SAP Predictive Maintenance and Service & SAP Asset Intelligence Network Ryan Weicker, Senior Support Engineer Digital Business Services, SAP America
Agenda Asset Intelligence Network AIN Overview Functions and Features Integration PdMS Overview Benefits Across the Maintenance Program PdMS Overview Asset Visualization Insight Providers Machine Learning Engine PdMS Customer Example 2
SAP Asset Intelligence Network Bringing together Business partners Models / Equipment Service Provider Manufacturer Operator Regulator SAP Asset Intelligence Network will provide a global registry of industrial equipment; built and shared between multiple business partners and used across the industry by all stakeholders. This will enable new collaborative business models resulting in true Operational Excellence. 3
SAP Asset Intelligence Network Enabling collaborative asset management Apps Apps for collaborative processing e.g. equipment lookup, announcements, service bulletins, performance improvement, spare parts management, obsolescence management Combined together to deliver Content Network A cloud portal of standardized content that defines and documents models and equipment, shared and stored, for a consistent definition between business partners. A secure network to connect multiple business partners for inter and intra company information exchange and collaboration. SAP Asset Intelligence Network 4
SAP Asset Intelligence Network Collaboration between manufacturers, service provider, and operators Network Content Apps Job Instructions Announcements Obsolescence Management Performance Improvement Spare Parts Equipment as a Service* Work Collaboration* Commissioning & Handover* Nameplate info Maintenance strategy Spare Parts Service bulletins Failure modes Recalls Bills of Materials Designs and drawings Design improvements Sensor definition Operating instructions Maint instructions Safety instructions Product training Manufacturer Service bulletin receipt Service bulletin processed Usage information Installation information Failure / incident data Design recommendations Risks and controls Measurement documents Telemetry Operator *planned 5
Enable OEM and Operator collaboration using a Digital Twin SAP Asset Intelligence Network OEM 1 20 MVA 3Phase Transformer Model Operator OEM 2 Digital Twin 20 MVA 3Phase Transformer - 1 20 MVA 3Phase Transformer - 2 20 MVA 3Phase Transformer - 3 20 MVA 3Phase Transformer - n OEM 3 20 MVA 3Phase Transformer - 1 20 MVA 3Phase Transformer - 2 20 MVA 3Phase Transformer - 3 20 MVA 3Phase Transformer - n Physical Assets Defined in SAP PM 6
SAP Asset Intelligence Network Building the network 1 2 An internal Installed Base / Asset Portal across multiple in-house systems An Install Base / Asset Portal where you invite key Service Providers and Manufacturers 3 Progressively benefit from an expanding network of contributors 7
The SAP Asset Intelligence Network Business value Operator view MANUFACTURER OPERATOR BUSINESS VALUE Manufacturer #A Nameplate Information Spare Parts Recommendations Maintenance strategies / tasks Documents and Drawings Reduce master data maintenance effort Reduce manual asset search effort Receive notifications, service work summaries and service bulletins Manufacturer #B Asset operator Establish one channel to many manufacturer s, EPCs and Service providers Manufacturer #C Service Provider Push communication and alerts to manufacturers / service providers Lower asset life cycle costs Pump Manufacturer A Flow Meter Manufacturer B Motor Manufacturer C Enabler for self-regulation Tracking of serialized components being installed into a major component (manufacturers orders subcomponents) 8
The SAP Asset Intelligence Network Business value - Manufacturer OPERATOR MANUFACTURER BUSINESS VALUE Usage information Maintain model (equipment) information once Operator #1 Send & receive data Specifications & drawings Get transparency into equipment usage Improve warranty and recall processes Operator #2 Manufacturer Specific customer commerce Improve customer relationships One solution for many customers Operator #n Recommendations & updates Product & service feedback Basis for collaboration and future business models Offer additional services and revenue Increase equipment portal reach Increase customer lifetime value Single source of truth / system of engagement 9
SAP Asset Intelligence Network Applications Admin Apps Business Partners Authorizations Templates Master Data Apps Models Equipment Locations Spare Parts Documents Instructions Process Apps Performance Improvement Obsolescence Report Error Code Lookup 10
