Nuclear Engineering and Design 224 (2003) 293–300
RELAP5-3D code validation in the neutron-dynamic analysis of transient processes taking place in RBMK-1500 reactors E. Uspuras, E. Bubelis∗ Laboratory of Nuclear Installation Safety, Lithuanian Energy Institute, Breslaujos 3, 3035 Kaunas, Lithuania Received 14 November 2002; received in revised form 8 April 2003; accepted 30 April 2003
Abstract The paper presents an evaluation of RELAP5-3D code suitability to model-specific transients that take place during RBMK-1500 reactor operation, where the neutronic response of the core is important. Certain RELAP5-3D transient calculation results were benchmarked against calculation results obtained using the Russian complex neutronic-thermal-hydraulic code STEPAN/KOBRA, specially designed for RBMK reactor analysis. Comparison of the results obtained, using the RELAP5-3D and STEPAN/KOBRA codes, showed reasonable mutual agreement of the calculation results of both codes and their reasonable agreement with the real plant data. © 2003 Elsevier Science B.V. All rights reserved.
1. Introduction The RELAP5 code was originally designed for PWR type reactors. However, during the last decade different versions of this code have been successfully applied for the analysis of other types of reactors, including RBMKs. Using RELAP5-3D code a thermal-hydraulic-neutronic RELAP5-3D model of Ignalina NPP RBMK-1500 reactor has been developed (Uspuras et al., 2003). Fairly comprehensive description of Ignalina NPP is presented in Almenas et al. (1998). The thermal-hydraulic part of Ignalina NPP RELAP5-3D model was previously validated against real plant transients, while the nodal kinetics part of Ignalina NPP RELAP5-3D model was validated against real plant data only in static neutron-physics calculations. The validation results were presented ∗ Corresponding author. Tel.: +370-37-401923; fax: +370-37-351271. E-mail address:
[email protected] (E. Bubelis).
in Uspuras et al. (2003). The benchmark calculation results are in reasonable agreement with the real plant data. In this paper, the RELAP5-3D code is evaluated for its suitability to model-specific transients that take place during RBMK-1500 reactor operation, where the neutronic response of the core is important. Two benchmark problem analyses that were performed during the on-going validation of the thermal-hydraulic-neutronic RELAP5-3D model of the Ignalina NPP RBMK-1500 reactor and presented here are: reactor power reduction and feedwater flow perturbation transients. The main reason for the selection of RELAP5-3D code for the analysis was that RELAP5/MOD3.2 code was not capable of predicting local effects taking place in such a big reactor core as that of RBMK-1500 reactor. The RELAP5/MOD3.2 code uses point kinetics, but that was not sufficient for the modeling of the selected transients. The main advantage of the RELAP5-3D code is its suitability
0029-5493/$ – see front matter © 2003 Elsevier Science B.V. All rights reserved. doi:10.1016/S0029-5493(03)00129-8
294
E. Uspuras, E. Bubelis / Nuclear Engineering and Design 224 (2003) 293–300
Nomenclature AZ-1,3,4,6 CPS FA INEEL MCP NPP RBMK
RRC “KI” U.S. DOE
emergency protections 1, 3, 4, 6 control and protection system fuel assembly Idaho National Engineering and Environmental Laboratory main circulation pump nuclear power plant Russian abbreviation for ‘Large Channel Type Water Cooled Graphite Moderated Reactor’ Russian Research Center “Kurchatov Institute” Department of Energy of the United States of America
to model-specific transients that occur during reactor operation, where the detailed neutronic response of the core and the local power effects depending on the local thermal-hydraulic characteristics of the core are important (i.e. in case of reactor power variations, feedwater flow perturbations, inadvertent control rod withdrawals, etc.). The two above-mentioned benchmarks were modeled using the RELAP5-3D code and the calculation results compared to the calculation results obtained using the Russian complex neutronic-thermal-hydraulic code STEPAN/KOBRA, specially designed for RBMK reactor analysis, as well as to the real plant data registered by the TITAN information computer system at Ignalina NPP. The agreement of the calculation results obtained using RELAP5-3D with those of STEPAN/KOBRA code, as well as with the real plant data, was evaluated using the adequacy standard, presented in the Guideline for performing code validation and issued by DOE International Nuclear Safety Center (DOE INSC, 1997). According to the standard, both excellent and reasonable agreement of the calculation results and the real plant data are considered as being acceptable. The agreement is judged to be excellent, when the code exhibits no deficiencies in modeling of a given behavior; major and minor phenomena and trends are predicted correctly; calculation results are judged to agree closely with the real plant data. The agreement is judged to be reasonable, when the code
exhibits minor deficiencies, although it provides an acceptable prediction; all major trends and phenomena are correctly predicted, but differences between the calculation and data are greater than deemed acceptable for excellent agreement.
