CDM Consolidated Methodology for Electric Sector Baselines – Issues

and Proposed Answers

Duane T. Kexel, Vice PresidentPower System EngineeringMadison, WI USAkexeld@powersystem.orgBuenos Aires, December 8, 2004

2

Background of My Comments

Dozens of Power Supply, IRP, and Feasibility Studies In Numerous Countries

Czech Standardized Electric Sector Baseline

Hungarian Biomass Project

Poland Standardized Sector Baseline

Flies, Birds and Elephants

3

The CDM Challenge

CostType II Type IShort ERs Excess ERs

Probability of E Probability of ETotal Costs

TransactionsForgone CostsCarbonCosts

Accuracy

Direction of IncreaseModeling Detail Depends On Time Horizon

Input Data Quality Aggregates Are More Accurate Than Detailed Data

Transparency Broader Data Most Public

Constituency Breadth Gaming Guard

4

CDM Quality Assurance Troika

CDM Meth Panel

Validators Verifiers

5

Baseline Uses

Ex Ante Allocation– Forgone Opportunities for Buyers– Underdelivery Penalties for Sellers– Inaccurate Methods Chill The Market

Ex Post Disbursement– Does Not Depend Totally On Forecasts– Should Make Maximum Use of Actual Data– Impacts Risk Split Between Buyer and Seller

Never Forget That The Common Goal of All Is To Reduce Emissions of GHG

6

Czech Electric Sector Standard Baseline Objectives Minimize Transactions Costs For Multiple Small Projects Consistent Comparative Standard for Renewable Electric

Generation Simple and Dynamic Recognize Key Features of Czech Grid Both Screening and Monitoring By Local CEA Staff/Consultants About One Day Per Project Per Year For Analysis Requires Only Readily Available Data With A Known Publication

History and Expected Future Defaults Are Defined In Case Data Are Not Available Conservative but Competitive

7

Key Czech Grid Considerations Impact of Temelin (2,000 MW Nuclear, 11,000 MW System) Addition

Additions Before 2010 Not Load Driven

8% Renewable Target by 2010

Many Small Renewables Not Dispatchable

Large Blocks of Must Run Units – IPP CHPs

Coal Dominance - Gas Limited

Marginal Not Average Fuel Mix Matters

Operating Margin - Not Build Margin is Relevant

8

CDM ACM0002 – Operating Margin EF is calculated as a combined margin (CM) of the operating margin(OM)

and the build margin (BM)

OM (a) – Simple OM – Weighted average of generation by fuel above non-displaceable sources

OM (b) – Simple Adjusted – Adjusts for “excess” non-displaceable sources above minimum load – then the weighted average of generation by fuel

OM (c) – Hourly Dispatch Using Top 10% Mix in Each Hour

In all cases, data from last available years or ex post monitoring is acceptable

9

OM(a) & OM(c1) Graphic Concepts

MW

Hg Hc

Hours

Non-Displaceable Baseload Generation

Intermediate Coal

Gas

10

OM (a) and OM (c1) Example

Numeric ExampleHg 1,000 Total Mix Above DisplacedHc 7,760 Fuel Fuel Mix Baseload MixMWg 200 Coal 98.0% 93.6% 88.6%MWic 300 Gas 2.0% 6.4% 11.4%MW P 10 Total 100.0% 100.0% 100.0%

MWhg 100,000 EfficiencyMWhic 1,464,000 Gas CT 34.0% 34.0% 34.0%Disp MWh 1,564,000 Coal 34.0% 34.0% 34.0%Gas Share 6.4% OM (a) Weighted 34.0% 34.0% 34.0%

MWhmg 5,000 Displaced Production Carbon Factors IPCC MT/MWh fuel inputMWhmc 38,800 Displaced Production Gas 0.200 0.200 0.200 Total MWhm 43,800 Displaced Production Coal 0.364 0.364 0.364 Gas Share 11.4% OM (c1) Weighted 0.361 0.354 0.345

MWbl 400 CEFs MT/MWh 1.06 1.04 1.02 MWhbl 3,504,000 Total MWh 5,068,000 Gas Share 2.0% Average Fuel Mix

11

OM (c1) Methodology OM (c) – Conceptually correct but literal data demands are excessive and

often not practical

Use of 3 year historic average can very seldom be justified.

Concept can be used but much more simply applied – that is (c1).

Czech method only needs the split of marginal hours by fuel and the average conversion efficiency for the class of marginal plants (Hg and Hc and maybe Hn).

