Industrial production management: measuring, arbitrating, winning

Introduction : quand les moyennes sabotent l'OPEX

In many workshops, 10 to 30% of capacity disappears. Not because machines are missing, but because operations are managed using fictional averages (see, for example, Hopp & Spearman, Factory Physics and Kleinrock, Queueing Systems). This loss does not come from a lack of effort, but from a system managed with averages that do not exist.

Result: teams compensate with emergencies, and OPEX drifts.

The number that stings: up to 25% of effective capacity lost when variability rises

When the utilization of a workstation increases and disruptions remain, waiting times explode. The literature on queueing theory formalizes this mechanism through Kingman's law (see Hopp & Spearman, Factory Physics, VUT section, and Kingman, “Formulae and approximations for GI/G/1 queues”). For a plant, the effect shows up as growing WIP and slipping lead times, without any sustained increase in throughput.

The shopfloor tension: customer promise, workshop saturation, then costly emergencies

The plan promises, the workshop saturates, and reality decides. Micro-stops on a bottleneck, a material delay, a missing operator qualification—and the day collapses into firefighting. You don't just pay time; you pay for instability.

Takeaway: a robust budget is built on flow scenarios, not on an average

An OPEX budget built on average times assumes a factory without variability—imaginary. A robust budget tests disruption scenarios and computes consequences on lead times, WIP, energy, and resource needs. Takeaway: model flows as a dynamic system, then budget on that system.

I. Defining industrial production management: decisions, scope, and the indicators that matter

Industrial production management groups the decisions that turn customer demand into executable work orders, then into conforming parts delivered. It covers planning, scheduling, release, execution monitoring, quality, maintenance, and the management of stocks and flows. It continuously arbitrates between throughput, lead time, cost, and robustness. When it drifts, the shopfloor compensates by consuming cash.

The KPIs that connect the shopfloor to the income statement

OEE (Overall Equipment Effectiveness, taux de rendement synthétique) measures equipment performance through availability, performance, and quality. FPY (First Pass Yield, qualité au premier passage) measures the share of parts that are conforming without rework. WIP (Work In Progress, encours) measures the cash tied up between stations. OTD (On-Time Delivery, livraison à l'heure) measures whether the promise is kept—penalties avoided and customer trust.

WCR (Working Capital Requirement, besoin en fonds de roulement) rises when WIP and inventories grow faster than cash collection. Energy consumption in kWh/part drifts when the factory restarts, reworks, and accelerates without stability. Production management can improve these metrics with the same levers: less variability, less WIP.

These KPIs are shopfloor indicators, but also direct levers on EBITDA, cash, and working capital.

II. The averages trap and Kingman's law

The average is reassuring because it hides things well. It produces clean budgets and plans that look feasible on a screen. It ignores a basic fact: disruptions do not cancel out over a day—they stack up on constrained stations. An upstream disruption shifts the arrival of a lot, then creates a wait at the next station; a downstream disruption prevents evacuation and creates blocking (blocage). Both add up and amplify WIP.

Kingman's law provides a simple approximation of waiting time in a queue with variability in arrivals and service (reference: Kingman, “Formulae and approximations for GI/G/1 queues”). Its message is still brutal: as utilization approaches 100%, waiting increases non-linearly. A 95% load leaves little margin to absorb even a small disruption. You end up with a “full” shopfloor and a “late” customer.

Congestion creates catch-up costs, so OPEX rises without added value. Unit cost drifts because you consume more energy, more labor, and more scrap for a similar throughput.

As long as you don't MANAGE flows, you suffer their consequences…

III. The 3 variabilities that eat margin: arrivals, process, product mix

Variability has several sources, and each requires a different lever. Material arrivals are rarely regular: batching, quality checks, and transport priorities add jolts that create component shortages, then sequence changes. Breakdowns and micro-stops add cycle-time variation; ramp-up (montée en cadence) adds a learning curve, so productivity changes week by week. Finally, product mix changes cycle times and required resources: the bottleneck can shift depending on the sequence, so scheduling decisions matter as much as capacity.

IV. Building “anti-variability” control: from endured budgets to scenario-based management

A robust OPEX budget starts with one question: what disruption must you absorb without falling into permanent firefighting? You then test flow scenarios and size margins and buffers (stocks tampons) in the right places. A buffer that is too small brings you back to emergencies; too large, it inflates working capital and hides the causes of variability. The right level comes from a flow model that integrates disruption distributions, not from a rule of thumb.

A planning freeze locks part of the plan over a short horizon to avoid the yo-yo effect. Priority rules resolve conflicts without endless meetings: customer due date, bottleneck criticality, changeover cost. Bottleneck-driven control aligns releases to the constrained station, then limits WIP. You often gain lead time without producing faster—because you wait less.

V. The 4 production types and their practical control modes

• Unit / project: finite loading, milestones, time buffers.
  Specialized resources, quality validations, and multi-supplier sub-assemblies.
  KPI: lead time and milestone adherence.

