© Presidion 2016
Formerly SPSS Ireland
Delivering significant efficiencies through
Predictive Maintenance in the Oil & Gas
Industry
Aberdeen, 10 March 2016
© Presidion 2016
1. Significant Financial Benefits linked to Predictive Maintenance
2. What Predicitve Maintenance is about
3. How Predictive Maintenance works
4. How to make a success of Predictive Maintenance
What I will cover
2
© Presidion 2016
Predictive Maintenance – Transforming of Maintenance
Business Model
3 Source: Roland Berger
2
75%
Reduction in breakdown
4
More data =
More
accuracy =
More value
1
$18 /hp p.a.
$9 /hp p.a
$13 /hp p.a.
3
15%Time spent on
Predictive
Maintenance only
75%
15%
Reactiveand
Preventive
© Presidion 2016
Predictive Analytics help connect data to effective action by
drawing reliable conclusions about current conditions and
future events
4
High
Low
Business
Value
Time
Past
Business
Intelligence
Sense and Response
Future
Predictive
Analytics
Predict and Act
Techniques and Data
• Optimisation, predictive modelling, forecasting, statistical analyis
• Structured/Unstructured Data, Internal/External Data, Massive
Data Sets
Driving Questions to be answered
• What will happen next? Why?
• Why is this happenin?
• What if?
• What's the optimal scenario for our business?
Techniques and Data
• Reporting, dashboarding, alerts, queries
• Structured Data, Manageable Data Sets
Driving Questions to be answered
• What happened last quarter / month /week?
• How many pumps did break down? How much did we
spend on maintenance? How much downtime on
these assets? How many preventive actions have we
completed?
• Where is the problem?
© Presidion 2016
How does Predictive Maintenance deliver?
Unearthing characteristics that lead to an
increased frequency of failures
Predicting impact or consequence scores to
enhance Alarms Management so that key alarm
events are prioritised
Identifying factors that increase ownership cost and
downtime over the life of a system
Identifying assets at risk of failure even
when they have no previous failure history
Mining free text from thousands of logs that describe the maintenance performed
on systems to accurately categorise maintenance reports and identify areas of risk
Finding patterns in maintenance operations
that could point to opportunities for
improvements
© Presidion 2016
What if I could tell you that a specific asset is 90% likely to fail
within one week for Reasons A, B and C?
6
Data Predictive Models -
Insights
Actions
(Work Order)
Anomaly Detection
Diagnostic Analysis
Recommendations and
decision support
What should be done next?
Priortiisation
What to attend to first
depending on fault severity?
Evaluate impact
Procurement & Supply Chain
Sensors
GIS
Data Historian
Asset Management
Maintenance Management
Automation
Change Management
Real Time
Other Sources
© Presidion 2016
How to deliver a Predictive Analytics Project succesfully
7
Determine Business
Objectives and Data Mining
Goals
1
Collect, describe, explore
and verify quality of Data
2
Select, clean, construct,
integrate and format data
3
Select, Generate, Build and
Evaluate Models
4
Evaluate how the results
help to achieve business
objectives
5
Integrate new knowledge
into your business
processes
6
Business
Understanding
Data
Understanding
Data
Preparation
Modelling
Evaluation
Deployment
© Presidion 2016
Asset Data
Availability
Criticality
of Failure
Trust in
Predictive
Technology
When Predictive Maintenance works well
8
Actionable Insights – Return on Investment
© Presidion 2016
Challenges
9
Heterogeneous Assets
Changing Operating Conditions
Interoperability
© Presidion 2016
End-to-End approach is required
10
2.1 Tactical Audit for
Advanced Analytics
2.2 Maturity
Assessment for
Advanced Analytics
2.3 Roadmapping and
Business Case
Development
3.1 Advanced
Analytics
Capability
Building
4.1 Performance
Monitoring and
Business Benefits
Realisation
2. Prepare yourself for success in Advanced Analytics
3. Improve
business
performance
4. Sustain
improved
performance
1.1 Advanced
Analytics
Transformation
Workshop
1. Learn how
Advanced
Analytics can
transform your
organisation
© Presidion 2016
What if you could deliver these numbers…
11 Source: us department of energy's o&m best practise guide,
Imagine if you add IoT data
4
10 xReturn on Investment
1
25%Increase in production
output
30%
2
Reduction in maintenance
costs
45%
3
Reduction in downtime
© Presidion 2016
www.presidion.com
Pierre.baviera@presidion.com
Q & A
Thank You

Predictive Maintenance for Oil and Gas

  • 1.
