Reliability Driven Asset Management Plant of the Year 2015 - Gábor Bereznai - October, 18 2016, Rotterdam
Plant of the Year winners in 2010 2
Small in the world Size of the logos are correlating to companies revenue. 3
But strong in the region 6 production units 23 mtpa refining capacity 2.1 mtpa petrochemicals capacity >2000 filling stations 4
Business drivers around the Economical crisis Source: Wood MacKinsey, Global Refinery View, Refining in Europe, Africa and FSU 5
Diverging world in Donwstream business Global refining and petrochemical business outside of Europe is growing New, large scale, high-tech refineries and petrochemical sites in Asia and the Middle East are increasing product export to Europe The European downstream business is pressurised by worldwide trends and not supported by local politics and regulations UNITED STATES Becoming a net exporter of oil products Huge investments targeting European markets Favourable legislation (incl. tax legislation) Cheap feedstock, 2470 kbpd, $95bn New and ongoing investments CENTRAL AND SOUTH AMERICA Strong demand growth Favourable legislation for local companies EUROPE Not growing with stagnating demand Record low margins, crude runs at 23-year low Still significant overcapacity Decreased competitiveness, EU regulations put further pressure on European DS business MIDDLE EAST Large new refining sites Additional ~2 mbpd refining capacity in 3 years (5x MOL capacity) Targeting the European market 2750 kbpd, $60bn New and ongoing investments ASIA-OCEANIA Strong governmental support New refineries in China, India, Vietnam, etc. 4070 kbpd, $125bn New and ongoing investments 6
Delivering Business Value: 500M$+500M$ in five years 7
Production loss due to UPDT 2,500,000 Production loss [USD] 2,000,000 1,500,000 1,000,000 500,000 [ 563 kedc ] [ 607 kedc ] 0 2009 2010 2011 2012 UPDT: Unplanned Down Time due to instrumentation causes FIMS: Field Instrumentation Maintenance System Source: MOL SAP-PM Maintenance System, 2014 2013 2014 Q1 8
Layers of Protection Computerised Maintenance Management Systems Master Asset Database Maintenance planning and sheduling Maintenance execution Evalution of Operative Maintenance Reliability analysis (MTBF, MTTR) Failure modes Maintenance Strategy Creation Definition of Group of Assets to Maintenance Risk evaluation and handling (Asset Policy) Strategy optimisation Asset Strategy Evaluation Detailed Reliability analysis Condition Monitoring Systems Risk Based Inspection Vibration diagnostic, Oil analysis, Thermo Control Valve and Instrumentation diagn., Data Collectors Off-line, diagnostic Corrosion database On-line, diagnosti c Off-line, diagnostic Vibration data collection On-line, diagnostic Off-line, diagnostic, Communicator RFID / PDA Serial digital comm. DCS systems Equipements Statical equipements Rotating machines Field Instrumentations 9
CMMS integration SAP-PM server SAP-PM user FIMS (AMS/PRM) user/ expert AMS DM + AMS FIMS servers Refinery Information System On-line FIMS subsystem (AMS) Predictive notifications can save 7k$-700k$ avoiding unit outages. DC MUX Alarm filtering in the SAP-FIMS interface Intelligent instrumentations of the DC units CFV-087 10
DCS and Smart Instrumentation in MOL Refining MOL has 58 units in the Refining 95% equipped with DCS and Safety PLC. (19000) The number of the non-smart pneumatic transmitters are decresing. 11
Integrated Maintenance Systems AM - AMS - PRM protocols (3382 pcs devices) AM - AMS - PRM protocols (4135 pcs devices) 2500 2500 2110 2114 2000 2000 Devices [PCS] 1500 1000 850 Devices [PCS] 1500 1000 1347 500 413 500 413 219 0 0 0 0 9 0 0 Honeywell AM Emerson AMS Yokogawa PRM 0 0 0 42 0 0 Honeywell AM Emerson AMS Yokogawa PRM HART FFB Wireless HART HART FFB Wireless HART 2010 2015 12
Improved organisational model is needed to achieve the Refinery objective Business rational of organisational model change: Align the organisation with refinery objectives Improve cooperation and eliminate silo operation Put more focus on key areas in order to improve our efficiency and performance based on Solomon benchmark (maintenance, operational availability, energy efficiency) 15
Peolpe behind the systems - Csaba Molnár-Valkó - October, 16 2016, Rotterdam 16
Man behind system Asset management system (AMS, PRM, FDM) Online equipment types Adding equipment / configuration ALERT setting / handling Valve diagnosis Pressure transmitter calibration Analytical instrument calibration Tranings / courses 17
Asset Management System AMS Asset Manager System PRM Plant Resource Manager FDM Field Device Manager 5 online plant: 8 online unit: 6 online unit: - DCU - GOK3 - CL5 - HDS - KBI Tags: ~ 2600 online ~ 5000 offline - MSA - REF4 - GOK1 - HGY1 - HGY2 - REF100 - KGÜ - BK4 Tags: ~1450 online - CL4 - CL6 - BEK5 - FCC - AV2 - PEM1 Tags: ~ 1600 Online 2010 2015 18
On-line equipment types - Transmitters - Pressure - Differential pressure - Level - Temperature - Flow transmitter - Valves - Control valave - ON-OFF valve (with positioner) 19
Adding device / Configuring Add new online device - Connecting new device to the system - Scanning and Denomination - Placing it in Plant structure - Setting all the 3-level alerts: - configaurated in the device - adjustable in the maintenance system - setting interface filter 20
Setting Alerts / Handling Device Alert handling DVC5000 (66 pcs parameters / 33 alerts) DVC 6200 (150 pcs parameters / 50 alerts) 21
Setting Alerts / Handling Setting Alerts in maintenance system 22
Setting Alerts / Handling Setting interface 23
Valve diagnostic ValveLink Online/Offline DTM based (Metso, Flowserve) Flowscan ValVue OVD 24
