Hydro Modeling Update April 7, 2020 Kevin Harris (CG) Production Cost Modeling Data Work Group (PDWG)
Hydro Modeling
Update
April 7, 2020
Kevin Harris (CG)
Production Cost
Modeling Data Work
Group (PDWG)
▪ Review Hydro Modeling
▪ Calculating Average Daily Operating Range
(DailyOpRange)
▪ Development of Columbia River
Assumptions
▪ Data Provided to WECC
2
Outline
Core slides are from PDWGAug 14, 2019 presentation
Review Hydro Modeling
3
▪ Options
• Hourly shape
• Proportional Load Following (PLF: K coefficient)
◦ Old method: User calc
◦ New method: User set average daily operating range by
month and PLF allocation (default 100%)
• Hydro Thermal Coordination (HTC: p coefficient)
◦ Old method: User calc
◦ New method: User set average daily operating range by
month and PLF allocation (default 100%)
• HTC allocation is 1 minus PLF allocation
4
Limited Energy Modeling
▪ PLF assumes Hydro generation the relative change in Hydro generation is equal to the relative change in load multiplied by a constant
▪ Calculation of K (the slope) can be simplified to
▪ GV can Calc K and p given
• Average Daily Operating Range (DailyOpRange) and PLF Allocation
• More info on calc can be found in spreadsheet Example_Hydro_K_and_p_Calc_v08.xlsx
5
Review Calc of K & p
𝐻𝑦𝑑𝑟𝑜 𝐺𝑒𝑛 𝐾′ =𝐻𝑦𝑑𝑟𝑜 𝐺𝑒𝑛 𝑅𝑎𝑛𝑔𝑒 (𝑎𝑣𝑔(𝑑𝑎𝑖𝑙𝑦max)−𝑎𝑣𝑔(𝑑𝑎𝑖𝑙𝑦 𝑚𝑖𝑛)) ∗ 𝑃𝐿𝐹 𝑆ℎ𝑎𝑟𝑒
𝐻𝑦𝑑𝑟𝑜 𝐺𝑒𝑛(𝑎𝑣𝑔)
𝐿𝑜𝑎𝑑 𝐾′ =𝐿𝑜𝑎𝑑 𝑅𝑎𝑛𝑔𝑒 (𝑎𝑣𝑔(𝑑𝑎𝑖𝑙𝑦max)−𝑎𝑣𝑔(𝑑𝑎𝑖𝑙𝑦 min))
𝐿𝑜𝑎𝑑(𝑎𝑣𝑔)
𝐾(𝑆𝑙𝑜𝑝𝑒) =𝐻𝑦𝑑𝑟𝑜 𝐺𝑒𝑛 𝐾′
𝐿𝑜𝑎𝑑 𝐾′
𝑫𝒂𝒊𝒍𝒚𝑶𝒑𝑹𝒂𝒏𝒈𝒆 = 𝑯𝒚𝒅𝒓𝒐 𝑮𝒆𝒏 𝑨𝒗𝒈 𝑫𝒂𝒊𝒍𝒚 𝑶𝒑 𝑹𝒂𝒏𝒈𝒆 (𝑨𝒗𝒈(𝒅𝒂𝒊𝒍𝒚𝐦𝐚𝐱)−𝑨𝒗𝒈(𝒅𝒂𝒊𝒍𝒚 𝒎𝒊𝒏))
How to Calculate
Average Daily Operating Range
(DailyOpRange)
6
▪ DailyOpRange: Average Daily Operating Range by month
• Something like ‘AvgDailyOpR’ would be a better label
▪ Included spreadsheet can be used to calculate DailyOpRange
• Example_Hydro_Calc_Hydro_OpRange_2019_0813.xlsx
• Contains 2018 data for Grand Coulee and Chief Joseph
7
How to Calculate Operating Range
▪ Data quality issues may exist with raw data▪ Review and correct any data issues▪ Data Issues: Abnormal spike in generation either
up or down• Max rating of Coulee 6,765 & Chief Joseph 2,456
8
Cleanup Raw Data
▪ Select the number of adjacent hour used to calc DailyOpRange for Hydro. • 1 hour: applicable for real-time runs
• 2 hours: 2 hour rolling average smooths out real time dispatch variability. Applicable for a day-ahead PCM
• 3 hours: 3 hour rolling average has additional smoothing of out real time dispatch variability. Applicable for a day-ahead PCM
9
Adjacent Hours for PLF/HTC
▪ Set GridView to use the
same adjacent hours for
PLF as used to calc the
“Avg Daily Op Range”
▪ Set in • ‘Simulation Control’ form
• ‘Generation’ tab
• Lower right
10
Adjacent Hours for PLF/HTC
▪ Grand Coulee for 2018• P max: 6,765 MW
11
Example Operating Range #1
▪ Lower Columbia (4 proj, 6,100 MW) for 2018• McNary, John Day, The Dalles & Bonneville
12
Example Operating Range #2
▪ Tab: DataIn_Hrly_Hydro• Import hourly hydro generation
• Assumes col 2 has the date/time for all data
• Row 5 is user set description of data
▪ Tab: Calc• Calc values used in “Results” tab
▪ Tab: Results• Charts monthly operations
◦ Use spinner Data1 and Adjacent hours
• Compare Two Shapes (Data1 & Data2)
◦ Use spinner Data2 and Month
13
Review DailyOpRange Spreadsheet
▪ PLF allocation set the percent of operating range allocated to PLF
▪ HTC allocation is equal to (1 – PLF Allocation)
▪ 1) Configure the use of “Hydro Dispatch Option” set to “Region Load – Wind – Solar”
• Set appropriate Wind & Solar Coefficient Factor by region
▪ 2) If LMP profile is a better match to Hydro operations shape than 100% PLF
• Lower PLF allocation to shift some DailyOpRange to HTC
• Iterate to best match Hydro operations
14
Selecting PLF Allocation
Development Columbia River
Assumptions
15
Issue▪ Over the years operation on the Columbia River
has changed based on various constraints: environmental, fish, recreation,...
