A major independent power producer's (IPP's) merchant fleet includes dozens of gas-fired combined-cycle plants of 3 x 1, 2 x 1, and 1 x 1 configurations. Nearly all have triple-drum heat-recovery steam generators (HRSGs) and similar steam cycle designs. The HRSGs, representing every major supplier, operate at three pressures: low (50 to 150 psi), intermediate (500 to 900 psi), and high (1,500 to 2,200 psi). For steam cycle makeup, virtually all plants use demineralized water with a resistivity higher than 10 megohms and a silica content of less than 10 ppb.
Given these fleetwide commonalities, one would expect individual plants' chemistry feed, monitoring, and control systems to be similar as well. But that was not the case five years ago. Individual plants' steam chemistry regimens were developed largely through evolution and trial and error. Parallel evolutions did produce some commonalities in treatment programs (all plants used phosphate, amine, and oxygen scavengers), but significant differences existed in feed systems, chemicals, sample panels, and operating philosophy.
Lacking standards, plant personnel and chemical vendors tended to use what worked at their last facility or what was used at nearby facilities. These varied approaches generally provided adequate results, but the lack of consistency in program specification created a hodgepodge of control limits and operating philosophies. Individual plants, lacking standards, often found it difficult to focus on those core parameters necessary to ensure long-term equipment reliability. Further, the varied programs made it extremely difficult for corporate or regional water specialists to troubleshoot problems at the plant level or to detect problems common to many plants.
Most plants used the chemistry monitoring software provided by their specialty chemical suppliers. Although these software packages usually performed acceptably, the lack of consistent monitoring packages made it virtually impossible to compare plants' performance. Data entered into these programs disappeared into the "vendor universe" and were lost to the IPP. Data extraction was difficult and time-consuming, and most of the data could only be examined locally using proprietary software. Specific plant information wasn't available to the rest of the company. This lack of shared data and information required each plant to reinvent the wheel. Each problem found was solved at the plant level; there was no easy method of communicating common problems and solutions to the fleet.
Most plants also relied on chemical suppliers for limits specification, troubleshooting charts, and control procedures. Though most vendors were diligent in providing this information, it often lacked the plant specificity required to make it useful to operators.
In addition, troubleshooting information often was lost in vendor service reports or languished in forgotten binders scattered about the plant. Plants relied primarily on "tribal knowledge" to combat chemistry monitoring and control problems.
For example, sample and analysis panels—the first line of defense in problem detection—often were abandoned in place because they "couldn't be maintained." Operators had low confidence in the analyzers and relied on wet tests alone to monitor and control chemistry. That being the case, problems were detected only by wet tests. Chemistry or pump problems often went undetected for many hours (the time between tests). In several cases, HRSG pH values dropped to extremely low levels and remained there for several hours prior to detection.
Most plants started with a maintenance schedule for sample panels, but many were abandoned early in commercial operation. Instrumentation and controls personnel, lacking familiarity with water analyzers, postponed maintenance. Thus, the sample panels became useful for obtaining grab samples only. Online data still went into the plant's data acquisition system, but alarms weren't active and readings were generally ignored.
As a result of these problems, plants had difficulty maintaining chemistry within limits or—worse—established limits that failed to adequately protect equipment. Operators wasted time and effort monitoring non-core parameters and performing frequent chemical testing at the expense of more important monitoring. Upsets were more common, took longer to detect, and resulted in greater damage than necessary.
Operator testing does have a role in monitoring, but it should not be the first line of defense in problem detection. Wet testing can find a problem, but only if it exists when the sample is taken. The rest of the time, the plant is "flying blind." For most plants, the averages appeared to be within target ranges (see table), but variability was extreme, and many target ranges didn't provide optimal equipment protection.
graph
Source: Dan Sampson
Typical chemistry of treatment programs "before"
Email
Comments
Print
Reuse