2. Reactor power reduction transient The reactor power reduction transient is initiated by the reactor shutdown operation, where the parameters of the reactor are strongly changed. Besides evident decrease of reactor power, such parameters as system pressure and MCP flow are changed significantly because of the total reactor power reduction and due to the operation of the reactor equipment. Reactor shutdown is a regular reactor operation procedure that is performed during the entire lifetime of the reactor. Usually, it starts by scram signal activation by the operator. After this signal, all 24 fast acting scram rods are fully inserted into the reactor core during ∼7 s from the top end switch position, while all the remaining of the control rods are fully inserted into the reactor core during ∼14 s from their present actual operation positions. The above described control rod insertion causes a sharp decrease of the reactivity and the total reactor core power. Usually, neutron field distribution in the reactor core during this transient is measured by in-core and lateral detectors. Measurement results are registered by the reactor information computer system. On March 29, 1999, Ignalina NPP Unit 2 was shutdown by the operator signal. Initial reactor core conditions and neutron flux behavior were registered by in-core detectors. The real operation conditions of Ignalina NPP Unit 2 (on March 29, 1999) were used for this benchmark. Reactor power was equal to 2065 MW(th) (as taken from the database), but just before the reactor shutdown it was decreased to 1204 MW(th). This reduced reactor core power was taken as the basis for the benchmark. Reactor core loading on this date consisted of 1251 FA with uranium–erbium fuel, 408 FA with 2.0% enriched uranium dioxide fuel, 1 water column and 71 manual control rods of 2477 type. The reactor scram calculation was performed in two steps:
E. Uspuras, E. Bubelis / Nuclear Engineering and Design 224 (2003) 293–300
295
Fig. 1. The total reactor core power behavior vs. time during the reactor shutdown transient.
• 24 fast acting scram rods were inserted into the reactor core from the top end switch beginning from the 48 s; • 13 s later all the remaining control rods were inserted simultaneously into the reactor core from their actual operation positions. Insertion velocity of the fast acting scram rods was assumed to be 120 cm/s, while the insertion velocity of all remaining control rods, except for the short-bottom control rods, was assumed to be 80 cm/s. Short-bottom control rods were assumed to be inserted with the velocity of 40 cm/s. Calculation results using RELAP5-3D code are presented for 220-s time period, including 48 s calculation of zero transient before the reactor scram signal initiation, to match the presented real plant data. Water flowrate at the reactor core inlet and the steam drum pressure were used as boundary conditions for the calculation. According to RELAP5-3D and STEPAN/KOBRA calculation results, the insertion of 24 fast acting scram rods beginning from 48 s of the reactor shutdown transient causes a sharp decrease of the total reactor core power. As seen in Fig. 1, during this first reactor shutdown phase, the results of STEPAN/KOBRA code show faster total reactor core power decrease than the RELAP5-3D code results. STEPAN code overestimates the efficiency of 24 fast acting scram rods in comparison with RELAP5-3D code and the plant data. This can be explained by different neutronic models used by both codes. Nodal model used by
RELAP5-3D code gives smaller control rod weight than finite difference model used by STEPAN code. From the beginning of the second phase of reactor shutdown (from time, t = 61 s), when AZ-1 signal is initiated and all the remaining control rods start to be inserted into the core from their actual operation positions, STEPAN/KOBRA and RELAP5-3D calculation results show exactly the same behavior of the total reactor core power versus time. So in general, STEPAN/KOBRA and RELAP5-3D codes calculate the total reactor core power behavior in time during the reactor shutdown transient in quite a similar manner and the calculation results correspond quite reasonably to each other. The total reactor core power (see Fig. 1) calculated by RELAP5-3D and STEPAN/KOBRA codes is the sum of immediate (prompt and delayed neutron) fission and decay (fission products and actinide) power in each fuel assembly location in the core and has nothing to do with the four lateral ionization chamber (5, 11, 17, and 23) readings, that were used for the total reactor core power determination during the reactor shutdown at the plant. The difference in the above-mentioned procedures for obtaining these total reactor core power values can be the reason of a certain difference that can be observed between the curves, especially beginning from ∼61 s of the transient. It should be noted here that the total reactor power determined at the plant based on the readings of the four lateral ionization chambers is always ∼30 MW(th) higher than the total reactor power value determined based on the readings of the
296
E. Uspuras, E. Bubelis / Nuclear Engineering and Design 224 (2003) 293–300
Fig. 2. Calculated “counter” reactivity behavior during the reactor shutdown transient.