SAP Asset Intelligence Network Equipment: Features Information Model Information Model Attributes Equipment Attributes Installation Information Life Cycle Information Structure and Parts Structure Spare Parts Documentation Model Documents Equipment Documents Instructions Announcements Monitoring Measuring Points Error Codes Improvement Cases Time Line 11
SAP Asset Intelligence Network SAP Asset Intelligence Network Content - Attributes Content - Attributes 12
SAP Asset Intelligence Network SAP Asset Intelligence Network Content Structure and Spare Parts Content Structure and Spare Parts 13
SAP Asset Intelligence Network SAP Asset Intelligence Network Content Spare Parts using 3D Visualization Content Integrated 3D Visualization 14
SAP Asset Intelligence Network SAP Asset Intelligence Network Content Work Instructions Content 3D Work Instructions 15
SAP Asset Intelligence Network SAP Asset Intelligence Network Content Documents Content Documents 16
SAP Asset Intelligence Network SAP Asset Intelligence Network Application Announcement Application Announcement 17
SAP Asset Intelligence Network SAP Asset Intelligence Network Application Measuring Points Application Measuring Points 18
Asset Intelligence Network Model Information in SAP EAM (PM) Side Panel The following information is available in the side panel: Model header information Characteristics (Attributes) Announcements Instructions Documents 20
Integration to SAP EAM (PM) View model information in PM side panel SAP Asset Intelligence Network SAP ERP PM Model 1 Equipment* Side Panel View of Model Info 2 Link Table requires SAP ERP 6.0 Enhancement Package 6 as a minimum and use of SAP Business Client. Customer is required to implement notes from SAP Service Market Place. 1) Find matching model in AIN 2) Link created 3) Option to create DMS documents from AIN Documents *could also be Functional Location depending on customer use 21
SAP Asset Intelligence Network EAM (PM) Integration Onboard PM Equipment / Functional Locations to AIN Ability to configure which remains as the master. Ability to configure at each attribute level. Equipment structure, documents & attributes (characteristics) are synchronized in a bi-directional way between PM and AIN AIN Announcement processing on EAM Manufacturer announcements processed by type POWL entry Work Item created for responsible user for Equipment per plant Notifications created Batch processing 22
Integration with SAP Master Data Governance (MDG-EAM) Use data out of SAP Asset Intelligence Network directly into a MDG Change Request (CR) e.g. equipment information is validated, enriched in MDG before create or change in SAP ERP. From MDG CR search in AIN e.g. for suitable model Provided as part of standard MDG EAM solution extension (by Utopia) 23
Agenda Asset Intelligence Network AIN Overview Functions and Features Integration Business Cases PdMS Overview Benefits Across the Maintenance Program PdMS Overview Asset Visualization Insight Providers Machine Learning Engine PdMS Customer Example 24
Technology is changing our approach to maintenance *Use of Maintenance Strategy Today Run to Failure Preventative Predictive *Use of Maintenance Strategy Future Run to Failure Preventative Predictive The Internet of Things is leading to increased use of predictive maintenance Although still relevant, preventative maintenance typically results in over-maintaining assets and high cost The goal of our solution is to enable a data science driven predictive maintenance in order to reduce unplanned failures *Proportion of maintenance strategies are for illustration purposes only and will vary based on many factors 25
Asset Condition Multiple Approaches to Predictive Maintenance Data science driven approaches are on the rise Data Science Driven Equipment Driven Human Driven More time to respond enables greater flexibility to dynamically plan maintenance events P P Potential Failure X-ray Radiography Battery Impedance Test Oil Analysis Hot to Touch P Why now? IoT/device connectivity Big data available for training models Declining hardware and software costs Massive computing power Audible Noise F Functional Failure P Potential Failure = First Indication of Failure Time Ancillary Damage T Total Failure 26
