abram gross yafeng peng jedidiah shirey...parameters: historic energy purchases, weather data,...
TRANSCRIPT
Abram Gross
Yafeng Peng
Jedidiah Shirey
Contents Context
Problem Statement
Method of Analysis
Forecasting Model
Way Forward
Earned Value
NOVEC Background (1 of 2) Northern Virginia Electric Cooperative (NOVEC) is a
distributor of energy to 6 counties in Northern Virginia.
Nearly 150,000 customers.
Mandated to meet all energy requests in pre-specified zones of obligation.
Primary means to service communities is through bulk energy market purchases.
NOVEC Background (2 of 2) Energy Purchases:
1) Bulk purchases via contracts 1 month to 3 years prior to delivery.
2) Spot purchases as needed to meet peak demand up to one day prior to delivery.
Problem Statement Warming trends have caused NOVEC to question
whether the current weather-normalization methodology is still the best available model.
NOVEC needs a new weather-normalization method that accounts for changing weather trends or a recommendation that the existing model is sufficient.
To what extent is the climate and its impact on energy demands changing?
Purpose & Objectives Purpose:
NOVEC needs to remove weather-effects from energy consumption to more accurately inform bulk purchases.
Explore and recommend a methodology to normalize monthly energy purchases.
Objectives: Characterize the relationship between economic
variables, weather, customer-base, and energy consumption.
Develop a normalization procedure to remove weather effects from energy demand.
Recommendation for improvement to current weather-normalization methodology.
Scope Data:
NOVEC monthly energy purchases data since 1983. Dulles weather data since 1963. Historic economic factors data since 1980s. 30 years of forecasted economic factors.
Model: Parameters: historic energy purchases, weather data, economics, and
customer-base. Predictions for energy consumption over a 30-year horizon. Regression: characterize dynamics between parameters. Weather-normalization: remove seasonal weather impacts on
NOVEC’s load. Forecast: facilitate testing of varied normalization methodologies. Ensure synergy with NOVEC’s existing suite of models
(regression, weather-normalization, forecast).
Key Definitions CDD: Cooling Degree-Day = MAX(ActualDegree – UpperBound, 0) * Duration.
HDD: Heating Degree-Day = MAX(LowerBound – ActualDegree, 0) * Duration.
Neutral zone: Upper and lower bounds for temperatures that do not impact load.
* Analysis will explore varied upper/lower bounds for temperature
30
40
50
60
70
80
Actual_Temperature Upper_Bound Lower_Bound
CDD
HDD30
40
50
60
70
80
Actual_Temperature Upper_Bound Lower_Bound
CDD
HDD
t
Weather Data Visualization (1 of 2) Historic weather data was
analyzed by average temperature per month to view trends.
Average temperature trending upwards from 1960 to the present for each month (graph is for July).
Weather Data Visualization (2 of 2) Variance was also evaluated and no change is seen.
Essential Elements of Analysis & Measures of Effectiveness
EEA 1: What is the rate of climate change? MoE 1.1: Seasonal trends and variability.
MoE 1.2: Rate of climate change.
EEA2: What econ-model best predicts NOVEC’s customer-base? MOE 2.1: Best subset of econometrics to gauge service demand.
MOE 2.2: Goodness of fit test for selected model.
EEA3: What impact does weather have on energy demand? MOE 3.1: Relationship between weather and load beyond base demand.
MOE 3.2: Changes in climate compared to per capita consumption.
MOE3.3: Goodness of fit test for selected model.
III. Forecast ModelII. Develop normalization methodologyI. Relate model parameters
Methodology
Residential
Commercial
CompositeCustomer-Base
GDP
Housing Stocks
Employment
Historic Power Demand
Monthly Demand
HDD Impact
CDD Impact
Remaining Power Demand
Weather-Normalization
Forecast
Assess Model
Basic Load
Economic Variables
Customer Combination
Per CapitaDemand
* Model forecast will be verified against 2012-2013 power demand and compared to existing model’s accuracy.
*
EEA1: Rate of Climate Change Historic weather data was analyzed to determine trends:
Seasonal fluctuations.
Overall rate of change.
Characterize uncertainty/variability.
