ISSN 1810-5408
Nuclear Science and Technology
Volume 9, Number 4, December 2019
Published by VIETNAM ATOMIC ENERGY SOCIETY VIETNAM ATOMIC ENERGY INSTITUTE NUCLEAR SCIENCE AND TECHNOLOGY
Volume 9, Number 4, December 2019
Editorial Board
Editor-in-chief
Tran Huu Phat (VINATOM)
Executive Editors
Vuong Huu Tan (VARANS) Tran Chi Thanh (VINATOM) Cao Dinh Thanh (VINATOM) Hoang Anh Tuan (VAEA)
Editors
Nguyen Kien Cuong (VINATOM) Le Hong Khiem (IOP) Nguyen Nhi Dien (VINATOM) Dao Tien Khoa (VINATOM) Nguyen Thi Kim Dung (VINATOM) Tran Hoai Nam (Duy Tan University) Ho Manh Dung (VINATOM) Dang Duc Nhan (VINATOM) Nguyen Nam Giang (VINATOM) Nguyen Hao Quang (VINATOM) Trinh Van Giap (VINATOM) Nguyen Mong Sinh (VINATOM) Le Ngoc Ha (108 Military Central Hospital) Tran Duc Thiep (IOP) Phan Son Hai (VINATOM) Dang Quang Thieu (VINATOM) Le Huy Ham (VAAS) Le Ba Thuan (VINATOM) Nguyen Quoc Hien (VINATOM) Nguyen Trung Tinh (VARANS) Le Van Hong (VINATOM) Tran Ngoc Toan (VINATOM) Nguyen Tuan Khai (VINATOM) Duong Thanh Tung (VARANS) Pham Dinh Khang (VINATOM) Nguyen Nu Hoai Vi (VARANS)
Science Secretary Hoang Sy Than (VINATOM)
...... Copyright: ©2008 by the Vietnam Atomic Energy Society (VAES), Vietnam Atomic Energy Institute (VINATOM). Pusblished by Vietnam Atomic Energy Society, 59 Ly Thuong Kiet, Hanoi, Vietnam Tel: 84-24-39420463 Fax: 84-24-39424133 Email: [email protected] Vietnam Atomic Energy Institute, 59 Ly Thuong Kiet, Hanoi, Vietnam Tel: 84-24-39420463 Fax: 84-24-39422625 Email: [email protected] ...... Contents
Bubble behavior in liquid-gas two-phase flow behind a cross-shaped obstacle in a vertical circular duct K. Takase, G. Kawasaki, K. Ueta……………………………………………..………………. 01
Evaluation of the Potential for Containment Bypass due to Steam Generator Tube Rupture in VVER-1000/V320 Reactor during Extended SBO sequence using SCDAP/RELAP5 code Nguyen Van Thai, Doan Manh Long, Tran Chi Thanh …………………………………… 09
Study on transmutation efficiency of the VVER-1000 fuel assembly with different minor actinide compositions Tran Vinh Thanh, Vu Thanh Mai, Hoang Van Khanh, Pham Nhu Viet Ha ……………. 18
Determination of in situ detection efficiency for IM-NAA of non-standard geometrical samples Nguyen Duy Quang, Trinh Van Cuong, Tran Tuan Anh, Ho Van Doanh, Nguyen Thi Tho, Ho Manh Dung……………………………………………………………………………. 27
Evaluating uncertainty of some radiation measurand using Monte Carlo method Bui Duc Ky, Nguyen Ngoc Quynh, Duong Duc Thang, Le Ngoc Thiem, Ho Quang Tuan, Tran Thanh Ha, Bui Thi Anh Duong, Nguyen Huu Quyet, Duong Van Trieu...... 34
Evaluation of image reconstruction algorithms in cone-beam computed tomography technique Tran Thuy Duong, Bui Ngoc Ha…………………………………………………………….. 41
Relative output factors of different collimation systems in truebeam STx medical linear accelerator Do Duc Chi, Tran Ngoc Toan, Robin Hill, Nguyen Do Kien …………………………….. 48
Nuclear Science and Technology, Vol.9, No. 4 (2019), pp. 01-08 Bubble behavior in liquid-gas two-phase flow behind a cross-shaped obstacle in a vertical circular duct
K. Takase*, G. Kawasaki, K. Ueta *Department of Nuclear System Safety Engineering, Nagaoka University of Technology, Japan E-mail: [email protected] (Received 07 November 2019, accepted 26 December 2019)
Abstract: Grid spacers installed in subchannels of fuel assemblies for nuclear reactors can promote heat transfer. However, the fluid velocity and bubble behavior are greatly affected as the cross- sectional area of the flow passage changes. Therefore, the void fraction distribution behind the obstacle that simulates the grid spacer shape simply was measured by using a wire mesh sensor (WMS) system. Moreover, a two-phase flow analysis was performed to investigate the effect of the obstacle on the bubble behavior in a vertical duct. Keywords: Bubbler, Two-phase flow, Obstacle, Vertical duct, WMS, Experiment, Analysis.
I. INTRODUCTION simulated spacer was visually observed and the void fraction and interfacial velocity Clarifying two-phase characteristics distributions just behind the simulated in a nuclear reactor core is important in spacer was measured. particular to enhance the thermo-fluid safety of nuclear reactors. Moreover, correct data II. EXPERIMENTAL METHOD on bubbly flow in subchannels with spacers are needed in order to verify two-phase flow 1. Experimental Apparatus models in conventional nuclear safety The experimental apparatus mainly analysis codes and validate predicted data consists of a measuring section and by current CFD codes like a direct two- water/air supply lines and is shown in Fig. phase flow analysis code (Douce, et al., 1. The measuring section includes a vertical 2010). Spacers installed in subchannels of flow channel with a diameter of 58 mm fuel assemblies have the role of keeping the made of an acrylic resin, inlet and outlet interval between adjacent fuel rods constant. plenums, and an air injection nozzle with Similarly, in case of PWR the spacer has 120 injection holes with a diameter of 0.6 also the role as the turbulence promoter. mm. Water flows through the inlet plenum When the transient event occurs in a nuclear reactor, two-phase flow is generated by into the flow channel and goes up through boiling of water due to heating of fuel rods. the measuring section to the outlet plenum. Therefore, it is important to confirm the Air is supplied from the air injection nozzle Fig. 2 Observed into the flow channel. As a result, water is rising bubbles in a bubbly flow behavior around the spacer. flow channel The purpose of this study is to make the mixed with air and then a water-air two- effect of the spacer affecting the bubbly phase flow is formed as can be seen in Fig. flow clear and obtain code validation data. 2. Bubbly flow conditions are determined So bubble dynamics around the simply by both flow rates of water and air.
