Assessing the Effect of Wind Power on the Demand Charge Using Monte Carlo Analysis

1
Assessing the Effect of Wind Power on the Demand
Charge Using Monte Carlo Analysis

• Laura Hildreth
• 25 April 2006

2
Overview

• I Background Information
• A PURPA
• B Electricity Industry
• C Monte Carlo Analysis
• II Data
• A Description
• B Analysis
• III Outcome/Conclusion

3
Public Utility Regulatory Policies Act of 1978

• Passed in 1978 as part of the National Energy Act
• Aimed to decrease dependence on foreign oil and
  diversify energy generation
• Created a class of nonutility generators called
  qualifying facilities (QFs)
• Small power production facilities
• Cogeneration facilities

4
PURPA (contd)

• PURPA requires utilities to connect QFs to the
  grid and to pay QFs rates for their excess
  electricity to not exceed the avoided cost of the
  utility
• This has spurred competition and the development
  of renewable resources such as the wind turbine
  at UMM

5
Electricity Industry

• Components of the electricity bill
• Energy charge-flat rate charged for electricity
  consumed
• Demand (capacity) charge-rate charged for maximum
  amount of electricity used in a billing period
• Total 125 0.02737kwh 7.91kw
• Our concern is the demand charge
• The UMM wind turbine should decrease this charge
  but the magnitude is unknown

6
Example

• The graph on the right shows hourly demand for
  January 2003
• Our goal is to determine how much the highest
  peak can be shaved with a wind turbine
• In this example
• total kwh demand799881
• max kw1671
• Total 125 7.911671 0.02737799881
  35235.35

7
Monte Carlo Analysis

• Monte Carlo Analysis is a statistical technique
  that simulates possible outcomes that could occur
  under the given data
• These simulations may yield results that are more
  extreme than what has been observed

8
Hypothesis

• Adoption of a behind-the-meter wind turbine will
  predictably affect energy charges only if all the
  power is consumed on-site if not, the size of
  the effect is an empirical question due to the
  spread between retail and wholesale rates. The
  effect on capacity charges (demand charges) is
  non-positive, but otherwise unknown.

9
Data Description

• Two sources of data
• UMM (demand data)
• West Central Research and Outreach Center (wind
  speed datasupply data)
• The data were taken on hourly intervals for March
  2001 through February 2005

10
Demand Data

• The data suffers from double dependency
• This implies that an observations is dependent on
  both the day and the hour
• Both effects must be captured for the analysis to
  be accurate

11
Demand DataDaily Effect
There is a trend in the data This is caused by
school being in session
12
Demand DataDaily Effect (contd)

• ANOVA was used to show a significant difference
  between the groups
• A general linear model was run to determine the
  effect of school being in session
• demand ß0 ß1insession e

13
Demand DataHourly Effect

• The data also have a time series element
• Auto-regressive (AR) models were fit to the
  original data
• Dt a0 a1Dt-2 a2Dt-1 e

14
Demand Data (contd)

• Using Monte Carlo simulations, 100 potential
  months were generated for each 12 months
• The simulations used the following regression
• demand ß0 ß1insession e
• Time series models were then fit to these data to
  get 100 complete months for each 12 months

15
Demand Data (contd)
16
Data AnalysisSupply Data

• The wind speed at 3 meters and 50 meters was
  extrapolated to 80 meters (there are 2 sets of
  supply data)
• The power curve of the turbine was used to
  determine the power generated from these wind
  speeds
• The plot on the right shows supply data
• These data were fit with gamma distributions

17
Supply Data (contd)

• AR models were then fit to this data
• Monte Carlo simulations created 100 months
• Instead of simulating a regression, values were
  simulated that could come from a gamma
  distribution with the given parameters
• These simulations were then fit with time series
  models

18
Data Analysis (contd)

• Both demand and supply were truncated on the left
• All negative values were recorded as zero
• From this data the difference between demand and
  supply was found
• Difference3 Demand Supply3
• Difference50 Demand Supply50

19
Data Analysis (contd)

• The data for demand, difference3, and
  difference50 were truncated on the right
• The top 5 of observations were ignored because
  they are extreme
• The maximum value for demand, difference3, and
  difference50 was found for each 100 months

