SUPi
DocsCore ConceptsDigital Twins

Digital Twins

Physics-based digital twin technology explained.

SUPi's digital twins are not static 3D models or simple dashboards. They are physics-based virtual replicas that simulate real-time stress, fatigue, thermal behavior, and degradation based on actual operating conditions.

Why Physics-Based?

Pure data-driven models break when conditions change. Physics-based twins understand why equipment behaves the way it does, making predictions dramatically more accurate in complex environments like offshore platforms, chemical plants, and wind farms.

Supported Twin Types

Asset TypePhysics Models
Rotating EquipmentVibration dynamics, bearing wear, shaft fatigue
Heat ExchangersFouling progression, thermal efficiency, tube degradation
PipelinesCorrosion modeling, fatigue cycling, pressure loss
ReactorsCatalyst deactivation, thermal runaway, yield modeling
TurbinesBlade stress, pitch optimization, power curve

Hybrid Modeling

SUPi combines first-principles engineering knowledge with machine learning:

  • Physics layer — encodes known thermodynamic, mechanical, and chemical relationships
  • ML layer — learns plant-specific patterns and compensates for unmeasured variables
  • Calibration — continuous Bayesian updating as new data arrives
# Digital twin definition
twin:
  asset_id: COMP-001
  type: centrifugal_compressor
  physics_model: rotating_machinery_v3
  ml_augmentation: true
  calibration:
    method: bayesian_online
    update_frequency: "1h"
  degradation_modes:
    - bearing_wear
    - seal_leakage
    - impeller_fouling