Connect981 – Content Dev

Glossary Tag: signal detection

  • Time grain

    Time grain is the level of time detail at which data is recorded, aggregated, displayed, or analyzed. It commonly refers to the size of the time interval used in a dataset or report, such as seconds, minutes, hours, days, weeks, or production shifts.

    In manufacturing and industrial systems, time grain affects how operational events are represented across MES, ERP, historian, SCADA, quality, and analytics workflows. For example, machine state changes may be captured at a second-level grain, while production reporting may be summarized by shift or by day.

    What it includes and excludes

    Time grain includes the temporal resolution of a record or metric. It does not, by itself, define the total time period being analyzed. A report can cover one month of data but still use an hourly or daily time grain.

    • Includes: the interval used for timestamps, aggregation, trending, and reporting
    • Excludes: the overall date range, retention period, or refresh frequency unless those are defined separately

    How it shows up in operations

    Time grain matters when comparing data across systems or deciding whether a metric is suitable for a given use. Finer grain data can show short stoppages, alarms, or process variation. Coarser grain data is more common for management reporting, scheduling, financial rollups, or longer-term quality trends.

    Examples in manufacturing include:

    • equipment downtime tracked by second or minute
    • OEE summarized by hour or shift
    • scrap trends reviewed by day or week
    • ERP production quantities posted by day or reporting period

    Common confusion

    Time grain is often confused with time range and sampling rate. Time range is the total period under review, such as the last 30 days. Sampling rate refers to how often a sensor or system captures raw observations, which may be finer than the grain used for reporting. Time grain is also different from update frequency, which describes how often a dashboard or interface refreshes.

    In analytics and data warehousing, the term may also be discussed alongside data granularity. Time grain is the time-specific part of that broader idea.

  • Local KPI

    A local KPI is a key performance indicator used to measure performance within a specific part of an operation, such as a machine, workcell, production line, department, shift, warehouse area, or quality function. It is narrower in scope than an enterprise or plant-level KPI and is usually owned by the team closest to the work.

    In manufacturing and regulated operations, a local KPI commonly refers to a metric that helps monitor day-to-day execution in a defined area. Examples include first-pass yield for one line, schedule adherence for one workcenter, changeover time for one packaging cell, or right-first-time performance for a specific process step.

    A local KPI is not simply any number visible on a dashboard. To qualify as a KPI, the metric is generally treated as an important signal tied to operational control, performance review, or escalation within that local context.

    How it is used in operations

    Local KPIs often appear in shift boards, MES screens, team huddles, visual management boards, and supervisor reviews. They are used to track conditions that operators, technicians, leads, or area managers can influence directly.

    • At the equipment or line level, a local KPI may track downtime, scrap, cycle time, or output versus plan.

    • In quality workflows, it may track defect rate, rework rate, inspection backlog, or deviation closure time for a specific area.

    • In warehousing or materials flow, it may track pick accuracy, staging delays, or replenishment response time for a zone.

    Well-defined local KPIs are often linked upward to broader site or enterprise measures, but they remain focused on a limited operational boundary.

    What it includes and excludes

    Local KPI includes metrics scoped to a specific team, process, asset group, or area of responsibility. It can be leading or lagging, depending on whether it signals conditions early or reports outcomes after the fact.

    It does not necessarily mean the metric is informal, temporary, or less important. Some local KPIs are tightly controlled because they support quality, throughput, traceability, or risk monitoring in a regulated environment.

    It also does not mean the metric is globally standardized. A local KPI may be unique to one area if it reflects that area’s process constraints or control needs.

    Common confusion

    Local KPI vs enterprise KPI: A local KPI applies to a limited operational scope, while an enterprise KPI is used across a site, business unit, or company.

    Local KPI vs metric: A metric is any measure. A KPI is a metric treated as materially important for monitoring or managing performance.

    Local KPI vs OEE: OEE is a specific composite performance measure. A local KPI can be OEE for one asset or line, but it can also be any other important area-level indicator.

    Manufacturing example

    If a plant tracks on-time delivery at the site level, a local KPI beneath it might track queue time at one bottleneck workcenter. The local KPI helps explain and manage the part of the workflow that contributes to the broader result.

  • Process capability index (Cpk)

    Process capability index (Cpk) is a statistical measure used to indicate how well a stable process can produce output within defined specification limits. It compares the process average and variation to both the upper and lower specification limits, then reports the side where the process is performing worst.

    Cpk is commonly used in manufacturing and quality control to summarize whether a process is both centered and consistent enough to meet engineering requirements. A higher Cpk generally indicates that the process output is farther from the nearest specification limit relative to its variation.

    What it includes and what it does not

    Cpk includes two core ideas: process spread and process centering. It reflects not only how much the process varies, but also whether the process mean is shifted toward one specification limit.

    Cpk does not itself prove that a process is in statistical control, and it does not replace measurement system analysis, sampling plans, or product acceptance decisions. It is a capability indicator, not a direct statement that every part is conforming.