©2019 Vietnam Atomic Energy Society and Vietnam Atomic Energy Institute BUBBLE BEHAVIOR IN LIQUID-GAS TWO-PHASE FLOW BEHIND A CROSS-SHAPED…
f58 mm
Wire mesh Sensor
Cross- direction Flow shaped
obstacle 2600 mm 2600
High speed
Camera
1150 mm 1150 20 mm or 100 mm 100 mm or 20 Air nozzle
Fig. 1. Outline of an experimental apparatus Fig. 2. Observed rising bubbles in a flow channel
2. Wire-Mesh Sensor System temperatures of water and air are measured at the water/air supply piping lines. The void fraction distribution at the horizontal cross- section in the flow channel is The WMS can obtain the void fraction measured with a wire-mesh sensor (WMS) distribution in a cross-section of the flow channel system (Prasser et al., 1998; 2000; 2001). by measuring the electric current from the Complicated behavior of an interface between transmitter-side wire layer through the fluid to the water and air of each bubble is visually receiver-side wire layer. Both wire layers consist observed with a high speed camera (HSC). of nine wires respectively and are installed in the Moreover, two-phase pressure loss is measured flow channel with the distance of 3 mm in the with seven differential pressure transducers flow direction. The wire diameter is 0.3 mm. installed in the arbitrary locations in the Appearance and measuring wires of the WMS vertical direction of the flow channel. In are shown in Fig. 3. A mearing method with addition, both flow rates, pressures and the WMS is as follows.
Layout of each wire of transmitter and Clipped layout of Appearance of a wire-mesh sensor receiver in the vertical direction each wire (Dimension: mm) in the horizontal Fig. 3. Appearance and layout directionof a wire -mesh sensor in a circular flow channel
2 K. TAKASE et al.
1) Current is given to the transmitter-side Moreover, the cross-correlation wire layer; coefficient used to obtain the interfacial velocity is obtained from the relationship 2) The current flows from the transmitter- between both data measured by a couple of side wire layer through the fluid to the the WMS layers and is expressed by the receiver-side wire layer; following equation. 3) The voltage is measured at the location where each wire intersects; and, n xtt x y y 4) Void fraction between both wires cor() t1 (3) n is obtained. 22 xtt x y y The following equation is used for t1 the conversion from the voltage to the Here, x and y represent time series void void fraction. fraction data measured with the WMS UU installed upstream in the flow direction and L i, j (1) ij, the WMS installed downward. The n shows UULG the number of comparison data. The cross- Here, is the local void fraction, (i, j) correlation coefficient, cor() has the value is the position where two wires intersect in from -1 to 1, and it shows 1 when both data of the cross-section of the flow channel, U [V] x and y have the positive correlation, -1 when is the voltage measured by the WMS at two- those are the negative correlation, and 0 when phase flow condition, Each of UL [V] and those have no correlation. UG [V] is the voltage measured by the WMS at the single-phase flow condition of liquid or gas. On the other hand, gas-liquid interface velocity is determined using the time-series void fraction data measured by a couple of the WMS layers which are installed upstream and downstream of the flow channel. Two void fraction data by both WMS layers are compared. In Fig 4, time difference, , in case that the cross-correlation coefficient between both void fraction data shows the highest match is identified. Finally, the gas-liquid interface velocity, ub, is calculated by Eq. (2) using S and . Here, S is the distance between both receiver-side wire layers. S u (2) Fig. 4. Example of time variation of void fraction b data at the different vertical positions
3 BUBBLE BEHAVIOR IN LIQUID-GAS TWO-PHASE FLOW BEHIND A CROSS-SHAPED…
3. Experimental Conditions Table II. Experimental Conditions of
Superfacial Velocity of Gas in Mixture, JG The present experiments were performed under the conditions of room Case JL1 JL2 JL3 JL4 temperature, atmospheric pressure and non- JL (m/s) 0.13 0.25 0.5 1.0 heated isothermal flow. Working fluids are water and air. Superficial velocities of water In order to investigate the influence of and air, J and J , are set to be 0.13-1.0 m/s L G an obstacle in the flow channel to the void for J , and 0.025-0.4 m/s for J . Experimental L G fraction distribution, the obstacle with the conditions of JL and JG are shown in Table I dimensions of the thickness 5 mm, the height and Table II. 50 mm and the horizontal distance 58 mm as Table I. Experimental Conditions of Superfacial shown in Fig. 5 was installed in the position Velocity of Water in Mixture, JL of 20 or 100 mm upstream from the WMS. The shape of the obstacle simply simulates Case JG1 JG2 JG3 JG4 that of the grid spacer and shows a cross- J (m/s) 0.025 0.05 0.1 0.2 G shape, but the dimensions are different.
Fuel rod
Grid spacer
Dimensions of a cross- Layout of fuel rods shaped obstacle with a grid spacer
Fig. 5. Appearance of a cross-shaped obstacle which simulates the shape of the grid spacer simply
III. RERULTS AND DISCUSSION. small bubbles are distributed around the wall. On the other hand, the force which The time-averaged void fraction includes the large lift force, etc. generates distribution in the radial direction in the to the center region of the flow channel in flow channel is indicated in Fig. 6. Here, in case of the large bubbles. As a result, many case of Fig. 6 (a) J is L and in case of L 2 large bubbles are distributed to the center Fig. 6 (b) J is L . When J is low, the void L 4 G region. fraction distribution shows the wall-peak distribution. On the other hand, when JG is Figure 7 is the bubble behavior observed high, it becomes the core-peak distribution. with HSC. Here, the superficial velocity
The force which includes the lift force, etc. condition in Fig. 7 (a) are L3 and G1. Similarly, generates towards the wall direction in case that in Fig. 7 (b) is L3 and G4. In case of Fig. 7 of the small bubbles. Therefore, many (a) small bubbles are observed. Meanwhile,
4 K. TAKASE et al. since coalescence of small bubbles is promoted slug flow with a shape of the bullet is observed with increasing the superficial gas velocity, the in case of Fig. 7 (b).
0.5 0.5 [-]
G1 G2 G3 G4 [-] G1 G2 G3 G4 G5 α 0.4 α 0.4
0.3 0.3
0.2 0.2
0.1 0.1
0 0 Time Time averaged voidfraction Time Time averaged voidfraction -20 -10 0 10 20 -20 -10 0 10 20 Radial position [mm] Radial position [mm] Radial distribution of void fractin in case of L2 Radial distribution of void fractin in case of L3 (a) In case of JL=JL2 (b) In case of JL=JL4
Fig. 6. Radial void fraction distributions obtained by the conditions of JL2 and JL4
(a) In cases of JL3 and JG1 (b) In cases of JL3 and JG4
Fig. 7. Observed bubble behavior at JL3, JG1 and JG4
Void fraction distributions in the flow Gris the small bubble (i.e., bubbly flow) channel with and without the obstacle when condition. On the other hand, Fig. 8 (b) and
JL=L3 and JG=G1 are indicated in Fig. 8. (c) are the results at each position of 20 or Here, since measuring data in the circular 100 mm behind the obstacle. The void flow channel are shown as a square figure, fraction becomes high just behind the four comers in the square figure are the obstacle at Fig. 8 (b). The reason for this is outside of the measuring area. The void that the bubbles separated at the leading edge fraction is shown with the color contour. of the obstacle coalesce just behind the Here, blue is 0, red is 0.2 and other colors obstacle. Furthermore, the void fraction show the value between 0 and 0.2. The void becomes uniform at Fig. 8 (c). This is fraction distribution of Fig. 8 (a) which is the because the distance of 100 mm is long and no obstacle case shows the wall peak because the distribution is flattened.
