Recommended Hydro Improvement April 11, 2017 A Continuation of 3/14 & 4/4 Meetings Kevin Harris, ColumbiaGrid TEPPC\Model Work Group - Chair

Overview Review How to Calc K and p Coefficients Modeling of the Columbia River Modeling Changes for the Columbia River Propose Lowering HTC on the Columbia River Hydro Dispatch Against Load – Wind - Solar Convert Select Daily Dispatchable Hydro from Fixed Hourly Shape to PLF

Approved Items

HTC p Factor Review of the 2026CC results show an input error GridView support two methods for evaluating HTC: Rectangle and Triangle method Rectangle Method divides by 4 Triangle Method divides by 2 Issue: P factor in the 2026CC are divided by 4. This reduces the HTC dispatch range by 50% Propose WECC: Correction all “p” factors to correspond to ‘Triangle Method’ by multiplying by 2

Castaic PS Current PS efficiency 85%, 75% and 65% Notes from CEC the average operational PS efficiency is 70% Proposed WECC lowering PS efficiency of Castaic to 71%, 66% and 61% (and adj the benefit cost ratio to 1.0)

Is Oroville and Thermalito PS No use at Edward Hyatt (Oroville) All 6 units modeled as PLF/HTC Duplicate units exist for Thermalito 2-4. It is modeled as a hourly shape and PLF Propose WECC consolidate modeling as Thermalito as PLF only

Aggregate Hydro Units by Plant I have a scenario that aggregates the Core Columbia river plus Lower Snake River units by plant This eliminates 197 units from the dataset ABB is no issues with this aggregation in the PCM or the round-trip process Propose WECC apply scenario that aggregate Hydro by plant when they become available or when WECC staff has time to create them ColumbiaGrid will provide a scenario to aggregate the Core Columbia and Lower Snake River

Review How to Calc K & p

How does PLF Work Proportional Load Following (PLF) assume the relative change in Hydro generation is proportional to a relative change in load The reference frame for load and Hydro generation is average monthly value

Calc Hydro K Avg daily operating range by month (MW) The difference between typical daily max minus min generation Max used the average of the 3 highest values/day Min used the average of the 3 lowest values/day Avg monthly generation (aMW) 𝑯𝒚𝒅𝒓𝒐 𝑲′= 𝑯𝒚𝒅𝒓𝒐 𝑮𝒆𝒏(𝒂𝒗𝒈 𝒅𝒂𝒊𝒍𝒚 𝐦𝐚𝐱 −𝒂𝒗𝒈 𝒅𝒂𝒊𝒍𝒚 𝐦𝐢𝐧⁡) 𝑯𝒚𝒅𝒓𝒐 𝑮𝒆𝒏(𝒂𝒗𝒈) 𝑯𝒚𝒅𝒓𝒐 𝑲′= 𝑯𝒚𝒅𝒓𝒐 𝑮𝒆𝒏(𝑶𝒑𝑹𝒂𝒏𝒈𝒆 (𝑴𝑾)) 𝑯𝒚𝒅𝒓𝒐 𝑮𝒆𝒏(𝒂𝒗𝒈) 𝑲(𝑺𝒍𝒐𝒑𝒆)= 𝑯𝒚𝒅𝒓𝒐 𝑲′ 𝑳𝒐𝒂𝒅 𝑲′ Load K’ is calc in the same way as Hydro K’

c p ∗C=𝑂𝑝𝑅𝑎𝑛𝑔𝑒(𝑀𝑊)*HTC(Share) → p= OpRange(MW)∗HTC(𝑆ℎ𝑎𝑟𝑒) 𝑐 𝑝 ∗𝐶 Calc HTC p HTC reshapes a share of PLF Hydro generation based on LMP The original equation for HTC Where: C – Plant Capacity A – Plant Operating Range (OpRange) This assumes a 50/50 split between PLF/HTC hence ½ Re-writing where HTC(Share) + PLF(Share) = 1 (Share of operating range) A/2:= Range of Plant Gen for HTC:= OpRange(MW)*HTC(Share) 4:= c(p) = Constant based on HTC method used: Rectangle method c(p) = 4 (2 passes with +/- = 4) and Triangle method c(p) = 2 (1 pass with +/- = 2) 4pC = 𝐴 2 → p= 𝐴 8𝐶 c p ∗C=𝑂𝑝𝑅𝑎𝑛𝑔𝑒(𝑀𝑊)*HTC(Share) → p= OpRange(MW)∗HTC(𝑆ℎ𝑎𝑟𝑒) 𝑐 𝑝 ∗𝐶

Calc K and p Consolidate K and p Calc where OpRange = Average Daily Operating range for Hydro and Load Avg monthly Hydro generation and load PLF(Share) + HTC(Share) = 1 (Share of operating range) c(p) = If HTC method is ‘Rectangle’ = 4 or 2 of ‘Triangle’ Max rating: Modeled (MW) Min rating: Modeled [ROR] (MW) 𝑲(𝑺𝒍𝒐𝒑𝒆)= 𝑯𝒚𝒅𝒓𝒐 𝑲′ 𝑳𝒐𝒂𝒅 𝑲′ 𝑯𝒚𝒅𝒓𝒐 𝑲′= 𝑯𝒚𝒅𝒓𝒐 𝑶𝒑𝑹𝒂𝒏𝒈𝒆 ∗𝑷𝑳𝑭(𝑺𝒉𝒂𝒓𝒆) 𝑯𝒚𝒅𝒓𝒐 𝑮𝒆𝒏(𝒂𝒗𝒈) 𝑳𝒐𝒂𝒅 𝑲′= 𝑳𝒐𝒂𝒅 𝑶𝒑𝑹𝒂𝒏𝒈𝒆 𝑳𝒐𝒂𝒅(𝒂𝒗𝒈) 𝒑= 𝑯𝒚𝒅𝒓𝒐 𝑶𝒑𝑹𝒂𝒏𝒈𝒆 ∗𝑯𝑻𝑪(𝑺𝒉𝒂𝒓𝒆) 𝒄 𝒑 ∗𝑴𝒐𝒅𝒆𝒍𝒆𝒅 𝑴𝒂𝒙

