Critical Condition Management on a Corporate Scale Lothar Lang Consulting Engineer - CSEE Lyondell Chemical Company Lyondell Is a Major Global Chemical Company A global leader in the manufacture of chemicals and polymers, building blocks for countless products that people around the world use every day Lyondell s products are vital to the goods that people depend upon to enhance their quality of life Lyondell includes three wholly owned businesses Equistar, Millennium and the Houston refinery 2 1
Fast Facts One of the world s 10 largest chemical companies Operations on five continents Nearly 11,000 employees Combined pro forma 2006 revenues of $22.2 billion* Leading capacity positions in all major products A Fortune 500 company Ranked among the industry best in on-the-job safety Continuous improvement in environmental performance, product quality and reliability measures 3 * Including 100% of refinery Lyondell at a Glance Segment Business Focus 2006 Revenues Geography Employees Ethylene, Co-Products & Derivatives (EC&D) Ethylene Propylene Polyethylene Acetyls $10.8 billion North America Propylene Oxide & Related Products (PO&RP) Propylene Oxide Styrene Monomer MTBE $6.7 billion Global 10,000 Inorganic Chemicals TiO 2 $1.4 billion Global Refining Refining of heavy, highsulfur crude oil $5.8 billion* North America 880 4 *Refining data represents 100 percent of post- and 58.75% of pre-acquisition Houston Refining sales and operating income 2
Operations & Control Systems Best Practice Team Motivation Human factors & process control systems have a large impact on Response to or prevention of critical conditions that impact reliability, safety, environmental performance, and quality Operational Excellence Reduce risk of wrong operator input or action Reduction of unit upsets Unit operating costs Appropriate level of automation Work practices and training systems to enhance operator s abilities Address operator workload Provide better information to operators, specifically under stress conditions (unit upset ) Address operator turnover through retirement Besides equipment performance, our reliability efforts also need to concentrate on human factors and process control capabilities 5 But How Do We Get There? Emergency and abnormal situations represent a unique challenge in the operation of all complex systems. On one hand, operators have to maintain a very high degree of proficiency in the relevant procedures, while on the other hand, they rarely have an opportunity to practice them. Aircraft have become very reliable. It is not uncommon today to meet retiring airline captains who have never had a serious emergency in their careers 6 3
What We Found A comprehensive approach for reducing human factors contributing to abnormal situations, with many parallels to our operations Focus Areas Handheld Devices Create Industry Benchmarks Automation Strategy Operator Training Simulators Critical Condition Management Checklists Scenario Drills 7 Critical Condition Management HCI (operator interface) Alarm management Control loop performance management Operator advisory system Software standard company-wide KPI company-wide Sharing of Best Practices 8 4
The Modern-day Problem Control rooms badly designed Alarm Management causing operators to react to problems, then becoming overwhelmed during a disturbance HCI causing similar problems to those on the panel: information overload - they may get confused, make mistakes, or panic Operators, often less experienced than their predecessors, are being required to manage larger numbers of valves/loops, with more associated data, thus increasing the potential for information overload 9 Success in Critical Condition Management Empower frontline (operator and support function) Heighten situational awareness Focus on key activities and key actors (prioritize) Automate use of information Share the information Engage the teams in continuous improvement process Drive intelligently by best practice and KPI 10 5
How Not to Design a HCI 11 Influences on Operator Intervention Success Orienting Information overload Vague or misleading information Inappropriate level of detail Vigilance decrement 30% Evaluating Insufficient knowledge Lack of experience or practice Conflicting priorities Inaccurate labeling or information presentation Excessive mental task 20% Acting Inadequate communications Deficient procedures Failure to follow procedures Inappropriate actions Inability to act Inadequate feedback Poor layout or information presentation Operator graphics (HCI) Assessing Policy & practice discrepancies Inadequate feedback Poor integration of information systems 30% 20% 6
We Can Do it Better 13 New OVERVIEW Example 14 7
Lyondell s Approach to Operator Graphic Design Document best practice Focus attention on critical information Assess situation at a glance Use company-wide 15 What Sort of Benefits Have Been Seen by Improving the HCI? Thorough tests were conducted using real operators and (simulated) known process upsets (NOVA), but with different sets of displays; one being ASM-compliant, the other being good operational displays, but not ASM-compliant. When using ASM-compliant displays: Tasks were completed 35% - 48% faster. Failures were successfully dealt with 26% more often. Presence of failure was recognized before the first alarm 38% more often. 16 8
Alarm Management Typical Phases Benchmark & assessment Baseline, KPI Alarm philosophy Alarm definition and governing rules Alarm rationalization Systematic review and documentation of alarm settings Implementation & Execution Control system configuration Continuous improvement Performance monitoring & optimization Reduce alarm frequency systematically by elimination of bad actors Dynamic alarm handling to address alarm flooding Maintenance Sustainability and MOC Phase I tool: logmate Phase II tool: UReason 17 KPI for Alarm Management Per Operator Average # of alarms per hour EEMUA (Engineering Equipment and Materials Users Association) guideline company goal 6 Average # of standing alarm 9 (longer than 24 hours) Peak alarm rate per 10 minutes 10 18 9