5 BUBBLE BEHAVIOR IN LIQUID-GAS TWO-PHASE FLOW BEHIND A CROSS-SHAPED…
Fig. 8. Void fraction distributions at JL3 and JG1
Fig. 9. Void fraction distributions at JL3 and JG5
(a) No obstacle (b) 20 mm behind (c) 100 mm behind
Fig.10. Three-dimensional void fraction distributions when JL=JL3 and JG=JG5
Void fraction distributions in the flow JL=L3 and JG=G5 are indicated in Fig. 9. channel with and without the obstacle when Here, blue is 0, red is 0.4 and other colors
6 K. TAKASE et al. show the value between 0 and 0.4. The void from the channel center to the side wall; r0 and fraction distribution of Fig. 9 (a) which is the r are the radius and radial distance from the no obstacle case shows the core peak because center of the circular flow channel.
G5 is the large bubble (i.e., slug flow) The flow velocity was JL=L3 and JG was condition. Figures 9 (b) and (c) are the results changed to three kinds of G1, G3 and G5. The at the positions of 20 and 100 mm behind the open symbol shows the result in the case of no obstacle. In Fig. 9 (b) the void fraction obstacle. When the flow rate is low, the void becomes low just behind the obstacle and high fraction distribution in the horizontal cross at the region except for the area just behind section of the flow channel becomes almost that. This is because large bubbles are divided uniform, and shows the trend of the core peak as by the obstacle. In Fig. 9 (c) the void fraction the flow rate increases. On the other hand, the becomes high at the core region because the solid symbol shows a case where there is an divided bubbles coalesce again. obstacle, and since the large bubble is divided as Figure 10 shows the results of three- the flow rate increases, a high void fraction is dimensional void fraction distribution in the seen near the wall of the flow channel. case of such large bubbles. Here, Fig. 10 (a) is Velocity distributions of the bubble the result when there is no obstacle. Figure 10 interface in the flow channel with and without (b) and (c) are the results at the positions of 20 the obstacle when and JG=G5 arc indicated in and 100 mm behind the obstacle, respectively. Fiq.12. Here, blue is 0 m/s, red is 1 m/s and Here, Fig. 10 (a) is found to be the slug flow. other colors show the velocity between 0 and Also, Fig. 10 (b) shows that large bubbles are 1 m/s. The velocity distribution of Fig.12 (a) divided into four. In Fig. 10 (c), the appearance which is the no obstacle case shows the core of becoming large bubbles is seen. peak because G5 is the slug flow condition. Figure 11 shows the void fraction Fig.12 (b) and (c) arc the results at each distributions in the radial direction of the flow position of 20 or 100 mm behind the obstacle. channel. Here, the vertical axis represents the In Fig.12 (b) the velocity distribution of the local void fraction and the horizontal axis bubble interface shows four peaks in the region represents the dimensionless radial distance where no obstacle exists. Moreover, the
Without an obstacle:
JG＝JG1
JG＝JG3
JG＝JG5 With an obstacle:
JG＝JG1
JG＝JG3 J ＝J G G5
Fig. 11. Radial void fraction distributions against different JG at the position of 20 mm behind the obstacle when JL=JL3
7 BUBBLE BEHAVIOR IN LIQUID-GAS TWO-PHASE FLOW BEHIND A CROSS-SHAPED… velocity distribution of the bubble interface tendency of the void fraction distribution, as becomes high in the core region as well as the can be seen in Fig. 12 (c).
Fig. 12. Gas-liquid interface velocity distributions in the flow channel with and without the obstacle when JL=JL3 and JG=JG5
IV. CONCLUSIONS 5) The gas-liquid interface velocity can Bubbly flow behavior, void fraction be predicted from the distribution of the void distribution and interfacial velocity distribution fraction by using the equation of the cross- around the simulated spacer installed in the correlation coefficient. flow channel were investigated experimentally and at non-heated and isothermal water-air REFERENCES two-phase flow condition. The results of the [1]. Prasser, H. M., Böttger, A. and Zschau, J., “A present study arc summarized as follows: New Electrode-Mesh Tomograph for Gas-
1) When JG is low, the void fraction Liquid Flows,” Flow Measurement and distribution in the flow channel without the Instrumentation, 9, 1998. obstacle becomes the wall- peak [2]. H. M. Prasser, H. M., Krepper E. and Lucas, distribution. When J is high, it becomes G D., “Fast Wire-Mesh Sensors for Gas-Liquid the core-peak distribution; Flows and Decomposition of Gas Fraction
2) When JG is low, the void fraction Profiles According to Bubble Size Classes”, becomes high just behind the obstacle. The Second Japanese-European Two-Phase Flow reason is that the bubbles separated at the Group Meeting, Tsukuba, Japan, September, leading edge of the obstacle coalesce just 25-29, 2000. behind the obstacle; [3]. Prasser, H. M., Scholz, D., and Zippe, C.,
3) When JG is high, the void fraction 2001, “Bubble Size Measurement Using Wire- becomes low just behind the obstacle and high Mesh Sensors”, Flow Measurement and at the region except for the area just behind Instrumentation, 12, 4, 2001. that. This is because large bubbles are divided [4]. Douce1, A., Mimouni1, A.., Guingo1, M., by the obstacle; Morel2, M., J. Laviéville1, J. and Baudry1, C., 4) Behind the obstacle, the distribution “Validation of Neptune CFD 1.0.8 for of the void fraction at the channel cross section Adiabatic Bubbly Flow and Boiling Flow”, is flattened as the distance from the obstacle CFD4NRS-3, September, 14-16, Washington increases; and, DC, USA, 2010.
8 Nuclear Science and Technology, Vol.9, No. 4 (2019), pp. 09-15 Evaluation of the Potential for Containment Bypass due to Steam Generator Tube Rupture in VVER-1000/V320 Reactor during Extended SBO sequence using SCDAP/RELAP5 code
Nguyen Van Thai1,*, Doan Manh Long2, Tran Chi Thanh2 1 Department of Nuclear Engineering and Environmental Physics, School of Engineering Physics, Hanoi University of Science and Technology 2 Vietnam Atomic Energy Institute E-mail: [email protected] (Received 05 November 2019, accepted 30 December 2019)
Abstract: A severe accident-induced of a Steam Generator (SG) tube releases radioactivity from the Reactor Coolant System (RCS) into the SG secondary coolant system from where it may escape to the environment through the pressure relief valves and an environmental release in this manner is called “Containment Bypass”. This study aims to evaluate the potential for “Containment Bypass” in VVER/V320 reactor during extended Station Blackout (SBO) scenarios that challenge the tubes by primarily involving a natural circulation of superheated steam inside the piping loop and then induce creep rupture tube failure. Assessments are made of SCDAP/RELAP5 code capabilities for predicting the plant behavior during an SBO event and estimates are made of the uncertainties associated with the SCDAP/RELAP5 predictions for key fluid and components condition and for the SG tube failure margins. Keywords: Containment Bypass, SBO, SCDAP/RELAP5.
