PI-BASED PERFORMANCE MONITORING SYSTEM FOR COMBINED CYCLE POWER GENERATION TECHNOLGY

1
PI-BASED PERFORMANCE MONITORING SYSTEM FOR
COMBINED CYCLE POWER GENERATION TECHNOLGY
4/21/2004
2
About Calpine

• Founded in 1984
• Headquartered in San Jose, Calif.
• Calpine has 87 energy centers in 21 states in the
  U.S., as well as in Canada and the United Kingdom
  with a total capacity of about 22,000 megawatts
• Uses clean, proven technologies natural gas
  combined-cycle and geothermal energy

3
A Large Portfolio

• Calpine offices
• Boston, MA
• Calgary, Canada
• Dublin, CA
• Ft. Collins, CO
• Houston, TX
• Jupiter, FL
• Northbrook, IL
• Portland, OR
• Tampa, FL
• Folsom, CA
• Corporate headquarters
• San Jose, CA

As of 4/21/2003
4
Growth History
5
Performance Monitoring Necessity

• Large, regionally diverse fleet requires central
  performance monitoring to provide
• the following
• Standard fleet reporting/comparison
• Operational benchmarking
• Asset improvement and degradation quantification

• Value
• Economic decisions based on operating performance
  data
• Internal knowledge and understanding of equipment
• Efficient and reliable operation

6
Why Use PI?

• Calpine chose to use OSI-PI as the foundation for
  our Performance Monitoring System
• Existing PI infrastructure eliminated additional
  capital expense
• Internal reliance and control
• PI provides all the required functional
  components of 3rd party systems
• Development team and end-users already familiar
  with PI tools/systems

• Value
• Zero additional capital cost associated with PI
• Zero new training
• Zero external reliance
• Zero risk

7
 
PERFORMANCE MONITORING SYSTEM ARCHITECTURE

• Consumable Products
• Wholesale Energy Products
• Retail Power
• Renewable Energy

• Overview of Calpine PI network
• PI Performance Equations
• PI Process Book as GUI
• PI Datalink and Excel employed as on-line QC

8
Calpine PI Infrastructure

• PI at Calpine
• Calpine standardized on PI as their operations
  historian in early 2001
• PI is installed at 86 of Calpines 89
  generating facilities
• The fleet PI infrastructure consists of 62
  production PI servers and 24 PI API nodes
• Calpine PI data for every site is available
  remotely to every Calpine employee, from plant
  personnel to executive management, because all
  Calpine PI systems are connected to the Calpine
  WAN

9
(No Transcript)
10
PI Performance Equations

• PI Performance Equation Subsystem used as
  backbone to calculate performance data
• Average of 350 Performance Equations per site
  constructed by Calpine Performance Engineers
• Calculations based on standard calculations as
  well as plant specific calculations
• Daily averaging functions used to easier
  identify performance trends and create summary
  reporting

11
PI Process Book as GUI

• PI Process Book used to view raw and calculated
  performance data
• 13 Monitoring Displays and 14 Trend Displays
• Multiple individual PDI files used as opposed
  to a single PIW file
• Navigation created using VB Code

12
C O M P OU N D A N N U A L G R O W T H R A T
E 1 9 9 7 T O 2 0 0 2
13
(No Transcript)
14
PI Datalink as On-Line Quality Control

• Data Quality Critical
• PI Datalink used to ensure data integrity
• Standard statistical evaluation of data used to
  evaluate data integrity
• PI Datalink used to summarize data and draw
  conclusions

15
 
TECHNICAL CHALLENGES

• Text Based Filtering in Performance Equations
• Getting the Most out of the Performance Equation
  Subsystem
• Correction Calculation Challenges

16
Text Based Filtering

• Identified a need to display why corrections
  were not being applied
• Solution must not affect summary data
• Early solution was to display -1 when
  corrections not applied

17
(No Transcript)
18
Text Based Filtering

• Solution
• Final solution was to display text when
  corrections not applied
• Text to be used in the tags must be present in
  the System Digital Set within the PI System
• Text is not taken into account when summary
  equations are applied

19
(No Transcript)
20
Maintaining the PI-PE Subsystem

• Found that 200 calculations is the approximate
  limit for a Performance Equation Scheduler
• Set up multiple schedulers to spread the load
• With the number of equations, and the
  complexity of the equations, Performance Equation
  Schedulers would fail from time to time
• Set up Performance Equation Schedulers to be
  restarted regularly to minimize impact.
• Used 3rd Party Software to notify when
  calculations are experiencing issues

21
Technical Challenges

• The Challenges
• Site instrumentation errors / lack of site
  instrumentation
• Each plant uses different tag naming conventions
• PI was originally configured primarily as a Data
  Historian PI tags typically have various levels
  of compression and exception enabled
• Gas Turbine based performance indicators vary
  drastically with load and ambient conditions

• The Solutions
• Correction, re-calibration, procurement of
  critical site instrumentation
• Standard PI tag naming convention applied to
  primary site tags (fuel, power, ambient
  conditions)
• PI compression and exception removed on critical
  performance tags filtering applied where
  required
• Load Normalization
• Apply OEM ambient corrections at normalized
  (base) load
• Use PI-based text substitution to allow
  functional averaging and Process Book trending

22
 
Performance Trend of CTG Corrected MW
23
Value Realization

• Examples
• Use PI based Performance Monitoring to conduct
  DWR contractual annual capacity tests on (15)
  simple cycle peakers in CA
• Quantify recoverable CTG degradation by analyzing
  pre and post offline water wash performance
• Remote monitoring identifies operational
  procedure mispractice and optimizes plant
  performance

• Value
• Eliminate dedicated on-site test personnel and
  temporary test instrumentation Estimated
  savings of 200K annually
• Accurate recoverable degradation data allows the
  sites to optimize CTG water washing based on
  market conditions
• In a low spark spread market, every optimization
  opportunity can significantly contribute to
  Calpines profitability

24
Performance Monitoring System vs. Performance Test
Calpine Simple Cycle Peaker Plant Performance
Test Results Comparison
Calpine 3 x 1 Combined Cycle Plant Performance
Test Results Comparison
The Performance Monitor calculations fall within
the 1 uncertainty band for power and slightly
outside the 1.5 uncertainty band for heat rate.
This is very good considering that the tests use
different instrumentation and the acceptance test
uses more measurements and more corrections in
their calculations.
25
Performance Quantification
26
Performance Quantification
27
Performance Quantification
CTG 2 Water Wash
CTG 1 Water Wash
CTG 2 Water Wash
CTG 1 Water Wash
28
Performance Reporting
29
Performance Reporting
30
Summary

• Calpine is successfully leveraging OSI-PI
  technology in a Performance Monitoring
  environment directly resulting in
• Increased plant availability and reliability
• Predictive maintenance scheduling
• A historical plant performance database that can
  be used to effectively optimize plant and fleet
  performance
• Better decision support data available to
  dispatch and trading
• INCREASED CALPINE PROFITABILITY !