Post on 12-Jan-2015
description
A collaboration of:
Using HANA To Add Value to
Electric & Gas Revenue Integrity
Gelyn M. Almanzar
Public Service Enterprise Group (PSEG)
Tracy Kirk
Public Service Enterprise Group (PSEG)
P • Introduction & Overview of PSE&G
S •HANA Business Case
E • SAP HANA POV for PSE&G
G • PSEG HANA RID Roadmap
H • Crawl, Walk, Run Approach
A • PSE&G HANA Next Steps
N • Lessons Learned Points to Take Home
A •Questions
Slide 2
PSE&G at a Glance
3
• New Jersey’s oldest and largest
regulated utility
• Provide service to 75% of New Jersey’s
population
• 2.4 million customers – 2.2 million
Electric customers and 1.8 million Gas
customers
• Robust Appliance Service and HVAC
competitive business
• Most Reliable Electric Utility in 2012 – 5th
time in last 8 years
• PSE&G ranks 3rd among all utilities in
installed solar capacity
Revenue Integrity Department (RID)
4
Prevent, identify, and correct any and all meter conditions causing potential lost revenue to PSE&G, including malfunctioning equipment, non-registering meters,
incorrect multipliers, human error, and theft of service
10 Investigators handle approximately 7000 cases
annually
The Employee Incentive Award Program offers a cash award to an office or field employee
providing a lead resulting in the recovery of lost revenue. This is a $300,000 program annually
Slide 5
Revenue Integrity Department (RID) Use Case
Meter
Corrected
Meter Goes
Bad
Meter
Identified
Once the bad meter is identified and corrected, the customer is billed for lost
revenue. Collecting this is not always successful.
After the meter is corrected, a new revenue stream is created.
Faster, more accurate identification of bad meters is the key to revenue recovery
• We engaged with SAP in a Proof of Value for HANA exercise
• SAP conducted a series of workshops with the Line of Business
• Captured high-level business requirements from the business
• Worked with business to identify ECC tables needed for RID Reporting
• Business Intelligence team worked with SAP to extract data and send
it to SAP HANA Labs to conduct a Proof Of Value
Revenue Integrity HANA Proof of Value
Slide 6
7
RID Manual Reporting and Analytics Challenges
6 months Extract X 4 hours = 24 hours 2 HRS Manually Emailed
8
RID Future State with HANA
9
SAP Proof of Value HANA Results
PSEG BI group packaged the extracted ECC tables and sent to SAP Labs
SAP produced HANA data models and provided a look into the basic
reporting and analytics needs for RID
SAP Table Tech Name SAP Table Description Number of Records
ETTIFN Installation Table 1,709,155,620
EABL Meter Reading Documents 164,311,266
EABLG Meter Reading Reason 165,344,425
EVER IS-U Contract 6,525,758
EANLH Installation Time Slice 7,429,901
PSEG RID HANA SAP - POV Results
Slide 10
2.6M Customers
3.7M Meter
200M Rows Data
6 Months Data
93 Seconds Response Time
SAP HANA In Memory System
PSEG HANA Roadmap
Slide 11
Crawl
Walk
Run Hardware Purchase
BI Team
Training
Investment
Proposal
System Readiness
Define 3
Project Tracks
Business
Requirement
Gathering
Project Plan
(Agile)
Extracted tables
from Prod system
Created CSV
Files
HANA CSV
Data Loads
HANA
Data Model
BOBJ Webi &
Dashboard Reports
Data Services
ETL
Jan Feb March Apr May June July Aug
PSEG HANA Investment - Crawl
Slide 12
Our IT SLT team decided to build HANA expertise in-house
Invested $50K in HANA training for BI group (6 resources) HA100 – HANA – Introduction & Overview
HA200 – Operation and Administration
HA300 – Implementation & Modeling
Purchased HANA
Purchased DELL R910 with Fusion IO Cards
Racked up DELL server in local Data Center (Newark,
NJ)
PSEG HANA Investment - Walk
Slide 13
In-depth Business Requirement Gathering Sessions
Interactive discussions with SMEs to finalize the business Logic
Project Planned with “Agile” approach
Defining the “Start to Finish” implementation path (3 Project
Tracks)
Get the System Ready
Data Identification and ETL Plan
Alternate Data Extraction Strategy (.CSV files) to mitigate the
delay in Data Services Deployment
Coordinate with SAP to resolve the Data Services Technical
Issues
PSEG HANA Investment - Run
Slide 14
File specification and coordination for Flat File extraction
Initial HANA Data Modeling – Evaluation of alternate Models
Iterative and interactive approach with Business for Data modeling to
avoid any “misses” in project Deliverables
Clear definition and strict adherence of Project Plan and Deliverables.