    How it is used in operations

    In production environments, Cpk is commonly calculated for critical dimensions, fill weights, temperatures, torque values, or other measurable characteristics with upper and lower specifications. Teams may review it during process qualification, ongoing process monitoring, supplier quality reviews, or continuous improvement work.

    In MES, QMS, SPC, or reporting systems, Cpk may appear as a quality metric tied to a part number, operation, machine, line, tool, or characteristic. It is often used alongside control charts and other process performance measures.

    Common confusion

    Cpk is often confused with Cp. Cp measures potential capability based on process variation alone and assumes the process is centered. Cpk adjusts for actual centering, so it is usually the more realistic index when the process mean is not exactly on target.

    Cpk is also sometimes confused with Ppk. While usage varies by organization, Ppk commonly refers to long-term or overall process performance using actual overall variation, whereas Cpk commonly refers to capability based on within-process variation under more controlled conditions.

    Practical interpretation notes

    • Cpk is most meaningful when the characteristic is measurable on a continuous scale and the process is reasonably stable.

    • The index depends on valid specification limits set by design or customer requirements.

    • Poor measurement quality can distort the result, so gage capability matters.

    • A single Cpk value does not explain the cause of variation or process shift.

    For example, a machining process for a bore diameter may show a lower Cpk if the average diameter drifts close to the upper tolerance, even if the overall spread has not changed.

  • Semantic KPI layer

    A semantic KPI layer is a business-facing definition layer that gives key performance indicators (KPIs) consistent meaning across data sources, applications, dashboards, and reports. It commonly defines how a KPI should be interpreted, calculated, named, filtered, and grouped so different users and systems refer to the same metric in the same way.

    In manufacturing and regulated operations, this layer often sits above raw data from MES, ERP, quality systems, historians, and other operational sources. It does not replace those source systems or the underlying transactional data. Instead, it organizes metric logic and business context so measures such as yield, scrap, downtime, first pass quality, or schedule adherence are calculated consistently.

    What it includes

    • Standard KPI definitions and naming conventions

    • Calculation logic, units, time windows, and aggregation rules

    • Business context such as plant, line, work center, product, shift, lot, or order dimensions

    • Data mappings that connect operational data fields to business terms

    • Governance elements such as ownership, approved formulas, and versioned changes

    What it does not mean

    A semantic KPI layer is not just a dashboard, a data warehouse, or a list of KPI names in a slide deck. Those may consume or display the layer, but the layer itself is the shared meaning and metric logic. It is also not the same as a general semantic data model for all enterprise data, although it may be implemented as part of one.

    Operational meaning

    Operationally, a semantic KPI layer helps align reporting between systems that track the same process differently. For example, an MES may record machine states, an ERP may record order completions, and a quality system may record nonconformances. The semantic KPI layer can define how those records contribute to metrics like OEE, throughput, rework rate, or right-first-time so downstream analytics use a common interpretation.

    This is especially relevant where KPI disputes arise from different formulas, timing assumptions, or data filters. A governed layer can document whether planned downtime is excluded, whether partial completions are counted, or which event codes roll up into a downtime category.

    Common confusion

    Semantic KPI layer vs. KPI dashboard: a dashboard presents metrics; the semantic layer defines what those metrics mean.

    Semantic KPI layer vs. data model: a data model structures data entities and relationships; a semantic KPI layer focuses on business meaning and reusable metric definitions, though the two often overlap.

    Semantic KPI layer vs. master data: master data manages core reference entities such as products or equipment; the semantic KPI layer uses those entities to define and contextualize performance measures.

  • operational baseline

    An operational baseline is a defined reference point for how a process, asset, production line, or system normally operates at a given time. It commonly includes the expected settings, conditions, performance ranges, and control context used to compare future operation against a known state.

    In manufacturing and regulated operations, an operational baseline may cover items such as standard cycle times, equipment parameters, approved process settings, expected throughput, normal alarm patterns, quality levels, or system configuration details. The exact content depends on what is being baselined: a machine, a production cell, a software environment, a plant utility system, or a broader operation.

    The term is descriptive, not necessarily fixed forever. A baseline can be revised when approved changes are made, but at any point in time it serves as the reference for detecting drift, assessing deviations, investigating issues, and evaluating whether current performance or configuration still matches the expected state.

    What it includes and excludes

    • Includes the documented or agreed normal state used for comparison.

    • Includes operational, technical, or performance attributes that are relevant to monitoring and control.

    • Excludes temporary conditions such as startup, shutdown, maintenance mode, or known abnormal events unless those are explicitly defined as separate baselines.

    • Excludes a target or aspiration by itself. A baseline is usually the current or validated reference state, not just a future goal.

    How it is used in practice

    Operational baselines are commonly used in daily management, process monitoring, quality review, and change control. For example, a plant may compare current machine performance against a baseline established after qualification, or compare current OT network traffic against a baseline of normal communications to identify unusual activity. In MES, ERP, historian, or monitoring environments, the baseline may be reflected in master data, approved recipes, version-controlled settings, or KPI thresholds.

