Skip to main content
NOVYRON

DELIVERY WORK

Selected delivery work, shown in engineering detail

We document delivery work in one consistent format: operating context, challenge, inputs, system boundary, workflow and outcome. Client identities, site details and performance figures remain confidential unless publication is approved.

PROJECT 01MANUFACTURING

Digital twin of an automated packaging cell

CONTEXT
High-volume packaging cell: two robots, conveyors, vision check, manual end-of-line
CHALLENGE
A planned product changeover threatened throughput; physical trials would cost production days
INPUTS
Cell layout, cycle and takt data, PLC signal logs, demand profiles
SYSTEM
Simulation & Twin module; scenario runs for three layout / buffer options
WORKFLOW
Model → validate against recorded weeks → compare scenarios → virtual commissioning of the selected option
OUTCOME
Commissioning-risk reduction; changeover decision made on evidence, parameters handed to the floor pre-validated
SIMULATION & TWINRelated solution

ARCHITECTURE

Twin architecture: layout and cycle data from CAD and PLC logs, plus demand data from the historian, feed a twin running three scenarios. The twin produces a comparison of units per hour, utilisation and queueing, and a validated parameter set handed to the floor.

RESULTS SHOWN AS CATEGORIES · QUANTITATIVE CLAIMS REQUIRE WRITTEN EVIDENCE

PROJECT 02MANUFACTURING · QUALITY

Visual inspection on a high-mix component line

CONTEXT
Component line with frequent changeovers and dozens of active part variants
CHALLENGE
Manual inspection missed intermittent surface defects; no structured record per unit existed
INPUTS
Multi-camera streams, part specifications, historical reject samples, synthetic data for rare classes
SYSTEM
AI Perception with review queues; Guidance gates for borderline lots
WORKFLOW
Detect with confidence per unit → queue low-confidence frames for QA verdicts → verdicts feed evaluation → weekly quality reports
OUTCOME
Quality consistency; every unit carries an audit-ready inspection record
AI PERCEPTIONGUIDANCERelated solution
PROJECT 03INTRALOGISTICS

Orchestrating a mixed robot fleet with the WMS

CONTEXT
Warehouse zone with AMRs from two vendors, a picking arm and conveyor handoffs
CHALLENGE
Vendor systems didn't coordinate; handover failures stalled lanes and every exception needed three consoles
INPUTS
Robot vendor APIs, WMS task queue, conveyor PLC signals, zone map
SYSTEM
Robotics Orchestration with vendor-neutral adapters; Guidance for exception approval
WORKFLOW
WMS task → dispatch across fleet → live state per participant → exceptions pause to safe state → operator resolves from one console, logged
OUTCOME
Coordination reliability; exceptions resolved in minutes from a single console
ORCHESTRATIONRelated solution

ARCHITECTURE

Orchestration architecture: WMS tasks, two vendors of AMR fleet, and a picking arm with conveyor PLC all connect to one orchestrator holding states and handovers. It drives a single live operations console and routes exceptions into a defined safe state.

VENDOR NAMES ILLUSTRATE INTEGRATION SURFACE, NOT ENDORSEMENT

REFERENCE SCENARIOS

Reference scenarios for the applied systems

These are engineering reference designs, not delivery work. Each one describes how an applied system behaves in a representative plant, written in the same format we use for real projects so the operating logic can be examined before any site visit.