Three methods of estimating marginal hours by fuel

12

OM (c1) vs OM (c)

OM (c1) Statistical method requires only monthly generation by fuel – Typically easily available and reliable data – Self testing.

OM (c1) Screening curve requires more data and calculation but dramatically less data than hourly generation by unit

Proxy unit hours are useful check if such units can be identified.

Statistical method is easily applied going forward with other methods as checks if results are bizarre. Default bounds can be reasonably set.

13

Screening Curve Hg-Hc SplitMW

Gas Coal

Cf

Gf

1 2 3 4 5 6 7 8 9 Hours (000)

% Pk

90%

80%

70%

60%Coal

50% Base and Intermediate Load

40%

30%

20%

10%

1 2 3 4 5 6 7 8 9 Hours (000)

Coal, Base and Intermediate Load

Other Baseload

Gas

14

Statistical Method of Hourly Split

0

1000

2000

3000

4000

5000

6000

7000

8000

1 2 3 4 5 6 7 8 9 10 11 12

Month

GW

h

Nuclear Hydro Renewables Coal Gas

15

Monthly Gas Gen = f(Total Gen)

SUMMARY OUTPUTDependent Variable: Gross Gas Generation

Regression StatisticsMultiple R 93.2%R Square 86.8%Adjusted R Square 85.5%Standard Error 27.16Observations 12

ANOVAdf SS MS F Significance F

Regression 1 48495.5103 48495.51 65.73865 1.04687E-05Residual 10 7377.016365 737.7016Total 11 55872.52667

Coefficients Standard Error t StatIntercept -554.3 92.8788 -5.97Total Generation 0.1179 0.0145 8.11So, 1 kW reduction for 8760 hours would lead to 8760 kWh reduction which would mean1032 kWh reduction in gas production or 11.8 %.

16

Model Fit For 2002 For Czech Rep

Actual Vs Estimated Gas Generation

0

50

100

150

200

250

300

1 2 3 4 5 6 7 8 9 10 11 12

Month

Actual Estimated

17

Proxy Unit Method

Look at average operating hours of peaking units on the system.

Would work well if system has conventional peaking CTs.

Czech case, IGCC and pumped storage hydro cover some of the peak. These are actually coal equivalent units.

Gas-fired CTs are minimal and the only sizeable unit is often used for voltage support rather than load.

For CR, it is more useful to track gas-fired generation.

18

Czech Results Comparison

Method Gas Hours Coal Hours Total Hours Gas % Coal %Screening Curve 803 7,957 8,760 9.2% 90.8%

Statistical 1,033 7,727 8,760 11.8% 88.2%

Average 918 7,842 8,760 10.5% 89.5%Recommended 1,000 7,760 8,760 11.4% 88.6%

2004Fuel Displaced Hours Percent

Gas 1,000 11.4%

Coal 7,167 81.8%

Nuclear 593 6.8%

Total 8,760 100.0%

19

What About OM (b)? More than gas and coal may be displaced.

It is critical to check for displacement of non-displaceable generation – “Above the knee”.

For Czech Republic, just completing 2,000 MW of new nuclear generation. Also, many coal-fired CHPs dispatched for heat load.

Czech method calculates from energy data like OM (b).

Still must decide on dispatch order of low-cost units. Must-run cannot be displaced. For CR, nuclear was chosen.

20

Czech Method for Excess Baseload

Ht% Pk Hg Hc Hn

90%Gas

80% Peakg

70%

60%Coal Slope = S

50% Intermediate Load

40%z z

30% Nuclear

20% Coal-Fired CHPS

10%ROR Hydro

1 2 3 4 5 6 7 8 9 Hours (000)S = z/Hn a 1z = S*Hn b - 2*HtA = S*Hn*Ht - SHn^2/2 c 2A/S2A - 2S*Hn*Ht + SHn^2 =02A/S - 2Hn*Ht + Hn^2 = 0 Hn = {-b +- (b^2-4ac)^0.5}/2a

A = z*Ht - z*Hn/2

21

Summary on OM Methods OM (a) will generally overstate emissions

Marginal rather than average fuel displacement should be used based on marginal hours.

Within fuel types, average unit efficiencies can be used and will provide conservative results with reasonably accessible data

OM (c1) is correct and preferable if data can be obtained and project can justify the transactions costs.

Three year historic average would only be reasonable in very few cases and could be very misleading.

