AEC CAES PLANT DELIVERS ON ITS PROMISES

PLANT PERFORMANCE AND LESSONS LEARNED

Lars Andersson $ Dr. Michael Nakhamkin Dr. Robert Schainker

Energy Storage and Power Consultants, Inc. Electric Power Research Institute

200 Central Avenue 3412 Hillview Avenue

Mountainside, New Jersey 07092 Palo Alto, California 94304-1395

Lee Davis $ Robert Meyer $ Herman Williams $ John Howard

Post Office Box 550

Andalusia, Alabama 36420

Abstract

The 110-MW compressed air energy storage (CAES) plant at McIntosh, Alabama now provides the performance, operational flexibility, and reliability promised by its developers.

Key lessons from plant commissioning and its operation are provided, giving valuable information for the next generation of CAES plants and for any gas-turbine-based plant integrating multiple components, or the thermodynamic features of reheat, intercooling, and recuperation.

Introduction

Alabama Electric Cooperative’s (AEC) 110-MW 26-hour plant is the first CAES unit in the U.S. and the second in the world (after the 290-MW 4-hour plant in Huntorf, Germany). Compared with the German plant, this CAES plant includes several conceptual innovations, of which the recuperator is one, significantly affecting plant performance, operation, and controls. It is the first U.S. combustion-turbine plant with intercooling, reheat, and recuperation. Integration with an underground air storage and unique operation further distinguish the plant design.

During initial operations, AEC, Electric Power Research Institute (EPRI), Energy Storage and Power Consultants (ESPC), and the equipment suppliers concentrated on raising reliability and availability of the CAES plant to those of the commercial, simple-cycle, combustion-turbine fleet. Had replacements to failing parts been readily available, the maturing process would likely have duplicated that of any new combustion turbine introduction.

Figure 1 shows the plant with the 240 feet long turbomachinery hall being the prominent structure.

Construction of the AEC plant began in February 1988 with commissioning in May 1991. Initial performance testing and functional demonstrations verified compliance with guarantees, and proved that the plant delivers specified services. With time, and resolution of problems, expected reliability and availability were demonstrated.

EPRI co-sponsored development of the CAES plant, the first domestic unit of its kind, as a demonstration of the technology. ESPC as EPRI’s contractor, conducted technical supervision of plant engineering, construction, shop, and field testing with emphasis on the novel and conceptual aspects of the CAES design. Principal tasks for ESPC included:

•  
• Conducting plant optimization and integration of turbomachinery and other thermal system components with the underground storage
•  
• Developing analytical programs and procedures, such as recuperator and cavern transient analyses techniques
•  
• Developing CAES plant control philosophy, emphasizing part load, start, and other operations
•  
• Developing test programs, instrument requirements, and procedures, accounting for the thermal cycle and CAES operating specifics.

EPRI and ESPC carried out most tests that validated the guarantees given by the turnkey contractor. After commissioning, AEC assisted by EPRI and ESPC, launched a program to raise unit reliability and availability to those of other combustion turbines in peaking service. Under their guidance, ESPC detailed this program, performed root cause analyses, and established causes to the initially low operating indices.

The technical-support program during early operations focused on:

•  
• Verification that the CAES concept works and delivers expected performance
•  
• Verification of proper integration of the thermal cycle components
•  
• Verification that new features are properly designed, controlled, and operated
•  
• Instrumentation of key plant components and developing methods to improve plant reliability.

This paper describes: CAES plant operation, focusing on the value it provides to AEC’s power system dispatcher; performance statistics, describing operation, availability, and starting reliability; ongoing programs to monitor performance, and; key lessons learned from operations to date. A bibliography refers the reader to earlier publications, describing lessons learned during engineering and construction of the plant.

CAES Plant Description

There are two principal modes of CAES plant operation: During low power system demand (off-peak periods), electrical energy drives the compressor train to charge the underground air storage, and; during periods of high system power demand (on-peak periods), the compressed air is withdrawn from the air storage cavern, heated in the recuperator and combustors before expanding through the turboexpander, driving the electric generator for power generation. With this schedule, the plant levels the utility’s load, effectively increasing it at night and decreasing it during the day. It permits more efficient use of the coal-fired generating units in the AEC system.

