implement or model-driven information systems at norfolk southern
DESCRIPTION
Implement OR Model-Driven Information Systems at Norfolk Southern. Clark Cheng Operations Research Norfolk Southern Railway Atlanta, Georgia INFORMS 2012 Annual Meeting Phoenix, Arizona. Agenda. Overview of OR Decision Support Systems at NS Success Factors General Technology Footprint - PowerPoint PPT PresentationTRANSCRIPT
Implement OR Model-Driven Information Systems at Norfolk SouthernClark ChengOperations ResearchNorfolk Southern RailwayAtlanta, Georgia
INFORMS 2012 Annual MeetingPhoenix, Arizona
Agenda
• Overview of OR Decision Support Systems at NS• Success Factors• General Technology Footprint• Examples of Technologies :
– Real-time System - SOA (Service Oriented Architecture)• Locomotive Assignment and Routing System
– Decision Support System – Client Server Architecture• Hump Yard Simulation Model and 3-D Animation
– BI/Analytics – GIS and Mobile Technology• GIS Map Dashboard on iPAD
• Q&A
What Do the Users Want?
They need an operable “car”, not just an fancy “engine”
Success Factors• Open Minded in Adopting New Technology
– SOA (Service Oriented Architecture)– Cloud Computing– Grid Computing– RIA (Rich Internet Application)– Big Data (Data growth by 20 times by 2020)
• Consistent Delivery Method– Minimize footprint of technologies– Better user experience (UX)– Browser based user interfaces (UI)– Client Server Architecture and Distributed System– Deployed on NS production servers with 24x7 support
• Rapid Turnaround– Close relationship with business users– Serve the business needs
• Senior Management Support– OR’s creativity and capabilities
General Technology Footprint• Browser-based Applications
– High availability and automatic updates • RIA User Interfaces
– (Java applet), GWT, Flex, Ajax • Back End
– Java EJB, Hibernate, Apache Axis 2– DB2, Teradata, MongoDB
• Middle Tier– WebLogic, Apache, Tomcat
• Special Purpose Software– Discrete event simulation framework– Business rule management system (BRMS)
Software Reliability and Quality Control
•Source Code Version Control•Merant PVCS
•Release Management and Incident Tracking•HP Quality Center
•Automated Testing•HP Load Runner
Locomotive Assignment & Routing System (LARS)
• Road Fleet Size Model– Determine the number of locomotives needed in the road
fleet for current operations and forecasted traffic demand• Short-Term Locomotive Surplus/Deficit Forecast
– Forecast locomotive surplus/deficit in 24 – 48 hours and improve the productivity of locomotives
• Shop Routing and Real-time Assignments– Model generates recommendations of optimal
assignments of locomotives to outbound trains– Consider locomotive operations at a more detailed level
than any human could possibly do
Software components and Environment:• Production, Quality Control, Development and Disaster Recovery deployment environment
Materialized viewsStored proceduresTriggersIndex PartitionsTable Partitions
IBM DB2 EETeradataJ
DBC
RPC
RHEL Linux GridLoad balancedCPLEX SolverTomcat serverSOAP Web service
VMware ESXPrivate Cloud
AuthenticationAuthorizationSAML/SSOUser RolesUser Groups
Application SecurityS
OAP
GIS applicationReal time messagingOperations dataField informationMaintenance data
Interface Enterprise AppsR
EST
Services Layer
Thin Clients
Traffic Manager
Apache Webserver
WebLogic Server
Locomotive Assignment and Routing System (LARS)
Software Components and Environment:• Production, Quality Control, Development and
Disaster Recovery environments
• Browser based RIA desktop and mobile client created using GWT, Flex, Ajax
Software components and Environment:• Production, Quality Control, Development and Disaster Recovery deployment environment
Materialized viewsStored proceduresTriggersIndex PartitionsTable Partitions