22

Select Additional Key OM Issues

Time Resolution of The Analysis

Load Curve Representations

Liberalization of Power Markets

Complete Calculation Method

23

Proper Time Resolution For OM

Fuel mix may change significantly by month if load changes significantly by month.

Production may vary dramatically by month for wind or hydro – generally not for biomass although the maintenance month could matter.

Practical candidate periods are annual, seasonal, monthly.

Hourly load data can be used to identify monthly distribution of peak hours and of low load hours.

24

Hgas and Hnuc by Month and YearTable 4-25

2002 2003 2004 2005Month High Load Hours Pct of Mo Hrs High Load Hours Pct of Mo Hrs High Load Hours Pct of Mo Hrs High Load Hours Pct of Mo Hrs

Jan 261 35.1% 211 28.4% 249 33.5% 296 39.8%Feb 177 26.3% 68 10.1% 148 22.0% 216 32.1%Mar 68 9.1% 26 3.5% 51 6.9% 98 13.2%Apr 0 0.0% 0 0.0% 0 0.0% 1 0.1%May 0 0.0% 0 0.0% 0 0.0% 0 0.0%Jun 0 0.0% 0 0.0% 0 0.0% 0 0.0%Jul 0 0.0% 0 0.0% 0 0.0% 0 0.0%Aug 0 0.0% 0 0.0% 0 0.0% 0 0.0%Sep 0 0.0% 0 0.0% 0 0.0% 0 0.0%Oct 13 1.7% 3 0.4% 11 1.5% 34 4.6%Nov 239 33.2% 127 17.6% 212 29.4% 273 37.9%Dec 275 37.0% 236 31.7% 266 35.8% 285 38.3%

1,033 671 937 1,203

2002 2003 2004 2005Month Low Load Hours Pct of Mo Hrs Low Load Hours Pct of Mo Hrs Low Load Hours Pct of Mo Hrs Low Load Hours Pct of Mo HrsJan 0 0.0% 0 0.0% 0 0.0% 0 0.0%Feb 0 0.0% 0 0.0% 0 0.0% 0 0.0%Mar 0 0.0% 0 0.0% 0 0.0% 0 0.0%Apr 0 0.0% 5 0.7% 5 0.7% 3 0.4%May 0 0.0% 88 11.8% 74 9.9% 67 9.0%Jun 0 0.0% 158 21.9% 136 18.9% 106 14.7%Jul 0 0.0% 254 34.1% 233 31.3% 202 27.2%Aug 0 0.0% 224 30.1% 194 26.1% 171 23.0%Sep 0 0.0% 68 9.4% 58 8.1% 51 7.1%Oct 0 0.0% 3 0.4% 0 0.0% 0 0.0%Nov 0 0.0% 0 0.0% 0 0.0% 0 0.0%Dec 0 0.0% 0 0.0% 0 0.0% 0 0.0%

- 800 700 600

25

Marginal Fuel Mix by Month/Year

Percent Coal Marginal Hours by Month

0.0%

20.0%

40.0%

60.0%

80.0%

100.0%

120.0%

Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec

2003

2008

26

Czech Rules On Time Granularity

If max and min monthly production of proposed project are within 10% of average, annual is okay.

If annual not okay, seasonal or monthly is required.

If within season variation about the mean is within 10%, seasonal is okay.

If annual and seasonal are not okay, monthly analysis is required.

Monthly is always accepted since it is more accurate.

27

Load Curve Representations

Relevant LDC for domestic generation is domestic consumption plus net exports.

Best basis is three year history of hourly domestic consumption.

Net exports can be adequately represented from data for peak and minimum loads plus annual total MWh.

Domestic load shapes normally shift slowly over time.

28

Load Curve Approximations

Linear is ideal for simplification but not adequate.

Piecewise linear was used for CR based on 2002 history and two segments.

Algebraic representation is very good for load factors between 40% and 60%

MWt = f(LF, Max, Min/Max, Pi, Hours)

29

Czech Piecewise Linear @ 360 Hrs

2002 Czech Load Duration Above 360 Hoursy = -0.4714x + 9352.1

R2 = 0.9861

0

1000

2000

3000

4000

5000

6000

7000

8000

9000

10000

360 1360 2360 3360 4360 5360 6360 7360 8360

Hours

MW

30

Net Exports From Peak, Min MW and Annual MWh

GW

12ba

10

d

8 c

6 Values For 2002e

4 f

2

1 2 3 4 5 6 7 8 9Hours

Domestic Load Curve Plus Net Exports

Domestic Load Curve (Gross Cons)

a = 11,205 MWb = 11,582 MWc = 8,144 MW d = 9,246 MWe = 5,742 MWf = 4,196 MW

8,400360

31

Displaced Emissions Calculation

32

ACM0002 – Build Margin

BM (1) – Ex Ante – Five Most Recent Power Plants Built, or– Most Recent Vintage That Accounts for 20% of MWh

BM (2) – Ex Post For 2008-2012, Then Ex Ante

Default Weighting of OM and BM is Equal butProponent Can Defend Alternative Weights.