• Job shop / small series: frequent rescheduling tolerated, but explicit priority rules by customer or margin.
  Lot sizes that limit changeovers without saturating WIP.
  KPI: OTD (On-Time Delivery, livraison à l'heure), WIP (Work In Progress, encours), lead time (délai de traversée), changeover rate, changeover time.

• Series / repetitive: line balancing, repeatable sequences, short buffers, rare rescheduling.
  KPI: OEE, OTD, changeover rate.

• Continuous / process: process stability, maintenance, capability, and energy managed as dominant variables.
  KPI: availability, kWh/part, scrap rate.

VI. The “5 Ps” translated into measurable levers

• People: multi-skilling and work standards reduce time dispersion. KPI: multi-skilling rate, OEE, FPY.

• Products: accurate routings and modularization reduce changeovers. KPI: shortage rate, OTD, bill-of-material change rate.

• Processes: capability and maintenance reduce rework loops and stabilize the bottleneck. KPI: FPY, scrap rate, mean time between failures.

• Plant (usine): layout and internal logistics dictate distances and the risk of starvation (affamement). KPI: distance traveled, logistics waiting rate, WIP.

• Planning: lot policy, buffer levels, and rescheduling frequency arbitrate between OPEX and working capital. KPI: schedule adherence, OTD, lead time.

VII. Production software and IS (information systems): ERP, MRP, GPAO, MES, APS

No tool compensates for an inconsistent control rule. The right stack depends on your variability and decision horizon.

MRP and GPAO (gestion de production assistée par ordinateur) handle variability poorly when they rely on fixed lead times and average yields: they then produce unstable promises and constant expediting. An APS arbitrates under constraints and helps stabilize sequences, but requires clean data and governance over priority rules.

An MES executes. An APS arbitrates.
If the rules are bad, they won't fix them—they'll amplify them.

VIII. Direct answers: definitions and methods in 60 words

Definition and objectives: what production control really measures and arbitrates

Production management organizes the transformation of demand into orders, then into conforming delivered products. It arbitrates throughput, lead time, costs, and robustness, with explicit control of variability. It is measured through OEE, OTD, WIP, lead time, scrap rate, and emergency costs. Its real objective is predictable performance—so protected margin.

Process: how to connect planning, scheduling, execution, and improvement

Planning sets volumes and capacity, then scheduling produces an executable plan under shopfloor constraints. Execution brings back shopfloor truth through events, times, stops, and quality. Continuous improvement reduces the dominant causes of variability and stabilizes rules. Control is measured via schedule adherence, OEE, FPY, WIP, and OTD.

4 production types: which control mode depending on product and demand

Unit and project require finite loading and milestones, with time buffers. Small series require explicit customer priorities and controlled lots. Series require leveling, balancing, and strict protection of the bottleneck. Continuous requires process stability, maintenance, and energy management. Choice is measured via lead time, WIP, OTD, OEE, and changeover cost.

5 Ps: which concrete decisions, which KPIs to measure

People: multi-skilling and standards, measured via OEE and quality.
Products: routings and variants, measured via shortages and changeover time.

Processes: capability and maintenance, measured via FPY and stops.

Plant (usine): layout and internal logistics, measured via WIP and waiting.

Planning: lots, buffers, and freeze windows, measured via adherence and OTD.

IX. Traps and countermeasures: 5 mistakes that blow up OPEX

1. Managing utilization instead of throughput: you load everything to 95% and get congestion. Countermeasure: manage bottleneck throughput and limit upstream releases with a release buffer. KPI: bottleneck throughput, upstream WIP, lead time.

2. Rescheduling too often: you constantly change the plan and break execution. Countermeasure: enforce a planning freeze over a short horizon and manage exceptions only. KPI: schedule adherence, number of priority changes per day, OTD.

3. Confusing inventory with availability: you increase overall stock and still stock out on the right components. Countermeasure: size targeted buffers by criticality and variability. KPI: shortage rate, OTD, inventory turns, WCR.

4. Treating symptoms instead of variability: you add overtime without reducing the causes of stops and scrap. Countermeasure: address the root causes that dominate bottleneck variability. KPI: stop-cause breakdown, FPY, scrap rate, OEE.

5. Investing without testing scenarios: you buy a machine to catch up on lead time, while WIP comes from a release rule. Countermeasure: test rules, buffers, capacity, and phasing before investing. KPI: ROI (return on investment), EBITDA, cash, working capital, OTD.

X. Conclusion: production management is first a flow discipline

Production management is not won with a “clean” plan, but with explicit rules that absorb variability. If you manage the bottleneck, limit WIP, and choose your buffers (stocks tampons) by scenario, you recover lead time, capacity, and cash—without necessarily investing.

