Dillygence
Managing a Flexible Factory: 5 Objections to the Liquid Factory
Managing a flexible factory: "Too expensive", "too complex", "unmanageable"... Discover how to address the 5 objections against the liquid factory
Flexible factory management: convince the executive committee with evidence, not opinions
In industry, 20 to 40% of manufacturing time is lost to waiting, changeovers, transfers and rework according to Hopp and Spearman's models. This waste inflates work-in-process (WIP) and sometimes triggers defensive purchases. Too often, the executive committee (CODIR, Comité de Direction) relies on static dashboards: operational management turns these opinions into quantified, simulated scenarios. Takeaway: a digital twin compares options, risks and financial impacts to make the debate measurable.
The debate: lean organizational flexibility vs equipment inflation
Flexibility is often pitted against higher capital expenditure (CAPEX, capital expenditure). Ignoring organizational levers—operation grouping, sequencing rules, multi-skilling—leads to unnecessary spending. Proposing organizational options first protects cash and can unlock hidden capacity. Dynamic simulation documents assumptions and consequences for the executive committee (CODIR).
1) Clarify the concepts: what you manage in the factory
Flexible production system, flexible line, “liquid” factory: scope and measurable promises
A flexible line produces multiple part numbers with controlled changeover times and stabilized quality. A flexible system organizes the workshop to absorb volume and mix variation without exploding work-in-process. A “liquid” factory extends this principle to the whole site, with reconfigurable cells depending on load. Promises are measured through service level, throughput time and usable capacity at the constraint workstation.
Management vs supervision vs planning: three loops, three common mistakes
Supervision shows real-time status. Planning anticipates. Management decides and links supervision and planning through action rules. Common mistakes: multiplying screens without rules, freezing a plan despite variability, optimizing locally at the expense of flow.
Mistake 1: confusing “seeing” with “acting”.
Mistake 2: confusing “forecasting” with “imposing”.
Mistake 3: optimizing a workstation and losing the flow.
The concept that changes everything: the roaming bottleneck and its effects
Depending on the mix, the bottleneck moves. This “roaming bottleneck” makes local optimization misleading and inflates work-in-process if you protect dedicated lines. Flexible management follows the real constraint, adjusts priorities, sequencing and buffers, and avoids poorly targeted purchases.
2) The management framework in 3 horizons: real time, short term, medium term
Real time: detect variability, decide fast, avoid local optimization
Goal: prevent a local incident from destabilizing the flow. Define thresholds, owners and standard actions for each critical alert. Measure the delay between alert and action: reducing it often cuts throughput time, especially if variability hits the constraint. Visual management must display an executable rule, not a decorative board.
Signal | Simple threshold | Expected decision | Measured result |
|---|---|---|---|
Constraint downtime | > X minutes | Maintenance priority + immediate support | Minutes lost at the constraint |
Upstream WIP before constraint | < target buffer | Speed up internal feeding | Risk of throughput break |
Cycle time drift | > Y % | Stop the cause + switch sequencing | Rate stability |
Short term (days–week): dynamic scheduling, simple rules, robust trade-offs
Choose which orders to pull forward and which skills to allocate. Limit exceptions with a small set of rules and controlled replanning. Protect the bottleneck with buffers sized to variability and reaction time, and validate this sizing by simulation before deployment.
Daily priority at the bottleneck (family, urgency, quality risk).
Limit the number of active manufacturing orders (OF, ordre de fabrication).
Adapt batch size based on changeover and quality.
Allocate operators based on skill coverage.
Medium term (month–quarter): capacity, product-mix scenarios, investment trajectory
The medium term sets hiring, training and investments. New product introduction (NPI, new product introduction) brings learning curves. Link throughput, work-in-process and capital expenditure to financial impacts: working capital requirement (BFR, besoin en fonds de roulement) and gross operating surplus (EBE, excédent brut d'exploitation). A 20% reduction in WIP on €5M of WIP already frees €1M of BFR.
