Dynamic Simulation of NEK Reactor Coolant Pump with a Bet Etimate Full Scale Model in APROS ABSTRACT Dejan Slovenc ZEL-EN d.o.o. Hočevarjev trg 1, 870, Krško, Slovenia dejan.lovenc@zel-en.i Ivica Bašić APoSS d.o.o. Repovec 3b 4910 Zabok, Croatia baic.ivica@kr.t-com.hr Samo Fürt GEN energija d.o.o. Vrbina 17 870, Krško, Slovenia amo.furt@gen-energija.i Thi paper preent the imulation of the Krško NPP (NEK) Reactor Coolant Pump (RCP) tart-up tet when pump i upplied from the normal bu (off-ite power ource) and the RCP coat down tet on the lo of offite power ignal. The imulation model wa developed in APROS - Advanced PROce Simulation environment. APROS i a multifunctional oftware for modelling and dynamic imulation of variou phyical procee in power plant which enable a coupled imulation of thermal-hydraulic, electrical and regulation ytem including reactor kinetic. The goal wa to create a imulation model which enabled imulation of the tranient proce occurring at the RCP tart-up and coat down tet. The entire integrated model conit of NEK primary ytem and a part of the econdary ytem. The RCP pump model, a a part of the reactor coolant loop, i upplied from the normal bu (electric power ditribution ytem connected to external power grid). The RCP moment and current lip characteritic were imulated and compared with the RCP vendor data to confirm the validity and accuracy of the APROS pump model. It i hown that, by uing the available characteritic data of the pump a well a the input obtained from the motor data heet, an accurate dynamic imulation of RCP tart-up i poible. Furthermore, the paper dicue the imulation of the RCP motor tart-up tranient and coat down of primary ma flow after a RCP trip. 917.1
917. 1 INTRODUCTION All nuclear plant have redundant emergency electrical ytem that are deigned to provide backup AC and DC power to the emergency afety equipment if the normal ource of electrical power are lot. If the normal AC power i lot, the emergency dieel generator (EDG) tart to provide electrical power to a afeguard electrical load centre (alo called witchgear or bu). Each bu then upplie power directly to important afety electrically-driven component. However, the capacity of EDG i not ufficient for driving the RCP pump and the reactor ha to be cooled down by a natural water circulation proce through the team generator (SG). Complete lo of reactor coolant forced flow i caued by imultaneou lo of electrical upplie to all RCP and lead to a rapid increae in the reactor coolant temperature. Thi increae could reult in departure from nucleate boiling (DNB) with ubequent fuel damage if the reactor are not tripped promptly and long core term decay heat removal etablihed. Thi paper decribe the partial validation of the NEK RCP pump model taking into conideration the real plant tet data. The main tak for thi analyi i to how the code capabilitie in the modelling a tart-up of induction motor with their nonlinear load. Our full NEK APROS imulation model conit of the primary and part of the econdary ytem. The model conit the Reactor Preure Veel (including the core and aociated neutron kinetic model), Reactor Coolant Leg with RCP, Preurizer, Reitance Temperature Detector Bypa RTD Bypa, Steam Generator 1, Steam Generator, Main Steam & Steam Dump, Feedwater 1 and Feedwater and part of the main electrical power ditribution ytem. The regulation ytem that maintain the model in the teady tate are alo modelled; Reactor Coolant Pump Power Lo, PORV & Safety Valve Control Sytem 1 and, Preurizer Liquid Level Control, Preurizer Preure Control, Rod Control, Steam Dump Control, Steam Generator and Feedwater Flow Control 1 and. The model conit of over 400 thermal hydraulic volume. The APROS NEK integrated model and the aociated teady tate evaluation have been already preented on ICONE conference 014 in Prague [1]. It i known that verification and validation hould be performed on all computer code and model ued for the determinitic afety analyi of NPP. The purpoe of validation (alo referred code aement) i to provide confidence in the ability of a oftware package to realitically or conervatively predict the value of the modelled parameter. The RCP model imulation (e.g. validation of moment and current lip characteritic) wa performed and reult compared with the RCP vendor data to confirm the validity and accuracy of the APROS pump model. APROS [], developed by the Reearch Centre VTT and Fortum Engineering in Finland, i a program package that allow making the dynamical imulation for engineering purpoe. The APROS tool i uitable for modelling and imulation of the dynamic of different procee (e.g. thermo-hydraulic, I&C, electric, neutron kinetic etc.) during all phae of nuclear power plant life pan from pre-deign tage to training, for mall imple model up to full cope imulator. Proceeding of the International Conference Nuclear Energy for New Europe, Portorož, Slovenia, September 8 11, 014