I. INTRODUCTION failure of steam generator tubes prior to the failure of one of other components (hot leg In 1990s, U.S. Nuclear Regulatory piping, pressurizer surge-line piping, and the Commission (NRC) and the Russian Federal reactor vessel) leads to discharge of some Nuclear and Radiation Safety Authority fission products into the steam generator Gosatomnadzor (GAN) agreed to work secondary system from where they may be together in BETA Project to perform a discharged to the environment through the probabilistic risk assessment (PRA) of Unit I pressure-relief valves. This sequence is of the Kalinin Nuclear Power Station (KNPS), potentially more risk-significant since it which is a VVER1000/V-338. Analysis within involves a containment bypass scenario. The the BETA Project involves different levels of a relative timing of these structural failures wide-scope PRA and the PRA model therefore affects the event sequence and represents the set of accident sequences whether the containment is bypassed. following the initial events (IEs) up to the end state of each sequence. The reports indicated This paper present thermal-hydraulic that the most significant release category for evaluations of VVER-1000/V-320 during offsite consequences is containment failure as a extended SBO event using the result of an IE with a leak from the primary to SCDAP/RELAP5 systems analysis code. In the secondary circuit [1]. In this aspect, steam general, the design features of VVER-1000/V- generator tubes comprise a majority of the 338 are similar to the standard VVER-1000/V- reactor coolant system pressure boundary and 320, excepted of the main circulation loops
©2019 Vietnam Atomic Energy Society and Vietnam Atomic Energy Institute EVALUATION OF THE POTENTIAL FOR CONTAINMENT BYPASS DUE TO STEAM… configuration which is equipped with main The turbine-driven auxiliary feedwater gate valves on the cold and hot legs (CLs and (TDAFW) system is assumed to independently HLs) [2]. The potential for “Containment fail, so no MFW or auxiliary feedwater (AFW) Bypass” is associated with natural circulation system is available; of superheated steam inside the piping loop A station battery life of four hours is that might induce creep rupture tube failure. assumed; after that time all automatic and SCDAP/RELAP5 predictions provide key fluid operator control of the pressurizer PORVs and and components condition for estimation of the SG secondary system PORVs is lost; SG tube failure margins. During the initial portion of the accident II. DESCRIPTIONS OF ACCIDENT scenario, buoyancy-driven coolant-loop natural SCENARIO AND PLANT BEHAVIOR circulation carries hot water from the core through the SGs, transferring heat to the SG The low-probability SBO base case secondary water inventory. The SG water accident event scenario results in a severe accident because none of the systems that inventory is boiled and the steam is released normally provide core cooling are assumed to through the SG PORVs. The secondary water be operable nor is any alternate equipment inventory declines and is eventually fully (e.g., security-related mitigation methods) depleted since the MFW and AFW systems are assumed to be available [4]. The accident event not operative. After that time, the core decay is initiated by a loss of off-site alternating power heats and swells the RCS water, current (AC) power, which immediately results increasing its temperature and pressure. The in reactor and turbine trips and the coast-down basic physical processes during this regard the of the four reactor coolant pumps (RCPs). transport of hot steam from the core outward Also, assumptions are made as follows: into the other regions of the reactor vessel (RV) and coolant loops. Two coolant loop The diesel-electric generators fail to natural circulation flow patterns that may be start and all AC power sources are lost; encountered subsequent to the uncovering and Letdown flow is isolated and the heat-up of the reactor core, based upon whether pressurizer level control and RCP seal injection or not the loop seals (the cold leg piping functions of the charging system are lost; connecting the outlets of the SGs to the inlets The high-pressure and low-pressure of the RCPs) remain liquid-plugged. safety injection systems are unavailable as a If liquid is cleared from a loop seal result of the AC power loss; (along with the region of the RV lower plenum The accumulator systems (four HAs) that extends above the bottom of the core are available for injecting coolant into the cold barrel, the "downcomer skirt"), hot steam is legs should the RCS pressure fall below the transported from the core through the HL, SG initial accumulator pressure, 5.9 MPa; tubes and cold legs in the normal direction of The main feedwater (MFW) flow stops flow (i.e., that seen during plant operation). and the motor-driven auxiliary feedwater This flow pattern transports the hot steam (MDAFW) system is unavailable as a result of directly (without benefit of mixing) through all the AC power loss; of the SG tubes, leading to SG tube failure
10 NGUYEN VAN THAI et al. prior to HL or pressurizer surge line failure. The issues of primary interest for However, if a loop seal remains liquid- containment bypass are: (1) do the loop seals in plugged, the more complex flow pattern all coolant loops remain liquid plugged and if develops instead. Hot steam is transported so (2) does the fluid mixing in the SG inlet through the upper portion of the HL cross plenum sufficiently slow the SG tube heat-up process so that the HL, pressurizer surge line section to the SG inlet plenum, where it is or RV will fail prior to a SG tube? [4]. The mixed with cooler steam emanating from analyses in this paper address these issues. circulations set up within the SGs, with some of the tubes flowing in the normal direction and the remaining tubes flowing in the reverse direction. The mixing process within the SG inlet plenum determines the temperatures of the steam entering the SG tubes and the steam that is returned to the RV through the lower portion of the HL cross section. Fluid mixing in the SG inlet plenum buffers the entry of hot steam into the SG tubes, thus delaying SG tube failure and making it more likely that some other component (HL, pressurizer surge line or RV) will be the first to fail. Fig. 1. RCS Layout of VVER-100/V-320 [3]
TURBINE 196
194 MSIV MSIV MSIV MSIV 187 287 387 487 186 286 386 486 190 490 189 489 182 184 BRU-A1 BRU-A1 484 482 REG REG REG REG 180 192 BRU-In1 Valve Valve Valve Valve BRU-In1 480 191 185 285 385 485 491 492
Steam Steam 169 469 dome dome
168 Separator Separator 468
Volume above Volume above 177 167 508 467 477 perforated sheets ACC ACC ACC ACC perforated sheets 152 PORV 452 Volume between 511 506 466 Volume between 176 136 166 116 416 436 476 tube bundles and tube bundles and JNG60 JNG50 JNG70 JNG80 perforated sheets perforated sheets 175 179 135 125 165 115 415 465 425 435 479 475 087 086 088 089 174 178 134 124 164 114 414 464 424 434 478 474 150 504 450
173 133 123 163 113 413 463 423 433 473
172 132 122 162 112 412 462 422 432 472
171 131 121 161 111 411 461 421 431 471 502 130 110 410 430 56
54
102 100 400 402 4 3 2 1 61-62 8 49-50 45-48 51-52 63-64 1 2 3 4
