1. Heat Rate Calc based on CEMS Data April 12, 2016
Kevin Harris, ColumbiaGrid
Paul Deaver, CEC

2. Overview Big Picture Data Source: CEMS Data Filtering Data
Station Service
Review IO Curves
Calc Generic IO Curve
Adjusting Full Load HR
Processing GT
Issues/Lesson Learned

3. Big Picture Purpose: To develop heat rate curve for PCM
Data Used: EPA CEMS data
Procedure:
Filter EPA CEMS data
Calculate IO Curve: 2nd order polynomial
Take into account station service by scale CEMS heat points to modeled heat points
Calc Avg Heat Rate: 𝐹𝑢𝑒𝑙 𝑖𝑛𝑝𝑢𝑡 𝑎𝑡 𝐶𝐸𝑀𝑆 𝐻𝑒𝑎𝑡 𝑃𝑜𝑖𝑛𝑡𝑠 𝑀𝑜𝑑𝑒𝑙𝑒𝑑 𝐻𝑒𝑎𝑡 𝑃𝑜𝑖𝑛𝑡𝑠
Error Check Results

4. CEMS Data EPA Continuous Emission Monitoring System (CEMS)
Track flue gas to provide information on compliance with emission standards
Processed hourly data from (~43,824 data points/unit)
Processed data for 287 units: 71 ST-Coal, 64 ST-NG, 83 GT and 69 CC
EPA CEMS Data can be found at:
ftp://ftp.epa.gov/dmdnload/emissions/smoke/
ftp://ftp.epa.gov/dmdnload/emissions/hourly/monthly/

5. Review both the IO Curve and Histogram when making filter decisions
Filtering Data
Why filter data: Noise/unreasonable operating points
Filtering is a multi pass process:
Pass 1: Used only whole hours of operation
Pass 2: X-Axis Gross Capacity
Max operating limit: Based on EIA-860 rating + station service + buffer
In some cases physical data did not exceed EIA-860 rating
Min operating limit: Based on observed lower operating limit in histogram
Pass 3: Y-Axis Fuel Burn
Iteration 1: Filter based on +/- IO Curve from Pass 2
Iteration 2: Based on calc IO Curve from Pass 3, iteration 1
Review both the IO Curve and Histogram when making filter decisions

6. Station Service
There are no generic rules when determining station service
Station Service was based on WECC 2015 power flow Case 25HS1a as follows:
ST-Coal: Individual units station service was used
ST-NG: Most of the ST-NG unit are retired by Value from WECC 2014 power flow Case 14LSa1e was used. The average of 12 units is 1 MW/unit. Used Zero
GT: For all practical purpose zero
CC: Values were all over the place for F-Frame CC with the following stats
Avg: 5.74 MW, Min: 0, Max 20.4
StDev 4.55 [StDev is 80% of Avg]).
Calc generic Station Service for F-Frame CC has little impact on CEMS calc full load HR. Used generic IO Curve for CC with CC 260 MW CC w/station service and a full load HR of This resulted in a HR increase of 0.049%. To be conservative use a 0.1% increase.

7. X-Axis (Filter, IO and Heat Points)
On the X-axis we have three primary components
Filter Points: Upper and lower point to filter CEMS data
IO Curve Points: Points on the X-axis to calculate fuel burn for the IO Curve
Heat Points: Modeled heat points
Max rating is based on the EIA-860 Sum/Win rating. The user selects Sum, Win or Avg(Sum/Win) rating based on available CEMS data
Station Service: IO Curve Point(Max) – Heat Point (Max)
Heat Points are scaled IO Curve Points based on ratio of:
Heat Point(Max)/IO Curve Point(Max)
Typically the following relationship exist:
Max Rating: Filter Points >= IO Curve Points >= Heat Points
Min Rating: Filter Points <= IO Curve Points <= Heat Points
The Y-axis filter excludes data outside the given +/- range around the previous Polynomial IO Curve
Given this filter will change the IO Curve this is iterated a minimum of twice to minimize any potential issues.

8. Example Filter: San Juan 4
San Juan 4 – ST-Coal New Mexico
Sum/Win/Min Rating: 507/507/150
Filter Points:
IO Curve Points: 275 – 550
Heat Points: – 507
Implied Station Service 43 MW
X-Axis Filter: 275 – 550
Y-Axis Filter: +/- 550 from Poly

9. A separate spreadsheet contains additional examples with charts
Example Heat Rate Calc
A separate spreadsheet contains additional examples with charts
2016_0419_HR_Examples_v01.xlsx

10. Example: Fort St Vrain GT
Fort Saint Vrain - GT in Colorado
EIA Sum/Win/Min Rating: 145/160/73
IO Curve Points = Heat Points implies no station service
Note: X-Axis Filter, clear IO trend to winter rating. Winter rating 160 is used over summer rating of 145
X-Axis Filter: 65 – 162
Y-Axis Filter: +/- 75 from Poly

11. Example: Magnolia CC Magnolia – CC in LADWP – 9/2005
EIA Sum/Win CT Rating: 161/165
EIA Sum/Win CC Rating: 301/305
Rule of Thumb: CT are 2/3 of CC implying CC should be 241.5/247.5
Assume EIA reported CC is CC+DB
Used CC Sum/Win Rating 242/248
Assume DB > MW
X-Axis Filter: 175 – 242
Y-Axis Filter: +/- 50 from poly

12. Example: Gila River CC 1a
Gila River 1a – PV – 7/2003
EIA Sum/Win CC Rating: 257.5/276.5
EIA Sum/Win CT Rating: 146/163
Heat Points: – 257.5
Implied Station Service 0 MW
Assume CC+DB: 295 DB
X-Axis Filter: 170 – 259
Y-Axis Filter: +/- 220/100 from poly
DB?

13. Error Check
If one of the following conditions is True the generic heat rate shape is used with calc full load Heat Rate
Incremental hear rates are always monotonically increasing: iHR(2) < iHR(3) < iHR(4)
Heat Rate (1) is greater than Heat Rate (Full Load): aHR(1) > aHR(Full Load)

14. Generic Steam IO Curve Unitize calculation for generic Steam Heat Rate
Criteria for selecting Steam units: four heat point without errors
y:= 𝑥 𝑥 2
𝑅 2 =0.9984

15. Generic CC IO Curve Unitize calculated for generic CC heat rate
Criteria for selecting CC units: four heat point without errors

16. Adjustment to Full Load Heat Rate
Full load hear rate were reviewed and adjusted if needed
Aligned or average sister units to same HR
Example Align: N Valmy unit 1 had a 9.4 HR while unit , changed unit 1 to 10.1
Example Avg: El Segundo 1a and 1b have a wide spread between the two full load HR but operating hours are about the same. Modeled HR were the avg of unit 1a and 1b.
For CC maintained a spread between technology (E-Frame, F-Frame, G-Frame, and H-Frame)

17. Processing GT In general CEMS data for most GT was thin
Processed CEMS data to obtain:
Full load heat rate
Operating range
Exception: NV Energy Clark GT were either on or off
Assume min rating is 55% of max
Model with two heat points and flat heat rate shape

18. Issues/Lessons Learned
A 3rd order polynomial resulted in additional error, therefore use a 2nd order
Additional filtering may be needed for 3rd order polynomial
What’s station service? This is still an open question
Units with a GT splitting out CEMS data by season improve IO Curve.
Duct Burner heat rate need improvement