Why it matters
Without AI, corrosion progress can be missed, treatment choices are delayed or misapplied, and insurance and loan processes face contention due to incomplete condition records and non-reproducible sampling strategies.
From design to safety
Design
AI-supported treatment planning
The system reads prior condition reports, corrosion panels, and material records to propose minimally invasive methods and inhibitor strategies aligned with reversibility and ethics. It defines data governance, workflow handoffs, and PACR-compliant documentation paths for auditability, including schema for XRF, PIXE, Raman, and colorimetric data, plus versioned condition histories.
- Portable XRF Analyzer
- Multispectral Imaging Camera
- Desiccation Monitoring Probe
- Corrosion Inhibitor Dispensing System
Build
Equipment fit-out and calibration
Hardware and software are commissioned, calibration data logged, and imaging paths established for reliable detection of chlorides, bronze disease fronts, and patina integrity. It records installation checklists, sensor alignments, and operator training completion, with baseline drift tests and documented pass/fail criteria.
- Portable XRF Analyzer
- Multispectral Imaging Camera
- Electrochemical Corrosion Sensor
- Desiccation Monitoring Probe
Operation
Ongoing condition surveillance
The AI ingests real-time readings, imaging, and prior reports to adjust risk scores, trigger targeted sampling, and generate treatment proposals for new condition reports. It links outputs to existing condition archives and treatment proposals, generating PACR-ready narratives and provenance stamps.
- Portable XRF Analyzer
- Multispectral Imaging Camera
- Electrochemical Corrosion Sensor
- Gas Chloride Quantification Kit
Maintenance
Routine recalibration and stock checks
Calibration logs, sensor drift notes, and inhibitor concentrations feed prompts for re-ordering, re-calibration, and updating treatment-protocol PDFs stored in archives. It enforces version control on all condition records, including supplier lot numbers and expiry dates.
- Desiccation Monitoring Probe
- Corrosion Inhibitor Dispensing System
- Gas Chloride Quantification Kit
- Portable XRF Analyzer
Safety
COSHH and ethics compliance guard
AI flags high-risk samples, requires sign-off by the accredited conservator and compliance team before any invasive sampling or new treatment protocol is applied. It maintains an auditable ethics trail and regulatory sign-offs, including sampling permits and inhibitor compatibility notes.
- Electrochemical Corrosion Sensor
- Desiccation Monitoring Probe
- Corrosion Inhibitor Dispensing System
- Multispectral Imaging Camera
Part by part
- 1
Portable XRF Analyzer
- Watch for
- chloride-rich corrosion layers and copper corrosion products diffuse along patinas
- AI action
- flag high chloride risk and suggest targeted sampling locations
- 2
Multispectral Imaging Camera
- Watch for
- areas hidden under patina or corrosion haze not visible in RGB
- AI action
- trigger enhanced imaging protocol and document findings in condition report
- 3
Gas Chloride Quantification Kit
- Watch for
- rising chloride concentration indicating active corrosion
- AI action
- generate sampling plan and site-specific irradiation or extraction steps
- 4
Electrochemical Corrosion Sensor
- Watch for
- current density spikes at corrosion fronts
- AI action
- issue alert and propose prioritised treatment areas
- 5
Desiccation Monitoring Probe
- Watch for
- humidity spikes causing desiccation of organic bindings
- AI action
- adjust HVAC setpoints and schedule re-evaluation in 24 hours
- 6
Corrosion Inhibitor Dispensing System
- Watch for
- inhibitor concentration drift
- AI action
- reorder stock and update application protocol
Worked example
Detecting Bronze Disease Before Severe Forward Spread
Illustrative scenario with sample data.
A private studio with a cast-bronze sculpture seeks early bronze disease detection to inform a reversible treatment plan.
Signal in
objectID: 1024, location: Gallery Hall, priorConditionReport: BRONZE_DISEASE_STAGE_2, chlorideLevel: 45 ppm, relativeHumidity: 58%, surfacePatina: intact, previousInhibitorBatch: B2023-11, imagingBaseline: SPECTRAL_2024_07
System out
AI-generated risk score: BRONZE_DISEASE_RISK_High; recommended sampling plan: chloride hotspot zones; provisional treatment proposal: apply reversible corrosion inhibitors; report updated to condition archive with provenance stamps and material safety data.
Human sign-off: Head Conservator signs off on sampling plan and treatment proposal; PACR-compliant condition report updated; client notified; inhibitor protocol versioned and included in safety dossier.
How to roll it out
Phase 1
Weeks 1 to 4
Baseline installation of imaging sensors, data pipelines, and initial AI training with existing condition reports.
Phase 2
Weeks 5 to 8
Calibration, validation against known samples, and integration with PACR-compliant documentation.
Phase 3
Weeks 9 to 12
Operational workflows established; first client project runs with monitoring dashboards and automated alerts.
Where people stay in charge
- Decisions on sampling and treatment must be approved by the accredited conservator or project lead.
- All AI-generated actions require documented human authorization before execution.
- Ethical and regulatory compliance decisions must be signed off by the studio’s safety officer or regulatory liaison.
Questions owners and crews ask
What data feeds does the AI use for bronze disease detection?
The AI integrates prior condition reports, imaging data, chloride measurements, humidity records, and inhibitor stock levels to generate risk assessments and treatment recommendations, with explicit provenance notes for each input.
Who approves AI-generated sampling plans?
An accredited conservator reviews the AI-generated plans, ensuring reversibility, minimal intervention, and COSHH compliance before any sampling, recording sign-off in the treatment proposal and condition report.
How are AI alerts documented in condition reports?
Alerts are logged as AI-generated notes with priority level, recommended actions, responsible technician, and links to source data within the treatment report, each item time-stamped and version-controlled.
What safeguards prevent misapplication of inhibitors?
The system requires human sign-off, references material safety data, and verifies inhibitor compatibility with the metal and patina before use, with audit trails for every re-application.