0
200
400
600
800
1000
1200
1400
1600
0
1000
2000
3000
4000
5000
6000
19
63
19
65
19
67
19
69
19
71
19
73
19
75
19
77
19
79
19
81
19
83
19
85
19
87
19
89
19
91
19
93
19
95
19
97
19
99
20
01
20
03
20
05
20
07
20
09
20
11
De
gre
e-D
ays
Be
low
65
°F
Heating Degree-Days - 1963 to 2011
Total HDD
Average HDD
Max HDD
0
200
400
600
800
1000
1200
1400
1600
0
1000
2000
3000
4000
5000
6000
19
63
19
65
19
67
19
69
19
71
19
73
19
75
19
77
19
79
19
81
19
83
19
85
19
87
19
89
19
91
19
93
19
95
19
97
19
99
20
01
20
03
20
05
20
07
20
09
20
11
De
gre
e-D
ays
Ab
ove
6
5 °F
Cooling Degree-Days - 1963 to 2011
Total CDD
Average CDD
Max CDD
EEA2: Economics Model Model to associate economic factors to customer-base
in under development.
0
20
40
60
80
100
120
140
0
10
20
30
40
50
60
70
80
Aug-89 Aug-90 Aug-91 Aug-92 Aug-93 Aug-94
De
man
d (
Mill
ion
s -
kWH
)
Cu
sto
me
rs &
Ho
usr
ing
Sto
cks
(Th
ou
san
ds)
Residential Demand & Metro-D.C. Econometrics
Housing Stocks
Customers
Demand
EEA3: Impact of Weather Changes to per capita energy demand by customer
type: residential and non-residential.
0
2000
4000
6000
8000
10000
12000
14000
16000
Jan-05 Jan-06 Jan-07 Jan-08 Jan-09 Jan-10
Ener
gy C
onsu
mpt
ion
(kW
hr)
Monthly Customer per Capita Demand (2005 - 2010)
Non-Residential
Residential
HDD
CDD
Model Development Pre-processing: Excel Workbook with VBA Macros.
Automated computations of HDD and CDD.
Allows user-defined “neutral” boundaries for calculating.
Ingests data as currently maintained by sponsor.
Automated linear transformations for regression models.
GUI design facilitates some simple regression models germane to VBA; limited capability.
Launch and export conditioned data to R; expanded capability.
Statistical Modeling in R: Executable file for regression analysis and improved visualization.
Preliminary Results Variability in yearly weather patters doesn’t appear to change.
Climate change is more apparent in winter and summer than fall and spring.0
500
1000
1500
2000
2500
3000
3500
1963
1966
1969
1972
1975
1978
1981
1984
1987
1990
1993
1996
1999
2002
2005
2008
2011
Degr
ee-D
ays B
elow
65
°F
Winter Heating Degree Days
DEC
JAN
FEB
0
200
400
600
800
1000
1200
1400
1600
1963
1966
1969
1972
1975
1978
1981
1984
1987
1990
1993
1996
1999
2002
2005
2008
2011
Degr
ee-D
ays B
elow
65
°F
Spring Heating Degree Days
MAR
APR
MAY
0
20
40
60
80
100
120
140
160
1963
1966
1969
1972
1975
1978
1981
1984
1987
1990
1993
1996
1999
2002
2005
2008
2011
Degr
ee-D
ays B
elow
65
°F
Summer Heating Degree Days
JUN
JUL
AUG
0
200
400
600
800
1000
1200
1400
1600
1963
1966
1969
1972
1975
1978
1981
1984
1987
1990
1993
1996
1999
2002
2005
2008
2011
Degr
ee-D
ays B
elow
65
°F
Fall Heating Degree Days
SEP
OCT
NOV
0
2
4
6
8
10
12
14
1963
1966
1969
1972
1975
1978
1981
1984
1987
1990
1993
1996
1999
2002
2005
2008
2011
Degr
ee-D
ays E
xcee
ding
65
°F
Winter Cooling Degree Days
DEC
JAN
FEB
0
