SCENARIO 01 · THERMAL PROCESSING · MULTI-GRADE DRYER
Moisture and energy optimisation in a multi-grade dryer
Soft sensing, constrained optimisation and anomaly detection around a continuous dryer whose grades each behave differently.
ILLUSTRATIVE CASE STUDYIllustrative scenario. This is not a named customer project and claims no achieved result. It shows how NOVYRON would structure a technically governed engagement.
OPERATING CONTEXT
A continuous dryer processes several grades whose inlet moisture and production rate both vary. Final moisture is confirmed by delayed laboratory samples, so operators hold conservative temperature and residence-time settings to stay safe.
CHALLENGE
- Long process dead time between an action and the resulting outlet moisture
- Grade changes that invalidate any one universal recipe
- An energy penalty from overdrying and from excess exhaust heat
- A requirement to preserve existing PLC protections and operator authority
SYSTEM BOUNDARY
NOVYRON supplies the software intelligence, integration and lifecycle layer. The process equipment, instruments, PLC/DCS logic, safety systems and approved operating procedures stay under the control of the customer and their equipment vendors.
WHAT NOVYRON WOULD BUILD
- Connect read-only historian tags, energy meters, recipe context and laboratory moisture
- Create a heat and mass-balance feature layer and a grade-specific soft sensor
- Forecast outlet moisture at the expected material residence time
- Rank permitted airflow, temperature and speed adjustments by constrained optimisation
- Detect model-residual changes tied to fouling, air leakage or sensor drift
MODELLING & DECISION LOGIC
A hybrid model combines physical evaporation features with gradient-boosted or sequence models. Prediction intervals widen when a new feed regime is detected, and the optimisation runs only inside engineer-approved recipe and equipment limits.
DELIVERY WORKFLOW
- S1Weeks 1–2: tag audit, residence-time mapping and KPI agreement
- S2Weeks 3–6: offline modelling and historical replay across grades
- S3Weeks 7–9: shadow predictions beside current operator practice
- S4Weeks 10–12: advisory pilot on a selected set of grades
- S5On acceptance: governed rollout with a recalibration schedule
DATA THE SCENARIO WOULD USE
Historian tags for inlet and outlet temperature, airflow, pressure, fan state, product rate, drum or conveyor speed and the active grade; steam, gas and electricity meters; laboratory or at-line moisture; downtime, maintenance and alarm records; and combustion-air or ambient conditions where they move the process. Tag timestamps, sample times and material residence time are reconciled before anything is trained.
MEASUREMENT PLAN
A pilot starts from agreed definitions, baseline windows and exclusion rules. These are the measures that would be agreed, not results that have been obtained.
- Moisture standard deviation and specification compliance
- Specific thermal energy per kilogram of water removed
- Overdrying and rework rate
- Throughput at equal moisture quality
- Warning lead time for abnormal thermal or airflow behaviour
Results would be evaluated against comparable products, feed conditions and operating constraints, with model predictions, operator actions and external disturbances all kept traceable.
MEASURED RESULTS — STATUS
No customer result has been measured or is claimed here. A real pilot would freeze a comparable baseline, define grade and feed-condition cohorts, and report prediction error, moisture variability, specific energy, throughput and off-spec rate with confidence intervals. A benefit would be published only after the customer validated it and permitted publication.
WHAT THE PILOT WOULD LEARN
Which disturbances actually dominate moisture variation; where the present strategy is conservative for no gain; how much warning a soft sensor can really give; and whether recommendations stay stable across grades and seasons. It also fixes the point at which the model should abstain because feed or sensor behaviour has left its validated domain.
DATA & GOVERNANCE
- Read-only and shadow-mode deployment come before any approved write-back.
- Access is role-based, and data and model versions are logged.
- Site safety systems and validated quality release stay authoritative.
- Customer data use, retention and model-training rights are documented.
- Every recommendation shows its assumptions, its confidence and its constraint status.
EXPECTED OUTCOME CATEGORY
Tighter moisture distribution and lower avoidable thermal load, subject to baseline quality, instrument capability and operating constraints. No percentage saving is claimed in this scenario.
DOMAINS USED IN THIS SCENARIO
Discuss a comparable process
Give us the process boundary, the tags you already have, the quality measures, the constraints and your current baseline. We will propose a data assessment and a shadow pilot — without a hardware-vendor change.