1-2 41-44 3-4 7 142 144 5-6 7-8 444 442 141 1 11 36 37-40 11 441 9 9 9-11 30 32 33 31 34 10-12 1 1 6 6 4 26 28 29 27
9 2 2 140 8 3 440 7
6 3 35 3 5
4
3
2
1 4 1 1 2 4
5 22-24 23-25 1 5
6 18-20 19-21 6 14-16 15-17
Fig. 2. Nodalization scheme of VVER-1000/V-320 Plant
11 EVALUATION OF THE POTENTIAL FOR CONTAINMENT BYPASS DUE TO STEAM…
III. PLANT NODALIZATION AND zero, the structure has not experienced any PARAMETERS SETUP WITH creep damage. If the value is one, the structure SCDAP/RELAP5 has failed due to creep damage. Two different theories (Larson Miller and Manson-Haferd) The SCDAP/RELAP5 computer code are applied in SCDAP/RELAP5 which are calculates the overall RCS thermal-hydraulic dependent upon the structural materials and the response for severe accident situations that range of stress and summarized in Table I. include core damage progression and RV heat- up and damage. The computer code is the result Table I. Equations for the time to creep rupture [5] of a merging of the RELAP5 and SCDAP computer codes. Prediction of structural failure are made based on the structure configuration, its material properties and the fluid conditions that are locally present. The code also includes models for calculating the creepture failure of structural components and these are used to predict failure time for the hot legs, pressurizer surge line, and SG tubes [5]. Since SG tubes are made from heat- resistant steel (08X18H10T) which is similar to The SCDAP/RELAP5 VVER- heat-resistant Stainless Steel, equations for 1000/V320 plant model represents the fluid creep rupture time of 316 Stainless Steel in volumes and structures in the core, RV and Table i are selected in this study. Estimated primary and secondary coolant system regions stress in SG tubes (cylindrical shaped in the plant as shown in Fig 1. The nodalization structure) is calculated as follows: diagrams for the final SCDAP/RELAP5 (P r P r ) VVER-1000/V-320 four-loop plant model are i i o o (2) provided in Fig. 2. ro ri In this work, creep rupture failure calculations are performed for the hot average SG tubes in which pressure and temperature conditions of fluid and heat structures in four SGs are used as inputs for the analyses. A parameter that measure creep damage is calculated each time step for each structure being monitored for creep Fig. 3. Model ofSG ubes rupture by following equation: IV. RESULTS AND DISCUSSION t (1) Dc t t Dc t A. Plant behaviors tr t The SCDAP/RELAP5 plant model was Where D t is creep damage at c run to a steady solution. The plant model using problem time t, t is time step at current the nodalization scheme to establish full-power problem time and tr t is time required for the steady-state conditions from which the SBO structure to fail by creep rupture at current state transient accident sequence is initiated of temperature and stress. If the value of Dc is (Figures.4&5).
12 NGUYEN VAN THAI et al.
Fig. 4. Steady-state conditions (Pressure) at primary Fig. 6. Transient condition of reactor power and and secondary loops pressure in PRZ
Fig. 5. Steady-state conditions (Mass Flow Rate Fig. 7. Transient condition of mass flow rate and Temperature) at primary and secondary loops through the primary loops and SG levels
The SBO base case event sequence was Buoyancy-driven coolant-loop natural simulated with SCDAP/RELAP5, starting from circulation carries hot water from the core time 250s when the loss of off-site power through the SGs, transferring heat to the SG occurs. Figure 5 showed the transient condition secondary water inventory until the SG heat of reactor power and pressure level in PRZ. sink is lost around 11000s (Fig. 7). It can be seen that after the SG heat sink is lost, the After a short period when the PRZ pressure the cooling afforded by system heat pressure initially falls and rises slightly due to loss to containment and RCP shaft seal leak the effects of the reactor and turbine trips, the flow is insufficient to remove the RCS heat PRZ pressure declines in response to the load, causing the RCS and PRZ pressure to cooling provided by heat removed to the SGs. increase (Fig. 6). After SG dry-out, the RCS The RCS fluid mass lost through the pressure increase is limited by multiple cycling pressurizer PORVs and SRVs and through the of the PORVs and by two cycles of the SRVs RCP shaft seal leakage paths depletes the during the period with the most-challenging RCS inventory. RCS pressurization conditions. This challenge
13 EVALUATION OF THE POTENTIAL FOR CONTAINMENT BYPASS DUE TO STEAM… is presented when the increasing temperatures occur after the time when SG tube structural cause the RCS fluid to swell, completely filling failures is experienced. the pressurizer with water. The fuel temperature increases after the heat sink is lost and the onset of fuel rod oxidation as well as fuel rod cladding rupture occurred as an unvoidable consequence (Fig. 8).
Fig. 9. Structure temperature and creep damage index
V. CONCLUSIONS
Fig. 8. Temperature of fuel pins and hot legs The paper presents preliminary assessments the VVER-1000/V-320 plant B. Creep Rupture Analysis of SG Tubes behavior during an SBO event with the Creep rupture model allows one to SCDAP/RELAP5 for key fluid and specify a "stress multiplier" in following components condition and for the SG tube meaning: a multiplier of 1.0 provides a creep failure margins. Further analyses will be rupture failure prediction based on no performed to evaluate the potential for degradation of the structural strength of the “Containment Bypass” in VVER-1000/V-320 material and multipliers greater than 1.0 reactor during extended Station Blackout represent degraded structural strengths. In this (SBO) scenarios. study, a stress multiplier from 1.0 to 5.0 (in increments of 0.1) are used for SG tube ACKNOWLEGEMENTS rupture prediction to investigate the spectrum This work was supported by the of material strengths, from undergraded to Ministry of Science and Technology (MOST) highly-degraded. R&D Project with contract number Typical results on Larson-Miller creep 05/HĐ/ĐTCB signed on 05/01/2018. rupture damage index for the hottest SG tubes is presented in Fig. 9. It was found that the .REFERENCES average-tube predicted failure margin is 2.3 [1]. Federal Nuclear and Radiation Safety and the hottest-tube predicted failure margin is Authority of the Russian Federation, Kalinin 1.7. Also, SG tube failure margin are VVER-1000 Nuclear Power Station Unit 1 insensitive to variations in the fuel damage PRA (Beta Project), NUREG/IA-0212, progression behavior primarily because these published by US NRC, 1992.
14 NGUYEN VAN THAI et al.
[2]. Design, Safety Technology and Operability [4]. C.D., Fletcher et al. SCDAP/RELAP5 Thermal- Features of Advanced VVERs, OKB Hydraulic Evaluations of the Potential for “GIDROPRESS’’, 2011. Containment Bypass During Extended Station Blackout Severe Accident Sequences in a [3]. S., Pylev. Assessment Study of Westinghouse Four-Loop PWR. NUREG/CR- RELAP5/MOD3.2 Based on the Kalinin NPP 6995, published by US NRC, 2010. Unit-1 Stop of Feedwater Supply to the Steam [5]. C. Allison et al., “SCDAP/RELAP5/MOD2 Generator No. 4, NUREG/IA-0167, published Code Manual”, Volume: I, II and III, by US NRC, 1999. NUREG/CR-5237, EGG-2555, June 1989.
