Predictive Analytics and the Digital Twin - · PDF file07.07.2016 · 1 © 2015...

1 © 2015 ANSYS, Inc. July 7, 2016 ANSYS Confidential

Predictive Analytics and the Digital Twin

Bernard DionCTO, ANSYS System Business Unit

Forum TERATEC29 June 2016


• Introduction

• Model-Based Systems Engineering (MBSE)

• Predictive Analytics and the Digital Twin

• Demonstration

• Q&A

Agenda


Unprecedented speed of technology evolution

The Age of Exponential Thinking

The era of “Accelerating Returns” and a new balance of “Price, Performance, Capacity and Bandwidth”

Performance doubles every 18-24 Months

Moore’s Law1975

“...We better disrupt ourselves in our own way before somebody does an Uber to us.“ Doug Oberhelman, CEO CATERPILLAR @ CIW, Oct15, 2015

http://www.caterpillar.com/en/news/caterpillarNews/innovation/caterpillar-disrupted.html


PRODUCTLIFECYCLE

USER

No

n-

Engi

ne

ers

Op

era

tio

nEn

gin

ee

rsD

esi

gnEn

gin

ee

rs

Evolution of simulation

An

alys

ts

OperateManufactureDesign

Advanced Technologies

Virtual Prototyping

Process Compression

Dynamic CAECollaboration

• Usability (UX Enhancement)

• Simulation on Demand

• MBE

• Performance

• ACT Apps• Embed Simulation


ANSYS enables systemsFrom comprehensive component-level design & simulation …

FLUIDS ELECTRONICSEMBEDDEDSOFTWARESTRUCTURES SEMICONDUCTOR


ANSYS enables systems… to complete systems simulation

VIRTUAL SYSTEM PROTOTYPES


Simulation Driven Product DevelopmentFrom build & test to virtual development

Platform based SDPD adoption enables enterprise level scalability for NPI acceleration


Digital PrototypeConnected Industrial Assets

Creation of the digital twin

Digital Twin

Predictive MaintenanceZero Unplanned Downtime

Asset Performance Management


Model-Based Systems Engineering (MBSE)


Components Design

Requirement

Analysis

Sub-systemsRequirement

Design

Analysis

SystemsSpecification

Functional

Logical

Model-Based System Engineering (MBSE): Product Lifecycle

Evolution of Model Based Engineering

Single Source of Truth Between Design Models, Manufacturing and Operations

Requirements Analysis

Product Definition

Commissioning

Operation & Maintenance

Changes & Upgrades

Integrated Design & Analysis

Multidisciplinary Optimization

Realization of MBE Digital continuity

Digital continuity & interoperability Digital Twin

Digital continuity: 3D ROM System ModelInteroperability B2B Model Share

Digital continuity: 3D MultiphysicsInteroperability Collaborative Platform


Model Based System Development

Model-Based System Engineering

Virtual Prototypes of Physical Components

Model-BasedEmbedded Software Development

Virtual Prototypingof Complete System


System Modeling & Simulation

Heat Exchanger

high_t

low_t

outlet_hinlet_h

outlet_c

inlet_c

U2_heat_ex

inletoutlet

U1_low_pump

inletoutlet

U3_h_pump

temp

temp

0.00 2.50 5.00 7.50 10.00 12.50 15.00 17.50 20.00Time [s]

300.00

350.00

400.00

450.00

500.00

550.00

600.00

Y2

[ke

l]

Ansoft LLC Simplorer1XY Plot 1Curve Info

U2_heat_ex.tc100TR

U2_heat_ex.th100TR

0

100

200

300

400

500

600

700

800

0 0.5 1 1.5 2

時間　（us）

電圧　（V）

0

100

200

300

400

500

600

700

800

電流　（A）

VCE-実測

VCE-SIM

IC-実測IC-SIM

Co-simulation with 3D Physics

Multi-Fidelity, Multi-Domain

Modeling

-22.50

60.00

0

25.00

50.00

0 240.00m100.00m

2DGraphSel1 NIGBT71.IC

0

474.00m

200.00m

400.00m

100.00 1.00Meg1.00k 3.00k 10.00k 100.00k

2DGraphCon1

GS_I...