IBM DB2 EETeradataJ
DBC
RPC
RHEL Linux GridLoad balancedCPLEX SolverTomcat serverSOAP Web service
VMware ESXPrivate Cloud
AuthenticationAuthorizationSAML/SSOUser RolesUser Groups
Application SecurityS
OAP
GIS applicationReal time messagingOperations dataField informationMaintenance data
Interface Enterprise AppsR
EST
Services Layer
Thin Clients
Traffic Manager
Apache Webserver
WebLogic Server
• Application delivery controlled and local traffic managed by F5 BigIP LTM (Local Traffic Manager)
Software Components and Environment:• Production, Quality Control, Development and
Disaster Recovery environments
Locomotive Assignment and Routing System (LARS)
Software components and Environment:• Production, Quality Control, Development and Disaster Recovery deployment environment
Materialized viewsStored proceduresTriggersIndex PartitionsTable Partitions
IBM DB2 EETeradataJ
DBC
RPC
RHEL Linux GridLoad balancedCPLEX SolverTomcat serverSOAP Web service
VMware ESXPrivate Cloud
AuthenticationAuthorizationSAML/SSOUser RolesUser Groups
Application SecurityS
OAP
GIS applicationReal time messagingOperations dataField informationMaintenance data
Interface Enterprise AppsR
EST
Services Layer
Thin Clients
Traffic Manager
Apache Webserver
WebLogic Server
• Clustered Apache web servers with WebLogic proxy maintains round robin affinity sessions
Software Components and Environment:• Production, Quality Control, Development and
Disaster Recovery environments
Locomotive Assignment and Routing System (LARS)
Software components and Environment:• Production, Quality Control, Development and Disaster Recovery deployment environment
Materialized viewsStored proceduresTriggersIndex PartitionsTable Partitions
IBM DB2 EETeradataJ
DBC
RPC
RHEL Linux GridLoad balancedCPLEX SolverTomcat serverSOAP Web service
VMware ESXPrivate Cloud
AuthenticationAuthorizationSAML/SSOUser RolesUser Groups
Application SecurityS
OAP
GIS applicationReal time messagingOperations dataField informationMaintenance data
Interface Enterprise AppsR
EST
Services Layer
Thin Clients
Traffic Manager
Apache Webserver
WebLogic Server
• Clustered WebLogic Application server on UltraSPARC T2
Software Components and Environment:• Production, Quality Control, Development and
Disaster Recovery environments
Locomotive Assignment and Routing System (LARS)
• Consumes enterprise security Authentication service to manage user access• Consumes enterprise Role based Authorization service to manage permissions
Materialized viewsStored proceduresTriggersIndex PartitionsTable Partitions
IBM DB2 EETeradataJ
DBC
RPC
RHEL Linux GridLoad balancedCPLEX SolverTomcat serverSOAP Web service
VMware ESXPrivate Cloud
AuthenticationAuthorizationSAML/SSOUser RolesUser Groups
Application SecurityS
OAP
GIS applicationReal time messagingOperations dataField informationMaintenance data
Interface Enterprise AppsR
EST
Services Layer
Application Service Layer (SOA):
• Consumes enterprise security Authentication service to manage user access
• Consumes enterprise Role-based Authorization (RBA) service to manage permissions
Locomotive Assignment and Routing System (LARS)
• Consumes enterprise security Authentication service to manage user access• Consumes enterprise Role based Authorization service to manage permissions
Materialized viewsStored proceduresTriggersIndex PartitionsTable Partitions
IBM DB2 EETeradataJ
DBC
RPC
RHEL Linux GridLoad balancedCPLEX SolverTomcat serverSOAP Web service
VMware ESXPrivate Cloud
AuthenticationAuthorizationSAML/SSOUser RolesUser Groups
Application SecurityS
OAP
GIS applicationReal time messagingOperations dataField informationMaintenance data
Interface Enterprise AppsR
EST
Services Layer
• Optimization models deployed on private cloud of 8 Linux servers provides a grid of computational resources