15 Nuclear Science and Technology, Vol.9, No. 4 (2019), pp. 16-26
Study on transmutation efficiency of the VVER-1000 fuel assembly with different minor actinide compositions
Tran Vinh Thanh1, Vu Thanh Mai2, Hoang Van Khanh1, Pham Nhu Viet Ha1* 1Institute for Nuclear Science and Technology, Vietnam Atomic Energy Institute, 179 Hoang Quoc Viet str., Cau Giay dist., Hanoi 100000, Viet Nam 2Hanoi University of Science, Vietnam National University, 334 Nguyen Trai, Thanh Xuan, Hanoi 10000, Viet Nam E-mail*: [email protected] (Received 06 November 2019, accepted 30 December 2019)
Abstract: The feasibility of transmutation of minor actinides recycled from the spent nuclear fuel in the VVER-1000 LEU (low enriched uranium) fuel assembly as burnable poison was examined in our previous study. However, only the minor actinide vector of the VVER-440 spent fuel was considered. In this paper, various vectors of minor actinides recycled from the spent fuel of VVER-440, PWR- 1000, and VVER-1000 reactors were therefore employed in the analysis in order to investigate the minor actinide transmutation efficiency of the VVER-1000 fuel assembly with different minor actinide compositions. The comparative analysis was conducted for the two models of minor actinide loading in the LEU fuel assembly: homogeneous mixing in the UGD (Uranium-Gadolinium) pins and coating a thin layer to the UGD pins. The parameters to be analysed and compared include the reactivity of the LEU fuel assembly versus burnup and the transmutation of minor actinide nuclides when loading different minor actinide vectors into the LEU fuel assembly. Keywords: VVER-1000 LEU fuel assembly, burnable poison, minor actinide transmutation.
I. INTRODUCTION environment, separation and transmutation of the plutonium and MAs in the used fuel are The radioactive waste and spent nuclear esssential [2]. It has been realized that the fuel discharged from nuclear power plants transmutation of these actinide into either causes a big issue for the countries holding short-lived fission products or valued fissile or such nuclear installations. It is widely stable isotopes can be accomplished in fast recognized that a light water reactor (LWR) reactors, subcritical reactors or thermal with electric capacity of 1000 MWe, on reactors [1,3-7]. average, produces 20-30 metric tonnes of spent nuclear fuel annually, which consist of The VVER-1000 reactor (the Russian approximately 95 wt% uranium, 1 wt% Pressurized Water Reactor, PWR) is plutonium, 4 wt% fission products and minor nowadays operated in various East actinide (MA) [1]. In the spent fuel, only with European and Asian countries [8,9]. In a very small amount of MAs, they still addition to the Western PWRs that have dominate the decay heat load to the repository been extensively studied for their MA and cumulative long-term radiotoxicity to the transmutation capabilities [10-14], the environment. To lessen the burden for disposal VVER-1000 is also considered as a and storage of spent nuclear fuel as well as to potential candidate for transmutation of reduce its cumulative radiotoxicity to the actinide in the spent fuel stock-pile and
©2019 Vietnam Atomic Energy Society and Vietnam Atomic Energy Institute TRAN VINH THANH et al. various methods of loading and burning insignificant change in the reactivity of the fuel transuranic elements in the Western PWRs assembly while providing considerable MA may be adopted similarly to the Russian transmutation rates. VVERs. In the past studies, transmuting the MAs in the burnable poison rods [15,16] or in II. CALCULATION METHOD some other locations in the PWR fuel The VVER-1000 LEU fuel assembly assemblies has been found technically feasible specified in the OECD VVER-1000 LEU and and recommended as potential transmutation MOX (mixed oxide) Assembly Computational methods for LWRs, in particular the unique Benchmark [18] is utilized in the present advantage of loading MAs to partially replace investigation to examine the possibility of MA the excess reactivity control functions of transmutation as burnable poison in the gadolinium and boric acid. VVER-1000 reactor. The configuration of the In a previous study [17], the feasibility LEU fuel assembly are shown in Fig. 1. The of MA transmutation in VVER-1000 LEU fuel LEU assembly consists of 300 fuel pin cells assembly [18] as burnable poison was studied with 3.7wt% 235U, 12 UGD pin cells with 235 and the results showed that the total MA 3.6wt% U and 4wt% Gd2O3, 18 water filled transmutation rate of ~20% could be obtained. guide tubes for control insertion and one However, only the MA vector of the VVER- central water filled instrumentation tube. The 440 spent fuel was considered. In the present LEU fuel assembly is modeled by the SRAC work, different MA compositions recycled code. The one-sixth of the LEU fuel assembly from the spent fuels of VVER-440 [4], PWR- modeled by the PIJ module of SRAC is 1000 [15] and VVER-1000 [19] with different presented in Fig. 2; the burnup calculations burnup levels and cooling time were therefore were performed with the BURN-UP module of employed in the analysis in order to estimate SRAC; and the 107 energy groups based on the effects of various MA contents in the spent the ENDF/B-VII.0 nuclear data library were fuel to the infinite multiplication factor (k-inf) used in the SRAC calculations. of the VVER-1000 LEU fuel assembly versus burnup as well as the MA transmutation In this investigation, we intend to efficiency. The SRAC code [20] was used for load the MAs in the UGD pins of the LEU modeling of the VVER-1000 LEU fuel fuel assembly for their transmutation assembly based on the ENDF/B-VII.0 library. without significant change in the fuel The comparative analysis was conducted for assembly configuration. The purpose is to the two models of MA loading in the LEU fuel investigate the transmutation capability of assembly: homogeneous mixing in the UGD the VVER-1000 LEU fuel assembly. To this (Uranium-Gadolinium) pins and coating a thin end, we consider two approaches to load the layer to the UGD pins. The MA loading into MAs into the fuel assembly while tuning the the LEU fuel assembly will be performed gadolinium content and boron without significant modification of the concentration: (1) mixing MAs assembly configuration to minimize the cost homogeneously with UO2 and Gd2O3 in the for fuel fabrication process and respective UGD pins; and (2) coating a thin layer of changes in reactor core design. The constraint MAs around the UGD pellets. In these for these MA loadings is to ensure cases, different vectors of MAs were
17 STUDY ON TRANSMUTATION EFFICIENCY OF THE VVER-1000 FUEL ASSEMBLY… employed including those recycled from the years of cooling [19]. The MA vectors recycled spent fuel of VVER-440 with 45 from the spent fuels of the VVER-440, PWR- GWd/tonne burnup and 5 years of cooling 1000 and VVER-1000 are given in Table I. The [4], PWR-1000 with 33 MWd/tonne burnup parameters to be investigated are the k-inf of the and 10 years cooling [15] and VVER-1000 LEU fuel assembly versus burnup and the MA reactors with 40 GWd/tonne burnup and 10 transmutation rates in the LEU fuel assembly.