33

Concerns With ACM002 BM Observed Additions Cannot Logically Represent What Was Displaced By The

CDM Project Since The Displacement Was Not Built. BMs Are Inherently Based on Forecasts.

Historic Additions Are Not Good Predictors of Future Additions Reliance on 20% of Generation Will Bias Results Toward Baseload Additions

When Many Renewable Projects Do Not Displace Baseload.

Default of 50-50 Seems Arbitrary And Will Seldom Be Reasonable For Specific Years.

Should Require A Demonstration of When BM Becomes Relevant And Then It Should Get 100% Weight. Before That, Weight Is Zero.

34

Principles of Correct BM

Expansion Plan Should Be Constrained Least Cost Path (CLCP)

– Additionality Typically Based On Investment Analysis– OM Based On Constrained Economic Dispatch– BM Should Have Same Foundation

Concept Is The Difference in CLCP With and Without Proposed Project

Many CDM Projects Will Provide Energy Only Which Is A Sticky Wicket

35

Constraints to CLCP Transmission Limits Can Corrupt The Timing, The Amount, And The Mix of

Capacity Additions

Must-Run Units Distort Simple Investment Sequence

Emission Limits Will Constrain Expansion Path Unless Monetized

Portfolio Standards Will Corrupt LC Mix

Poland – 25,000 MW Peak, Need To Replace 10,000 MW by 2010 – May Replace 5,000 MW

But Subject To These, It Is Still The Best Predictive Assertion

36

Illustrations of Problems Czech Case

– 2000 MW Nuclear Unit Is Last Addition – Accounts for about 16% of Total Generation in 2003.

– With 20% Rule or With Last Five Plants BM Would Be Mostly Nuclear and Some Coal For ROR Small Hydro

– In Fact, BM is Not Relevant For CR Until 2010 or Later– Displaced Future Plants Are Certainly Not Nuclear

US – Most Recent Plants are Merchant Gas CCs - Future plants are now Utility Coal

CLCP Is Always Alternating Between Fuels, Technologies, And Duty Cycles. Long Term Extrapolation of Recent Past Would Clearly Not Be the Least Cost Plan

37

Proposed Minimalist Solution Start With Load Vs Capability Analysis To See If BM Is Relevant For Period of

Interest – If no Capacity Change In That Period Only OM Matters.

Compare Optimum Mix With Existing Mix For Select Future Years To Establish Likely Sequence of Additions.

Establish Addition Sequence of Each Type (Peaking, Int, Base) by Year.

Find Least Cost Option For Each Type In Each Year

Determine What Type or Mix The Proposed Project Displaces

Say Biomass Displaces Baseload Coal in 2012. Then Use 100% Weight for OM through 2011 and 100% Weight for Coal Thereafter.

38

Optimum Mix AnalysisMW

Gas CT

Gas CCCoal

Cf

Gcc

Gct

1 2 3 4 5 6 7 8 9 Hours (000)

% Pk

90%

80%

70%

60%

50%

40%

30%

20%

10%

1 2 3 4 5 6 7 8 9 Hours (000)

Baseload

Peak

Intermediate

39

Generic Competitors By Type

Baseload – – Clean Coal– CC Gas or Oil– Other Renewables

Intermediate– Small Clean Coal– CC Gas

Peaking– CT Gas– Storage Hydro

40

Conclusions On State of the Art

Art Striving To Become Science CDM Meth Panel Process Is Sound But

– Should Retain Flexibility Based On Relatively Small Case Law Base– Should Continue To Broaden Case Base

Substantial Carbon Is Still At Issue Dialogue Is Expanding And Needs To Expand Much Further OM Rules Are Now Reasonable Although I Would Hope OM(c1) Will Be

Recognized To Avoid Need for Dispatch Data Three Year History Is Not Very Reliable Basis – Should Be Dynamic BM Is Much More Complex And Should Still Evolve 50-50 Weighting Is Weak Link And Should Not Be The Default