Question 1

Where in the ISA-95 stack does your ML inference actually sit — level 2, 3, or 4?

Accepted Answer

Inference belongs at level 3 (MES / plant operations), occasionally bridged through the level 3.5 DMZ to a level 4 (ERP / cloud) training cluster. Level 2 (PLC / SCADA real-time control) stays untouched — we never co-locate inference on the PLC and never write back synchronously to control loops. A predictive-maintenance model reads from the historian (level 3), publishes a work-order recommendation to FactoryTalk ProductionCentre or Opcenter (level 3), and lets the CMMS push it to operators. A demand-forecast or yield-optimisation model runs at level 4 against the S/4HANA or Oracle Fusion data and posts adjusted MRP suggestions back. Anything that closes a loop on a regulated process passes through human approval first.

Question 2

OPC-UA versus MQTT Sparkplug B — which do you choose and how do you secure it?

Accepted Answer

OPC-UA is the right choice when you need rich type information, historical access (HA service set), and explicit security contexts — we run Basic256Sha256 with mutual X.509 certs issued from the plant CA, with audit logging on the OPC-UA server. MQTT Sparkplug B is the right choice for wide-area, low-bandwidth, high-cardinality edge-to-cloud flows — its Birth/Death certificates and unified namespace (UNS) topology make it easier to handle thousands of intermittently-connected devices. We typically run OPC-UA south of the DMZ (level 2 to level 3) and bridge to Sparkplug B for cloud-side analytics, TLS 1.2+ with mTLS on the broker (HiveMQ or EMQX), and tagged ACLs per edge gateway. Both modes are wrapped in IEC 62443 zone-and-conduit policy.

Question 3

How do you segment AI pipelines under IEC 62443 without strangling the model?

Accepted Answer

IEC 62443 zone-and-conduit thinking maps cleanly onto the Purdue model. We place training infrastructure at level 4 (or in the customer VPC), the inference and feature store at level 3, and a data-diode or one-way replication into the level 3.5 DMZ from level 2 historians. Conduits are firewalled, certificate-pinned, and monitored — typically Claroty or Nozomi on the OT side feeds anomaly alerts into the same SIEM. Models never call out to the public internet; updates arrive as signed offline bundles, scanned at the boundary, and registered in a private MLflow. Latency budget for inference is in the 100-500ms range for advisory workloads, which fits the segmentation without contortion.

Question 4

For pharma manufacturing under FDA 21 CFR Part 11, how do AI decisions get treated as electronic records?

Accepted Answer

Any AI output that contributes to a GMP decision — golden-batch pass/fail, in-process release, deviation flag — becomes a Part 11 electronic record. That means: attributable (signed by the model version and the human who approved it), legible, contemporaneous, original, accurate (ALCOA+), with audit trails on every change, secure e-signatures, and validated change control. We pin model versions, hash the feature vectors, and write an immutable audit log (typically into the same eBR system — MasterControl, Veeva Vault QualityDocs, or a plant historian configured for Part 11). The model itself is validated as a Computer System under GAMP 5 — risk-categorised, IQ/OQ/PQ executed, with a continued process verification (CPV) plan tied to FDA Process Validation Stage 3. Pharma never auto-acts; the system recommends and a qualified person approves.

Question 5

What kind of historian tag fidelity do you actually need for predictive maintenance to work?

Accepted Answer

Vibration-based bearing health and RUL prediction needs raw or near-raw waveform data, not aggregated historian tags — typically 1-25.6 kHz sampling from accelerometers, captured by an edge box (NI cDAQ, OSIsoft PI Connector for high-speed, or an Ignition Edge with the Sparkplug module). For motor-current signature analysis, 10-100 Hz is enough. The PI / Ignition tags that most plants log at 1 Hz with compression are fine for OEE and process anomaly detection, but useless for incipient bearing-fault detection — you cannot see a 2 kHz bearing pass frequency in 1 Hz data. We do an information-content audit before quoting a predictive-maintenance project: if the historian compression is dropping the relevant frequencies, we install or commission edge capture first and start the model later.

Question 6

On a conveyor visual-inspection line, how do you tune defect detection accuracy vs false-reject rate?

Accepted Answer

It is a Pareto curve, not a single number. We start by labelling 5-10k images per defect class with the plant QA team, train a two-stage detector (a fast classifier for present/absent, then a localiser for defect type and bounding box), and calibrate the threshold against a labelled holdout. On a typical electronics or food-packaging line we target recall above 99% on safety-critical defects (cracked seal, foreign object) — that comes at the cost of a false-reject rate around 1-3%, which is acceptable because rejects go to a rework lane, not the bin. For cosmetic defects we tune the other way: lower recall, near-zero false-reject, because over-rejecting blows up the scrap rate and operators will disable the system within a week. The model goes live in shadow first — it flags, the human inspector still decides — for two to four weeks before we hand over the gate.

Question 7

How do you detect golden-batch deviation across ISA-88 batch records?

Accepted Answer

A golden-batch model learns the multivariate envelope of the recipe phases — temperatures, pressures, agitation, addition timings — from a curated set of historically good batches. We pull the batch genealogy from the ISA-88 batch executor (DeltaV Batch, Rockwell PlantPAx, Siemens SIMATIC BATCH) as B2MML or via the OPC-UA historical access service, segment by recipe phase, and fit either a multivariate PCA / PLS model (the regulated-pharma default — explainable, easy to validate) or a more flexible variational autoencoder for non-regulated chemicals. At runtime each new batch is scored phase-by-phase; deviations beyond the Hotelling T-squared or SPE control limits surface to the batch-record reviewer with the exact tags driving the deviation. For commercial pharma the model is an aid, not an arbiter — release-by-exception still requires a qualified-person signature against the eBR.

Question 8

How does AI-driven traceability and multi-tier genealogy actually shrink recall scope?

Accepted Answer

A traceability graph links every finished serial number back through its work-order, the operator and machine that built it, the in-process parameters from the historian, the component lots consumed (per BOM), and from there to the upstream supplier lots — across however many tiers the customer can give us data for. When a defect is found in the field, the query is "which serials share this combination of supplier lot, machine, shift, and process-parameter band?" — instead of recalling six months of production we recall the affected sub-population, often two orders of magnitude smaller. We build the graph in Neo4j or TigerGraph for relational manufacturers, or in a columnar store with materialised joins where the volumes are higher (automotive Tier 1, semiconductor). For regulated industries the graph is read-only from validated source-of-truth systems (MES, ERP, LIMS) — we do not become the system of record, we become the system of inference on top of it.

Plant-floor AI: OEE from 60% to 75%+, defect recall above 99%, IEC 62443-clean

MES, ERP, SCADA, historian

What we build on the plant floor

Predictive maintenance with RUL prediction

Visual quality inspection on conveyor lines

Golden-batch deviation across ISA-88 records

OEE root-cause from PLC tag history

Multi-tier traceability & supply-chain genealogy

Operator-assist generative agents

OT/IT data engineering for regulated plants

OPC-UA + MQTT Sparkplug B pipelines

MES & ERP read-side integration

Historian access - PI, AVEVA, Ignition

Time-series + waveform processing

Genealogy, BOM & batch records

IEC 62443 zone-and-conduit deploy

Questions plant engineers ask

Where in the ISA-95 stack does your ML inference actually sit - level 2, 3, or 4?

OPC-UA versus MQTT Sparkplug B - which do you choose and how do you secure it?