Modeling the Core Columbia River

Operational Change Starting in 2011 Slide was originally present 11/3/2015 Starting in 2011 the annual average daily operating range decrease by 2,224 MW (38%) Any forecast run should reflect this reduction in operational flexibility Take-a-way: To calc 2009 Hydro the OpRange needs to be calc based on post 2011 operation (2011-14 used)

Example Historic Operations Slide was originally present 8/31/2015 Example operation for January 2010 through 2013 Aggregated and some individual projects Hydro generation are proportional to load Take-a-way: Hydro gen from individual projects are not always proportional to load but the aggregate Net Columbia, Upper Columbia and Lower Columbia are Load Used:= 100% of BPA + 100% of MidC + 6% of CAISO Per unit of generation the operating range of Upper Columbia is greater than the Lower Columbia

Summary of Calc Procedure Columbia River operation were split into aggregate system (Upper & Lower Columbia) OpRange: Calc based on average OpRange based on 2011-14 operation for 2009 monthly generation w/2008 used for Jun-Aug Hydro K’ calc based on allocated OpRange *PLF(Share) and avg monthly generation Calc p based on allocated OpRange *HTC(Share)

Modeling Changes for Columbia River

Provided to WECC All Core Columbia River units converted to PLF Hydro data for 2009 w/ 2008 used for Jun-Aug. When appropriate a PLF/HTC split of 50/50 Calc based on region load equal the sum of 100% of BPA, MidC (CHPD, DOPD, GCPD) + 5% CAISO Provided: Annual load weighting of BPA 60.7%, CHPD 4.3%, DOPD 2%, GCPD 5.9% + CAISO 27.1%

Current Modeling of Columbia River Provided monthly PLF/HTC variable for all plants None of the fixed hourly shapes were converted to PLF This results in a mismatch of input assumption used in the dataset Propose WECC completing the modeling change on the Core Columbia by changing the fixed hourly shapes to PLF

Hydro Dispatch to Multi Regions Hydro dispatch to multi regions uses a scaled unit against the region load This makes it difficult to calc an appropriate K when load factor of the regions are not similar A clean relationship exist of historic Hydro operation to BPA load. See posted file ‘ Columbia Riv Gen HTC PLF Compare v01’ Propose WECC eliminate multi regions assignment for NW entities with regions outside of the NW

Propose Lowering HTC on the Columbia River

Core Columbia with 50/50 PLF/HTC From posted spreadsheet ‘ Columbia Riv Gen HTC PLF Compare v01’ 50% HTC shifts morning ramp energy to afternoon ramp in most months PLF only misses the afternoon ramp in when NW peaks in the morning (Mar) 20% HTC is a reasonable compromise that balances 50% HTC with 0% HTC (100% PLF)

Core Columbia with 50/50 PLF/HTC Posted spreadsheet ‘ Columbia Riv Gen HTC PLF Compare v01’ compares average weekday Hydro generation shape vs. historic Hydro operation and load For both absolute and unitized values Overall the 50/50 PLF/HTC results in a significate shift in avg hourly generation shape when compared to historic Propose WECC lowers PLF/HTC modeling assumption on the Core Columbia River from 50/50 to 80/20 Long term: Work with BPA to determine what an appropriate Hydro flexibility exist and provide to the HMTF with updates when new information becomes available

Hydro Dispatch Against Load – Wind - Solar

Hydro Dispatch Load - Wind Slide was originally present 8/31/2015 Relative to load, wind generation serves up to 64% of BPA daily load in 2014 or 20% of annual load Changing from “Load” to “Load – Wind” increases the deviation in daily a factor of 2.7 The expanded daily StDev directly impacts the daily allocation of Hydro generation. This impact can be amplified when K is > 1 Installed Wind Capacity in BPA for 2014 4,515 MW

Summary Load – Wind for BPA Issues occur if wind/solar is intermittent and a high percentage of area load The relationship for K breaks down

Region Load – Wind - Solar Conceptually Hydro is dispatched against net load In CA daily wind generation is stable relative to PNW wind In CA daily solar generation is stable relative stable except for some days in the winter intra monthly allocation of load – Solar should not be a problem

Hydro Dispatch Propose WECC test or implement the use of “Region Load – Wind –Solar” with the following setting: In all case BPA Wind Coefficient Factor:= 0.0 (0%) Option 1: Solar/wind Coefficient Factor of 50/50% Option 2: Solar/wind Coefficient Factor of 100/50% Option 3: Solar/wind Coefficient Factor of 100/100%

Convert Select Daily Dispatchable Hydro from Fixed Hourly Shape to PLF

Hourly Shape Hydro Dispatch Previously (3/14/2017) we covered the advantages of converting Hydro that has daily shaping capability from fixed hourly shape, based on 2005, to PLF The primary advantage is when Hydro is dispatching against “Load – Solar” the daily peak generation shifts in response to the net load The posted spreadsheet from the 3/14/2017 meeting was in PDF format This format is difficult to review

Hourly Shape Hydro Dispatch Propose the following: Convert the listed units in “Convert Fixed to PLF 2026CC1.7 v02.xlsx” from fixed hourly shapes to PLF For any plant the owner/operator disagrees with will not be converted

Other Hydro Suggestion? harris@columbiagrid.org (503) 943-4932