Fig. 1. Configuration of the VVER-1000 LEU fuel assembly
Fig. 2. One-sixth model of the VVER-1000 LEU fuel assembly by SRAC
18 TRAN VINH THANH et al.
Table I. MA vectors used in the analysis
MA vector (Fraction - at.%) Isotope 237Np 241Am 242mAm 243Am 242Cm 243Cm 244Cm 245Cm 246Cm VVER-440 48.89 31.56 0.11 14.65 0.001 0.049 4.43 0.26 0.05 PWR-1000 41.80 47.86 0.0 8.62 0.0 0.0 1.63 0.09 0.0 VVER-1000 0.0 83.75 0.10 13.16 1.22 x 10-6 0.03 2.73 0.23 3.59 x 10-6
III. MA TRANSMUTATION IN THE Additionally, in the case of loading MAs VVER-1000 FUEL ASSEMBLY recycled from spent fuel of VVER-440 reactor, the excess reactivity was generally higher at A. Homogeneous mixing of MAs in the the early burnup steps and became smaller than UGD pins the reference case after about 7 MWd/kgHM As the MAs are homogeneously mixed as gadolinium burned out. in the UGD pins of the VVER-1000 LEU fuel assembly, the gadolinium content and boron The gadolinium content was therefore concentration were adjusted with varying increased from 2 to 2.5 wt% to expect a content of MAs in order to maintain the decrease of the aforementioned high excess reactivity of the fuel assembly. It is because reactivity at the early burnup steps and the the MAs can act as burnable poison and thus boron concentration was adjusted to 400 ppm can partially replace the functions of the with respect to the MA content of 10 wt%. As gadolinium in the UGD pins and boric acid in can be seen in Fig. 3, adjusting the gadolinium the coolant for excess reactivity control of the content to 2.5 wt% and the boron concentration fuel assembly [15,16]. In this calculation, the to 400 ppm could lead to a comparable cycle content of MAs was loaded up to 10 wt%; the length while still keeping the excess reactivity content of the gadolinium was reduced from somewhat lower than the reference case. 4.0 wt% in the reference case to 2 wt%, 2.5 The gadolinium content was further wt%, 3 wt% and the boron concentration was increased from 2.5 to 3 wt% and the boron reduced correspondingly to compensate the concentration was adjusted to 350 ppm with negative reactivity insertion by the MAs. respect to the MAs content of 10 wt%. It was The results of the k-inf of the VVER- found that the behaviour of the k-inf versus 1000 LEU fuel assembly versus burnup were burnup in these cases is very similar to those illustrated in Fig. 3 for cases with MA content with the gadolinium content of 2.5 wt% as of 10 wt%. The gadolinium content was first previously mentioned. However, the cycle reduced to 2 wt% and the boron concentration length when loading 10 wt% of MA with the was decreased from 600 ppm (reference case) gadolinium content of 3 wt% and boron to 450 ppm with respect to the MA content of concentration of 350 ppm was further 10 wt%. It was found that the fuel cycle length improved and became almost identical to the when loading 10 wt% of MAs and decreasing reference case. Moreover, the excess reactivity the gadolinium content to 2 wt% and the boron of the LEU fuel assembly at the beginning of concentration to 450 ppm was substantially the cycle was also further reduced in reduced as compared to the reference case. comparison to the reference case.
19 STUDY ON TRANSMUTATION EFFICIENCY OF THE VVER-1000 FUEL ASSEMBLY…
Fig. 3. The k-inf of the LEU fuel assembly versus burnup when loading 10 wt% of MA and reducing GD to 2 wt% (upper), 2.5 wt% (middle) and 3 wt% (lower)
20 TRAN VINH THANH et al.
Table II. Transmutation capability in case of homogeneous loading 10 wt% of MA VVER-440 MA vector PWR-1000 MA vector VVER-1000 MA vector
Isotope Mass reduced Initial Mass reduced Initial Mass reduced Initial after 306 days amount after 306 days amount after 306 days amount (g) (g) (g) (g) (%) (g) (%) (g) (%) 237Np 896.78 140.19 15.63 765.09 118.35 15.47 0.00 __ __ 241Am 580.05 223.76 38.58 877.75 313.70 35.74 1536.57 482.05 31.37
243Am 269.52 49.80 18.48 158.24 26.79 16.93 241.75 38.26 15.83
244Cm 81.54 -42.09 -51.62 29.94 -28.17 -94.09 50.18 -39.56 -78.84 245Cm 4.79 -4.94 -103.14 1.65 -3.10 -187.40 4.17 -3.90 -93.42 Total 1832.67 366.73 20.02 1832.67 427.57 23.33 1832.67 473.97 25.86
The results illustrated in Fig. 3 also length is mostly unaffected when loading with imply that the MAs with the content of up to different MAs vectors. 10 wt% can be loaded into the VVER-1000 The transmutation of MA isotopes is LEU fuel assembly without significantly shown in Table II for the cases when loading affecting the fuel cycle length by means of 10 wt% of MAs and adjusting the gadolinium reducing the gadolinium content and the content to 3 wt% and the boron concentration boron concentration to offset the negative to 350 ppm. It can be seen that the reactivity insertion by the MAs. For the MA concentrations of 241Am and 243Am decreased loading up to 10 wt%, it was found that the with fuel burnup while those of 244Cm and lower excess reactivity and equivalent cycle 245Cm accumulated with fuel burnup. The length as compared to the reference case can concentration of 237Np decreased with burnup be obtained with the gadolinium content when loading the VVER-440 and PWR-1000 reduced to around 2.5-3.0 wt% and the boron 237 MA vectors. After 306 days, the Np concentration reduced to around 350-400 concentration was reduced ~15.63 % when ppm. As a result, loading 10 wt% of MA is loading the VVER-440 MA vector and ~15.47 recommended for the sake of excess reactivity % when using PWR-1000 MA vector. The control and high loading amount of MAs 241Am concentration reduced ~38.58 %, ~35.74 while keeping almost the same cycle length % and ~31.37 % while the 243Am concentration with the reference case. reduced ~18.48 %, ~16.93 % and ~15.83 % in It is found that the case of loading MA correspondence with loading VVER-440, vectors from the VVER-440 shows the highest PWR-1000 and VVER-1000 MA vectors. k-inf while that from the VVER-1000 exhibits Meanwhile, those of 244Cm and 245Cm the lowest k-inf. This also makes the excess increased ~51.62%, ~94.09 %, ~78.84 % and reactivity at the beginning of the cycle when ~103.14 %, ~187.40 %, ~93.42 % loading the MA vector from the VVER-440 corresponding to VVER-440, PWR-1000 and spent fuel higher than the two others. VVER-1000 MA vectors. The results However, Fig. 3 indicates that the fuel cycle demonstrate that the transmutation of MAs