▪ The court throughout the 2011 biological opinion and required re-do of operation to comply to operating constraints
Problem▪ How to model 2009 (2008 Jun-Aug) the latest
observed Columbia River operation?
16
Development of Columbia River Modeling
Assumptions
17
Key Columbia Rive Projects
▪ Average daily operating range show
considerable operation flexibility at all
generation levels (<2011)
18
Change in Operation <2011
▪ Average daily operating range is less than pre
2011 and as generation increase the operating
range decreases (2011+)
19
Change in Operation 2011+
▪ Are individual projects proportional to load:
Nov-2017? (Not always)
20
Individual Projects: Nov-2017
▪ Area aggregated projects proportional or in
line with load shape? (Overall Yes)
21
▪ Individual projects are not always
proportional to load
22
Individual Projects: Oct-2016
▪ Aggregated projects are proportional or in
lone with load
23
Aggregate Projects Oct-2016
▪ Chart Daily Max Gen vs. Avg Daily Gen
▪ Chart: Daily Min Gen vs. Avg Daily Gen
▪ Year-over-year operation are consistent operations
24
Calc Daily Operating Range #1
25
Calc Daily Daily Operating Range #2
Lower Columbia River
Upper Columbia River
▪ Assume average daily operating range
(DailyOpRange) is equal to calc range range
based on average daily generation by month
▪ Calc OpR for 2011-20189
• Include years where target monthly generation is
within the limits of the polynomial data
▪ Model avg OpR: Where current years carry
more weighted that later years
26
Calc Daily Daily Operating Range #3
▪ By aggregated system (Upper Columbia, Lower Columbia and Lower Snake River), calc average daily operating range (DailyOpRange):
• Set appropriate adjacent hours value for calc
• Calc average by month by year (Daily Max[(Gen(Mo(i))] – Daily Min[Gen(Mo(i)])
◦ Where Gen(Mo(i)) is in aMW
• Average by month OpR for system with weighting 2011-2018
• Calc a individual plants allocation based on its’ typical OpR divided by Aggregated typical OpR
27
Allocate Operating Range to Plant
▪ Its assumed the monthly daily generation has a normal distribution
▪ Assume Pmin equal to 10% probability of occurring (z:= 1.2816)• Avg Gen – StDev(Daily Gen) * -1.2816 (Z at 0.1)
• Include only years where target monthly generation is within +/- 3% of the polynomial data
• Average available min ratings with weighting
• Error check: Calc value > 0 and is less than min used in calc DailyOpRange
• Calc a individual plants allocation based on its’ typical OpR divided by Aggregated typical OpR
28
Calc Monthly Min Gen (Pmin)
▪ Assume Pmin equal to 90% probability of occurring (z:= 1.2816)
• Avg Gen – StDev(Daily Gen) * 1.2816 (Z at 0.9)
• Include only years where target monthly generation is within +/- 3% of the polynomial data
• Average available min ratings with weighting
• Error check: Calc value < P max and is greater than max used in calc DailyOpRange
• Calc a individual plants allocation based on its’ typical OpR divided by Aggregated typical OpR
• By plant add additional capacity for spin
29
Calc Monthly Min Gen (Pmin)
Data Provided to WECC
30
▪ Original hourly Hydro generation data
provided for 2008 & 2009
• From individual operator by plant
• CAISO provided hourly generation by aggregate
system (river system)
31
Provided Data
▪ Monthly max EIA-860
▪ Based on provided hourly Hydro generation data
• Monthly average daily operating range: Monthly Avg(Daily Max: Avg(Large(1), Large(2)) – Daily Min: Avg(Small(1),Small(2))
• Monthly generation and minimum gen
▪ Set GV “Adjacent hours for PLF” to:= 2
32
Data to WECC
Columbia River projects processed as described in section 3 “Development of Columbia River Assumptions“
▪ Values calc based on section: “Development
Columbia River Assumptions”
33
Data for Columbia River