Skip to content

Data Integration

NTWIST products are only as good as the data they read. This section describes what data we need, where it usually lives, and how integration is sequenced during deployment.

Principles

  • We deploy against the data that exists today. We do not require historian or LIMS cleanup as a precondition.
  • No data leaves the customer environment by default. Process data, recipes, and operating knowledge stay inside customer infrastructure.
  • Integration is iterative. We start with a minimal viable signal set and add depth as the use case matures.

Data sources by application

MineMax (mining)

Application Typical sources
OreMax Resource and grade control models, ore movement records, stockpile surveys, plant feed assays, LIMS
DynaMax Resource and grade control models, ore movement records, ROM pad surveys, reclamation activity logs, mill weightometer data
MillMax Plant historian (throughput, power, recovery, reagent usage, control tags), upstream feed characterization, circuit configuration; optionally LIMS, equipment availability, cost data
PlanMax OreMax / DynaMax inventory models, short-term mine plan, constraints and targets; optionally lab and mineralogy, plant model curves
HaulMax Fleet management system, dispatch, fuel system, equipment availability, mine plan

Nexus iMES (manufacturing)

Capability Typical sources
Trusted Scheduling Work orders, routings, machine capabilities, labor availability, inventory readiness
Shop-Floor Execution Published schedule, task instructions, workstation assignments
Master Data Management ERP item master, routings, BOMs, HR shift data, tooling lists
KPIs and Insights Schedule history, execution events, downtime reasons, work-order outcomes

Common

  • Identity: customer identity provider (SAML, OIDC, LDAP, Active Directory)
  • Audit and observability: customer log destination (SIEM, ELK, Splunk, Sentinel) where one exists

Supported protocols and systems

Historians

  • OSIsoft PI (AF SDK and PI Web API)
  • Aveva PI and Aveva Historian
  • Cogent DataHub
  • Canary Historian
  • Wonderware Historian
  • Custom historians via REST or SQL

Plant-floor protocols

  • OPC UA (preferred)
  • OPC DA (legacy, supported with tunneller)
  • Modbus TCP
  • MQTT
  • Direct PLC drivers for Allen-Bradley, Siemens, and Schneider where required

Enterprise and business systems

  • SAP (ECC, S/4HANA) via IDoc, BAPI, or OData
  • Oracle ERP via REST or database integration
  • Microsoft Dynamics 365
  • Custom ERPs via REST, SOAP, or database integration
  • LIMS systems via REST, file drop, or direct database

Identity

  • SAML 2.0
  • OpenID Connect
  • LDAP and Active Directory

Messaging and event buses

  • Apache Kafka
  • RabbitMQ
  • MQTT brokers
  • Azure Service Bus
  • AWS SQS and SNS

Integration sequencing

A typical integration sequence during a first-product deployment:

  1. Identity. SSO integration with the customer identity provider, week one or two.
  2. Tag inventory. Inventory of the time-series tags relevant to the use case, week two.
  3. Connector deployment. Stand up the historian or PLC connectors on the plant gateway, week two or three.
  4. Backfill. Backfill historical data to support model calibration, weeks three to five.
  5. Enterprise systems. Integrate ERP, LIMS, fleet management as required by the use case, weeks three to six.
  6. Continuous ingestion. Switch from batch backfill to continuous streaming, week five or six.
  7. Validation. Side-by-side validation against the customer's existing reporting, weeks five to eight.

Data quality and gaps

It is normal for a first integration to uncover data quality issues: missing tags, incorrect units, time-zone mismatches, sensor drift. We document each finding, decide jointly with the customer whether to fix at source or compensate downstream, and we do not let this block deployment progress.

Outbound data

By default, no operational data leaves the customer environment. Optional outbound integrations are supported, with customer consent and a documented data flow:

  • Aggregated KPI reporting to a customer-controlled BI tool.
  • Optional NTWIST analytics or benchmarking services, where the customer explicitly opts in.

All outbound flows are logged and auditable.