Average # of Alarms per Operator at a Site October Average: 2.3 Data captured by logmate application 19 Peak Alarm Rate per Operator at a Site, for December 20 10
Benchmarking Data for 4 th Quarter 06 21 Path Forward - Alarm Management Establish KPI and baselines at all sites Drive with continuous improvement program Periodically review KPI, bad actors and action items Need a Champion (preferentially from operations) Leads the review process Drills down using logmate to identify worst (3-5) bad actors Involves appropriate people to address the identified issues Follows up to ensure quick implementation Start dynamic alarm handling 22 11
Benefits from Dynamic Alarm Handling (OASYS-AM from UReason) Two phased approach: Provide Real-Time Reduced Alarm Lists to Operators Uses alarm and configuration data stored/collected by logmate Recognizes alarm relationships, chattering alarms and nuisance alarms Provides operators with reduced alarm lists independent of the DCS Prevents Alarm Floods by Predicting Process Upsets Correlates alarm and process data to predict process upsets Provides operators with alarm advisories (checklists ) independent of the DCS 23 Alarm Data during a Plant Upset 24 12
Alarm Data during a Plant Upset (continued) 25 Playback of Alarm Data Using a Rule for Reducing Six Most Frequent Chattering Alarms significant alarm reduction after initial alarm burst Data generated using UReason 26 13
Playback of Alarm Plant Using a Rule for Reducing Six Most Frequent Chattering Alarms Data generated using UReason Original two most frequent alarms are no longer among ten most frequent alarms 27 Phases for Control Loop Performance Management Benchmark & assessment Define KPI Establish baseline Analyze systematic review of worst performers and highest impact loops Hardware transmitters, control valves, speed control, analyzers Tuning Control structure Implementation & execution MOC, capital, turnaround timing Continuous improvement Performance monitoring & optimization Publish and review KPI Maintenance and sustainability External Focus Company standard PlantTriage from Expertune Internal Focus 28 14
KPI for Control Loop Performance % of loops not in normal operation (i.e. manual or cascade open) more than 10 % of the time % of loops running more than 5 % of the time at their limits (saturation) % of loops that are oscillating % of loops with valve issues 29 % Control Loops in Normal Mode IV 2006 Industry Benchmark 30 15
CPM KPI February to June 2006 REACTION DISTILLATION 35 30 25 20 15 10 5 0 02/15/06 03/17/06 04/16/06 05/16/06 06/15/06 20 15 10 5 0 02/15/06 03/17/06 04/16/06 05/16/06 06/15/06 50 40 30 20 10 0 02/15/06 03/17/06 04/16/06 05/16/06 06/15/06 % Loops with valve issues Rx average ~ 9% Distillation average ~ 25% <5% good, >10% needs improvement % Loops above oscillating threshold -tuning Rx 0% - 4% Distillation 0% 4% <5% good tuning not current problem % Loops above oscillating threshold - load Rx 20% - 50% Distillation 20% - 40% >20% needs improvement control strategies 31 Energy Savings Finishing Unit 2005-2006 Btu/gal Fin feed vs Fin feed for various time periods 26000 24000 2nd Q 2005 22000 20000 3rd Q 2005 4th Q 2005 1st Q 2006 BTU/gal 18000 16000 2nd Q 2006 Poly. (3rd Q 2005) Poly. (4th Q 2005) 14000 Poly. (1st Q 2006) Poly. (2nd Q 2006) 12000 Poly. (2nd Q 2005) 10000 50 60 70 80 90 100 110 120 130 140 Fin Feed, gal 32 16
Key Benefits of Process Control Loop Monitoring and Performance Improvements Smoother operation Less process upsets lead to increases in availability and rate Running closer to process constraints allows increase of BDP (max. rates) Reduced operator input Reduced variability caused by differences in experience Less biased operation Less alarms Reduced shutdown cost Improved basis for DMC applications Reduced upsets Less mechanical damage Major impact Valve wear Pumps Column internals Efficiency Planning Causes of many reliability problems are process control related and not strictly mechanical Maintenance 33 Why Operator Advisory System? One Problem: Data Overload Trends are a valuable tool for operators to identify potential critical conditions The human brain gains insight more easily from visual representations like graphs than from numbers No easy way to graph large amounts of data Conventional maximum of 4 or 5 variables per graph 435 x-y graphs needed to show the interactions of 30 variables Difficult to visualize data sets of high dimension But we want a proactive operating philosophy 34 17
Suggestion of an Operator - Advisory Interface Process Move <Tag> <Descriptor> Process Objective <Tag> <Descriptor> NOW FUTURE Advisor shows where to position the process handles to meet the process objectives AND optimize the process economics 35 Path Forward Critical Condition Management Implement tools at all sites Alarm management: Logmate (TiPS) phase I CPM: PlantTriage (Expertune) Establish KPI and baselines at the sites Drive with continuous improvement program (CIP) Share best practices (Lyondell user groups ) Establish OE standard Provide dynamic alarm management and enhanced knowledgebased operator support systems Implement UReason phase II Operator graphics, operator advisories 36 18
Success in Critical Condition Management Accelerate implementation through the use of best practice and standard tools Sustain the gain through performance monitoring (KPI), CIP, and maintenance management Support Operations to be More proactive More vigilant Adopted by operations Prevention of shutdowns and avoidance of process upsets Faster recovery from process upsets and critical conditions (trips ) 37 19