    © Presidion 2016 FormerlySPSS Ireland Delivering significant efficiencies through Predictive Maintenance in the Oil & Gas Industry Aberdeen, 10 March 2016
  • 2.
    © Presidion 2016 1.Significant Financial Benefits linked to Predictive Maintenance 2. What Predicitve Maintenance is about 3. How Predictive Maintenance works 4. How to make a success of Predictive Maintenance What I will cover 2
  • 3.
    © Presidion 2016 PredictiveMaintenance – Transforming of Maintenance Business Model 3 Source: Roland Berger 2 75% Reduction in breakdown 4 More data = More accuracy = More value 1 $18 /hp p.a. $9 /hp p.a $13 /hp p.a. 3 15%Time spent on Predictive Maintenance only 75% 15% Reactiveand Preventive
  • 4.
    © Presidion 2016 PredictiveAnalytics help connect data to effective action by drawing reliable conclusions about current conditions and future events 4 High Low Business Value Time Past Business Intelligence Sense and Response Future Predictive Analytics Predict and Act Techniques and Data • Optimisation, predictive modelling, forecasting, statistical analyis • Structured/Unstructured Data, Internal/External Data, Massive Data Sets Driving Questions to be answered • What will happen next? Why? • Why is this happenin? • What if? • What's the optimal scenario for our business? Techniques and Data • Reporting, dashboarding, alerts, queries • Structured Data, Manageable Data Sets Driving Questions to be answered • What happened last quarter / month /week? • How many pumps did break down? How much did we spend on maintenance? How much downtime on these assets? How many preventive actions have we completed? • Where is the problem?
  • 5.
    © Presidion 2016 Howdoes Predictive Maintenance deliver? Unearthing characteristics that lead to an increased frequency of failures Predicting impact or consequence scores to enhance Alarms Management so that key alarm events are prioritised Identifying factors that increase ownership cost and downtime over the life of a system Identifying assets at risk of failure even when they have no previous failure history Mining free text from thousands of logs that describe the maintenance performed on systems to accurately categorise maintenance reports and identify areas of risk Finding patterns in maintenance operations that could point to opportunities for improvements
  • 6.
    © Presidion 2016 Whatif I could tell you that a specific asset is 90% likely to fail within one week for Reasons A, B and C? 6 Data Predictive Models - Insights Actions (Work Order) Anomaly Detection Diagnostic Analysis Recommendations and decision support What should be done next? Priortiisation What to attend to first depending on fault severity? Evaluate impact Procurement & Supply Chain Sensors GIS Data Historian Asset Management Maintenance Management Automation Change Management Real Time Other Sources
  • 7.
    © Presidion 2016 Howto deliver a Predictive Analytics Project succesfully 7 Determine Business Objectives and Data Mining Goals 1 Collect, describe, explore and verify quality of Data 2 Select, clean, construct, integrate and format data 3 Select, Generate, Build and Evaluate Models 4 Evaluate how the results help to achieve business objectives 5 Integrate new knowledge into your business processes 6 Business Understanding Data Understanding Data Preparation Modelling Evaluation Deployment
  • 8.
    © Presidion 2016 AssetData Availability Criticality of Failure Trust in Predictive Technology When Predictive Maintenance works well 8 Actionable Insights – Return on Investment
  • 9.
    © Presidion 2016 Challenges 9 HeterogeneousAssets Changing Operating Conditions Interoperability
  • 10.
