Applying Analytics in the industry_Steel.pptx

Advance analytics solutions for steel industry

1. Applying Analytics in the Industry

2. Applying Analytics in the Industry

3. Applying Analytics in the Industry

4. Applying Analytics in the Industry

5. Applying Analytics in the Industry

6. *Peter Sondergaard , SVP Gartner, 2011 DATA IS THE OIL OF THE 21ST CENTURY, AND ANALYTICS IS THE COMBUSTION ENGINE* IT’S ALL ABOUT INTELLIGENT, AUTOMATED & DATA-DRIVEN DECISIONS!

7. In industrial companies manufacturing brings 3.5-5% of the total digital value of 6-10% Source: McKinsey

8. KPI’s influenced by analytics within the industry Uptime (Reliability) Performance (Yield) Loading (Planning) Theoretical maximum output when operating 24 hours per day, 365 days per year First pass yield (Quality) Actual output Maximum output when operating during scheduled operating time Output when operating at full speed when available to operate Output at actual operating speed % of total calendar time scheduled for operation % of scheduled operating time available to operate Speed at which the line runs as a % of its designed speed % of Good Units produced as a percentage of Total Units Started TEEP 1 OEE Planned downtime Breakdowns Minor stops Speed loss Production rejects Rejects on startup Scheduled non-operational days/hours Description Description Main drivers 1 Total Effective Equipment Performance

9. Data Insight Action Value The analytics Equation

10. Data Insight Action Value The analytics Equation

11. The Analytics Equation – The data Data Sources Quality data Reports Images Video Process data Process ontology Maintenance data (incl. non structured txt-files) Manufacturing data Value Data Action Insight

12. Data Action Value The analytics Equation Insight

13. Citizen Data Scientist Statistician Data Scientist Machine learning models The Analytics Equation – The discovery The Subject matter expert AND the Machine learning expert Process Engineer Chemical Engineer Quality Manager Automation Engineer Industrial Engineer First principles models Subject matter expertise Machine learning expertise Insight Data Action

14. The Analytics Equation – The Analytics REACTIVE PROACTIVE Optimization What is the best that can happen? Prescriptive Analytics Raw data Clean data Standard reports Ad hoc reports What happened? Descriptive Analytics Reporting /Access Advanced Analytics Visual Analysis Statistical modelling Why did this happen? Diagnostic Analytics What will happen next? Forecasting Predictive modelling Predictive Analytics Data mining, AI & ML techniques Genetic algorithms & reinforcement learning techniques Future Past Insight Data Action Forecasting and time-series models

15. Data Value The analytics Equation Insight Action

16. The Analytics Equation – The deployment Edge Analytics At device/sensor level Smart sensors - Monitor equipment on the platform, and take action. M2M communication to optimize operational process Between sensor, machine or human interface In-Motion Analytics Intelligently integrate quality / maintenance data with real-time streaming data Strategic Data Integration At-Rest Analytics Data Insight Action

17. Integration into day-to-day way of working

18. Data Insight Action Value The analytics Equation

19. Vision without execution is hallucination. Thomas A. Edison

20. From Vision to Value - Common challenges … Analytics is not about the most complex algorithm! Data scientist vs. engineer Data silo’s and data quality Unclear where analytics can make a difference Focus should be business relevance not model performance Stuck in POC, not able to rollout. How to scale? Don’t start with the end in mind

21. From vision to value – The frame work Quantify Assess Achieve Sustain Step 1 Assess the potential Step 2 Diagnostic Step 3 Industrialize Step 4 Scale

22. Assess the potential Step 1 Assess the potential Quantify business potential 1 Analytics assessment 3 Data assessment 2 Feasibility assessment 4

23. Model Explore Prepare Evaluate Act Implement Ask Find Fast or Fail Fast… Advanced analytics process optimization model Ad hoc data prep Decision on industrialization Give insight on business impact MES ERP LIMS Sensor 4 3 1 2 Step 2 Diagnostic

24. Model Explore Prepare Evaluate Act Implement Ask Put the model in action Step 3 Industrialize Advanced analytics process optimization 2 Integration into day- to-day way of working 4 Performance tracking 5 Model deployment 3 Full cycle runs every X minutes Plant sensors and data systems 1 MES ERP LIMS Sensor

25. Scale Increasing scope Increasing analytics maturity Asset/Process Plant Company Descriptive Diagnostic Predictive Prescriptive Scale your data-driven decisions Get the bad actor Huge potential for analytics 2 1 Full impact potential can be captured by deepening the approach and scaling it across production units Step 4 Scale 1 2

26. Takeaways Data needs analytics as lever to create value Analytics is more than a fancy algorithm  Value is generated through Data - Discovery - Action Start small, find value fast and scale even faster!

27. Is analytics the holy grail? Does It solves all our problems… George Box, (18 October 1919 – 28 March 2013) British statistician Essentially,  all models are wrong,   but some are useful.

28. References