Who decides what: production, logistics, maintenance, quality, methods
Define a decision matrix. Production manages operational priorities; logistics manages feeding; maintenance prioritizes based on flow impact; quality defines exceptions; methods standardize work and times.
Production: priorities, sequencing, manufacturing order (OF) arbitration.
Logistics: feeding the bottleneck, storage areas.
Maintenance: interventions based on throughput loss.
Quality: deviation rules and checkpoints.
Methods: standards, changeover preparation.
3) Five executive committee objections, five answers testable by simulation
Objection 1: capital expenditure goes up
Answer: measure time losses and transfers, then simulate operation groupings and alternative sequencing. Organizational modularity often unlocks hidden capacity. Launch capital expenditure (CAPEX) only if simulation shows a net economic gain greater than organizational options.
Objection 2: management becomes chaos
Answer: formalize a few rules—customer cadence, elastic buffers, bottleneck priorities—and allow deviations only if the constraint changes. Visual management must display action signals and executable thresholds. Testing these rules in a digital twin reduces controversy.
Objection 3: teams refuse multi-skilling
Answer: structure multi-skilling through cross-training, authorizations and planned rotations. Measure skill coverage and tie it to the risk of throughput break. Team leaders manage the skill ramp-up.
Objection 4: WIP saturates the workshop
Answer: move from push flow to pull flow, limit active manufacturing orders (OF, ordre de fabrication) and size buffers to variability and reaction time. Little's Law links WIP, throughput time and throughput: reducing WIP frees cash, but the effect depends on the constraint and management rules. Simulation validates buffer size.
Objection 5: changing the layout is too risky
Answer: use a virtual test bench to stress each scenario with breakdowns, absences and mix variation. Splitting the relayout into stages and defining stop criteria avoids paralysis. Simulation provides measurable paths for the executive committee (CODIR).
4) Data and tools: connect each building block to a management decision
MES, ERP, APS, BI: what each system brings—and what it doesn't solve
The MES (Manufacturing Execution System) tracks execution and traceability. The ERP (enterprise resource planning) manages inventories and transactions. The APS (advanced planning and scheduling) computes constraint-based plans. BI (business intelligence) structures reporting. No tool decides without defined arbitration rules.
Real-time supervision: actionable alerts, thresholds, expected response time
Design alerts that trigger one clear action, with an identified owner and a measurable response time. Limit the number of alerts to avoid alert fatigue. Measuring the delay between alert and action often helps reduce throughput time, especially when the alert concerns the constraint.
Releasing manufacturing orders and managing quality constraints: rules, exceptions, traceability
Limiting active manufacturing orders (OF, ordre de fabrication) based on downstream capacity and target buffer controls WIP. Defining quality checkpoints and deviation rules prevents a late investigation from becoming an invisible bottleneck. Traceability prevents workarounds.
The digital twin: move from reporting to scenario-based arbitration
The digital twin compares sequencing policies, batch sizes, resource allocation and buffers, then measures total throughput, WIP and throughput time. It documents assumptions and results, builds a reusable scenario library, and accelerates multi-site decisions.
Executive committee question | Compared scenarios | Observed measures | Decision |
|---|---|---|---|
Should we buy a machine? | Organization vs capital expenditure | Throughput, WIP, lead time, costs | Invest if net gain is proven |
Which scheduling rule? | 2–3 simple rules | Service, stability, changeovers | Most robust rule |
What buffer size? | 3 bounded levels | Constraint starvation vs congestion | Explicit trade-off |
5) Minimum viable KPIs for a reconfigurable, flexible factory
Service and plan KPIs: OTIF, plan adherence, rate stability
OTIF (on-time in-full) measures customer service. Plan adherence compares planned vs executed. Rate stability quantifies output variability. Each KPI must trigger an action when a threshold is crossed.
Flow and cash KPIs: WIP, throughput time, turns, BFR impact
Work-in-process (WIP) is immobilized cash. Throughput time is total time and customer risk. Turns measure capital efficiency. Report these KPIs by area and by family to detect local congestion.