• Provides and consumes services to start and control CPLEX grid servers
Locomotive Assignment and Routing System (LARS)
Application Service Layer (SOA):
• Consumes enterprise security Authentication service to manage user access• Consumes enterprise Role based Authorization service to manage permissions
Materialized viewsStored proceduresTriggersIndex PartitionsTable Partitions
IBM DB2 EETeradataJ
DBC
RPC
RHEL Linux GridLoad balancedCPLEX SolverTomcat serverSOAP Web service
VMware ESXPrivate Cloud
AuthenticationAuthorizationSAML/SSOUser RolesUser Groups
Application SecurityS
OAP
GIS applicationReal time messagingOperations dataField informationMaintenance data
Interface Enterprise AppsR
EST
Services Layer
• DB2 Enterprise Edition serves as transactional database and NCR Teradata as data warehouse
Locomotive Assignment and Routing System (LARS)
Application Service Layer (SOA):
• Consumes enterprise security Authentication service to manage user access• Consumes enterprise Role based Authorization service to manage permissions
Materialized viewsStored proceduresTriggersIndex PartitionsTable Partitions
IBM DB2 EETeradataJ
DBC
RPC
RHEL Linux GridLoad balancedCPLEX SolverTomcat serverSOAP Web service
VMware ESXPrivate Cloud
AuthenticationAuthorizationSAML/SSOUser RolesUser Groups
Application SecurityS
OAP
GIS applicationRealtime messagingOperations dataField informationMaintenance data
Interface Enterprise AppsR
EST
Services Layer
• Provides locomotive location service to GIS system
• Consumes locomotive pre-assignments decisions from Field Operations
• Consumes real-time events reported by trains passing scanners and GPS beacon from locomotives
Locomotive Assignment and Routing System (LARS)
Application Service Layer (SOA):
Hump Yard Simulation (YardSIM)
• Analyze yard capacity and evaluate the impact of changes in traffic pattern, infrastructure, resource, and operating policy.
• A Decision Support System:– Web Interfaces– Discrete-event Simulator– 3-D Animation– Yard Layout Editor– Process Modeler and Rule Engine, and – What-if Scenario Management
Yard Simulation Architecture
Scenario & Report
Manager
Database
TrainsTrains ConsistTrip PlanYard LayoutScenario ConfigurationYard Operating RulesReportsStatistics
Simulator&
Optimizer
UI (Browser) Application Database
3D Animator & Live Statistics
View 1
3D Animator & Live Statistics
View n
Dispatcher & Client Manager
Simulator&
Optimizer
Built on the Discrete Event Framework
3D Animation using the JAVA 3D APIs.
BRMS used for Rules, Workflow and Event Processing.
Client-Server model with multiple clients connecting to the server in parallel
Yard Simulation with 3D Animation
Yard Simulation Multiple View Windows
Class Tracks
Pullback TracksYard Top View
Receiving Tracks
Operations Research GIS Dashboard• Objective
– Monitor real-time train status– Track locomotives and equipment
• Features:– Interactive GIS maps – Accessible through web browser– Train status summary– Train route display– Locomotive location and status– Locomotives due for maintenance– Equipment location and status
External Data Sources Client
MQ Series
MQ Series Locomotives
(LARS)
Tonnage Rating (Web Service)
Trains & Schedule(OPD)
Car Distribution(CDM)
History(Teradata)
Geo-Location
Data Visualization
BING Maps(Base Layer)
NS BP Layers(Stations & Scanners)
NS Core Layers(Rail Infrastructure)
- Flare- Symbol- Chart- Star Burst- Route
Renderers
ESRI Flex API
Mapping Services
Proxy and Application Layer
DNS – IP Clustered App and Web Servers
App Server Weblogic 10G3
Web Server IPlanet
App ServerWeblogic 10G3
Web ServerIPlanet
AJAX Request
Response (XML)
GIS Server
ArcGIS 10.3 Server
REST
Operations Research GIS Map Dashboard Architecture
Questions?