• Measure reality: WIP (Work In Progress, encours), OTD (On-Time Delivery, livraison à l'heure), OEE (Overall Equipment Effectiveness, taux de rendement synthétique), FPY (First Pass Yield, qualité au premier passage).

• Decide with a constraint: one bottleneck, one priority rule, one planning freeze.

• Test before deploying: flow scenarios, then OPEX, WCR (Working Capital Requirement, besoin en fonds de roulement), and lead time (délai de traversée) trade-offs.

Dillygence combines industrial expertise and a digital twin (jumeau numérique) to test these production management scenarios before acting, and turn decisions into measurable gains in capacity, lead time, costs, and carbon footprint.

FAQ — Production control

What is production management?

Production management covers all decisions that convert an order into shopfloor operations, then into conforming shipped products. It includes planning, scheduling, release, execution monitoring, quality, maintenance, as well as inventory and flow management. Its role is to hold the throughput–lead time–cost triad, while integrating real variability. It is managed through OEE, OTD, WIP, lead time, and scrap rate.

What are the key objectives of production management?

The objectives are easy to state, but hard to achieve: meet the customer promise, produce at the right cost, and limit cash tied up. Concretely, that means increasing useful bottleneck throughput, reducing lead time, containing WIP, and stabilizing OTD. It also targets reducing emergency costs, scrap, and energy drift (kWh/part). A good objective ties a shopfloor KPI to a readable impact on the P&L (compte de résultat) and working capital.

What are the main processes in production management?

You typically find a chain of processes: industrial planning (volumes/capacity), scheduling and sequencing (executable plan), release, execution and traceability, quality, maintenance, then inventory management. Each block must operate with stable decision rules compatible with shopfloor variability. Without reliable execution data, planning is just an intention. Without continuous improvement, the same instabilities return and OPEX degrades.

What are the 4 production types?

Typically: unit/project, job shop small series, repetitive series, and continuous (process). These models do not impose the same trade-offs between lead time, changeover costs, stock levels, and stability. Scheduling rules and buffers must match the production type, otherwise priorities contradict each other. Same software, same teams—different results if rules stay copy-pasted.

What are the 5 Ps of production management?

The “5 Ps” structure the control levers: people, products, processes, plant (usine), and planning. People covers skills, multi-skilling, and standards. Products covers routings, variants, and bills of materials. Processes covers quality, capability, and maintenance. Plant covers layout and internal logistics. Planning covers lot sizes, buffers, and rescheduling cadence.

How do you align production management with industrial strategy and investments?

Alignment comes from quantified scenarios linking capacity, lead times, and costs to financial results (EBITDA, cash, working capital). You compare expansion, automation, retrofit (rénovation), and make or buy (faire ou acheter) on explicit assumptions about volumes, mix, and variability. Then you phase implementation by planning buffer capacity and, if needed, a double-run (double production). Investment should follow a quantified demonstration—not an emergency peak.

How do you choose the right planning models in production management?

“Average-based” models can be enough when load is far from saturation and variability is low. As soon as utilization rises or the mix becomes unstable, you need dynamic models and scenarios. The right model integrates breakdowns, scrap, changes, and material delays, then outputs robustness indicators—not just a plan. Validation then happens through a short measurement loop on the shopfloor.

How do you quickly diagnose production management issues on the shopfloor?

Start by locating the real constraint: queues, blocking (blocage), and starvation (affamement) often provide the verdict. Then measure WIP by zone and schedule adherence on the bottleneck. Identify the dominant variability (arrivals, process, or mix), otherwise you will treat symptoms. Change one rule at a time and track one KPI—otherwise diagnosis stays a belief.

How do you stabilize day-to-day scheduling in production management?

Stability comes from a short-horizon planning freeze combined with explicit, few priority rules. Protect the bottleneck, limit releases, then manage exceptions rather than “rebuilding the world” every morning. Document deviations and address root causes—otherwise emergencies return in cycles. Recommended tracking: schedule adherence, WIP, and OTD.

How do you standardize production management across multiple sites?

Standardizing first means speaking the same language: same KPI definitions, same calculation rules, same scopes. Then you must integrate context differences (product mix, quality requirements, automation) to avoid an unrealistic standard. Think “network” too: rare tooling, critical skills, and shared bottlenecks across sites. Finally, without synchronization between ERP, MES, and APS, comparison becomes fragile.

How do you evaluate ROI for a production management project?

ROI is calculated in EBITDA, cash, and working capital—not only local gains. Convert WIP reduction into released cash, OTD improvement into avoided penalties, and scrap reduction into lower material and energy costs. Test sensitivity to volumes, mix, and variability, then frame transition risks. A serious ROI gives a results range and its assumptions—not a single number.

Dillygence combines industrial expertise and a digital twin (jumeau numérique) to test planning and scheduling scenarios before acting, and turn those decisions into measurable gains in capacity, lead time, costs, and carbon footprint.