in-core detectors. Taking this into account, the total reactor core power, calculated by RELAP5-3D and STEPAN/KOBRA codes, agree reasonably well with the plant data. Fig. 2 shows the comparison of the “counter” reactivity (i.e. reactivity measured by lateral ionization chambers) as calculated by STEPAN/KOBRA and RELAP5-3D codes. Calculated “counter” reactivity, presented in Fig. 2, is the reactivity obtained from the measurements of the four out-core detectors (2, 8, 14, and 20), situated in the water tank of the biological shield of the reactor. Plant data are presented in Fig. 2 as well. As in the previous figure, one can see that during the first phase of reactor shutdown process, STEPAN/KOBRA code overestimates the efficiency of 24 fast acting scram rods in comparison with RELAP5-3D code. As a result of this, the calculated “counter” reactivity by STEPAN/KOBRA code decreases faster beginning from time, t = 48 s, than the reactivity calculated by RELAP5-3D code or the plant data. In the second phase of reactor shutdown, when AZ-1 signal is initiated and all the remaining control rods start to be inserted into the core from their actual operation positions, RELAP5-3D code overestimates the efficiency of CPS rods without 24 fast acting scram rods in comparison with the STEPAN/KOBRA code. As a result of this, the “counter” reactivity calculated by RELAP5-3D code decreases faster beginning from time, t = 61 s, than the reactivity calculated by STEPAN/KOBRA code. This can be explained by different neutronic models used by both codes, as well as by differences in the modeling of different
reactor core components. For example, RELAP5-3D code models gap between fuel pellet and cladding as being constant, while STEPAN/KOBRA code models the above-mentioned gap as being variable, depending on fuel temperature and fuel burnup. There are also other differences (i.e. different nodalization schemes, different thermal hydraulic modeling of the fuel channels, etc.) in the models that can results in certain differences of the calculation results as obtained by both codes. At the final reactor shutdown point (beginning from time, t ≈ 75 s), both codes calculate the total reactor shutdown effect value to be approximately the same. “Counter” reactivity, calculated by RELAP5-3D and STEPAN/KOBRA codes, corresponds reasonably well to the plant data at the beginning of the transient, but later, beginning from ∼75 s, RELAP5-3D and STEPAN/KOBRA calculation results indicate smaller reactivity increase than the real plant data. The reactivity increase after ∼75 s of the transient can be explained by feedback effects (e.g. power reactivity effect, etc.). The reactivity increase due to the feedback effects is more clearly seen in the plant data than in the calculation results. This can be explained by the fact that RELAP5-3D and STEPAN codes models the out-core detectors as being situated in first fuel assemblies row nearest to the radial reflector, while at the plant they are situated in the water tank of the biological shield of the reactor. The real out-core detectors might “see” different neutron flux and reactivity deviations than the modeled out-core detectors by both codes. Despite this, the calculated “counter”
E. Uspuras, E. Bubelis / Nuclear Engineering and Design 224 (2003) 293–300
reactivity behavior in time corresponds reasonably well to the plant data. In STEPAN/KOBRA case, the total reactor shutdown effect is equal to ∼19.0β, while in RELAP5-3D case, the total reactor shutdown effect is equal to ∼19.1β. The worth of 24 fast acting scram rods is equal to ∼2.6 and ∼2.8β, respectively. In general, the “counter” reactivity behavior in time, calculated by STEPAN/KOBRA and RELAP5-3D codes, is very similar and the final reactor shutdown effectiveness values correspond quite reasonably to each other. Thus, in the case of reactor power reduction transient, RELAP5-3D and STEPAN/KOBRA codes show reasonable mutual agreement of the calculation results of both codes and their reasonable agreement with the real plant data.