20
Outcome

• The graph shows the maximum value of the groups
• Notice that demand is higher than difference3 and
  difference50

21
Outcome (Average Differences)
22
Outcome (contd)

• Statistically the groups are different
• This is not surprising since the data was
  truncated which leads to small variance
• For most months, Difference50 was greater than
  Difference3
• Adoption of a wind turbine has the potential to
  significantly reduce the demand charge
• From the previous table, expected yearly
  reduction is
• 7.916588.899 52118.19 for difference3
• 7.917676.715 60722.82 for difference50

23
Limitations of This Analysis

• The supply data is obtained through several
  estimations
• Other time series estimates may be equally (or
  more) appropriate
• Other methods may be more appropriate, especially
  spatial models
• Potential for overestimation and underestimation

24
Sources

• Bird, Lori et al. Policies and Market Factors
  Driving Wind Power Development in the United
  States, Energy Policy, Vol 33 Issue 11, July
  2005, pg 1397-1407
• Blankinship, Steve. Can IPPs Play the Coal Game?
  Power Engineering, Vol 109 Issue 7, July 2005, pg
  28-32
• Boyd, James. The Regulatory Compact and
  Implicit Contracts Should Stranded Costs Be
  Recoverable?, Resources for the Future, October
  1996
• Braeutigam, Ronald. The Effects of Rivalry with
  Price Regulation of Electric Power Generation,
  Journal of Regulatory Economics, volume 11, pages
  119-137, 1997
• Brennan, Timothy J, et al. A Shock to the System,
  Resources for the Future, USA, 1996
• Brennan, Timothy J and James Boyd. Stranded
  Costs, Takings and the Law and Economics of
  Implicit Contracts, Journal of Regulatory
  Economics, Volume 11 pages 41-56, 1997
• Brennan, Timothy, Karen Palmer and Salvador
  Martinez. Alternating Currents Electricity
  Markets and Public Policy, Resources for the
  Future, USA, 2002
• Diggle, Peter. Time Series A Biostatistical
  Introduction, Oxford University Press, UK, 1990
• Energy Information Administration. The Changing
  Structure of the Electric Power Industry2000 An
  Update, October 2000, http//www.eia.doe.gov/cneaf
  /electricity/chg_stru_update/toc.htmlhttp//www.ei
  a.doe.gov/cneaf/electricity/chg_stru_update/toc.ht
  m
• Guey-Lee, Louise. Energy Information
  Administration, Renewable Electricity Purchases
  History and Recent Developments. 1998,
  http//www.eia.doe.gov/cneaf/solar.renewables/rea_
  issues/renewelec_art.pdf
• Hirsh, Richard. Power Loss, MIT Press, USA, 1999
• Hirst, Eric. How Stranded Will Electric Utilities
  Be?, Public Utilities Fortnightly, 15 Feb 1995,
  page 30
• Hirst, Eric and Jeffrey Hild. The Value of Wind
  Energy as a Function of Wind Capacity, The
  Electricity Journal, Vol 17, July 2004, pg 11-20
• Monney, Christopher Z. Monte Carlo Simulation,
  Sage Publications, USA, 1997
• Timney, Mary M. Power for the People, ME Sharpe,
  USA, 2004
• Vestas Wind Systems, last accessed 16 Feb 2006,
  http//www.vestas.de/pdf_ls/ok-1204-4926820VES20
  BL20V8220UK.pdf
• Woolf, Fiona and Jonathan Halpern. Integrating
  Independent Power Producers into Emerging
  Wholesale Markets, World Bank, December 2001
• Yajima, Masayuki. Deregulatory Reforms of the
  Electricity Supply Industry, Quorum Books, USA,
  1997
• Zucchet, Michael. Electricity Industry
  Restructuring Renewable Resources in a
  Competitive Environment, Congressional Digest,
  August-September 1997

25
Thank You

• Arne Kildegaard
• Engin Sungur

26

• Questions?

27
ANOVA and GLM Estimates
28
Demand AR Components
29
Supply Parameter Estimates

• Supply3 Data Supply50 Data

30
AR Supply Estimates

• Supply3 Data Supply50 Data

31
Outcome (Average Maximum Values)