strukton rail big data expo 2016

21
David Vermeij 21 september 2016 1.0 POSS® Switch Analytics new ways to prevent failures

Upload: bigdataexpo

Post on 16-Apr-2017

256 views

Category:

Data & Analytics


2 download

TRANSCRIPT

Page 1: Strukton rail   big data expo 2016

David Vermeij21 september 20161.0

POSS® Switch Analyticsnew ways to prevent failures

Page 2: Strukton rail   big data expo 2016

Presentation

• Introduction

• Welcome to Strukton’s world

• POSS® monitoring system

• Switches (point machines)

• Data science forpredictive maintenance

Page 3: Strukton rail   big data expo 2016

Introduction

D.J. (David) Vermeij MSc.Manager R&D - Strukton Rail

Education1992-1998 TU Delft : fac. of civil engineering, dep. of railway engineering

Experience1998-2006 Movares : project manager HSL2006-2012 Strukton Rail : tendermanager maintenance (procurement)2012-2015 Strukton Rail : contract manager maintenance (operation)2015-present Strukton Rail : manager R&D

Responsible for the innovation program at Strukton RailPartner in European Research Projects

Page 4: Strukton rail   big data expo 2016

Strukton

1921 Established under the name: “ NV Spoorwegbouwbedrijf ”, as a subsidiary company of Dutch Railways (NS)

1974 Renamed to “ Strukton ” after the merger with the Danish company Christians & Nielsen

2010 Acquisition by Oranjewoud NV, a private entity of Sanderink Invest

Page 5: Strukton rail   big data expo 2016

Strukton’s World

Railsystems

813

CivilInfra-structure618

Technique & Buildings

345

Revenu per market 2014(millions of Euros)

Strukton Rail develops, builds, installs and maintains rail systems, ensuring optimum availability, reliability, safety and measurability .

• Maintenance, renovation and new construction of railways and rail systems

• Machines• Railway safety• Railway acquisition and data management• Train systems

Page 6: Strukton rail   big data expo 2016

Railway maintenance

Facts and Figures24/7 maintenance (NL)

Operational Asset Management

• Define & Select: Your network and your network needs

• Measure & Monitor: How to get the data you need

• Data Management & Interface: How to manage data and make it user-friendly

• Analyse & Interpret: How to make sense of all that d ata

• Organise & Plan: How to put that data to good use

• Maintain & Feedback: How to keep improving your network.

Page 7: Strukton rail   big data expo 2016

The challenge

Introduction of Performance based contracts (PGO) :

• Competition : need for continuous improvement

• Focus on availability : reduction of failures

• Focus on costs : preventive maintenance

Page 8: Strukton rail   big data expo 2016

POSS® monitoring

Page 9: Strukton rail   big data expo 2016

Worldwide

Strukton’s Preventive Maintenance and Fault Diagnosis System

Over 10.000 assets are monitored by POSS®

Page 10: Strukton rail   big data expo 2016

point monitoring

Frequently obstructed movements due to:

• Poor adjustment of rolling construction

• Lack of grease on slide chairs

• Bent blades

• Electrical problems

(worn-out brushes, motor, etc.)

Page 11: Strukton rail   big data expo 2016

Fault diagnosis

Locking problem Motor problem

Functional point model

Page 12: Strukton rail   big data expo 2016

Data analysis

Page 13: Strukton rail   big data expo 2016

The next step

Railway maintenanceTo achieve further reduction in failuresand optimize maintenance (costs),we need to make the next step towards predictive maintenance

To enable this we need a set of tools to detect and predict failures in a veryearly stage in order to be able to plan/act (as possessions are restricting).

The predictions are to be accurate, reliable and provide information about whatfailure mode is going to occur.

Data Science ?

Page 14: Strukton rail   big data expo 2016

2015: Introducing Data Science

2015 : Pilot on POSS dataGoal: Discover what Data Science can bring to railway maintenance

Page 15: Strukton rail   big data expo 2016

2015/2016: POC

2015/2016 : Proof of ConceptGoal: development of a failure prediction model

Testcase:algoritm based on T-7dys behaviour

Page 16: Strukton rail   big data expo 2016

2016: Production

2016 : Development towards ProductionGoal (1): System architecture and code

Goal (2): Development of a ‘pipeline’

Page 17: Strukton rail   big data expo 2016

2016: Failure prediction

2016 : Failure predictionGoal (3): Improvement of algoritm

Page 18: Strukton rail   big data expo 2016

Lessons learned

• There´s lots of data

• Valuable information can be extracted

• (mechanical) failures can be predicted

• Enormous potential,we´ve just started

Think big, start small

Connecting data science todomain knowledge is essential

There’s always a correlation…. The challenge is to predict whatfailure mode will occur

Reliability and availability of (labelled) data is crucial

Page 19: Strukton rail   big data expo 2016
Page 20: Strukton rail   big data expo 2016

Movie

46299 - Strukton - Point Monitoring_15092016_1080P.mp 4

Page 21: Strukton rail   big data expo 2016

Contact

www.struktonrail.com

Inquiries:Yves Kusters

Mobile : +31 – 6 [email protected]

Westkanaaldijk 23542 DA UtrechtThe Netherlands

www.struktonrail.com