The Internet of Things Benefits the Entire Maintenance Program Run to Failure Preventive On-Condition Predictive Replacements driven by more logical utilization rates WHEELS & BRAKES Energy Dissipation versus Mileage Preventive Drive scheduled maintenance based on the right utilization metric Advanced condition monitoring techniques BEARINGS Vibration Analysis versus Oil Analysis Program On-Condition Near real-time condition monitoring BATTERY Data science based health indicators Installed battery = Normal battery Predictive Leverage data science-based health indicators The Internet of Things improves existing strategies and enables new data science driven maintenance approaches 27
Customer Example Train Operator * Run to Failure Preventive On-Condition Predictive Company Owns and operates a fleet of around 2,000 electro-trains, 2,000 locomotives and 30,000 coaches and wagons 40% of maintenance is currently reactive * The maintenance strategy proportions are for illustration purposes only and not reflective of actual customer percentages 28
Customer Example Train Operator Run to Failure Preventive On-Condition Predictive Solution Improve effectiveness of maintenance programs Data fusion between IT and OT data Remote train diagnostics Improved Program Effectiveness Starting with Improvements to Preventative Maintenance Plans BRAKES Energy Dissipation versus Mileage Benefits DOORS Open/Closure Cycles & Times versus Mileage Engineering rules and predictive models Dynamic planning of maintenance schedules Higher asset availability & passenger satisfaction Projecting 100M Euro savings per year in maintenance operations costs when fully implemented 29
Agenda Asset Intelligence Network AIN Overview Functions and Features Integration Business Cases PdMS Overview Benefits Across the Maintenance Program PdMS Overview Asset Visualization Insight Providers Machine Learning Engine PdMS Customer Example 30
SAP Predictive Maintenance and Service Decision support across the ecosystem & asset lifecycle DESIGN PURCHASE SUPPORT BUILD DISPOSE OPERATE & MAINTAIN OEM Dealer Service Provider Owner/Operator R&D Procurement Production Aftermarket Sales Service Fleet Operator Improve asset reliability and up-time Monitor quality of purchased components Improve manufacturing processes Comply with contract service level agreements Increase customer satisfaction and loyalty Deliver the value added service at the right price Decrease maintenance costs Increase asset up-time Decision support to ALERT, DISCOVER AND REMEDY Machine Data Business Data Combining IT & OT data gives machine data context 31
SAP Predictive Maintenance and Service Solution components and value drivers Actions Logistics & Maintenance Execution Systems Business User Enables a data science driven approach to condition monitoring Insights SAP Predictive Maintenance and Service Asset Health Control Center Asset Health Fact Sheet Domain Expert Flexible extension concept for customers to build industry or customer specific models and analytics Alerts Machine Learning Engine Data Scientist A scalable Machine Learning Engine that drives data science insights into our business processes Raw Data SAP SAP Leonardo IoT Foundation SAP Leonardo SAP Leonardo for Edge IoT Edge Computing Machine Data Business Data Data Manager Flexible visualizations across equipment structures End-to-end process integration Alert, Discover, Remedy 32
SAP Predictive Maintenance and Service System and component level visualizations Logistics & Maintenance Execution Systems Asset Health Control Center SAP Predictive Maintenance and Service Asset Health Control Center Asset Health Fact Sheet Machine Learning Engine Asset Health Fact Sheet SAP Leonardo Foundation SAP Leonardo for Edge Computing Machine Data Business Data 33
SAP Predictive Maintenance and Service * Health Status Overview is an example of a custom Insight Provider built using SDK 34
SAP Predictive Maintenance and Service Asset Health Control Center Asset Health Control Center Asset View Houston Refinery New Orleans Refinery Asset View 35
SAP Predictive Maintenance and Service Asset Health Control Center Asset Health Control Center Asset View Asset Hierarchy Houston Refinery New Orleans Refinery Asset View Asset Hierarchy 36
SAP Predictive Maintenance and Service Asset Health Control Center Asset Health Control Center Asset View Houston Refinery Asset Hierarchy New Orleans Refinery Asset View Asset Hierarchy 37
SAP Predictive Maintenance and Service Asset Health Control Center Asset Health Control Center Asset View Asset Hierarchy 38
SAP Predictive Maintenance and Service Asset Health Control Center Asset Health Control Center Insight Providers Asset View Asset Hierarchy Insight Providers 39
SAP Predictive Maintenance and Service Asset Health Control Center Asset Health Control Center Insight Providers Remaining Useful Life 2 5 12 8 10 18 13 22 16 32 Asset View Asset Hierarchy Insight Providers 40