917.3 MODELLING THE REACTOR COOLANT PUMP START-UP.1 The power ytem outlay Figure 1 repreent a implified NEK electrical power ditribution ubytem that both RCP model are connected to. During normal operation, main generator G1 upplie electrical network through the main tranformer GT1 and GT. It alo upplie onite power upply over unit tranformer. Both RCP are upplied from Non Cla 1E M1 and M 6.3 kv bue. Alternatively, if normal AC power i lot, both RCP can be energized from the tation auxiliary tranformer T3, which i connected to the 110 kv RTP Krško and directly to independent external power ource. In thi cae, the plant i on iland operation mode eparated from the ret of the electric power grid and control ytem are ued to control frequency and voltage drop during the RCP pump tart-up. Figure 1: NEK implified electrical power ytem model in APROS code Proceeding of the International Conference Nuclear Energy for New Europe, Portorož, Slovenia, September 8 11, 014
917.4. Reactor coolant pump The APROS model of RCP preent a vertical, ingle tage, centrifugal, haft eal pump driven by air cooled, three phae induction motor with nominal active power of 5.4 MW. The motor i connected to a 6.3 kv bu. The pump inertia i provided by a flywheel connected to the haft above the motor to increae the total rotating inertia, auring a longer pump coatdown. Interaction of the pump impeller and the fluid i decribed by empirically developed curve (homologou curve) relating pump head and torque to the volumetric flow rate and pump peed. The pump model i modelled to cover all poibly zone of operation, normal and during the tranient. General input data table for RCP are collected in below table. Table 1: RCP pump model attribute Attribute Value Unit Nominal rotation peed 1485.7 rpm Nominal volumetric 6.35 m 3 / flow Nominal head 8.9 m Nominal hydraulic 3038 Nm torque Nominal active power 541 kw Nominal motor torque 34879 Nm Nominal voltage level 6300 V Moment of inertia 694 kg m (pump and motor) Start torque 1.1 of nom. Maximal torque.45 of nom..3 Baic model of the induction motor The APROS pump module calculate the motor torque a a function of the peed (lip), frequency and voltage [3]. The pump model i connected to the 6.3 kv bu and voltage drop at the time when the imal moment occur i le than one percent of nominal. The bu frequency i contant during the imulation. The electromagnetic part of the model i baed on the dynamic equilibrium of the motor and hydraulic torque where pump peed i calculated with the aid of the both torque and with the aid of inertia of the entire pump-motor et. The motor torque can be calculated with Eq. (1), where i the lip (relative motion between the rotor and the magnetic field), i the lip when torque get it imum value, T i the torque imum value, U i current voltage and Un i nominal voltage level. The K coefficient i ued to correct the torque by Eq. (1) during the tranient, when motor i tarted and peed i near zero. Becaue the motor torque curve i non-linear ome extra linearization term have been added to the torque derivative according to the peed, frequency and voltage to enure convergence. Alo the dynamic current lip characteritic can be calculated. The motor tarting current depend on active and reactive power conumed by the motor. The reactive power i calculated with the help of a power factor v peed curve determined by interpolation between tart and normal power coefficient attribute. T m = U U n K T (/ ) ( /) (1) Proceeding of the International Conference Nuclear Energy for New Europe, Portorož, Slovenia, September 8 11, 014