Well integrated team work with HANA and BOBJ resources
Refine the HANA and BOBJ development for Performance and “look
and feel” of the reports / Dashboards
Resolve Data Services issues and Optimum Strategy (Full / Delta) Data
extraction using Data Services
Switch from HANA modeling to Data Extracted using Data Services
Slide 15
Data Extraction
• 16 ECC Tables
• 240M Rows
• CSV Files Created
Manual Load
• Largest Table 58M Records
• Initial Load Average 10hrs
Data Service ETL
• Largest Table 58M Records
• Initial Load Average 4hrs
• Delta Monthly load average 7 minutes
Model / Reports
• Created 8 Models
• BOBJ Webi 5 monthly Reports
• 2 Xcelsius Dashboards
HANA Data Extraction / Load and Models
Plan Build Execute
2.6M customers
367M 13 3.7M
rows of data
months of history
meters 60 seconds
26 hours
Standard System
In-Memory System
1500x faster
PSEG RID POC HANA Results
Anticipated Business Benefits
Slide 17
Increase the number of cases identified
Shorten overall cycle time by reducing time to identify cases
Increase percentage of good leads
Increase revenue by identifying and working high value leads first
Save on Incentive Award program
Greatest Insights Engage SAP as a preferred Consultant Partner
Identified areas of expertise needed, e.g. ECC, HANA, Data Provisioning, BO, etc. are required to produce a HANA application that can’t be handled by a single area.
The sizing of the HANA server affect the sizing of the BO servers.
Optimizing the Services configuration on BO 4 start only service that are required and stop those not required.
Webi reports have to consider segmenting the data so they can be processed by the BO server.
HANA allowed us to load data using complicated business rules processing large amounts of data that couldn’t be previously performed
Need to better understand the SAP Router process and what is required to permit SAP to access your system.
SAP Modeling Expert may help you to model but will not necessarily have the business expertise to produce a business ready model
There are always unforeseen obstacles and roadblocks encountered when utilizing new technologies
What did we get right? System Check and HANA expertise:
SAP performing the system check prior to starting the project and engaging SAP Consulting Service to assist, e.g. Data Services, Modeling, etc
Data Extraction Strategy: Extracting Data as .csv files (while configuring Data Services)
In Depth, Interactive Discussions with Business: To understand business scenarios and rules.
Business Expectations: Setting appropriate and relevant business expectations (Data and Performance)
Proof of Value (Crawl, Walk and then Run): Deliver limited scope and then extend.
Go Forward strategy for Data Extraction: Got Data Services fully configured and working to load data into HANA
Agile Project Approach: Continuous engagement of business through design, build and data validation
What would we do differently?
• Build a flexible project plan to accommodate the fact that many tasks
and durations are not known since it is the first time we are using this
technology
• Get commitment from other areas (Basis, UNIX, etc.)within IT
department to help resolve issues more quickly
• Isolate the project team, and have dedicated HANA project resources
committed to HANA solution
Potential Future PSEG Use Cases
Operations
Profitability Analysis at the
appliance level
Annual Work Planning
Outside leak investigation review
Spatial representation of
operational data
Optimized scheduling of meter
route
Customer Ops & Service
Accelerate Sales Stats process
by improving the extraction
performance of bill header data
Campaign planning and
tracking
Customer segmentation
Improve worry free contract by
better analysis
Proactive management of bill
shock
Outage management / Storm
Restoration
Finance & Trade
Rate Case Analysis
BPC on HANA
Improve FERC Reporting
Risk analysis of trade positions
Generation dispatch from trading
floor based on complex models
including causal factors
POD analysis for regulatory
reporting
A collaboration of:
Gelyn M. Almanzar PSEG
Gelyn.Almanzar@pseg.com
Tracy Kirk PSEG
Tracy.Kirk@pseg.com