SAP Predictive Maintenance and Service Asset Health Control Center Asset Health Control Center Insight Providers Asset View Asset Hierarchy Insight Providers 41
SAP Predictive Maintenance and Service Asset Health Control Center Asset Health Control Center Insight Providers Asset View Asset Hierarchy Insight Providers 42
SAP Predictive Maintenance and Service Asset Health Control Center Asset Health Control Center Insight Providers Asset View Asset Hierarchy Insight Providers 43
SAP Predictive Maintenance and Service Asset Health Control Center Asset Health Control Center Insight Providers * Health Status Overview is an example of a custom Insight Provider built using SDK Asset View Asset Hierarchy Insight Providers 44
SAP Predictive Maintenance and Service Asset Health Control Center Asset Health Control Center Insight Providers Asset View Asset Hierarchy Insight Providers 45
SAP Predictive Maintenance and Service Asset Health Fact Sheet Asset Health Fact Sheet Equipment View Asset Heath Control Center Remain ing Useful Life 2 5 Serial #12345 1 2 1 8 2 2 3 2 Equipment View Asset Health Control Center 46
SAP Predictive Maintenance and Service Asset Health Fact Sheet Asset Health Fact Sheet Equipment View Asset Heath Control Center Remain ing Useful Life 2 5 Serial #12345 1 2 1 8 2 2 3 2 Equipment View Asset Health Control Center 47
SAP Predictive Maintenance and Service Asset Health Fact Sheet Asset Health Fact Sheet Equipment View Asset Health Control Center 48
SAP Predictive Maintenance and Service Asset Health Fact Sheet Asset Health Fact Sheet Insight Providers Equipment View Asset Health Control Center Insight Providers 49
SAP Predictive Maintenance and Service Asset Health Fact Sheet Asset Health Fact Sheet Insight Providers Equipment View Asset Health Control Center Insight Providers 50
SAP Predictive Maintenance and Service Asset Health Fact Sheet Asset Health Fact Sheet Insight Providers Equipment View Asset Health Control Center Insight Providers 51
SAP Predictive Maintenance and Service Asset Health Fact Sheet Asset Health Fact Sheet Insight Providers Equipment View Asset Health Control Center Insight Providers 52
SAP Predictive Maintenance and Service Machine learning challenges High dimensional data SOLUTION Feature construction/selection requires data scientists & domain user collaboration No labeled failure data Anomaly detection and reinforcement through user feedback Rare failure events Failure prediction using ensemble learning Outdated models, human scale Model management, continuous learning and scoring Use case specific algorithms Extensibility and integration of new algorithms 53
SAP Predictive Maintenance and Service Machine Learning Engine Supervised learning enables failure predictions like Remaining Useful Life Finds contributing factors to failure events Expert feedback Models change as operational environment changes Extensibility for out-of-the-box algorithms Possibilities to deploy new R based algorithms Unsupervised learning detects anomalies Enables Health Scores New Algorithms Extensibility Failure Prediction Trigger prediction when algorithm detects a specific combination of input variables Reinforcement* Domain expert feedback Anomaly Detection Trigger anomaly alert when the algorithm detects an abnormal pattern Continuous Improvement & Learning Model Management Tools Failure Prediction *PdMS roadmap Item 54
Agenda Asset Intelligence Network AIN Overview Functions and Features Integration Business Cases PdMS Overview Benefits Across the Maintenance Program PdMS Overview Asset Visualization Insight Providers Machine Learning Engine PdMS Customer Example 55
Caterpillar - manufacturing division Goals: Monitoring health of connected assets Leveraging machine learning and telemetry-based statistics to trigger automated predictive notifications Improve availability and overall performance Benefits: Reduce loss of production due to unplanned downtimes Planned maintenance becomes dynamic responding to health signals and not to a fixed schedule Reduce maintenance cost Personas: Superintendent Maintenance planner Manufacturing Engineer Maintenance Technician Phases: #1: 1 plant 1 critical asset #2: 1 plant more assets #3 and beyond : more plants more assets 56