917.5.4 RCP motor protection limitation The following ignal provide the neceary protection againt a complete lo of forced reactor coolant flow: RCP power upply Undervoltage or Underfrequency Reactor Coolant Flow-Low RCP Breaker Open The undervoltage RCP bu trip et point are 70 % of nominal bu voltage and the underfrequency et point are 47.7 Hz [4]. The entire pump-flywheel-motor et time contant amount to 1 for 100 % of the nominal voltage and i increaed to 4 at 80 % of the nominal voltage. A further decreae of voltage reult in a coniderable increae of the motor et time contant and may activate the delayed overcurrent protection [5]. 3 SIMULATION AND FIELD TESTS RESULTS Our aim i to imulate the pump tart-up at two different voltage level and to how that the APROS pump model i capable to accurately imulate real AC induction motor (momentlip and current-lip) characteritic in comparion with manufacturer and vendor data. Upon thi analyi two field tet were imulated, Coat down tet and pump Start-up tet, to confirm the validity of RCP pump model. 3.1 Coat down tet The complete lo of reactor coolant flow ( Coat down tet ) wa analyed for the aumed lo of the RCP (initially in operation in a full cale imulation model at teady tate in the range of the reactor at 100% power). Figure how the primary circuit ma flow rate tranient repone after RCP trip between the APROS imulation and USAR [6]. The ma flow rate falling trend i a little different and after 10 econd of imulation APROS ma flow rate reached 500 kg/. Thi can be explained by the fact that the primary circuit flow reitance between APROS and thee ued in USAR analyi cannot be compared. Proceeding of the International Conference Nuclear Energy for New Europe, Portorož, Slovenia, September 8 11, 014
917.6 3. Pump tart-up tet Figure : RCS Ma flow during cot down tet compared between model in APROS and USAR [6] During the APROS RCP model validation, the pump tart-up ha been imulated and compared to the RCP motor characteritic (moment-lip and current-lip) obtained by manufacturer hown in Figure 3 and Figure 4 [7]. The difference between motor and hydraulic torque i ufficient to accelerate a RCP to it operating peed in a very hort time. A a reult, the motor take up a high current, with the rik to overheat and damage the motor winding. If there are no force which accelerate or low down the peed, the peed i ettled down to the poition where both torque are equal. Simulation of both RCP tart wa initiated in the integrated APROS NEK model right after Cot down tet. After pump trip the peed doe not ended at zero becaue the rotor wa not conidered to be frictionle. RCP i rotating becaue of the etablihed natural loop circulation driven by the water denity difference between core and SG. The bue M1 and M voltage level wa changed from 6.3 kv to 6 kv to et the ame initial condition a in teting condition at manufacturer. The new nominal voltage level i 6 kv and 80% of nominal voltage level mean 4.8 kv. The frequency during imulation wa et to 50 Hz. The nominal rotation peed i le than the ynchronou peed, which i 1500 rpm. The tart and imal torque value aumed in imulation, needed to decribe typical induction motor moment peed characteritic, are 1.1 and.45 time more than nominal. Proceeding of the International Conference Nuclear Energy for New Europe, Portorož, Slovenia, September 8 11, 014
917.7 Figure 3: Comparion of the RCP torque v peed curve between APROS imulation and vendor data [7] In the cae where the nominal voltage level i 6000 V the imulated top torque at the time when motor top moment (lip i ) occur i 80 % of the nominal torque. The deviation of the APROS imulation from the characteritic curve are negligible at tart and more than 10 % when motor reached it imal moment. A decreae in voltage level to 4800 V (80 % of rated) during the tart-up caue the motor torque curve to droop and the time needed for tartup changed from 10 to 18 econd. In accordance with Eq. (1), the torque drop varie to the quare of the voltage. A the reult of decreaing in voltage level een from Figure 3 the trend of torque peed curve in APROS imulation are the ame. Proceeding of the International Conference Nuclear Energy for New Europe, Portorož, Slovenia, September 8 11, 014
917.8 Figure 4: Comparion of the RCP current v peed curve between APROS imulation and vendor data [7] The reult of the imulation in Figure 4 how that the current peed characteritic can be accurately imulated by the APROS pump model and compared with the vendor characteritic curve. During the tart-up, the imal current value reache up to 650 % of the nominal current in cae of the nominal voltage. A coniderable voltage drop on bu decreae the tarting current and the time to reached nominal peed extend from 10 to 18 econd. 3.3 Obervation of the reult During the imulation it wa oberved that the imal torque i alway bigger than the referenced vendor characteritic. Becaue of thi deviation, it wa invetigated how coefficient K, that correct the torque given by Eq (1), affect the torque calculation during pump tart up. The mathematical model of the motor model in the APROS code calculate the motor torque a a function of the peed-lip during the tranient. The motor imal torque i re-calculated to how the correpondence between the imulated and theoretically calculated value. Firtly, the given rated torque Tr may be calculated from the nominal active power and rotation peed attribute lited in Table 1. Eq. (): 3 544 10 60 T r = = 34878 1485.77 Nm () Proceeding of the International Conference Nuclear Energy for New Europe, Portorož, Slovenia, September 8 11, 014