    © Presidion 2016 End-to-Endapproach is required 10 2.1 Tactical Audit for Advanced Analytics 2.2 Maturity Assessment for Advanced Analytics 2.3 Roadmapping and Business Case Development 3.1 Advanced Analytics Capability Building 4.1 Performance Monitoring and Business Benefits Realisation 2. Prepare yourself for success in Advanced Analytics 3. Improve business performance 4. Sustain improved performance 1.1 Advanced Analytics Transformation Workshop 1. Learn how Advanced Analytics can transform your organisation
  • 11.
    © Presidion 2016 Whatif you could deliver these numbers… 11 Source: us department of energy's o&m best practise guide, Imagine if you add IoT data 4 10 xReturn on Investment 1 25%Increase in production output 30% 2 Reduction in maintenance costs 45% 3 Reduction in downtime
  • 12.

Editor's Notes

  • #2 My name is Pierre Baviera, CEO of Presidion, Specialised Company in delivering Advanced Analytics Solutions – leveraging data to deliver Business Benefits in Return in Investment Big Data, IoT will create more and more data – data without insights and actions are worthless and at Presidion we are passionate about bridging this gap. Alex Kemp I am going to talk to you about what predictive maintenance delivers for those that have already adapted this approach. Savings are sustensial, the technique is proven and well beyond hype. We are about to witness a transformation of the maintenance business model.
  • #4 Pumps - Hour of Production per Annum 50% cheapper thant Reactive Maintenance 30% cheapper than Preventative Maintenance 75% reduction breakdown in Energy Companies Would like to double the time spent / effort spent on Predictive Maintenance
  • #5 Business Intelligence – as a paralell Business Intelligence very much suits the reactive and preventive maintenance world Story – 14 pumps broke down last quarter. Out of these 2 2 brand new broke down and this was a real surprise Downtime was 30 hours. This is how it impacted our production by a reduction of 10% of throughput equating $400,000. These were the reasons why it broke down. As a result this is how we are going to review of preventive maintenance policy? Predictive analytics – Predictive maintenance? These 4 pumps have 90% probabilty to breakdown in the next 48 hours – these are the most probable reasons. 2 of these pumps will have limited impact on production. 2 others are critical and will put the production down as awhole. A workorder is created to address these… Image is like moving from rules based – conditioning Information to prescriptive maintenance
  • #6 A Major US-based Oil Company saved tens of million of dollars by preventing oil well collapse Oil and Gas produced in Australia getting 87% accuracy and 48-hour warning about potential equipement failure – saving millions of dollars by minimise downtime A global Oil & gas Company unearthing USD 11 Million by optimisting the production of 12 wells; 97% accuracy in detecting underperformaing wells and allowing to make adjustments US Oil and Gas Producer saved USD53Millions by identifying key reasons linked to stuck pipe situations and getting 85% accuracy in predicting these situations The list of case study goes on and on and on… How do you go about answering these questions
  • #7 So … That is what Predictive Maintenance is about… Internal/External, stuctured/unstructured data – we all have more data than we think Predictive modelling – pick out what’s unusual, what’s important, identifying the lead indicators to a particular event Act – your work-order Think big, start small with actiona plan to deliver big
  • #8 Predictive Maintenance is not magic – there is a proven best practice way of of delivering predictive insights this is CRISP-DM – Presidion participated to its design To make it work you need the right involvement of Line of Business People, Data People and IT people Insights that don’t deliver against a business problem are worthless Insights that cannot be deployed are worthless Always starts with the Business Question / Business Goals and loop back to the Business and deployment – this where you people from the Business will be involved
  • #9 Preditive Analytics and Predictive Maintenance are proven technologies, these are beyond hype – it is about thinking big and starting small and building momentum Asset Data – volume/quality/availability You have much more data than you assume It is also about low frequency hihg impact Criticality of failure is key
  • #10 Our experience has shown that our challenges are about…There are not only about Oil&Gas..We help our customers overcome these challenges So you need to think about Identifying key business questions or a problem that needs solving through data Identiying key assets Identiying the data that we use Identifying process that might be changed Identifying key stakeholders Identiyfing Performance Indicators Also different organisations are at different stages It does not have to be overwhelming and this can
  • #11 Learn and educate at the beginning In your situation what process and data will deliver the biggest bang for your bucks – business case and roadmap (think Big) Think Big – start small Sustain