1. Electricity Generation Mix by US Industrial Sectors Disaggregating Electricity Generation and Modeling Interstate Transfers Joe Marriott, Carnegie Mellon University Advisor Scott Matthews InLCA/LCM September 2003

2. September 25, 2003Joe Marriott - Carnegie Mellon University2 Why Disaggregate? Primary contributor to environmental impact is electricity generation Primary contributor to environmental impact is electricity generation Impacts by generation type varies a lot Impacts by generation type varies a lot –e.g. Hydro versus coal This variation should be reflected in LCA results This variation should be reflected in LCA results For example: Aluminum manufacturing For example: Aluminum manufacturing –Industry sector has emission numbers reduced due to plants in WA, which has 80% hydroelectric generation

3. September 25, 2003Joe Marriott - Carnegie Mellon University3 Currently in LCA Use aggregate US generation mix to calculate emissions, GWP, etc. Use aggregate US generation mix to calculate emissions, GWP, etc. Net Electricity Generation by Energy Source, 1999 SourceGeneration, % Coal50.99% Petroleum3.22% Gas15.31% Nuclear19.72% Hydroelectric8.32% Other2.41%

4. September 25, 2003Joe Marriott - Carnegie Mellon University4 Disaggregate Results Rather than a single sector: Rather than a single sector: “Electric services (Utilities)” Have multiple electricity sectors: Have multiple electricity sectors: “Electric services (Utilities, Coal)” “Electric services (Utilities, Nuclear)” etc. Each industrial sector would receive a specific mix of these disaggregated electricity sectors Each industrial sector would receive a specific mix of these disaggregated electricity sectors

5. September 25, 2003Joe Marriott - Carnegie Mellon University5 Making It Happen Data is not readily available Data is not readily available –Need complete facility-level transaction data for all US industrial sectors –No central repository of this data –Economic data not necessarily a good estimate Instead, assign a specific generation mix to each industrial sector using: Instead, assign a specific generation mix to each industrial sector using: 1.“Locations” of industry sectors from BEA 2.State generation mixes from DOE

6. September 25, 2003Joe Marriott - Carnegie Mellon University6 Sector Allocation to States Need percentage of each industry sector located in each state Need percentage of each industry sector located in each state –Available using a tool developed at Carnegie Mellon by Iavor Kostov and Scott Matthews Economic Census location data from BEA used for placement Economic Census location data from BEA used for placement –Then uses number of employees and shipments as measures of size and intensity to weight various sectors in various states

7. September 25, 2003Joe Marriott - Carnegie Mellon University7 Include Interstate Trading Lots of electricity transfer in the United States, especially following deregulation in 1996 Lots of electricity transfer in the United States, especially following deregulation in 1996 Currently, interstate electricity transfer ignored, but it’s a big part of the market Currently, interstate electricity transfer ignored, but it’s a big part of the market –30% of California power is imported –West Virginia exports 60% of theirs These numbers have a significant impact, so new generation mixes are created for each state These numbers have a significant impact, so new generation mixes are created for each state

8. September 25, 2003Joe Marriott - Carnegie Mellon University8 Western US: Net Imports (TWh) +2.7 -70.9 +2.8 +3.4 +14.6 -10.0 +10.7 +16.6 +10.8 -5.6 +26.9

9. September 25, 2003Joe Marriott - Carnegie Mellon University9 Model: Linear Optimization Using 2 large matrices (23 x 28), find out where imports likely came from by minimizing distance (hops) traveled Using 2 large matrices (23 x 28), find out where imports likely came from by minimizing distance (hops) traveled Classic transportation/distribution cost minimization problem Classic transportation/distribution cost minimization problem This still isn’t what is actually happening on the grid, but it’s a pretty good estimate This still isn’t what is actually happening on the grid, but it’s a pretty good estimate

10. September 25, 2003Joe Marriott - Carnegie Mellon University10 State “Hops” for California 1 0 1 2 3 2 2 1 2 3 3 3 3 4 4 4 4 1

11. September 25, 2003Joe Marriott - Carnegie Mellon University11 CACA COCO CTCT DCDC DEDE FLFL GAGA IAIA IDID ILIL LALA MAMA MDMD MEME MIMI MNMN MOMO MSMS MXMX NCNC NJNJ NVNV NYNY OHOH OROR VAVA VTVT WIWI AL 6464411353263845214256537264 AR 4264532242164843112264532363 AZ 11967653243961154341581853594 CN 33243752125231113599422122411 IN 7443343251443613235336316242 KS 4164543232364843133353544363 KY 6342332251342623124235316142 MT 32887762135871042354672752683 ND 43887762236871031355663643682 NE 4164543122364832133353533362 NH 1282549779681415767863102411516 NM 21857643232871054231472763684 OK 3365643232265843122373534463 PA 8522154473521424347318118223 RI 11815498796814357681063 24 527 SC 8563421464463856336158548265 SD 4276665122576931244563643573 TN 5352421242252733113145425153 TX 3277644343174954221363754374 UT 21867653143861054342571752584 WA 231088774155 812644637829617105 WV 7432243352431524236226217143 WY 31866652134851042243562642483 Complete US Hop Count

