Dinner keynote at Wharton May 9th 2011 @ 11th Annual Strategy and the Business Environment Conference (SBE) jointly with the 3rd Annual Research Conference Alliance for Research on Corporate Sustainability (ARCS)
3. SUPPLIER CUSTOMER DISTRIBUTIONCENTER CFO CRO Sourcing CUSTOMS/ REGULATORY AGENCY CUSTOMS/ REGULATORY AGENCY Customs Operations COO Fulfillment Receiving Manufacturing EVERY 2ND DOLLAR OF WORLD TRADE RUNS ON SAP Suppliers and Customers Exports Imports Export Compliance GLOBAL ENTERPRISE
11. What Can ICT Industry Do? “The ICT (Information and Communications Technology)industry is responsible for 2% of global CO2 emissions. ICT solutions have the potential to be an Enabler to reduce 30-50% of the 98% CO2 emitted by non-ICT industries.”
19. By 2020 it will be 35 Zetabyte (IDC, UC Berkeley and UC San Diego)
20.
21.
22.
23.
24. Big Iron - Commodity HPCDesign by SAP Enterprise Supercomputer - 1/30 Price of Mainframe 5 X 4U Nodes (Intel XEON x7560 2.26Ghz) 160 cores (320 Hyper-threads) 5 X 32 5 TB memory total, 30TB solid state disk 160 GB/s InfiniBand interconnect per node Scalable coherent shared memory (via ScaleMP) Developers don’t need additional skills for in-memory Data base becomes data structures Scalable DB on virtualized HW – Alternative to Cloud
25. Warren Powell et al. Princeton University - Operations Research and Financial Engineering Optimal Learning & In-Memory Handle Uncertainty
26. Solve Very Compute Intensive ProblemsLike Stochastic Optimization @Princeton Juggle intermittent energy from wind, solar & volatile electricity prices to meet time-varying loads – Princeton has the algorithms With BigIron we can reduce compute time from days to minutes! Wind speed Load Electricity prices
27. Modeling uncertainty in power scheduling The effect of modeling uncertainty in wind 2% wind 40% wind Uncertain forecast Perfect forecast Constant wind
28. Modeling Uncertainty In Power Scheduling Designing energy portfolios…. … is like building a stone wall. You can do a perfect job with a perfect forecast. The challenge is dealing with uncertainty.
29. John Williams et al. MIT Auto ID Lab Multithreading Real Time Event Platform
30. Rapid Growth of Events and Messaging Platforms Verizon and T-Mobile: 2-3 days to generate phone bill iTunes: 24 hours to generate bill Uninterrupted Growth of online billing systems (Hulu, Netflix…) Dynamic Pricing on SmartGrid requires design of infrastructure capable of ingesting millions of events in quasi-real time Goal: Design a multi-threaded system that produces the electricity consumption bill of a city of 1M households 8 hours seconds A Comparative Study of Data Storage and Processing Architectures
31. Smart Meter Reading Problem Data Generation Data Persistence Data Processing
33. Multithreading Real Time Events & Messaging Platform Platform that handles billions of events/day AND large numbers of threads on one machine (> 1 million), e.g. Siemens 500k events/s RDBMS (used by today’s MDUS vendors) provides good query performance but does not scale to millions of households (8 h) Prototype for SmartGrid allowing to ingest smart meter data in real time, do dynamic pricing (4 buckets), store in DFS & do real time analytics Bill for 1 M households in seconds A Comparative Study of Data Storage and Processing Architectures
35. California Statewide Cumulative Investment Through 2020 To Achieve Renewable Portfolio Standard Goals Governor Schwarzenegger signed Executive Order S-21-09 to adopt regulations increasing California's Renewable Portfolio Standard (RPS) to 33% by 2020. Need to forecast financial and operational impacts before investing
36. CalPower – A Hypothetical California Utility with 15% Renewable Generation Today CalPower generation portfolio today CalPower RPS goal in ten years 2010 2020 Total Renewables: 33% Total Renewables: 15% Geo Thermal: ?% Geo Thermal: 4% Biomass: 3% Biomass: ?% Solar: 3% Solar: ?% Natural Gas: ?% Wind: 5% Wind: ?% Nuclear: 18% Natural Gas: 48% Nuclear: ?% Coal: ?% Coal: 19% Traditional Technologies: 85% Traditional Technologies: 67%
37.