Rated output is 110 MW and the underground air storage supports continuous generation at 100-MW load for 26 hours. Based on a load-management analysis by AEC, the compressor and turboexpander are selected so one hour of generation at 100-MW requires about 1.6 hours of compression to maintain mass balance in the air-storage cavern. Figure 2 depicts the plant heat-and-mass-balance diagram, with generation-mode parameters at rated load and compression-mode parameters at average cavern conditions. The plant includes many features unique to power plant design, including:

•  
• A turbomachinery train that comprises nine individual components, which are assembled into a single, rigid string
•  
• HP and LP turbines with reheat combustion
•  
• Multiple-section compressor train with intercooling
•  
• A recuperator that reclaims turbine exhaust heat for preheating of motive air
•  
• Integration of the turbomachinery and air system with the air storage
•  
• Unique operations with unprecedented transients
•  
• Synchronous condenser operation.

Performance Testing

All performance and cycle-related parameters were verified with a performance-test and operability-verification program. Extensive testing (fifty-six performance and functional tests) demonstrated compliance of plant-performance characteristics with those guaranteed by the turnkey contractor. Important performance guarantee verification tests included:

•  
• Heat rate and air consumption tests at six loads, ranging from 10 to 110-MW, and operating with either natural gas or distillate oil. Figure 3 shows very close correlation between tested plant-heat rates and performance guarantees, despite a recuperator effectiveness that is below the guarantees. The corrected plant-heat rate in the figure represents expected plant performance, with the recuperator operating as guaranteed. The difference between guaranteed and corrected plant-heat rates thus represent the performance margin available with this turbomachinery. Figure 4 shows guaranteed and tested air consumption, with and without the recuperator in service (plant design allows air-side bypass during operation, at the expense of higher heat rate). Tested air consumptions with the recuperator in service duplicate guarantees. It is better than guarantees with the recuperator bypassed.
•  
• A compressor train power consumption test, which at 48.9 MW meet guarantees.
•  
• An energy consumption test completely recharged the storage cavern. Tested results, accounting for available margins identified during the test, exceeded the guarantee by about 0.5-percent (guaranteed not to exceed 2,084 MWh).
•  
• A storage capacity test that supported the guarantee of 2,600 MWh of available, uninterrupted energy storage.
•  
• Recuperator performance tests at six loads. At rated load, performance was 71.8-percent effectiveness, below the guarantee of 75.0-percent. Mechanical integrity tests verified unimpeded thermal expansion, acceptable tube stresses, absence of in-service air-side corrosion, and that tubes will not crack in service (a problem encountered with vintage recuperators).
•  
• Emission tests for criteria pollutants which met project requirements using existing LP combustors (which are planned to be replaced in November of this year). To meet the specified opacity with liquid fuel, requires a fuel additive.

Operations

Operations demonstrated that this plant works without limitations on startup, shutdown, transfer between modes, etc. Since commercial operation began, about 152 GWh of energy has been produced (through July 1997).

The AEC system primarily serves residential customers, with one, very large industrial consumer. Consistent with the residential demand, the CAES plant operates during high load in the summer months. With two peaks occurring each day in the winter, the CAES plant is started to operate during the fast load changes for morning and afternoon peaks. During spring and fall, load factors are typically much lower. With its very rapid start-and-load characteristics (synchronizes in five minutes and loads normally to rated output in an additional seven minutes, or three and one-half minutes during an emergency), the CAES plant’s operating flexibility is unique among frame-sized combustion turbines. Capitalizing on this feature, it has frequently been started several times during a day, ramped to a high load, and after thirty minutes unloaded and returned to hot standby. On several occasions, this fast start capability operation supported AEC’s power system when problems occurred in base load generation units and where the CAES plant served as an emergency unit. It also offers flexibility to the dispatcher to pursue best costs when purchasing short-term power on the spot-market.

Availability

Although the CAES plant has two principal operating modes, availability of interest relates to the generation mode. A reliability, availability, and maintainability (RAM) analysis shows that 95-percent generation availability is attainable with a mature CAES plant, having proper spare parts support. A performance objective has been 90-percent availability.

Table 1 shows annual average availability-related indices for 1994 through 1996. Figure 5 shows twelve-months rolling average availability, by month in 1997, up to and including July. Availability in April reflects the annual scheduled outage of the unit. The forced outage rate for 1997 is near zero.