    Common confusion

    Operational baseline is often confused with a performance target. A target states the desired result, while a baseline states the reference condition used for comparison.

    It can also be confused with a configuration baseline. A configuration baseline usually focuses on approved technical components, versions, or settings. An operational baseline is broader and may include how the process or system behaves in use, including expected operating ranges and performance patterns.

    In some contexts, people also use the term similarly to standard work or a golden batch, but those are narrower ideas. Standard work defines the approved method for performing tasks, and a golden batch refers to a model production run or parameter profile. An operational baseline may incorporate aspects of both without being limited to either one.

  • KPI council

    A KPI council commonly refers to a cross-functional governance group responsible for overseeing key performance indicators, including how KPIs are defined, calculated, reviewed, and changed over time. In manufacturing and regulated operations, it often exists to keep performance reporting consistent across functions such as production, quality, maintenance, supply chain, and finance.

    It is not a KPI itself, and it is not just a reporting meeting. The term usually describes a standing forum or decision body that manages KPI ownership, data definitions, thresholds, review cadence, and escalation rules. Depending on the organization, it may be formal with documented charters and approval workflows, or informal but still used as the place where metric disputes and updates are resolved.

    How it is used in operations

    In operational settings, a KPI council often reviews questions such as:

    • Which KPIs are official and who owns them
    • How each KPI is calculated and from which systems the data is sourced
    • Whether plant, line, shift, supplier, or enterprise views use the same definitions
    • How targets, thresholds, and exception rules are set
    • What to do when ERP, MES, QMS, or manual reports show conflicting values

    For example, a KPI council may decide whether first-pass yield, on-time delivery, scrap rate, or schedule adherence should be measured at the work center, work order, or site level, and which source system is considered authoritative for each metric.

    What it includes and excludes

    A KPI council usually includes governance activities around metric standardization, review, and change control. It may also support metric rationalization, meaning the removal of duplicate or low-value measures.

    It does not usually perform day-to-day data entry, direct production execution, or root cause analysis itself, although it may trigger those activities when KPI results indicate an issue.

    Common confusion

    KPI council is sometimes confused with a daily management meeting, performance review meeting, or steering committee. A daily management meeting focuses on current performance and immediate actions. A KPI council focuses more on metric governance, consistency, and lifecycle management. It may also be confused with a data governance council. A data governance council usually has a broader scope that covers master data, data quality, access, and policies beyond performance metrics.

    In system and reporting contexts

    Where MES, ERP, QMS, historian, or analytics platforms are integrated, a KPI council often helps define the official business meaning of metrics so dashboards and reports use the same logic across systems. This is especially relevant when the same operational signal can be calculated differently by different applications or departments.

  • operational visibility

    Operational visibility commonly refers to the ability to see, understand, and track what is happening across an operation as work is planned, executed, measured, and escalated. In manufacturing and regulated environments, it usually includes timely access to information about production status, materials, equipment, labor, quality events, and workflow conditions.

    It is not limited to a single dashboard or report. It depends on the availability, context, and reliability of operational data across systems such as MES, ERP, quality systems, maintenance platforms, shop floor devices, and connected machines. The term usually implies that people can view the state of operations clearly enough to recognize progress, delays, exceptions, and emerging risks.

    What it typically includes

    • Status of orders, batches, jobs, or work orders

    • WIP location and flow through routing steps

    • Equipment state, downtime, alarms, or utilization signals

    • Material availability, shortages, and staging status

    • Quality checks, holds, nonconformances, and rework status

    • Shift performance, bottlenecks, and schedule variance

    • Traceability-related context such as lot, serial, or genealogy data when relevant

    Depending on the organization, operational visibility may be near real time, shift-based, or based on periodic synchronization between systems. It does not necessarily mean full automation or a complete digital thread.

    How it appears in practice

    In day-to-day operations, operational visibility often shows up through dashboards, escalation boards, status views, alerts, exception queues, and reports that combine data from multiple sources. For example, a supervisor may use it to see which work orders are stalled, which machines are down, or which lots are on quality hold. A planner may use it to see whether material shortages or labor constraints are affecting schedule adherence.

    In regulated manufacturing, the term can also extend to visibility into controlled records and evidence trails, but it is not the same as document control, audit readiness, or compliance management.

    Common confusion

    Operational visibility is often confused with traceability. Traceability focuses on the historical record of where a product, material, or action came from and where it went. Operational visibility focuses more broadly on the current and recent state of operations.

    It is also commonly confused with observability. In software and infrastructure contexts, observability usually refers to the ability to infer system behavior from logs, metrics, and traces. Operational visibility is broader and more business-process-oriented in manufacturing.

    Another related term is performance management. Performance management uses metrics and targets to assess outcomes. Operational visibility is the underlying ability to see the signals and conditions that inform those assessments.