooth chart may be helpful for presentation while an incident review needs the original gaps. The exported record should make both the source evidence and any derived display values distinguishable.

Define zero and unavailable separately

A valid zero can describe stopped flow, no output or another real condition. A missing source, failed communication path or invalid conversion is different. The application should not substitute zero merely because a numeric field is required.

The same distinction applies to unchanged values. A stable process, a frozen instrument and repeated transmission of an old record can look similar on a chart. Retain age, quality and source identity so the reviewer can decide which investigation is appropriate.

Reconcile a representative cross-system sequence

Select a known event or operating interval and obtain the independent source records. Compare identities, units, acquisition times, arrival times and final interpretation. Include a deliberate delayed-upload or replay exercise where the approved test method allows it.

Agree the expected ordering and any timing tolerance before the comparison. The tolerance should reflect the actual clock and acquisition capabilities. If the available evidence cannot distinguish two events in time, preserve that limitation rather than infer a causal order from a sorted display.

Control corrections and calculation versions

An authorised correction should identify the original record, changed field, reason, evidence and effective time. Keep the source observation available where appropriate. A silent historical rewrite can make later reviews impossible even when the corrected value is technically more useful.

Version the calculations that create totals, ratios or event classifications. The same raw data can produce a different result after a rule change. A reproducible report needs the applicable rule and source population, not only a saved screenshot of the final number.

Hand over an explicit data contract

The acceptance pack should describe field names, units, time semantics, quality states, identities and correction behaviour. Demonstrate an export containing normal, missing, delayed and revised records. Have another authorised reviewer reconstruct the selected report outside the application.

Assign ownership of the contract and the process for reviewing new devices or interfaces. When the system expands, verify that each new source supports the required meanings. Trustworthy reporting comes from preserving what the evidence can establish, including the gaps and uncertainties that remain.

Connect the decision to the project

Use Industrial iot, Shop floor data integration, Iot solutions to place this task within the wider installation. Bring a recent operating record and identify the decision that the proposed measurement should support. A focused discussion should establish the first boundary, the responsible site participants and the evidence needed before expanding the scope.

Agree who maintains the point register and who can change interpretation rules. Preserve the previous version when a device, product, reporting period or responsibility changes. During handover, demonstrate an export that another authorised person can reconcile without relying on the dashboard operator to explain missing fields.

Discuss the scope with addanode using the actual site information, available interfaces and proposed acceptance record.

Sources and method

NIST SP 800-82 Revision 3 provides OT-system context. The field contract and reconciliation tests are proposed engineering methods, with actual clock, storage and interface capabilities verified for the selected equipment.

Frequently asked questions