Digital Twins Intro

Posted on Fri, Jul 15, 2022 Industry Note Industry 4.0

A digital twin = a virtual representation of a real-world system

the system can be a physical asset, a plant, or a process

a digital twin combines data from multiple sources, providing a better approximate of the real object

Everyday example: Google Maps as digital twin of the earth’s surface

Layers

Simulation and optimization: route optimizer

Real-time data: current and historical road traffic

Digital copy of physical assets: museum/restaurant locations + schematic map + aerial photos

Different types of digital twin:

Product digital twin

Production plant digital twin

Procurement and supply chain digital twin

Infrastructure digital twin

Building a product digital twin:

example: Jet engine

Decision logic: what actions need to be taken in real-time based on digital twin simulations outputs?

flag engine for maintenance due to predicted component failure

Data visualization: what data outputs need to be visualized? what is the best format for visualization? summary dashboards or parametric models (such as CAD)

real time data used to highlight failure points on product workbench (could be CAD)

Modeling and simulation: what use cases need to be simulated? what type of model should be implemented? how can simulation accuracy be improved with real world test or operational data?

surrogate model that combines real time data with FEA and FMECA tools, simulation scenarios for wear to fuel, air and electronic systems

Data storage and processing: where is the data stored? what are the processing requirements and where to process the data?

data stored on private cloud network and fed into automated reliability performance model

Data capture: what data is needed? how can it be measured? what sensors are required?

real-time vibration, temperature, and pressure data using sensors on jet engine

Types of simulation modeling:

  1. discrete event based

    describes system as sequence of discrete events, such as manufacturing assembly process

  2. agent based

    describes systems as group of agents and their behaviors, such as group of autonomous vehicles on a traffic lane

  3. system dynamics

    describes system on a high-level in abstract form, ignores fine details, such as marketing campaign effectiveness

Model a digital twin

  1. define problem and target KPI
    1. example problem: high machine A downtime of “x hours” leading to low plant output
    2. example KPIs:

      throughput: number of units per plant or per machine

      overall equipment effectiveness: uptime/downtimes per machine

      inventory: number of units in work-centers and storages

    3. expected impact:
      1. reduction in machine “A” downtime by “X%” leading to “Y%” of improvement in margins
      2. additional unlocked capacity of “Z” units
  2. clean and structure data
    1. steps: decide data granularity, collect raw data, convert it into model input
    2. example of data:
      1. material data: type, category, thickness, weight, machine sequence
      2. machine data: number, cycle-time, planned down-time, unplanned down-time
      3. storage data: capacity by type, transporter access
      4. transporter data: capacity by type, speed
      5. shop-floor rules: machine trigger, batch sizes, sequence restrictions, waiting times
    3. raw data converted into custom probability distributions to incorporate inherent system uncertainty in the model
  3. analyze data and map process
    1. process mapping: by conducting structured interviews of on-ground employees and process owners
    2. macro-process flow → exploded view of sub-process for logic building
  4. develop initial model and improve

  5. calibrate model
    1. model output is generated from multiple Monte-Carlo simulation runs, when system reaches the steady state
  6. scenario analysis and implementation for impact

    feasibility-impact metrics analysis helps decision making

    a manufacturing context example

    1. capacity
      1. improve cycle-time of machine
      2. increase number of machines/transporters
      3. improve speed of transporters
      4. increase storage yard capacity
    2. reliability/maintenance
      1. reduce MTOF of a particular machine
      2. reduce MTOF of all upstream machines
      3. reduce MTBF of a particular machine
    3. planning
      1. change material flow sequence for a particular product type
      2. change daily product-mix into the plant
      3. improve material arrangement in storage yard
    4. combined
      1. improve cycle-time of one machine and reduce downtime of another machine
      2. improve storage flow movement and speed of one transporter