Solution components & PdMS microservices HANA Machine data Maintenance records Rule services and rules ML models BODS PdMS R and scores HRF SAP Predictive Maintenance and Service Asset Health Control Center Asset Health Fact Sheet SLT ERP Gateway Hub Automatic PM notifications triggered by PdMS Alerts using odata PDMS HANA Server SAP HANA DB 1.0 SPS12 Revision 122 or higher SAP HANA Rules Framework (HRF) SAP XS Advanced (XSA) 1.0.34 SAP PdMS On-Premise 1.0 FP02 + Patch 2 SAP PdMS SDK RAM = 350 GB; Disk = ~400GB of Disk Space; OS RHEL 6.7 or higher R Server: RAM = 16 GB; Disk = ~60 GB of Disk Space; OS RHEL 7.3 R Version 3.3.2 Rserve Version 1.8.5 Machine Learning Engine Hana Rules Framework 57
PdMS @ CAT A Dynamic System Data Integration Fully automated initial/ delta loads from OT/IT Frequency of delta loads managed by job scheduler. Control table mechanism to manage individual Machine data system API modules. Job logs to capture processing history of data extraction/load run. Emails notifications in case of job failure. Plant Maintenance notifications created in backend ERP system using OData via SAP Gateway Hub system. Machine Learning Principal Component Analysis: PdMS READINGS table to be populated with raw data; outliers removal Running instructions data, Time series data. Align time series and impute data. Pivot view of all sensors. Aggregate views Training & Scoring Scheduling Weibull Analysis: Workorder/notification data from maintenance system. Fetch required info from PdMS EVENTS table as a dynamic view. Project scoring view for next X units of time. Train model based on dynamic input view. Schedule training & scoring on a periodic basis. User will always have a projection for next X units of time based on all analysis of work order data. Rules Telemetry Rules: Driven by sensor upper and lower bound limits Driven by direct machine alerts severity Machine Learning Scores rules: Driven by anomaly scores above severe and critical thresholds Driven by probability of failure above severe and critical thresholds Maintenance records rules: PM notifications not attended to during past X units of time PM work orders not attended to during past X units of time 58
Continuous Data processing Scheduling [ machine data * ] [ SAP PM data ] [ telemetry/sensors data ] Ingest raw data Align sensor time series [ rule service view ] Remove duplicates and outliers* Load delta data Impute sensor values Filter timestamps where asset is not operating Apply HRF rules Create PdMS Alerts Pivot data [ equidistant-imputedfiltered-pivoted table ] Trigger HRF Actions [ call.xsjslib ] Create PM notification [ rest end point ] BODS [ data for training ] [ data for scoring ] XSA scheduler PdMS Executor Train model ** Score model [data sci. scores] XSC scheduler [ model ] ** outliers are removed only for freedom elog data ** model training uses different job execution frequency than model scoring 59
Thank you.
SAP Leonardo IoT - The Big Picture UI Layer (SAP Leonardo Bridge) PEOPLE Enterprise Management The Digital Core Procurement R&D Supply Chain Planning Manufacturing Logistics Sales After Sales Service PROCESSES Network Log. Hub Connected Goods SAP Leonardo IoT Apps Track & Trace Connected Mfg. Vehicle Insights Predictive Main. Asset Intelligence Network APPLICATIONS SAP Leonardo IoT Edge SAP Leonardo IoT Foundation SAP Cloud Platform / SAP HANA Platform PLATFORM Things / Physical Layer PRODUCTS THINGS 61
PdMS Solution Architecture Safety Control DCS PLC/SCADA Devices Advance Control Plant Databases Manual Data Device Connectivity Device Management Portal Data Ingestion Landing Zone Files Messages Batch Stream Ingestion Pipeline Transformations Rules TimeSeries Database Data Archive Insight Provider ERP / CRM PdMS Application Insight Provider Business process integration Insight Provider ERP/CRM On-demand replication Exploration Zone Data Fusion Production Zone Data Fusion Extensions Real-time replication Key Figures Rules Predictive Models Derived Signals Transport Key Figures Rules Predictive Models 62
PdMS Solution Architecture Safety Control DCS PLC/SCADA Devices Advance Control Plant Databases Manual Data TELIT, SAP PCo, IoT SERVICES* Device Connectivity Device Management Portal SAP DATA SERVICES BIG DATA HUB* SAP IQ, OSISoft PI, HADOOP/VORA* Data Ingestion Landing Zone Files Message s Batch Stream Ingestion Pipeline Transformations Rules TimeSeries Database Data Archive HANA SMART DATA STREAMING Insight Provider ERP / CRM PdMS Application Insight Provider Business process integration Insight Provider ODATA / HCI / SAP PO* UI5 & XSA Extensions ERP/CRM On-demand replication SAP HANA & HANA RULES FRAMEWORK R Real-time replication Exploration Zone Data Fusion Key Figures Rules Predictive Models Derived Signals Transport Production Zone Data Fusion Key Figures Rules Predictive Models SAP HANA & HANA RULES FRAMEWORK R * Planned 63