12. September 25, 2003Joe Marriott - Carnegie Mellon University12 CACA COCO CTCT DCDC DEDE FLFL GAGA IAIA IDID ILIL LALA MAMA MDMD MEME MIMI MNMN MOMO MSMS MXMX NCNC NJNJ NVNV NYNY OHOH OROR VAVA VTVT WIWI AL 6464411353263845214256537264 AR 4264532242164843112264532363 AZ 11967653243961154341581853594 CN 33243752125231113599422122411 IN 7443343251443613235336316242 KS 4164543232364843133353544363 KY 6342332251342623124235316142 MT 32887762135871042354672752683 ND 43887762236871031355663643682 NE 4164543122364832133353533362 NH 1282549779681415767863102411516 NM 21857643232871054231472763684 OK 3365643232265843122373534463 PA 8522154473521424347318118223 RI 11815498796814357681063 24 527 SC 8563421464463856336158548265 SD 4276665122576931244563643573 TN 5352421242252733113145425153 TX 3277644343174954221363754374 UT 21867653143861054342571752584 WA 231088774155 812644637829617105 WV 7432243352431524236226217143 WY 31866652134851042243562642483 Complete US Hop Count

13. September 25, 2003Joe Marriott - Carnegie Mellon University13 CACA COCO CTCT DCDC DEDE FLFL GAGA IAIA IDID ILIL LALA MAMA MDMD MEME MIMI MNMN MOMO MSMS MXMX NCNC NJNJ NVNV NYNY OHOH OROR VAVA VTVT WIWI AL 246 3 33 AR 4 4 AZ 16 CN 13 1 3 5 01234 IN 5 1 8 14 KS 1 1 2 KY 2 0 7 9 MT 3 9 0 2 14 ND 17 219 NE 1 1 03 NH 5 5 NM 9 1 11 OK 12 3 PA 84 1 4 16 RI 0 0 SC 4 4 SD 3 3 TN 0 0 TX 1 0 1 UT 2 2 WA 17 WV 14 22 20 56 WY 233 26 70513104246595181431172101540530227014 Completed Optimization, Showing Electricity Transactions in TWh

14. September 25, 2003Joe Marriott - Carnegie Mellon University14 Generating the Generation Mix Have the % of imports for each importing state Have the % of imports for each importing state –Example: CA imports 30% Have the % of that imported amount that came from each exporter Have the % of that imported amount that came from each exporter –AZ: 34%, MT: 3%, NM: 19%, UT: 4%, WA: 36% Know the % mix of the amount from each importer Know the % mix of the amount from each importer –Arizona: 45% Coal, 10% Gas, 35% Nuclear, 10% Hydro Multiply these 3 sets of percentages, normalize with existing generation mix to get the new values Multiply these 3 sets of percentages, normalize with existing generation mix to get the new values

15. September 25, 2003Joe Marriott - Carnegie Mellon University15 The Next Generation (Mix) in CA

16. September 25, 2003Joe Marriott - Carnegie Mellon University16 Electricity Allocation to Sectors Now, apply each state’s generation mix to the percentage of all the industrial sectors in the state Now, apply each state’s generation mix to the percentage of all the industrial sectors in the state –20% of all widgets are manufactured in CA, so 1/5 of the widget sector will have CA’s generation mix Then sum the generation types across all sectors and states Then sum the generation types across all sectors and states –Each sector now includes part of the generation mix for each state it’s located in

17. September 25, 2003Joe Marriott - Carnegie Mellon University17 Results of Modification DescriptionCoal, %Gas, %Oil, %Hydro, %Nuclear, % Household laundry equipment manufacturing 8320213 Oil and gas field machinery and equipment 41441210 Jewelry and silverware manufacturing 233414916 Aircraft manufacturing 301633116 Cellulosic organic fiber manufacturing 4932241 Industries with the highest specific electricity generation mix values (by energy source) in the U.S. economy

18. September 25, 2003Joe Marriott - Carnegie Mellon University18 Aircraft Manufacturing

19. September 25, 2003Joe Marriott - Carnegie Mellon University19 Dirty Laundry? And How!

20. September 25, 2003Joe Marriott - Carnegie Mellon University20 United States: Well Oiled

21. September 25, 2003Joe Marriott - Carnegie Mellon University21 Results Trend Towards Average

22. September 25, 2003Joe Marriott - Carnegie Mellon University22 Contributions Disaggregating adds accuracy to a critical sector in terms of environmental impact Disaggregating adds accuracy to a critical sector in terms of environmental impact Industrial sector generation mixes answer some interesting questions Industrial sector generation mixes answer some interesting questions –Which sectors are vulnerable to shifts in fuel price or technology change? –What is the potential impact of carbon taxes on the US economy? Import-export estimate an intuitive substitute for complete transmission grid analysis Import-export estimate an intuitive substitute for complete transmission grid analysis Most industrial sectors have a mix close to the US average mix Most industrial sectors have a mix close to the US average mix –Some interesting sectors have significant differences