Simulation-Based Test

Embedded Control / SW

Reduced Order Models


ROM creation example: Generic underhood test case

• 3D thermo-aerodynamic

• 2 parameters :

- Inlet velocity from 5 m/s to 20 m/s

- Exhaust Temperature from 700K to 900K Inlet velocity

Exhaust pipe


Reduced Order Model generation process (learning process)

• 50 calculations were done– 10 were used to create the model

– 40 others used for verification

• ROM Model creation is 9 seconds

• Average error on velocity : 0,06 m/s

• Average error on temperature : 0,49 K– Quantitative differences in between

Fluent calculation and ROM model are presented on the worst verification point

0

5

10

15

20

25

650 700 750 800 850 900 950

inle

t1-v

elo

city

exhaust temperature

5

6

1 1

1 12

1

3

4

test


Static Pressure Comparison

ROM

Absolute DifferenceTexhaust = 723KVcar = 14,3 m/s

Fluent

2 hours on 16 cores cluster 3 seconds on laptop


SafetyAnalysis

SystemDesign

SafetyAnalysis

SystemRequirements

Translation and Merge

System Design and Safety Analysis

FTAFMEAetc.


System Simulation and Safety Analysis (FMEA)Injecting and analyzing faults

0

0 0

CONST

n_ref

3600/60*pi*2

RMxprt

A

B

C

N

ROT1

ROT2

+

F

SM_ROT1

w

+

VM_ROT1

MASS_ROT1

0.005kgm2

TF_ROT1

(10+0.25*sin(2*pi*120*Time))*(-VM_ROT1.OMEGA/n_ref.VAL)

rg

T

FM_ROT1

Inverter

z1

z4

z2

z5

z3

z6

Va

Vb

Vc

Inverter_SystemLevel

A

AM1

A

AM2

A

AM3CONST

SystemTime

Time

CONST

TP

5e-005

CONST

ustmax

0.01

CONST

idref

0

motorAngle

Ia

Ib

motorSpeed

SystemTime

id_ref

n_ref

ustmax

TP

z1

z2

z3

z4

z5

z6

Normal Behavior Sensor Failure


Simulation has been used so far to Engineer better products and improve

Bottom line.


Predictive Analytics and the Digital Twin


Industrial IoT (IIoT) is a Logical Evolution of Product


System Simulation coupled with IoT can now be used to Better Operate

Products and improve Top Line.


Physical Simulationof all the Mechanical & Fluids Components of the BOP System

Engineering Simulation: Example Oil Field Blow Out Preventer


Complete System Simulation,including Physics, Electronics

and Embedded Software Controls=

Digital Twin of the BOP

Engineering Simulation: scaling to System LevelVirtual Sensors


As Is To Be

Test

Operate

Design

Transformational journey to digital twin


Predictive MaintenanceZero Unplanned Downtime

Asset Performance Management

Digital twin opportunity

360o Digital Insights

Predictive Analytics Unaddressed Failure Modes

30% of failure modes can be reliably

predicted by BDA, today

ANSYS powered Digital Twin

As Designed

As Operated

Big Data Analytics


Dat

a sc

ien

ce c

om

ple

xity

Data completeness, breadth, quality

Industrial analytics value map evolution

Basic Reporting


GE Predix* and ANSYS Industrial IoT Global Architecture

Tools

Platforms

Apps

Critical controls & HMI

Sensors and Actuators

DisplayApp

3D-ViewApp

AssetApp

GE Predix App Store

Digital Twin App

ANSYS ACT App Store

ANSYS IIoT Platform

EKMModels

JobsDataW

eb A

PI

Simplorer

ROM

3D Solvers

Monitoring&

Optimization

SCADE Suite & Display

Asset

* Similar architecture has been set up with PTC ThingWorx

DDS,…


• Life • Performance• Diagnostics• Optimize

As OperatedAs Designed

Digital Signatures

Design Decisions

Virtual Sensors

Operational Decisions

• Cost• Weight• Efficiency• Robustness

Digital Twin Summary

MODEL-BASED ENTERPRISE & SYSTEMS ENGINEERING INTEGRATED IOT ASSETS & ECOSYSTEMS


Q&A

Predictive Analytics and the Digital Twin - · PDF file07.07.2016 · 1 © 2015...

Documents

Transcript of Predictive Analytics and the Digital Twin - · PDF file07.07.2016 · 1 © 2015...