A Real-time & Predictive Process Safety Dashboard

0 A Real-time & Predictive Process Safety Dashboard Dr.ir. Bert Knegtering, Honeywell SMS Process Safety congress – Dordrecht, 30 May 2018 2015 Honeywell International

1 Vision / Business case Offering a ‘safety dashboard’ for the process industry Real-time information wrt the actual, previous and upcoming (predictive) safety ‘level’ Combine data / information sources beyond the ‘DCS/SIS’ Deterministic as well as probabilistic How to make use of IoT and BIG data? HOW SAFE IT IS AT A CERTAIN PLACE AT A CERTAIN MOMENT IN TIME? 2015 Honeywell International

BP Texas, March 2005 2015 Honeywell International

3 Buncefield oil depot, December 2005 2015 Honeywell International

Buncefield, December 2005 2015 Honeywell International

5 Dynamic safety level Actual safety level Where are we? How close to failure are we? Based on which factors do we know the actual level? ? Acceptable safety level (required by law) Time 2015 Honeywell International

Common characteristics of major accidents 1. None of the accidents were the result of unknown ‘new’ physical or chemical hazardous situations. In all cases, chemical substances were involved of which its hazardous characteristics are known for decades. 2. None of the accidents happened due to a single problem or failure, but in multiple perspectives, flaws, lacks and deficiencies are observed, which together have formed the fundament for the accident. 3. The fundament for having these accidents is mainly characterized by technical, organizational and human factors. The remaining issues related to the complexity of the process installations are observed as well but not dominating. 2015 Honeywell International

7 BP-Texas investigation Major conclusion: Better measuring! 2015 Honeywell International

8 A relative unsafe situation Maintenance behind schedule Over-rides on various actuators Various leakages Welding activities Production backlog Corrosion problems Postponed plant shut-down 2015 Honeywell International

9 Dynamics Seasons Aging, wear-out Production capacity Office hours / weekend Wind directions Construction work Maintenance work Inspection rounds Vegetation, bush and trees (Gexcon) What is the actual safety level at a specific moment in time at a particular place? 2015 Honeywell International

10 Safety indicators Safety-KPI’s, risk parameters, Safety factors, etc. 1989: Center for Chemical Process Safety (CCPS), OSHA 1910.119 Process safety management of highly hazardous chemicals Tripod Condition Survey: Basic Risk Factors 2008: CCPS - Lagging and leading indicators 2010: CCPS - Guideline for Process Safety Metrics But also: British Health & Safety Executive ASM consortium (safety metrics review) American Petroleum Institute (practice 754) Step Change in Safety (UK oil and gas industry) Hopkins, thinking about process safety indicators 2015 Honeywell International

11 CCPS Safety Metrics, 2010 - 1 21 categories, 393 factors Process safety culture Compliance with standards Workforce involvement Process safety competency Stakeholder outreach Process knowledge management Hazard Identification & Risk Analysis Operating Procedures Safe Work Practices Asset Integrity & Reliability Contractor Management Training & Performance Assurance Management of Change Operational Readiness Conduct of Operations Emergency Management Incident investigation Measurement & Metrics Auditing Management Review & Continuous Improvement Incident Reporting Centre for Chemical Process Safety – Guideline for Process Safety Metrics Wiley, 2010 ISBN 978-0-470-57212-2 2015 Honeywell International

12 CCPS Safety Metrics, 2010 - 2 Extensive system with 21 categories and 393 indicators Too much, not practicable and not realistic What then to do? Research still required Risk-Safety Modeling / Analyzing Methods Is it going into the right direction? (Why serious accidents in the process industries continue to happen ) 2015 Honeywell International

13 Risk-accident models Visual - Conceptual 2015 Honeywell International

14 Safety KPI’s Leading & Lagging metrics (CCPS process safety metrics 2010) 2015 Honeywell International

Technology - The SIS; your last line of defense DCS SIS Near miss 2015 Honeywell International

16 The Bow-Tie redefined Long term dynamic Mid term dynamic Short term dynamic People Organization Technology Safety Culture Limited (slow) Long term improvement Limited mid-term improvement 2015 Honeywell International Near miss Structural long term improvement

Many bowties – Shell HEMP experiences ‘Infinite’ variety of hazardous event scenario’s 2015 Honeywell International

Bowtie-model with time dimensions weeks days hours minutes seconds pressure temperature vibration noise Contamination 2015 Honeywell International

‘Remote’ large datasets – precursors? ‘Large data sets’ ‘Small data sets’ Big Data Data warehouse SAP ERM ERP SCADA (DCS) SIS Near miss Prediction ‘MDM’ (Master Data Management) 2015 Honeywell International Analysis

Indirect precursors vs. direct indicators ‘Large data sets’ Remote. ‘Small data sets’ Big Data Belling -cat Group James Bond 007 Prediction ‘MDM’ (Master Data Management) 2015 Honeywell International Analysis

21 Tracking traffic with GPS-enabled cell phones Crowd sourcing - Traffic information 2015 Honeywell International

22 What to measure? What do you want to know? Short term; Traditional HAZOP parameters; level, flow, pressure, temperature, More to date; dust, acoustic, vibration, weather, BIG DATA? Mid-term; Seasons, corrosion, aging, maintenance, testing, BIG DATA? Longer term; Aging, wear-out, knowledge, culture, BIG-DATA? 2015 Honeywell International