917.9 With the aumption that the frequency drop at the time when the imal moment occur i le than one percent, the voltage drop at that time i 597 V (taken from the APROS imulation model) and the correction factor K which correct torque i 1.18 the imal torque could be calculated a follow in Eq. (3) and Eq. (4). U 597 Tm = K T U (/ ) ( /) 6000 n 1.18.45 1 1.865 (3) T, calculated =T T r m = 34878.865 = 9994 The correction factor K at the time when torque get the imum value can be calculated a follow: K =1+ (T /T where: 1-1) 1.18 Nm T1 = 0.18 i the value of torque calculated with Eq. (1) when peed i zero T = 1.1 i the real torque when peed i zero. i the lip at the time when the torque get it imum value () In accordance with the Eq. (1) at the time torque reached it pick value (.45 time of nominal) value are a follow: T m, T m ( ) ( ) M M top r y =.45 where a quadratic equation ha two real olution: y - 4.9 y +1 = 0 Now the outcome for y i: y 1 = 4.687 (8) Thi mean that the lip at the time when torque get it imum value i =nom y 1 0.045 where the nominal lip i: nom n n n r 1500 1485.7 0.0095 1485.7 The imulated imal torque according to APROS Figure 3 i about 99779 Nm. There i le than 1 % mimatch between the calculated and imulated torque imal value. It eem that our calculation how that correction the factor K, which i ued to correct torque given by Eq. (1), i valid only when the rotation peed of pump i near zero. Thi mean that the input in the NEK APROS pump model are correct but the APROS calculation make decribed 1 y (4) (5) (6) (7) (9) (10) Proceeding of the International Conference Nuclear Energy for New Europe, Portorož, Slovenia, September 8 11, 014
917.10 difference. The found mimatche between the theoretical induction motor model and the model applied in APROS regarding the motor top torque calculation have been ent to the APROS developer for further invetigation and correction. 4 CONCLUSION The preented reult how that, in the cae where the motor voltage level i 100% and 80 % of nominal, the moment-lip and current-lip characteritic can be atifactorily imulated by an APROS pump model. The main tak of thi RCP model invetigation and validation i to how the APROS code capabilitie in the modelling a tart-up of induction motor with their nonlinear load characteritic. It wa oberved, during the imulation and after the invetigation, that the current motor imal torque calculation in APROS are wrong and the correction factor K, that correct the torque during pump tart-up, need to be invetigated and corrected by the developer. Now we are in contact with the developer to collaborate in the olving of thi iue. ACKNOWLEDGMENTS Thi work wa partial funded by the European Union (European Regional Development Fund). REFERENCES [1] T. Polcah, K. Debelak, I. Bašić, L. Štrubelj, Nodalization of NEK In APROS and Steady State Simulation, Proceeding of the 014 nd International Conference on Nuclear Engineering ICONE, July 7-11, 014, Prague, Czech Republic [] Apro overview, http://www.apro.fi/en/ [3] FORTUM & VTT, Pump model-uer manual, APROS certification package, 009 [4] B. Glaer, B. Krajnc, NEK-Technical pecification, Krško, 011, tab..-1 [5] Pavel Omahen and Ferdinand Gubina, Simulation and field tet of a RCP pump emergency tart up by mean of remote ga unit, Elektroinštitut M. Vidmar, Faculty for Electrical Engineering and Computer Science, Ljubljana, Slovenia, Yugolavia [6] USAR NEK, Chapter 15 rev16, 15.3. COMPLETE LOSS OF REACTOR COOLANT FLOW [7] Krško Nuclear Power Plant, Reactor coolant pump characteritic and motor deign information, Part 1&, Savke elektrarne of Ljubljana-Slovenia and Elektroprivreda of Zagreb-Croatia [8] APROS Nuclear documentation, VTT and Fortum, 01 Proceeding of the International Conference Nuclear Energy for New Europe, Portorož, Slovenia, September 8 11, 014