Simulation Guide
Knowledge Guides /Why Do Simulation Monitor Values Disagree? Align Lists, Charts, Alerts, and Exports
Last updated: · New Horizon Digital Outsourcing Team
When monitor lists, charts, and exports disagree, first check whether they refer to the same object, field, and time, then examine units, display sampling, and aggregation. Connect output, display, and storage through a shared field dictionary and retain original records. A smooth curve or an absent alert alone does not establish that the original data contained no anomaly.
Connect Displays through Field Identity
- Record a stable identifier, owning object, source node, units, and data type for each monitored field, then associate its display name, chart label, and export column. Names may change for readability while field identity remains traceable. If values are accessed by array position, changes to output order must trigger checks of display and storage configuration.
- Use easily distinguishable teaching data to check mappings field by field: for example, provide different input traces for identically named fields on two objects and verify that lists, charts, and files do not swap them. These values check mapping only and do not represent actual equipment values or performance. Repeat the check after replacing objects, reordering fields, and reopening the project.
Distinguish Recorded Samples from Display Refresh
- Algorithm outputs, file storage, and display refresh may use different frequencies; document timestamps and sampling rules for each. If a chart uses subsampling, averaging, or interpolation, its displayed values must have an explainable meaning. Align sample times before comparing values rather than assuming a computation error because two windows currently show different numbers.
- After zooming or scrolling a chart, check that historical data, current data, and selected-point time are clearly distinguished. Retain an original time series and compare it with the processed display to identify brief changes hidden by subsampling. Record display optimization acceptance separately from solver accuracy so visual smoothness does not substitute for numerical checks.
Specify Alert Rules and Missing-Data Behavior
- Alert configuration should specify the field, units, inclusive or exclusive bounds, and whether evaluation uses original or display-processed values. Lists and charts should reference the same rule version and provide a text status for inspection. Set thresholds from the task's own basis rather than copying demonstration values or another project's internal criteria.
- Define separate states for missing, stale, invalid, and pause-frozen data. If the last value remains visible, show its timestamp and validity so it is not mistaken for a current normal reading. Check boundary values, sustained excursions, recovery, and interrupted data so alert outcomes can be traced to explicit samples and rules.
Trace Display Values Back from Exports
- Exports should include run identity, object and field identities, time, units, and validity, linked to the field configuration version used in the run. Select a record containing normal changes, a pause, and an abnormal state, reconstruct displayed values using the same sampling rules, and compare them with lists and charts. Screenshots document appearance but do not replace reviewable data.
- Acceptance should also cover no selected fields, deleted fields, configuration loading failure, and state after a new run. Errors should identify missing fields or mismatched configuration instead of leaving charts from an earlier run on display. After closing a run, verify that file writes have completed so its display, alerts, and recorded data form a consistent delivery.