REPRESENTATIVE PLANTS · NOT CLIENT DATA · NO PERFORMANCE FIGURES OR GUARANTEES

SCENARIO 01PROCESS WATER · BOILER FEED

A demineralisation train whose cycles keep getting shorter

CONTEXT
Representative plant producing low-conductivity water for a steam boiler and critical process lines: raw water → pre-treatment → RO → cation and anion beds → mixed-bed polishing, on shift operation with variable demand
CHALLENGE
Anion bed cycle length shortened over three consecutive cycles under comparable raw water, while conductivity alarms, regeneration records and RO trends sat on different screens — so each shift formed its own theory
INPUTS
PLC/DCS tags, conductivity and flow analysers, chemical dosing and tank levels, regeneration recipe versions, laboratory results, operator shift notes
SYSTEM
AI Perception for the deviation signal; Guidance for the investigation card and audit trail; Orchestration for the regeneration state machine; Edge & Cloud Operations for OPC UA connectivity and store-and-forward
WORKFLOW
Flag the shortening trend against comparable cycles → open one event card holding RO permeate conductivity, caustic consumption, rinse duration and analyser calibration status → verify the instrumentation first → then examine the pre-treatment trend and recipe version → record the engineering decision on the card
WHAT SURFACES
Investigation completeness: process, maintenance and quality read the same evidence in one place, and the closed card becomes the reference for the next similar event. No cause is asserted before the instruments are verified
AI PERCEPTIONGUIDANCEORCHESTRATIONRelated solution

ARCHITECTURE

Demineralisation scenario architecture: the RO train, the cation and anion beds with their conductivity analysers, and the regeneration recipe and chemical records all feed one cycle model that tracks trend and forecast. It produces a single event card on one timeline, and a data-trust signal that limits what the analytics will assert when an instrument is suspect.

REFERENCE SCENARIO · NOT CLIENT DATA

SCENARIO 02LIQUID SEPARATION · CROSSFLOW MEMBRANE

A ceramic UF train where pressure climbs at constant flux

CONTEXT
Representative industrial plant separating a variable-turbidity, high-organic liquid on multi-channel ceramic membranes, running in clarification or concentration mode depending on the campaign
CHALLENGE
During one campaign, transmembrane pressure rose faster than in comparable cycles while flux was held at target. Raising pressure further risked energy and deposition; cleaning early would spend a cycle for nothing
INPUTS
Feed flow, temperature and solids proxy; crossflow flow and pump speed; feed, retentate and permeate pressures; permeate flow and quality; crossflow and auxiliary power; cleaning records and campaign identity
SYSTEM
Simulation & Twin for the operating envelope; AI Perception for the deviation signal; Guidance for the investigation card; Orchestration for cleaning and backpulse recipes; Edge & Cloud Operations for edge buffering
WORKFLOW
Compare the pressure slope against the envelope for the same feed class → open an investigation card with volume processed since the last clean, campaign, temperature and crossflow energy → verify the pressure instrumentation → examine the hydraulic path and feed character → choose between a short hydraulic clean and a full chemical clean, and record why
WHAT SURFACES
A cleaning decision made from cycle evidence rather than the calendar, and a per-cycle record tying flux, pressure, energy, cleaning and operator notes to one cycle identity
SIMULATION & TWINAI PERCEPTIONGUIDANCERelated solution

ARCHITECTURE

Ultrafiltration scenario architecture: feed and campaign data, the pressure and flux pair with crossflow, and the cleaning history with its recipe and recovery all feed an operating envelope held per feed class. The envelope produces an investigation card with first checks, and a per-cycle record tying flux and energy together.

REFERENCE SCENARIO · NOT CLIENT DATA

ILLUSTRATIVE CASE STUDIES

How an engagement would be structured, end to end

Each scenario below takes one process domain and walks through it the way a project actually runs: operating context, challenge, system boundary, modelling approach, delivery workflow and the measurement plan that would decide whether it worked.

ILLUSTRATIVE SCENARIOS · NOT NAMED CUSTOMER PROJECTS · NO ACHIEVED RESULT IS CLAIMED

CLIENT CONFIDENTIALITY

Client identities and sensitive site details are withheld unless publication is approved. Quantitative results are shown only when they can be supported and released; otherwise we state the verified outcome category.

Bring us your operating context

Share the process, interfaces, constraints and target outcome. We will use the same structure to frame a technical discussion.

Discuss a ProjectSee the five modules