Since commercial operation begun, early forced outages significantly affected station availability. These outages were caused by:

•  
• HP combustor mechanical failures from low-cycle fatigue
•  
• LP expander rub from tight clearances
•  
• Explosion damage after liquid fuel was admitted in the absence of a flame
•  
• Cavern motive-air-flow liner collapse from a flow transient
•  
• Explosion after trapped liquid fuel was injected during a fuel purge sequence
•  
• LP combustor thermal damage following trip when natural gas expanded from manifolds in the absence of cooling-air flow.

All recurring forced-outage causes are identified and rectified, and the average generation-mode availability consistently reaches its target.

Generation-Mode Starting Reliability

The RAM analysis also showed that expected generation-mode starting reliability for a mature CAES plant could be 95-percent. Initially low, the monthly starting and running reliabilities improved in 1996, especially so during the second half of the year. Table 2 summarizes reliability-related indices for 1994 through 1996.

Figure 5 also depicts the starting reliability by month (average over the preceding twelve months) for 1997, up to and including July. During those seven months, 157 successful starts were made.

The initially low starting and running reliabilities improved as causes (to low reliability) were found and rectified. The low reliability (which does not relate to the CAES technology itself) was attributed to a conservative control system (the design includes more than 200 trip sensors) and to additional precautions by the turbine designer in view of the new technology, and the planned remote operation of the plant.

Improvements to these indices were accomplished by a combination of:

•  
• Reducing the single sensors capable of causing a trip. For remaining trip sensors, signal quality is verified, voting logic included where feasible, or corroborating measurements used to validate the need before triggering the trip. The latter is particularly significant for this turbomachinery, with its many bearings and vibration sensors capable of inducing a trip
•  
• Main-flame-acquisition system logic improvements to more reliably detect flame at all loads, and for both fuels. Fuel permissive now relies on a combination of scanner output and direct temperature measurements
•  
• Tuning of control logic, set-points, and dwell timers, especially as it relates to torch-flame
•  
• Differentiating between hot and cold inlet-air starting logic for better control of motive air flow (stoichiometric fuel-to-air mixture) during all light-off conditions
•  
• Indexing of combustors and injectors with respect to optimum light-off performance and intensified maintenance of flame-acquisition system, control logic, and air-valve positioning devices.

After correcting all these, the reliability indices are now approaching those of the simple-cycle, combustion-turbine fleet.

Lessons Learned

The following, selected lessons evolved from operations of this plant and are unique to this technology. These findings will be useful to subsequent CAES facilities:

• A new component was the HP combustor. Test rigs which replicate service conditions do not exist, so shop testing was conducted at reduced conditions. Field measurements confirmed shop test results for efficiency, flame characteristics and emissions. This proved that extrapolations of test results to real engine conditions are valid
•  
• Unique to the CAES plant is that it shuts down by closing both air and fuel valves, compared to only fuel valves in conventional combustion turbines (which provides an inherent flowpath purge through compressor coast-down flow). As a result, a CAES plant needs special purging provisions which were identified, designed, and tested. Particulars with respect to this are:

  Maintain air flow to support combustor cooling as remnant natural gas is expelled from manifolds

  Thoroughly purge injectors from remnant fuel to avoid coking from post air-isolation heat soak

  Drain low points in purge-air manifolds before purging liquid-fuel lines and injectors

  Limit the rate of liquid fuel purging to what can be combusted when entering the combustor

  Corrosion-protection coatings for in-line, carbon-steel components, in place of stainless steel, need to be erosion resistant, if selected

• Unrestricted discharge of cavern air may cause the cavern flow liner to collapse. All possible transients must be identified, effects evaluated, and appropriate designs provided to preclude this from happening
•  
• A small turbine-air control valve would improve control during combustor light-off and low-load operation. The existing, full-ported valve is difficult to consistently position accurately for low flow control
•  
• A recuperator, designed for the maximum cavern pressure, would simplify the motive-air-control system and would improve upon the time required to synchronize the generator with the grid (would not require pre-start pressurization of the recuperator and associated motive-air piping).

Plant Performance and Component Condition Trending

Initially, EPRI and ESPC developed a data-acquisition system (DAS) to support the plant performance testing. This system and software uses a personal computer, connected to the plant’s distributed control system through a read-only data port. It allows the user to specify plant process data and to archive that information on the PC’s hard disc, floppies, or magnetic tapes for later data manipulation and analyses using conventional spreadsheet programs.

For example, the data from which Figures 3 and 4 are based, was produced using this program, logging about 180 data points at one-minute interval during the thirty minute test periods. Limited by the speed of the logging computer (386-25 MHz), data collection capacity is about 500 measurements per minute.

