> > >Keeping technology alive.

Author:

Date:

Reference:

Algorithms and Processes for Maintenance Optimization

V1

F. Langmayr / E. Lichtenegger

10 2019

> > >10.10.2019 2

▪ Introduction and Motivation

▪ Information Merging

▪ Recommendation System

▪ Process Development

Content

© Uptime Engineering

> > >10/10/2019Uptime Engineering GmbH 3

The Experts for the Reliability Process

We provide reliability solutions

Consultancy and Software for

Analytics, Diagnostics and Prognostics

We support the entire lifecycle of mechatronic systems: verification & validation & operation.Our methods are based on a broad expertise in failure physics and applied statistics.

We have implemented more than 100 projects with leading OEMs and fleet operators.We optimize product development.

We support cost reduction via optimization of service and maintenance.

by Franz Langmayr et al. Employees European Locations

Founded in 2010 A team of 16 Based at 2

> > >

Uptime‘s Vision & Mission

Uptime‘s Visionis keeping technology alive

Uptime‘s Missionis to extract maximum value from datato support the reliability of mechatronic systemsthroughout their entire lifecycles

10/10/2019Uptime Engineering GmbH 4

> > >

CoreCompetences

B2B Software Reliability Engineering

Physics of Failure Applied Statistics

10/10/2019Uptime Engineering GmbH 5

> > >10.10.2019 6

Business Case: Unmanned operation in harsh environment

© Uptime EngineeringSource: Wikipedia

> > >10.10.2019 7

Sustainable Cost Reduction Performance Improvement

Objectives

© Uptime Engineering

Risk Mitigation Failure Minimization

> > >10.10.2019 8

The Potential of MODEL-BASED ANALYTICS:

© Uptime Engineering

Max. Availability

Operation Planning

Evaluate Load Factors

Minimize MTTR

Efficient Usage

Equipment Distribution

Benchmark Load Spectra

Minimize Idling

Max. Lifetime

Protective Operation

Alarm and Warn

Operate Preventive

Equipment

Health State

Agile Operation

Indicate Outliers

Trigger Service

Cost Transparency

Resource Allocation

Calculate Cost/Benefit

Buy /Lease

Risk Management

Identification

Mitigation

Verification

CustomerExpectation

Use Cases

Tasks

> > >10.10.2019 9

▪ Introduction and Motivation

▪ Information Merging – ROMEO Project

▪ Recommendation System

▪ Process Development

Content

© Uptime Engineering

ROMEO Project- The Objectives

Increase wind farm reliability and decrease the number of failures leading to downtime.

Increase the life time of key turbine components.

Reduce the WT O&M costs through the reduction of the resources required for annual inspections of the turbine.

Reduce the O&M costs associated to foundation through reduction in jacket substructures inspections.

Greater reliability, less repairs, more safety

This project has received funding from the European Union’s

Horizon 2020 research and innovation programme

under grant agreement No 745625.

10.10.2019Uptime Engineering GmbH 11

Uptime Engineering in ROMEO

Roles and responsibilities

▪ Integration of multiple data sources

▪ Analysis and combination of information

▪ Centralized O&M Platform for access by multiple stakeholders

▪ Support of maintenance process

▪ Reporting and communication on multiple channels

Making information usable

10.10.2019Uptime Engineering GmbH 12

Uptime Engineering in ROMEO

Roles and responsibilities

Degradation Analytics Engines

Historical and real time data• SCADA Data• Inspection Data• Environmental Data• Operational Data

Uptime Data Examination• Event Based Reasoning• KPI Calculation• HARVEST physical models

O&M Information• Analytics Modules Results• Monitoring• Visualization and analysis

Uptime Advisory Generation• Failure Diagnosis• Maintenance Process• KPI (Availability, MTTF,

Statisitics)

Data acquisition and cloud-based analytics

Data Merging End user application

> > >10.10.2019 13

▪ Introduction and Motivation

▪ Information Merging – ROMEO Project

▪ Recommendation System

▪ Process Development

Content

© Uptime Engineering

> > >10.10.2019 14

Machine Learning / Neural Network Knowledge/Physics-based Models

The Solutions

© Uptime Engineering

For both approaches: top-quality, risk-related (“wise”) data are essential for reliable results

INPUT

▪ SCADA▪ Events

Trained network

Failure

Prognosis

Physics-based

Models

INPUT

▪ SCADA▪ Events▪ Rules▪ System Info

OUTPUT

▪ Detection

OUTPUT

▪ Detection▪ Diagnosis▪ Prognosis

> > >10.10.2019 15

State DETECTIONIs there anything remarkable?

System Supervision

and

Pattern Recognition

and

System Response Models

→ Failure Indicators

→ Alarm & Warning

DIAGNOSISHow did it come?

Domain Failure Knowledge

and

Reasoning Engine

and

Reliable Observation

→ Failure Modes

→ Problem Solving

PROGNOSISWhat next?

Physics of Failure Models

and

Lifetime Models

and

Load History

→ Risk Prediction

→ Recommendation

The UPTIME Engineering Route to Predictive Maintenance

© Uptime Engineering

> > >10.10.2019 16

State DETECTION | System Response Modelling

© Uptime Engineering

Residual-Trend-Analysis delivers outlier detection with high sensitivity AND extremely low frequency of false alarms

Time [10 min]

Bea

rin

g Te

mp

erat

ure

[°C

]= Measured

= Model

Res

idu

al [

°C]

Time [10 min]

Subsequent Outliers, Running average, Slope, …

Fixed Alarm Level: not sufficient

> > >10.10.2019Uptime Engineering GmbH 17

DIAGNOSIS | Model based Reasoning Engine

Turbine Operation

Model based Reasoning

Formalized expert

Knowledge

FeedbackUE Knowledge base

Root Cause Observations : Diagnosis : Maintenance Task

UE Reasoning EnginePropositional Logic

Reduced Power AND NOTWear of blade AND NOTCurtailment ANDGenerator de-rating

Formalized ObservationsDiagnosis:Generator

Overheating: Loss of insulation

State: Generator overheated Maintenance

> > >

Cavitation

Wear

HCF

TMF

KNOWLEDGE BASEDMONITORING

EventsCorrelations

Patterns

-----------------------------

MODEL BASEDANALYTICS

DamageFailure Probability

Alarm

10.10.2019 18

PROGNOSIS | Damage Accumulation terminates Lifetime

© Uptime Engineering

> > >10.10.2019 19

▪ Introduction and Motivation

▪ Information Merging – ROMEO Project

▪ Recommendation System

▪ Process Development

Content

© Uptime Engineering

> > >10.10.2019 20

Service Process Re-Engineering

▪ Identify objectives and BUSINESS CASE(S)

▪ Map organisation in PROCESS

▪ INVOLVE all contributing parties

▪ Provide BENEFIT for each involved party

SERVICE PROCESS | Design

© Uptime Engineering

> > >

SERVICE PROCESS | Digitization for CBM

Service Staff

Spare Parts

Equipment

Issues & DiagnosisOptimised Service

FeedbackReasoning

EngineKnowledge

Base

Time Series DB

10.10.2019© Uptime Engineering

> > >10.10.2019 22

Quick Wins

▪ Elimination of O&M weak-points

▪ Value from available data

Corporate Development

BENEFIT

© Uptime Engineering

▪ Technical & process lessons learned

▪ Feedback drives quality and involves the staff

Hands-on Approach

▪ User-acceptance via integration of service staff

▪ Focus on risk-drivers

Analytics & Process

▪ Results trigger the CBM-process

▪ Functional extension acc. to process evolution

> > >

www.uptime-engineering.com