3. Feedwater flowrate perturbation transient Void reactivity coefficient measuring is one of regular procedures used at a nuclear power plant of RBMK type. During this measurement, the feedwater flowrate changes. It causes a reactivity perturbation. Automatic rods change their insertion depths to compensate this reactivity change. The experimental procedure for void reactivity coefficient measuring consists of two stages. At the first stage, the perturbation of feedwater flowrate is performed and reactivity change due to this perturbation is compensated by the movement of automatic control rods. At the second stage, the worth of these control rods is being calculated. Dynamic calculations to repeat the experimental results of void reactivity coefficient measuring were performed for Unit 2 (on November 26, 1998) core conditions. Before the simulation of the transient, initial automatic regulator positions in the reactor core condition files were corrected for the calculation according to the reactor core condition before the experiment. The results showed the insertion depth of ∼300 cm for each automatic regulator. At the first stage of the void reactivity coefficient measurement, feedwater flowrate was increased by ∼(205–210) t/h per reactor core side and caused the decreasing of coolant temperature at the inlet to the reactor core by 1 ◦ C. This led to the reactor core neutron field distortion and inserted the negative reactivity into the reactor core. Four ionization chambers
297
located in the lateral water tank (Nos. 3, 9, 15, and 21) measured neutron field change. Four automatic control rods located in cells (32–33, 16–33, 16–17, and 32–17) were changing their positions to compensate for the reactivity change. Each automatic control rod operated according to a signal coming from one lateral ionization chamber located in the annular water tank around the reactor core serving as a biological protection shield. Positions of all other control rods were not changed during this experiment. Table 1 presents the initial/final calculated and measured automatic control rod positions for the simulated transient. According to the RELAP5-3D calculation results, automatic regulators average shift is more than that obtained during the experiment (33.75 and 12.5 cm, respectively). This difference can be explained by the fact that reactor core is more “transparent” for neutrons in nodal model of RELAP5-3D code, than in finite difference model of the STEPAN code. RELAP5-3D code gives smaller control rod weight than the STEPAN code, so for the negative reactivity insertion compensation in RELAP5-3D case the control rods move a bigger distance than in STEPAN/KOBRA code calculation. Since the real transient data is not available for the comparison of each parameter presented below, the calculation results obtained using RELAP5-3D code were compared only with the calculation results obtained by RRC “KI” using STEPAN/KOBRA code. Calculation results presented in Fig. 3 indicate, that the increase of feedwater flowrate by ∼200 t/h leads to the total reactor core power decrease from the initial power value. However, initial power increasing by ∼5 MW was obtained by RELAP5-3D and STEPAN/KOBRA codes. This can be explained by the fact, that at the very beginning of the coolant temperature decrease (due to the additional feedwater supply through auxiliary feedwater pumps in the amount of ∼200 t/h), pressure in the primary circuit also decreases. Pressure decrease propagates with the speed of sound, while the front of the colder coolant, moving through the primary circuit, reaches the core region only after a certain period of time. This initial pressure decrease leads to coolant density decrease in the reactor core region, which results in the power increase by ∼5 MW. Afterwards, after a certain period
298
E. Uspuras, E. Bubelis / Nuclear Engineering and Design 224 (2003) 293–300
Table 1 Initial/final calculated and measured automatic control rod positions Gfw (t/h), 210/205
Location of automatic control rods (cm)
Initial (in calculations) Calculation by RELAP5-3D code (final) Calculation by STEPAN/KOBRA code (final) Measurement (initial/final)
16–33
32–33
16–17
32–17
Average value
300 285 286 300/290
300 226 274 300/290
300 280 284 300/290
300 274 274 300/280
300 266.25 279.5 300/287.5
Fig. 3. Total reactor core power vs. time during the feedwater flow perturbation transient.
of time, reactivity starts to decrease, because by then the front of the colder coolant has already entered the core region and coolant density increased again. Fig. 4 shows that the average signal deviation of the four lateral fission chambers from the set-point value at the start of the transient is equal to zero. After the stabi-
lization of the transient this signal deviation decreases to ∼(−0.01). This corresponds to the reality, because automatic regulation rods start to move up/down when signal deviation from the set-point value reaches ±1%. As it is indicated in Fig. 5, both codes overestimate the total insertion depth of four automatic
Fig. 4. Summary signal of four lateral fission chambers during the feedwater flow perturbation transient.