Recently, the DAS software was reprogrammed to record data in support of trending performance and component conditions. About three hundred data points are archived at ten-minute intervals, around the clock. Again, conventional spreadsheet programs reduce the monthly data base to manageable proportions, calculate key indicators and display those indicators relative to their baseline equivalents for a number of load groups. The baseline data is generally from the initial performance tests, when equipment was in its new and clean condition.

As examples of these trends, Figure 6 shows compressor train power consumption and Figure 7 shows high-pressure expander efficiency, two key indicators of plant and principal equipment conditions. Additional trends which are routinely produced and evaluated include:

• Heat rate and air consumption
• Combustor performance
• Rotating component efficiencies (low-pressure turbine and compressor sections)
• Recuperator effectiveness and pressure losses
• Intercooler pressure losses and terminal temperature differences
• Journal bearing amplitudes, rotor position, and pad temperatures (relative to alarm setpoints).

The trends are compared with conventional limits for performance degradation. Inspection or predictive maintenance to restore performance is scheduled, when warranted.

True degradation is distinguished from false indications by manually reviewing coincident process parameters supporting the trend. Likely diagnostics are possible, also by reviewing combination of indicators. For example, review of the HP expander efficiency and supporting data can focus inspections toward fouling, seal wear, bowed nozzles, damaged blades, etc.

For example, Figure 6 reveals that specific motor power and energy consumption have increased over time, while cavern pressure remained relatively constant, an effect which was traced to a sticking air-system valve, increasing system losses substantially. Figure 7 shows that efficiency visibly degraded over the last year and an inspection of that component has been scheduled, in part because of findings from this trend. The upper graph in the figure is the monthly trend of relative performance. The lower graph in the figure compares monthly averages with those recorded six and twelve months ago. For convenience, data is presented by load into three groups. Stability of the measurements is represented by the stock-type graph, where the vertical line represents plus and minus one standard deviation of the monthly measurement population, with the horizontal line being the average. A long vertical line thus represents unstable data, frequently the case from many starts and short runs. A short line suggests steady-state operation. The vertical bars in the bottom of the figure are the average load, expressed as a percentage of the maximum load for each group.

Conclusion

Initial operation preceding commercial acceptance demonstrated successful integration of cavern, turbomachinery, motive-air system and control philosophy and that the plant worked as specified with respect to starts, load ramps, transfer between operating modes, etc.

An extensive performance test and functional demonstration program also showed compliance with guarantees given. With time, plant availability and starting reliability were also demonstrated.

Consequently, Alabama Electric Cooperative’s CAES plant now delivers on its promises, which include:

• Performance as guaranteed
•  
• Operations as guaranteed
•  
• Availability consistent with the simple-cycle, combustion-turbine fleet
•  
• Unit reliability (starting and running), approaching that of simple-cycle combustion turbines.

As a result, the 110-MW CAES-plant design now represents a matured technology, and by applying the lessons learned with the AEC plant, future units should be capable of delivering expected reliability and availability at time of commercial operation.

Bibliography

Operating Experience and Lessons Learned at Alabama Electric Cooperative’s 110-MW 26-Hour CAES Plant, Presented at the Power-Gen Conference, Anaheim, CA 1995.

110-MW-26Hr CAES Plant Performance Test Results and Initial Reliability Indices, Proceedings of the 55th American Power Conference, Chicago, IL. 1993.

First U.S. CAES Plant Initial Startup and Operation, Proceedings of the 54th American Power Conference, Chicago, IL 1992.

History of First U.S. Compressed-Air Energy Storage (CAES) Plant (110 MW 26h), EPRI Publication TR-101751, Palo Alto, CA:

•  
• Volume 1 Early CAES Development, Final Report December 1992
•  
• Volume 2: Construction, Final Report April 1994
•  
• Volume 3: Part 1; Operations and Lessons Learned Testing, Operations and Maintenance, and Overview, Draft Final Report February 1997 (unpublished).

Standard Compressed-Air Energy Storage Plant: Design and Cost, EPRI Publication TR-103209, Palo Alto, CA., Final Report March 1994.

CAES Plant UNIRAM Reliability Prediction Model, EPRI Publication TR-102563, Palo Alto, CA., Final Report September 1993.

Compressed Air Energy Storage -- Combustor Development and Operation, Rural Electric Research (RER) Project 86-5, Final Report January 1996.