📚 Course ContentsChapter 2 KPI definition and hierarchy in industrial operations

2. KPI definition and hierarchy in industrial operations

Industrial KPI analytics begins with a deceptively simple question: what exactly are we measuring, and why? A KPI is not merely a number collected from a PLC or MES. It is a formally defined ratio, rate, or duration with an explicit numerator, denominator, time basis, and scope. Without that rigor, two engineers can compute "OEE" from the same line and arrive at 62% and 78% — both technically correct, both useless for comparison. This chapter establishes the standard definitions of the core operational KPIs, shows how they relate mathematically, and explains how they map upward to business objectives.

Core Operational KPI Definitions

The foundational KPIs in discrete and process manufacturing fall into three families: effectiveness (OEE and its components), flow (throughput and cycle time), and quality/reliability (yield, scrap rate, MTBF, MTTR), plus the cross-cutting energy intensity.

OEE (Overall Equipment Effectiveness) is the product of three independent loss factors:

OEE = Availability × Performance × Quality

Availability = Run Time / Planned Production Time
Performance  = (Ideal Cycle Time × Total Count) / Run Time
Quality      = Good Count / Total Count

Each factor isolates a distinct loss category. Availability captures breakdowns and changeovers; Performance captures speed loss and minor stops; Quality captures defects and rework. Because OEE is multiplicative, a line at 90% on all three factors yields only 72.9% — this is why improving the weakest factor usually produces the largest gain.

Throughput is the rate of good units produced per unit time (e.g., units/hour). It differs from nameplate capacity: throughput is measured, capacity is specified. Cycle time is the inverse relationship — the time to produce one unit, typically expressed in seconds. For a line running at 1,800 units/hour, the effective cycle time is 2.0 s/unit. Yield is the ratio of good output to total input, while scrap rate is its complement (1 − yield) when rework is excluded. These must be defined at a stated process boundary; first-pass yield and final yield are different metrics.

MTBF (Mean Time Between Failures) and MTTR (Mean Time To Repair) characterize reliability and maintainability:

MTBF = Total Operating Time / Number of Failures
MTTR = Total Repair Time / Number of Failures
Availability (inherent) = MTBF / (MTBF + MTTR)

Note the consistency requirement: MTBF uses operating time, not calendar time. Mixing the two inflates MTBF and corrupts the availability figure.

Energy intensity normalizes consumption against output — kWh per unit, or kWh per tonne — making it comparable across production volumes. Absolute kWh readings are misleading when demand fluctuates.

How the KPIs Interlock

These metrics are not independent. Availability derived from MTBF/MTTR feeds directly into OEE. Throughput and cycle time are reciprocals constrained by the same run time. Scrap rate reduces quality, which reduces OEE and simultaneously raises energy intensity because energy is consumed on scrapped units. The SVG below maps these relationships and their upward linkage to business objectives.

Mapping KPIs to Business and Operational Objectives

A KPI without a mapped objective is a vanity metric. The mapping discipline is straightforward: every operational KPI should trace to at least one business objective, and every business objective should have at least one operational KPI that can move it. Consider a bottling plant with a business objective of reducing cost per case by 8% over two quarters. The operational decomposition might look like this:

Business Objective	Operational KPI	Target	Owner
Reduce cost per case 8%	OEE	72% → 80%	Production Manager
Improve delivery reliability	Throughput	1,800 → 2,000 u/h	Line Supervisor
Reduce warranty claims	First-pass yield	94% → 97%	Quality Engineer
Cut energy cost per unit	Energy intensity	0.42 → 0.36 kWh/u	Energy Manager
Increase asset uptime	MTBF / MTTR	MTBF +20%, MTTR −15%	Maintenance Lead

The mapping is not decorative. It determines which dashboards get built, how alarms are prioritized, and which improvement projects get funded. When OEE is the lever for cost per case, the OEE dashboard must decompose into the three loss factors so that teams can act on the dominant loss rather than staring at a single percentage.

Practical Definition Discipline

Before any KPI reaches a dashboard, lock down five attributes: formula (numerator and denominator with units), time basis (shift, day, rolling 24 h), scope (single machine, line, plant), data source (which tag, table, or manual entry), and exclusions (planned maintenance, no-demand idle). A common failure mode is comparing OEE across lines where one includes changeover time in planned production time and another excludes it. The fix is a written KPI definition sheet signed off by operations and engineering.

Common pitfall: Computing MTBF with calendar time instead of operating time. If a machine runs 16 h/day and fails twice in a 30-day month, calendar-based MTBF = 360 h, but operating-time MTBF = 240 h. The inflated figure understates failure frequency and leads to under-maintenance.

Rule of thumb: If a KPI cannot be traced to a business objective within two mapping steps, question whether it belongs on the executive dashboard. Keep it in the engineering layer for diagnostics, but do not elevate it.

With definitions and hierarchy established, the next step is instrumentation: ensuring the underlying tags, counters, and timestamps actually produce these KPIs reliably. That is where data quality and sampling strategy become the limiting factor, and it is the subject of the following chapter.