Pressure Transmitter Calibration Process - Adding device, scanning, denomination - Test-scheme assigning, placing in the Plant structure - Check Out - Calibration - Check IN - Report template assign - Report preparation 25
Analyser equipment calibration Process - Offline TAGS download from AMS to PDA - Device identification with RFID - Calibration (manual) - Check IN from PDA - Report generation - Notice posting to ERP 26
Skills / Competance A high level knowledge of systems (AMS, PRM, FDM) Database handling at user level Knowledge of specific devices Calibration knowledge Valve diagnostic on a really high level 27
Delivering Business Value from Digital Transformation Operational Availability Maintenance Efficiency Energy Efficiency Yield improvement - Tibor Komróczki October, 18 2016, Rotterdam
Digital Transformation of New & Next Downstream program Safety & Asset integrity (PSM) Interlock statuses Integrity Operating Windows Corrosion control (HTHA) Alarm management Preventing coke steam eruption Energy Energy monitoring and management Energy KPI breakdown Column energy efficiency dashboard Hydrogen, utilities - energy balances Flaring Yields Product quality Analyser reliability (Argus) Yield Accounting via Sigmafine (PI AF based) Operational Availability Energy Efficiency Maintenance Efficiency Yield improvement Operational optimization Operating envelopes NG (natural gas) and fuel gas demand forecasting Normal mode of control loops APC control monitoring Diesel sulphur Asset optimization Reliability from Proactive Coker yield & Predictive optimization Advanced Analytics SAP PM Integration Health Score in PI AF CBM on all rotating equipment PSA Pressure Swing Adsorbers Chillers Heat Exchangers Electrical Infrastructure
Advanced Analytics and IoT Oil & Gas Downstream Increase productivity and efficiency across all major business units through the best practices for data harmonization Condition based & predictive maintenance Alarm Management System (alarm rationalization) Mode Base alarm, new alarm logics Inferential and descriptive statistics Analyze control loops behavior Energy modelling optimization Deeper understanding of technological processes - Alternative crude oil usage as feed; yield optimization
Advanced Analytics Intelligence VALUE Cloud How can we make it happen? Analytic Ascendancy Model Data What happened? Descriptive Analytics Why did it happen? Diagnostic Analytics What will happen? Predictive Analytics Prescriptive Analytics Gartner, March 2012 DIFFICULTY
Digital transformation People Enablement of Contextual Data Based Decision Making & Management Improving skill knowledge capture changing paradigms about data - leading by example Process Success Technology Reengineering the workflow around enhanced & consistent data
Machine learning Find the optimal mixture of different feeds into the Delayed Coker process Achieve minimal level of coke yield Diesel Hydrotreater unit product sulfur content estimation based on available data Data analysis and feature selection for modelling Azure ML predictive model building and scoring Azure ML technology adaptation compare laboratory, online analyzer, APC soft sensor and ML data Evaluation of results and data visualization
The Importance of Having an OT Data Infrastructure Rapid development and scalability of applications Reinforce the use of data and analytics based decision making Support cultural change and normalization Leverage advanced technologies including advanced analytics and IOT to accelerate business value Enable sustainable business value in the 21st century
Machine Learning Architecture Current & Future Field DCS SCADA Real-time data & Meta data E-Logbook LIMS PI System Opralog NICE Laboratory data Natural Info Center PI Integrator for BA 35
Business issues in Delayed Coker Unit Increasing coke yields from 25.39% to 27.43% (+2.04 %) from 2012.01.01 and 2016.03.01. Average monthly steam eruptions in 2012-2015 period was 3.85, in the first month of 2016 it was 15.5 (4X increase) Steam Eruptions 1 % Coke yield decreasing ~ $6M/year benefit in Danube Refinery!
Coke Yield & Explosion Blue: coke yield (output value) Red: steam eruption ~In case of > 3100 t Furnace feed input the coke explosion likelihood is increasing Blue histogram: Row count coke cycle Between the 2550-2800 t intervallic the coke yield could be decreased without coke explosion
Training series Competence improvement Improve Azure ML and Data Analyst competences Participants: Process technologist, Developer technologist, Automation, operation and energy management; IT specialist (with superlative PI and statistics knowledge) Overview of tools (Azure ML environment, R, Python) Regression methods, interpretation of models, regression tree, linear regression, evaluation methods Steps of real data mining projects Deeper analyses of Delayed Coker unit Support Delayed Coker Feed Blender project
Ongoing Project OSISoft PI SAP PM Connection Support Condition Base maintenance
Challenge Critical Availability Problems Hydrogen Production Plants (HPP) are critical units in the refinery Pressure Swing Adsorbers (PSA) are critical equipments in unit operation Cyclic operation Heavy load on valves (9-10 open-close hourly) 1.2 MUSD loss in three years due to PSA valve failures UPTIME program: 97 % Operational availability
Architecture Roles of components PI Server Process database Online analysis of process information Calculation of asset health Asset condition Running hours Performance User Interface PI Coresight PI DataLink Connection (WebLogic) Calculated asset health Maintenance related information SAP PM Maintenance database Management of maintenance processes Creation of work orders or notifications Trigger maintenance strategies based on asset health
Future Project PI Connector for HART-IP
PI Connector for HART-IP