Capacity KPIs: OEE at the bottleneck, changeover time, availability, yield
OEE (Overall Equipment Effectiveness, TRS—taux de rendement synthétique) combines availability, performance and quality; measuring OEE at the bottleneck remains the priority. Changeover time guides effort on mix; availability guides maintenance. Yield links losses and causes to prioritize improvement work.
Quality KPIs: scrap, rework, cost of poor quality and induced delay
Scrap and rework consume capacity and margin. Cost of poor quality aggregates these losses in euros to support arbitration. Induced delay measures the impact on throughput time.
Anti-trap reading: when a “good” KPI drives a bad decision
A KPI in isolation can mislead: a high OEE outside the bottleneck doesn't improve throughput. Short-term OTIF can hide cost of poor quality. Rank KPIs by impact on total throughput, WIP and service, then test side effects through simulation.
Trap: optimize OEE everywhere. Response: focus energy on the bottleneck.
Trap: reduce WIP everywhere. Response: keep a bounded buffer to protect the bottleneck.
Trap: chase OTIF with expediting. Response: stabilize arbitration rules.
6) Three mini-cases: when management becomes truly adaptable
Mini-case | Context | Implementation | Impact |
|---|---|---|---|
Mini-case 1: unstable product mix | Multi-part-number workshop with customer expedites. | Bottleneck priorities, limit active OF, adjust batch sizes, simulation. | WIP reduced by 20–30% and throughput time cut without machine investment. |
Mini-case 2: ramp-up | Volume increase on one family, constraint workstation. | Simulate resource allocation, cross-training and calibrated buffers. | CAPEX deferred and better plan adherence. |
Mini-case 3: staged relayout | Saturated workshop and long transfers. | Phasing, simulations and temporary routing rules. | Continuity maintained and reduced transfers and waiting. |
7) Deploy flexible factory management in 8 steps, without a Rube Goldberg machine
Step 1: flow/capacity diagnosis and constraint mapping
Identify families, routings and cycle times; locate the bottleneck and measure its stability by mix. Quantify WIP by area to build a baseline the executive committee (CODIR) can challenge. Without this baseline, each function produces its own reality.
Step 2: minimal reliable data and pragmatic instrumentation
Choose few but reliable measures: changeover times, major downtimes, good quantities, WIP by area and operator hours. Simple governance on definitions avoids number wars. Approximate data leads to random decisions.
Step 3: prioritization rules and buffer logic
Define the rule that protects the bottleneck, set target buffers and alert thresholds, formalize OF release and document exceptions. Size buffers to variability and reaction time; test via simulation. An unbounded buffer quickly becomes a parking lot again.
Step 4: dynamic scheduling loop and exception management
Run replanning at a fixed frequency, limit changes and measure their impact on plan adherence. Treat quality and maintenance with a defined protocol. Without a protocol, the exception becomes the rule.
Step 5: shopfloor execution rituals and reaction standards
Install short rituals on daily trade-offs with playbooks for breakdowns, quality drifts, internal shortages or absenteeism. Track and reduce the delay between signal and action. This gain often shortens throughput time if variability hits the constraint.
Step 6: dynamic simulation and a digital twin to decide trade-offs
Build a clear flow model, test mix variation, breakdowns and team sizing, then compare effects on throughput, WIP and lead time. Goal: align workshops and leadership on measurable results.
Step 7: stage the master plan and defer CAPEX if not profitable
Build a staged plan, prioritize organizational gains and changeover improvements. Launch capital expenditure only if simulation proves a return greater than organizational options. Keep reversible options while assumptions remain unstable.
Step 8: standardization and multi-site continuous improvement
Harmonize KPI definitions, buffer logic and prioritization rules, while keeping local latitude for machine constraints. Capitalize scenarios in the digital twin and create a continuous improvement loop with thresholds and traceable decisions.
8) Traps & countermeasures: 5 mistakes that ruin flexible management
Decorative KPIs: tracking indicators without action.
Countermeasure: link each KPI to an arbitration rule and an owner.Real time without decisions: multiplying ignored alerts.