23 Overview - #1 Data Data sources SAP BPCS Maximo SIS Weather stations Hand-held devices Test reports Human observations Big Data Activities Welding Testing Maintenance Modifications Road blocks Location of People Vehicles / trailers 2015 Honeywell International Symptoms Equipment: Vibration Smoke Leakages HH-L/F/P Low-voltage Human observations: Humidity Vibration Dust Smoke Corrosion Noise Leakages Discoloration Physical failure / error / problem Functional failure / error / problem Bearing damage Wear-out Fire Run-away Valve stuck Power-down Effect: Device failure Available / Unavailable Safe / Dangerous Detected / Undetected

24 Overview - #2 Performance of the Safety Management System (SMS) KPI Characteristics Data & Information Identification of Safety-KPI’s Measurable Controllable? ‘System Safety-KPI’s oriented’ Inputs, Safety KPI’s Safety management system Modelling 2015 Honeywell International Data collection / Measurement: Automatically Manually Instantaneously Periodically Safety Barometer - Explosions - Injuries, FAR, LTI - Production loss - .

25 Dynamic risk parameter - time : mapping Matrix Process safety indicators / risk parameters inventory matrix years months weeks days hours minutes seconds pressure temp How many and which indicators / parameters? SIS DCS corrosion leakages SAP vibration ? Weather IoT / BIG data Unknown (black swans) 2015 Honeywell International Uncertainty (unsure) Probabilistic (approximately) Deterministic (precise)

26 Data transition – Safety Level model List of S-KPI’s (to be selected) Measurand e.g. LTI Pressure Data source e.g. DCS SIS Actual (real time) value HH-P Weight Safety Factor System (importance) Model (bow-tie) 12 p.d. ‘8’ ‘2’ LTI ‘.’ ‘.’ ‘9’ ‘.’ ‘.’ Alarm freq. LeakAges p.w. 2015 Honeywell International DCS SAP 12 p.d. Real-time (actual) Safety Level ‘6’ & / or ‘4’ etc. SL ‘5’ Orange Trend analysis ‘5’ Time line

27 Dashboard Real time Safety-KPI’s View S-KPI-1 45% HH-L S-KPI-4 42% HH-L S-KPI-7 15% HH-L Overall Real Time Safety Level 63% of Max. S-KPI-2 55% HH-L S-KPI-5 53 HH-L S-KPI-8 85% HH-L S-KPI-3 25% HH-L S-KPI-6 23% HH-L S-KPI-9 28% HH-L 2015 Honeywell International

28 Safety KPI’s – Overall Safety Level Unacceptable Safety Level Overall Safety Level Acceptable Safety Level Time 2015 Honeywell International

29 Dashboard ‘landscape’ real-time SL-view SL – Trend S-KPI-5-HH-L SL – Trend S-KPI-3-HH-L SL – Trend S-KPI-2-H-L S-KPI-8-H-L 2015 Honeywell International

30 Research challenges Innovation areas Parameter identification (Safety-KPI’s) Measurable / measurement Data sources Data selection / filtering / security Data (warehouse) storage Modeling / analyzing methods-tools Disorder measurement Uncertainty handling Real-time measurement Control/Improvement strategy 2015 Honeywell International

31 More info? Please contact Dr. Bert Knegtering - – TÜV FSExp, CFSE Honeywell Safety Management Systems Global Center of Excellence (CoE) for Safety Solutions Burgemeester Burgerslaan 40 5245 NH Rosmalen The Netherlands 31-73-6273273 office 31-6-20607213 mobile www.honeywellsms.com 2015 Honeywell International

Questions? WHERE WE ARE HEADED - 2011 & BEYOND 2015 Honeywell International

33 Safety-KPI’s critical stage definitions To be calibrated (site-, plant-, unit specific): S-KPI Normal H-level HH-Level Shut-down Collapse Pressure 50 55 PSI 60 PSI 65 PSI 70 PSI 80 PSI 10 sec. 20 sec. 35 sec. 50 sec. 25 per day 50 per day 100 per day ? 1 hour 2 hours 4 hours ? #7 problems per week #15 problems per week #25 problems per week ? 1 month 2 months 4 months ? 90% on time 75% on time 60% on time ? 4 weeks 6 weeks 8 weeks ? 85% on scheme 75% on scheme 60% on scheme ? Half year One year Two years ? Time Alarms Corrosion Maintenance Training 10 per day #3 problems per week On schedule On scheme Time flow - dimension 2015 Honeywell International

34 Ultimate goals and achievements summary ‘Prevention of hazardous events by Prediction of hazardous situations and circumstances, based on long term, mid term and short term focal points and strategies’ 2015 Honeywell International

35 Trends – Time measurement 1. Time-depending trends Short term Mid term Long term 2. Correlations between process parameters Strong / weak correlation 3. Degree of Disorder Data pollution / noise Uncontrolled situation / process 4. Unexpected deviations Uncertainty Black swans / white spots (unknown) 2015 Honeywell International

Safety indicators Safety-KPI's, risk parameters, Safety factors, etc. 1989: Center for Chemical Process Safety (CCPS), OSHA 1910.119 Process safety management of highly hazardous chemicals Tripod Condition Survey: Basic Risk Factors 2008: CCPS - Lagging and leading indicators 2010: CCPS - Guideline for Process Safety Metrics .