TURBINE PROTECTION SYSTEM PERFORMANCE by Margaret C. Campbell Project Engineer Foster-Miller, Incorporated Waltham, Massachussetts based control, the majority of the installed turbine protection As a Project Mechanical Engineer with systems rely on the high pressure pneumatic EHC system. To Foster-Miller,Incorporated, in Waltham, supplement the protection afforded by the this system, most Massachusetts, Margaret C. Campbell suppliers also provide a comprehensive package of turbine provides engineering and managerial sup­ supervisory instrumentation (TSI) systems. These TSI systems port to power industry related programs. can be designed, at the request of the customer, to initiate pro­ Her work has included reliability studies of tective actions. nuclear turbine protection systems,design Power plant experience indicates that the turbine protection of pneumatic systems fo r robotic controls, systems often initiate turbine trips when operating conditions and designof innovative naturalgas piping do not warrant a trip. In Table 1, the Institute of Nuclear Power systems. Operations (INPO) reports the listing of subsystems and compo­ Before joining Foster-Miller, Ms. Camp­ nents that are typically associated with turbine trip events [2]. bell was employed with Stone and Webster Engineering Corpora­ tion,where she was a Systems engineer involved in such activities Table 1. Subsystems and Components TypicallyAssociated with as the startup of auxiliary boilers and associated systems at the Thrbine Trip Events. Millstone III Nuclear Power Plant. Percent ofTotal Ms. Campbell received a B.S.M.E. degree in 1982,from the Subsystem Turbine Trip Events University of Massachusetts. EHC System 47 Governor/Control Valves 9 Intercept/StopValves 6 ABSTRACT Instruments 5 The need to protect a turbine generator from abnormal Pressure Regulator 5 operating condidtions is a given. Power plant operational histor­ Bearings 3 ical data indicate that malfunction of the turbine protection sys­ Oil System 3 tem (TPS) has been the initiator of many power plant trips and Shafts/Blades 3 attendant losses of plant availability. Many of these documented Miscellaneous or Unknown 20 events have been considered to be caused by improper or faulty action of sensors leading to false trip conditions. Whether in a petrochemical or nuclear power plant, the de­ Clearly, it would be most desirable to eliminate spurious tur­ sign and operation of the TPS is fairly consistent. It is logical to bine trips, but not at the expense of proper protective functions. consider, therefore, that tbeunreliabilities found within the TPS The dangers both to operating personnel and to plant equip­ at nuclear power plants will be similar to those found at pet­ ment posed by turbine overspeed, shaft vibration, loss of lub­ rochemical facilities. ricating oil, etc., are very real, and should be reliably detected Statistical methods of analysis to quantify the unreliability of and avoided through comprehensive protection systems. False TPS components are considered, along with the magnitude and trip conditions, however, should be positively identified; and root causes of lost availability attributable to the spurious func­ needless turbine trips should be avoided to the maximum extent tioning of the TPS at nuclear power plants; identification of cor­ possible. rective actions such as sensor replacement, multiple logic and Research reported herein was sponsored by the Electric artificialintelligence; and methods of analysis to quantifythe re­ Power Research Institute and the New York Power Authority to liability improvements which will result from these corrective investigate the numerous false turbine trip events which have actions. plagued the nuclear industry, and to suggest equipment and/or techniques which can improve TPS performance and reliability INTRODUCTION in a cost beneficial manner. Nuclear industry data bases report that turbine related fail­ DISCUSSION ures account for approximately one unplanned forced outage per plant per year [1]. Of the many turbine subsystems and compo­ Root Cause Analysis nents, the turbine protection system (TPS), or more specifically, To identify the magnitude and root causes of lost availability the electrohydraulic controls systems (EHC), has been iden­ attributable to the spurious function of the TPS, detailed tified as a principal source for spurious turbine trips. The EHC analyses were conducted using industry data bases of nuclear system is the backbone of the turbine protection system, con­ power plant operating histories. Data from approximately 80 necting critical turbine components with the automatic turbine plants covering the years 1978 through 1985 were included in trip relay. Although new turbine protection systems are imple­ the evaluation. In all, over 340 years of nuclear power plant tur­ menting wholly digital systems with distributed microprocessor bine operating history was studied. A statistical analysis of the 67 68 PROCEEDINGS OF THE SEVENTEENTH TURBO MACHINERY SYMPOSIUM data, augmented by an industry survey and utility