Countermeasure: define thresholds, owners and response time; reduce alerts to actionable signals.Frozen scheduling: refusing any replanning.
Countermeasure: create a dynamic scheduling loop at a fixed frequency.Bottleneck ignored, then roaming bottleneck not tracked: managing misleading averages.
Countermeasure: identify the current bottleneck and steer maintenance, logistics and priorities toward it.Technology added on an uncontrolled flow: automating an unstable flow.
Countermeasure: stabilize OF release, buffers and sequencing, then validate by simulation.
Conclusion
Flexible management is judged by arbitration speed when the mix changes and the bottleneck moves—not by pretty dashboards. The executive committee (CODIR) wants three simple measures: total throughput, work-in-process (WIP) and throughput time, compared across 2–3 opposing, quantified scenarios.
When these scenarios show 20–30% less WIP at constant service level, you free up BFR without buying a machine.
Dillygence delivers this management approach through its digital twin, turning shopfloor intuition into measurable and controllable results.
FAQ — Flexible factory management
What is a flexible production line?
A flexible line produces multiple part numbers without a noticeable deterioration in lead time, quality or cost. It reduces the impact of changeovers through controlled changeover times and sequencing rules. Its effectiveness is judged on service, WIP and throughput time.
What is flexible factory management?
Management brings together decisions and trade-offs that adjust priorities, sequencing, resources and buffers to maintain service despite variability. It differs from supervision (seeing) and planning (forecasting) and relies on explicit rules and controlled replanning loops.
What are the goals of flexible factory management?
Stabilize total throughput, bound WIP and control throughput time despite mix variation; improve plan adherence and reduce cost of poor quality. Translate these gains into financial impacts for the executive committee (CODIR).
Which KPIs should be tracked to manage a flexible factory?
Track OTIF, plan adherence, rate stability, work-in-process (WIP), throughput time, OEE at the bottleneck, changeover time, scrap and rework. Each KPI must trigger a defined action and be segmented by area and family.
How should investments be sized for flexible factory management?
Compare organizational scenarios and investments through simulation. Estimate capacity freed through sequencing, buffers and multi-skilling, then launch capital expenditure only if simulation proves a durable financial advantage.
How does flexible factory management reduce costs and throughput times?
Reducing WIP frees BFR and floor space. Limiting expediting and overtime reduces operating costs. Placing controls in the right place reduces rework and scrap, therefore improving lead time when the actions treated weigh on the constraint.
How do you implement flexible factory management day-to-day?
Define simple rules for releasing manufacturing orders (OF, ordre de fabrication), target buffers and bottleneck priorities; install short rituals and actionable alerts. Measure plan adherence and response time, then improve continuously.
How do you manage performance in real time?
Deploy thresholds on bottleneck downtime, WIP by area and cycle time drifts; link each alert to a standard action and measure the delay between alert and action. Prioritize protecting the constraint to preserve total throughput.
How do you manage a ramp-up?
Test scenarios for resource allocation, cross-training and buffers through simulation; check impacts on the constraint and track plan adherence. Launch capital expenditure only if organization isn't enough.
How do you standardize management across multiple sites?
Standardize KPI definitions, buffer logic and prioritization rules, while allowing local adaptations for machine constraints. Capitalize scenarios in the digital twin to replicate effective solutions.
How do you limit the risk of a transformation without stopping production?
Split the switch-over into stages, run a pilot flow and define stop criteria and rollback plans. Use dynamic simulation to stress each scenario with realistic variability before any physical change.
What ROI can you expect from flexible factory management?
ROI comes from BFR freed by lower WIP, useful capacity gained by reducing changeovers and waiting, and lower cost of poor quality. A credible ROI requires auditable assumptions and post-deployment tracking on total throughput, service and cash.
Traps & countermeasures (recap): 1) KPIs without action → link each KPI to a rule; 2) alerts without decisions → thresholds and owners; 3) frozen plan → dynamic scheduling loop; 4) bottleneck ignored → measure OEE at the bottleneck; 5) automating an unstable flow → stabilize then simulate.