38.
39. Step 1: Use GridLAB-D To Model Objective & Constraints Today’s Power Sale Portfolio Goal – Year 2020 Weather Model Renewable Portfolio Standard 33% Natural Gas 48% 2.4 GW Coal 19% CalPower’s Load Models Constraints Total Peak Capacity Other 7% Maximum Wind Maximum Coal Wind 5% Nuclear 18% Solar 3% (GW)
41. Step 3: Drill Down Analysis Of Exception Days And Risks
42. Step 3: Drill Down Analysis Of Exception Days And Risks
43. Exception Day Risk Mitigation Strategies Use stored power to close the gap Decrease demand in response to supply drop 1. Adopt demand response OPEX Exception Day Risk 2. Invest in power storage technologies CAPEX Exception Day Risk
44. RPS Study Takeaway: GridLAB-D Solution Provides Larry The Answers He Needs Plan C Comprehensive model of utility operations, including the distribution level. Can model distributed generation, and can model loads at high resolution to make more precise forecasts of operations KPIs (e.g. CAIDI, CO2) and financial KPIs (OPEX, CAPEX). SAP User Experience Team helps business customers access results, and increase precision of their KPI forecasts. Plan B Plan A 2020 Portfolio C Questions: Peak Total Capacity: 5GW CAPEX:$1405/MWh OPEX: $167/MWh Total Cost:$15,566M Total CO2 emission:5MT Avg. CAIDI:1.63 Hours 2020 Portfolio B Questions: Larry’s questions answered Peak Total Capacity: 5GW CAPEX:$15,306.77 M OPEX: $368/MWh Total Cost:$15,566M Total CO2 emission:5MT Avg. CAIDI:1.63 Hours “Which plan offers the best expected total cost?” Questions: 2020 Portfolio A “Which plan minimizes financial and service quality risks?” Peak Total Capacity: 5GW CAPEX:$1,414.04 M OPEX: $13,726.04 M Total Cost:$15,140.08 M Total CO2 emission: 145,765,543.95 T CAIDI:1.63 hours/year “How do we mitigate these risks?”
46. Thank You! Contact information: Paul Hofmann SAP Labs, Palo Alto paul.hofmann@sap.com www.paulhofmann.net
Editor's Notes
Mobile banking in SA has almost doubled to 44% up from 27% in one year. 12% are now sending money from phone to phone. In Africa mobile minutes are currency.M-PESA (mobile pesa=money) is a branchless banking service, very successful in Kenya used by over 10 M people or 50% of the adult population; joint venture between Vodafone and Safaricom. Example of Prahalad’s developing countries leap frog the West.A story from a friend’s servant in SA. His pregnant wife having complications late in the night and they needed to get to the doctor. This was very late in the night and their residence was far from town. Neither the gentleman nor the lady had airtime to call the doctor. However the lady was a registered customer for Mobile Banking service. With no other option left to get airtime, a solution lay in her hands under her thumb. She got onto her phone and was able to buy airtime and called the doctor who diagnosed the issue on phone and the lady was relieved of the pain.Moore’s law for data creation - amount of data is doubling every 18 month
Larry will be using gridlabd to forecast the performance of different capacity plans in the next ten years, the first step he needs to do is to build case and as the input to the simulation tool. This is a summary of the goals and constraints Larry want gridlabd to know. You‘ll see here the first thing is calpower‘s current capacity portfolio, and larry will also import calpower‘s current load models build by the power engineers. The input also includes thirdparty data such weather forecast, and maybe data from the utilities existing databases such as ERP system or GIS systems. And also Larry sets some constraints to the case such as the maximum wind capacity cannot be larger than 3 GW. After building the case and, goals are cearly set, what larry is going to is to run the simulation in GridlabD, and he‘ll get the simulation results in the output dashboard.
Executive summary of some of the best case portfolios and the KPI value associated with each plan. Here Larry is able to compare the plans and have a clear overview of the advantage and disadvantages of each plan.
Look at the exception days, find a big gap on this day.
To summarize: In these case studies, we illustrate GridLAB-D‘s capability to forecast financial and operations KPIs for electric power utilities.