50
100
150
200
250
300
350
1963
1966
1969
1972
1975
1978
1981
1984
1987
1990
1993
1996
1999
2002
2005
2008
2011
Degr
ee-D
ays E
xcee
ding
65
°F
Spring Cooling Degree Days
MAR
APR
MAY
0
200
400
600
800
1000
1200
1400
1963
1966
1969
1972
1975
1978
1981
1984
1987
1990
1993
1996
1999
2002
2005
2008
2011
Degr
ee-D
ays E
xcee
ding
65
°FSummer Cooling Degree Days
JUN
JUL
AUG
0
50
100
150
200
250
300
350
400
450
1963
1966
1969
1972
1975
1978
1981
1984
1987
1990
1993
1996
1999
2002
2005
2008
2011
Degr
ee-D
ays E
xcee
ding
65
°F
Fall Cooling Degree Days
SEP
OCT
NOV
0
500
1000
1500
2000
2500
3000
3500
1963
1966
1969
1972
1975
1978
1981
1984
1987
1990
1993
1996
1999
2002
2005
2008
2011
Degr
ee-D
ays B
elow
65
°F
Winter Heating Degree Days
DEC
JAN
FEB
0
200
400
600
800
1000
1200
1400
1600
1963
1966
1969
1972
1975
1978
1981
1984
1987
1990
1993
1996
1999
2002
2005
2008
2011
Degr
ee-D
ays B
elow
65
°F
Spring Heating Degree Days
MAR
APR
MAY
0
20
40
60
80
100
120
140
160
1963
1966
1969
1972
1975
1978
1981
1984
1987
1990
1993
1996
1999
2002
2005
2008
2011
Degr
ee-D
ays B
elow
65
°F
Summer Heating Degree Days
JUN
JUL
AUG
0
200
400
600
800
1000
1200
1400
1600
1963
1966
1969
1972
1975
1978
1981
1984
1987
1990
1993
1996
1999
2002
2005
2008
2011
Degr
ee-D
ays B
elow
65
°F
Fall Heating Degree Days
SEP
OCT
NOV
0
2
4
6
8
10
12
14
1963
1966
1969
1972
1975
1978
1981
1984
1987
1990
1993
1996
1999
2002
2005
2008
2011
Degr
ee-D
ays E
xcee
ding
65
°F
Winter Cooling Degree Days
DEC
JAN
FEB
0
50
100
150
200
250
300
350
1963
1966
1969
1972
1975
1978
1981
1984
1987
1990
1993
1996
1999
2002
2005
2008
2011
Degr
ee-D
ays E
xcee
ding
65
°F
Spring Cooling Degree Days
MAR
APR
MAY
0
200
400
600
800
1000
1200
1400
1963
1966
1969
1972
1975
1978
1981
1984
1987
1990
1993
1996
1999
2002
2005
2008
2011
Degr
ee-D
ays E
xcee
ding
65
°FSummer Cooling Degree Days
JUN
JUL
AUG
0
50
100
150
200
250
300
350
400
450
1963
1966
1969
1972
1975
1978
1981
1984
1987
1990
1993
1996
1999
2002
2005
2008
2011
Degr
ee-D
ays E
xcee
ding
65
°F
Fall Cooling Degree Days
SEP
OCT
NOV
Way Forward & Risks Data visualization and preliminary analysis completed.
Website design completed. Still need to finalize content.
Model design/creation under construction. Major focus over next couple of weeks.
Excel calling R needs to be completed.
Customer transformation
Final model fitting
Forecast
Earned Value
0.00
10000.00
20000.00
30000.00
40000.00
50000.00
60000.00
Do
lla
rs
Earned Value
Earned Value
Planned Value
Actual Cost
CPI & SPI
0.00
0.50
1.00
1.50
2.00
2.50
3.00
3.50
4.00
Week 1 Week 2 Week 3 Week 4 Week 5 Week 6 Week 7 Week 8
Ra
tio
CPI and SPI
CPI
SPI
References www.novec.com
NOVEC 25th Anniversary History Film http://www.youtube.com/watch?v=2qfeOKnPPGg
“Improving Load Management Control for NOVEC”; Kozera, Lohr, McInerney, and Pane. http://seor.gmu.edu/projects/SEOR-Spring12/NOVECLoadManagement/team.html