Research and methods
Make the assumptions visible.
The research programme examines how evidence, constraints, readiness, data and human judgement interact in complex industrial systems.
Research agenda
Six open questions under active investigation.
Constraint economics in high-mix manufacturingHow should attention and capacity be allocated when the active constraint moves across products, processes and suppliers?
- Practical relevance
- High-mix plants rarely have one fixed bottleneck; capacity spent on the wrong resource does not protect delivery.
- Current evidence
- Theory of Constraints, queueing and operations-research literature, and synthetic high-mix scenarios.
- Design proposition
- A constraint model that makes migration explicit will allocate attention better than fixed-bottleneck heuristics.
- Assumptions
- Load and demonstrated capacity can be estimated per resource with acceptable error.
- Unresolved issue
- How frequently constraints must be re-estimated in volatile mix before the model misleads.
- Validation status
- Concept model with synthetic demonstration; not validated on operational data.
- Next test
- Historical-data replay against known bottleneck migrations.
- Sources
- Goldratt (Theory of Constraints); operations-research capacity literature.
Evidence-gated NPIWhat evidence should be sufficient for a programme to advance, and how should accepted exposure remain traceable?
- Practical relevance
- Gates often advance on calendar pressure while the underlying evidence remains incomplete.
- Current evidence
- Design-science literature, NPI and quality-gate practice, evidence-based management.
- Design proposition
- Gates expressed as decisions under explicit evidence and accepted exposure will produce more defensible advance/hold choices.
- Assumptions
- Required evidence objects can be defined per gate in advance.
- Unresolved issue
- How to weight qualitative expert evidence against quantitative capability data.
- Validation status
- Research prototype logic; not validated with programme outcomes.
- Next test
- Expert review of gate criteria and a usability study with programme leaders.
- Sources
- Evidence-based management; design-science research methodology.
Readiness as a network conditionHow can the weakest supplier, material, process, capacity or approval dependency be surfaced before commitment becomes failure?
- Practical relevance
- An average readiness score can look acceptable while a single dependency is programme-stopping.
- Current evidence
- Systems-engineering dependency analysis and digital-twin readiness literature.
- Design proposition
- Representing readiness as a dependency network will expose commitment risk that an averaged score hides.
- Assumptions
- Critical dependencies can be identified and scored consistently.
- Unresolved issue
- Calibration of dependency criticality across different programme types.
- Validation status
- Concept architecture; predictive validity not established.
- Next test
- Synthetic testing followed by controlled data integration.
- Sources
- Systems-engineering literature; closed-loop manufacturing research.
ERP as decision infrastructureHow can ERP signals become timely operating decisions rather than late administrative records?
- Practical relevance
- ERP holds the data but frequently reports state after the decision window has closed.
- Current evidence
- ERP-enabled operations practice and decision-latency literature.
- Design proposition
- A decision layer that maps ERP signals to explicit actions, owners and thresholds will reduce decision latency.
- Assumptions
- Signal thresholds can be defined that are neither too noisy nor too quiet.
- Unresolved issue
- How to keep thresholds calibrated as demand and supply patterns shift.
- Validation status
- Concept; integration accuracy not tested.
- Next test
- Define exception thresholds and test against historical transaction logs.
- Sources
- Operations-management and ERP practice literature.
Human–AI industrial governanceWhere can AI improve sensing and synthesis while preserving human authority, context and accountability?
- Practical relevance
- Autonomous claims outrun evidence in safety- and contract-critical industrial work.
- Current evidence
- Human-centred AI research and human–AI decision-making studies.
- Design proposition
- Bounding AI to sensing and synthesis, with human decision rights, will preserve accountability without discarding useful assistance.
- Assumptions
- Decision boundaries can be specified and enforced in the interface.
- Unresolved issue
- How to keep boundaries stable as model capability increases.
- Validation status
- Design principle applied across the artefact; not empirically tested here.
- Next test
- Task-based study of decision quality with and without bounded assistance.
- Sources
- Human-centred AI research; strategic decision-making literature.
Judgement under pressureHow should thresholds, trade-offs and invalidation conditions be defined before urgency distorts the decision?
- Practical relevance
- Under deadline pressure, teams often redefine acceptance criteria after the fact.
- Current evidence
- Decision-science and evidence-based management literature.
- Design proposition
- Pre-committing thresholds and invalidation conditions will reduce post-hoc rationalisation.
- Assumptions
- Teams will accept pre-committed criteria and record them.
- Unresolved issue
- How rigid pre-commitment should be when genuine new information arrives.
- Validation status
- Design principle; not validated with field decisions.
- Next test
- Controlled decision exercise comparing pre-committed and ad-hoc criteria.
- Sources
- Decision science; evidence-based management.
Research object template
Every research note must display:
- research question
- practical problem
- evidence base
- design proposition
- concept artefact
- assumptions
- limitations
- validation status
- next test
- sources
Editorial standard
Each note distinguishes:
- fact
- source
- interpretation
- model
- assumption
- synthetic example
- unresolved question