interviews, ation would justify an expenditure of roughly $700,000 on TPS led to identification of areas within the TPS which could benefit improvements. fromretrofit of corrective actions. Potential Corrective Actions The data were divided into groups based on the cause of the event. These groupings were further manipulated to obtain the Numerous hardware modifications are available to increase equivalent outage hours expected to result from each type of TPS reliability. The following presents possible corrective ac­ spurious event. The expected outage hours were determined by tions for TPS problem areas identified through the preceding examining the performace parameter, equivalent full power statistical analysis, and some methods of quantifyingthe reliabil­ hours (EFPH) lost, which is recorded for each event. The EFPH ity improvement to be gained through implementation of the lost is the percent capacity lost due to an event multiplied by suggested corrective actions. the duration of the event in hours. For example, an event that Improved Sensor Performance causes a 20 percent power derating and lasts for five hours would have an EFPH of one hour. TPS reliability can be improved by replacing certain trouble­ The EFPH is a function of the outage duration and is therefore some sensors whose malfunctions were frequently noted. A sen­ essentially "repair time." From experience it is known that re­ sor may be replaced-in-kind or with an improved sensor whose pair time data follow the exponential probability distribution. design is intended to preclude previously observed difficulties Switch failure data are shown plotted in histogram form in Fig­ inherent in the older design. Replacement of a troublesome sen­ ure 1, and the curve drawn along the histogram indeed follows sor with a new (but same model) sensor will provide immediate an exponential probability distribution. Since the failure data performance improvement; however, if the original sensor mal­ are exponential, the EFPH expected to result from the occur­ functions are due to worn materials or other similar problems, rence of a switch failure is best represented by the median, not long-term performance improvement of this sensor can only be the mean, of the data. Using the median value of each root cause realized through application of a more aggressive preventive grouping, the amount of EFPH expected to result from each fail­ maintenance program. ure type was determined. Finally, the EFPH expected to be lost Beyond replacement-in-kind is replacement of a problem­ annually by each operating nuclear plant was calculated. The re­ prone sensor with a unit which is technically superior, e.g., re­ sults of these calculations are given in Table 2. placing a level sensor having moving internals with a unit which An economic analysis of the worth, in terms of replacement senses level via a conductance measurement. A typical example power costs due to more reliable TPS operation, led to the con­ is shown in Figure 2. In the latter case, there are no moving clusion that elimination of six to eight hours of lost power gener- parts, and, therefore, less likelihood of sensor failure or false measurement. Table Lost Power Generation due to Spurious TPS Events. 2. Expected EFPH Expected EFPH Lost Root Cause Lost per Event (hr/yr/plant) Vibration 32.2 1.02 Circuit Card 24.0 1.04 Stop Valve 13.0 2.02 Main Steam Bypass Valve 13.0 0.59 Svvitch 12.0 0.76 Personnel Error 12.0 2.76 Other Electrical 8.9 0.33 Relay 7.5 0.19 Control Valve 5.0 1.48 Turbine Overspeed Test 2.9 1.44 Valve Test 1.1 5.20 Figure Hydratect Resistivity Electrode. 2. 40 § t; Another sensor that often malfunctions due to mechanical [5 30 a wear is the shaft-riding vibration probe. Shown in Figure 3, the fE typical shaft-riderhas a spring-loaded tip which rides against the � 20 shaft surface. It was the failure of this spring at one nuclear plant 3 U-:;;: that resulted in a forced outage which lasted for about 100 hr, 10 and in replacement power on the order of $120,000. One corrective action that could improve TPS reliability and 0 offer greater equipment diagnostic capability would be the re­ placement of the shaft-riding vibration detector with the eddy 0 20 40 60 80 100 120 current type proximity probe. Intuitively, the noncontacting EQUIVALENT FULL POWER HOURS LOST proximity probe is expected to provide better performance than Figure 1. Histogram of Switch Failures. the traditional contacting type vibration probe. In addition to TURBINE PROTECTION SYSTEM PERFORMANCE 69 COLLAR GAP TO BE SET WITH DIMENTION SHOWN WHEN PROBE IS INSTALLED. DECREASE IN GAP WIDTH INDICATES TIP WEAR. SHAFT (WHEN GAP WIDTH IS RE- BRG GAP 1/32" r-··-y PLACE RIDER TIP) UPPER BEARING : 1 r--J RIDER TIP I SPRING i • 1 - --� CHECK INSULATION RESISTANCE BETWEEN RIDER TIP AND PROBE CASING WITH MEGGER. A MINIMUM OF OHMS500v IS CONSIDERED SATISFACTORY.10,000 INSTRUMENT HEAD MTG B Figure Shaft-Riding Vibration Probe. 3. mechanical wear, the shaft-riding probe design has some other Since the mercury-wetted relay fails in the as-is position, the disadvantages.