E. Uspuras, E. Bubelis / Nuclear Engineering and Design 224 (2003) 293–300
299
Fig. 5. Total insertion depth of four automatic regulators during the feedwater flow perturbation transient.
regulators during the transient. RELAP5-3D code overestimates this value by 85 cm, while STEPAN/ KOBRA code overestimates it by 32 cm (see Table 1). The reason for this difference was already mentioned above in the text. As could be seen from Figs. 3 and 4, the starting time moment of simulation for STEPAN/KOBRA and RELAP5-3D cases is different. This is due to the fact that STEPAN/KOBRA code models this feedwater flow perturbation transient simply by changing coolant temperature by 1 ◦ C at the core inlet. This corresponds to the increasing of feedwater rate by ∼200 t/h. In RELAP5-3D code feedwater flow perturbation transient is modeled directly by changing feedwater flowrate by ∼200 t/h. This also causes coolant temperature decrease by 1 ◦ C at the core inlet. So during the modeling of this transient using RELAP5-3D code, there exists some delay between the start of feedwater flowrate change and the decrease of coolant temperature at the core inlet. That’s why there can be observed the time difference between the moments of the decrease of the total reactor core power in RELAP5-3D and STEPAN/KOBRA cases. Zero time point in the presentation of STEPAN/KOBRA results is shifted to the right to make the comparison of both calculation results easier and more evident. Thus, in the case of feedwater flow perturbation transient, RELAP5-3D and STEPAN/KOBRA codes give reasonable mutual agreement of the calculation results and their reasonable agreement with the real plant data.
4. Conclusions A thermal-hydraulic-neutronic RELAP5-3D model of the Ignalina NPP RBMK-1500 reactor has been developed and validated against the real plant data. The validation of the model has been performed using operational transients from the Ignalina NPP. The benchmark problem analyses that were performed during the validation of the RELAP5-3D model of Ignalina NPP RBMK-1500 reactor and presented here are: reactor power reduction and feedwater flow perturbation transients. The above-mentioned benchmarks were modeled using the RELAP5-3D code and the calculation results were compared to the calculation results obtained using the complex neutronic-thermal-hydraulic code STEPAN/KOBRA, as well as to the real plant data registered by the TITAN information computer system at Ignalina NPP, where such data was available for comparison. In general, both codes proved to be able to reproduce experimental results with certain accuracy. Comparison of the calculation results obtained, using the RELAP5-3D and STEPAN/KOBRA (specially designed for RBMK reactor analysis) codes, showed reasonable mutual agreement of the calculation results and their reasonable agreement with the real plant data.
Acknowledgements The authors would like to acknowledge the technical (from INEEL and ANL), financial support and
300
E. Uspuras, E. Bubelis / Nuclear Engineering and Design 224 (2003) 293–300
access to the code RELAP5-3D provided by the U.S. DOE, the U.S. International Nuclear Safety Program. Especially we want to thank experts from INEEL, Dr. James Fisher and Mr. Paul Bayless, as well as Mr. John Ahrens from ANL, who helped a lot with their advice and sharing their experience. All this made it possible to bring together the analytical methodology developed in the U.S. with the knowledge of the unique RBMK-1500 systems available in Lithuania. The authors would like to acknowledge also the technical support and access to the RBMK-1500 materials macroscopic cross-section library provided by RRC “Kurchatov Institute”. Especially, we want to thank RRC “KI” experts Dr. A. Krajushkin, Mr. A. Balygin and Dr. A. Glembotskiy for their valuable advice. We also want to extend our thanks to the administration and technical staff of Ignalina NPP, for providing information regarding operational procedures and operational data. The authors would like
to acknowledge the sharing of experience by Dr. A. Kaliatka from the Lithuanian Energy Institute and his active participation during the simulation of the above described transients. This work was supported by the U.S. Department of Energy, National Nuclear Security Administration. The U.S. Government makes no endorsement of the results of this work. References Almenas, K., Kaliatka, A., Uspuras, E., 1998. Ignalina RBMK-1500: A Source Book, Extended and Updated Version. Ignalina Safety Analysis Group, Lithuanian Energy Institute, Kaunas, Lithuania. DOE International Nuclear Safety Center, September 1997. Adequacy Standard. Guideline for Performing Code Validation within DOE International Nuclear Safety Center (INSC). Uspuras, E., Kaliatka, A., Bubelis, E., 2003. Development of Ignalina NPP RBMK-1500 reactor RELAP5-3D model. Nucl. Eng. Des. 220 (2), 159–178.