21 STUDY ON TRANSMUTATION EFFICIENCY OF THE VVER-1000 FUEL ASSEMBLY… recycled from spent nuclear fuel in the VVER- 1000 fuel assembly is feasible from neutronic viewpoint and the total transmutation rate higher than ~20% can be achieved. Besides, it is noticed that in case of loading the VVER- 1000 MA vector without 237Np, the transmuted 241 amount of Am was much larger compared with the two other cases since the initial Fig. 4. Coating a thin layer of MA to the UGD pellet loading amount of this isotope was more than two times larger. This explained why the case The results of the k-inf of the VVER- of loading the VVER-1000 MA vector 1000 LEU assembly versus burnup when exhibited the highest total MA transmutatiton coating MAs to the UGD pins and reducing mass and efficiency as can be found in Table the gadolinium content and boron II. It is also worth noting that more than 90% concentration are shown in Fig. 5 in relation of the radiotoxidity of MAs from long time to the reference case. It was found that the storage spent fuel (more than hundred years) cases of reducing only the gadolinium come from 241Am (half-life of 432 years). content led to a significantly lower excess Thus, with the significant amount of 241Am reactivity at the beginning of the cycle and that was transmuted in the VVER-1000 fuel a considerably shorter cycle length. This assembly, it could contribute to a significant behavior of the k-inf versus burnup is reduction of radiotoxicity level of the similar to the cases of homogeneous loading as above mentioned. For that reason, the radioactive waste. boron concentration was reduced to 400 B. Coating a thin layer of MAs to the ppm, 350 ppm, and 300 ppm with respect to UGD pins the gadolinium content of 2 wt%, 2.5 wt%, In the case of heterogeneous loading of and 3 wt%. It is worth noting that the MAs in the UGD pins of the VVER-1000 LEU amount of boron concentration reduction in these cases was about 50 ppm larger than fuel assembly, MAs were coated as a thin layer the respective ones of homogeneous loading at the outside of the UGD pellets as shown in due to the self-shielding effect of MAs. The Fig. 4. The thickness of the cladding was kept excess reactivity at the early burnup steps unchanged and the outer radius of the UGD when reducing the gadolinium content to 2 region was reduced to accommodate the layer wt%, 2.5 wt%, and 3 wt% was generally of MAs. For the purpose of MA burning and lower than the reference case; except that it keeping the fuel cycle length, the MA content was slightly higher for the case of the VVER- of 10 wt% was selected in this investigation. 440 MA vector with the gadolinium content The MA coated layer (see Fig. 4) equivalent to of 2 and 2.5 wt% (Fig. 5). Sooner or later the homogeneous loading with 10 wt% of MA is k-inf in the three cases became smaller than 0.01981 cm thick. Similar to the case of the reference case. However, the cycle length homogeneous mixing, the gadolinium content with gadolinium content of 2.5 and 3 wt% and boron concentration were also reduced to was almost the same with the reference case compensate the negative reactivity insertion by while that with gadolinium content of 2 wt% the MAs. was somewhat shorter. Consequently,
22 TRAN VINH THANH et al. reducing the gadolinium content to 3 wt% and homogeneous and heterogeneous loadings was boron concentration to 300 ppm is relatively small. However, the transmutation recommended when coating with 10 wt% of mass in the case of heterogeneous loading was MA to the UGD pellets. The transmutation of slightly higher than that with homogeneous MA isotopes when coating with 10 wt% of loading, in particular for the case of VVER- MAs and reducing the gadolinium content to 440 MA vector. Table III also signify that the 3 wt% and boron concentration to 300 ppm is highest total MA transmutation mass and given in Table III. Comparing Tables III and efficiency was again achieved for the case of II, it is shown that the difference in the loading the VVER-1000 MA vector as transmutation rate of MA isotopes between compared to the two other cases.
Fig. 5. The k-inf of the LEU fuel assembly versus burnup when coating a layer of MAs to the UGD pins and reducing GD to 2 wt% (upper), 2.5 wt% (middle) and 3 wt% (lower)
23 STUDY ON TRANSMUTATION EFFICIENCY OF THE VVER-1000 FUEL ASSEMBLY…
Table III. Transmutation capability in case of heterogeneous loading of 10 wt% MA VVER-440 MA vector PWR-1000 MA vector VVER-1000 MA vector
Isotope Mass reduced Initial Mass reduced Initial Mass reduced Initial after 306 days amount after 306 days amount after 306 days amount (g) (g) (g) (g) (%) (g) (%) (g) (%) 237Np 896.79 150.40 16.77 766.06 118.35 15.45 0.00 __ __ 241Am 580.05 238.34 41.09 877.12 317.87 36.24 1536.83 494.08 32.15
243Am 269.51 51.13 18.97 157.98 24.03 15.21 241.56 34.39 14.23
244Cm 81.53 -41.27 -50.62 29.87 -25.87 -86.61 50.12 -36.42 -72.67 245Cm 4.79 -7.81 -142.27 1.65 -3.44 -208.80 4.17 -4.85 -116.46 Total 1832.67 391.79 21.38 1832.67 430.93 23.51 1832.67 484.39 26.43
IV. CONCLUSIONS level of the radioactive waste since more than 90% of the radiotoxidity of MAs from In this study, the efficiency of MA long time storage spent fuel (more than transmutation as burnable poison in the hundred years) come from 241Am (half-life VVER-1000 LEU fuel assembly was of 432 years). Furthermore, the case of examined using the SRAC code for the MA loading the VVER-1000 MA vector is homogeneous and heterogeneous loading highly recommended as it could lead to the parterns with different vectors of MAs highest 241Am transmutation mass as well as recycled from the spent fuel of VVER-440, the highest total MA transmutation mass PWR-1000, and VVER-1000 reactors. The and efficiency. gadolinium content and the boron concentration were reduced correspondingly In addition, it was shown that the MAs to compensate the negative reactivity loading in combination with the reduction in insertion by MA loading. The results show gadolinium and boron concentration could help that, with 10 wt% of MAs loading, 2.5-3.0 facilitate the excess reactivity control at the wt% of gadolinium content and 400-350 ppm beginning of the fuel cycle without significant of boron concentration were recommended effect on the cycle length. Moreover, the MA for homogeneous mixing MAs in the UGD coating approach could increase slightly the pins while 3 wt% of gadolinium content and MA burning efficiency when comparing with 300 ppm of boron concentration were homogeneous MA mixing model because of recommended for heterogeneous loading of the self shielding effect on MAs, especially for MAs in the UGD pins. It was also found that the VVER-440 MA vector. Besides, the results 237Np, 241Am, and 243Am could be indicate that, although loading of different MA significantly transmuted with a transmutation vectors slightly affected the fuel cycle length, rate as high as ~40% for 241Am. With this loading the MA vectors with lower amount of transmutation capability, burning MAs in the 237Np and higher amount of 241Am could help VVER-1000 fuel assembly could contribute significantly reduce the excess reactivity at the to a significant reduction of radiotoxicity beginning of the cycle.
24 TRAN VINH THANH et al.
Further investigation on transmutation reactivity swings,” Nuclear Technology, vol. of MAs and radiotoxicity reduction at a full 205, no. 11, pp. 1460–1473, 2019. core level and MOX core of the VVER-1000 [8]. Vladimir Sebian, Vladimir Necas, Petr Darilek, reactor is being planned. Transmutation of spent fuel in reactor VVER- 440, Journal of Electrical Engineering, Vol. 52, ACKNOWLEDGMENTS No. 9-10, 299-302, 2001.
This research is funded by Vietnam [9]. B. R. Bergelson, A. S. Gerasimov, G. V. National Foundation for Science and Tikhomirov, Transmutation of actinide in power reactors, Radiation Protection Technology Development (NAFOSTED) Dosimetry, Vol. 116, No. 1–4, pp. 675–678, under grant number 103.99-2018.32. 2005, doi:10.1093/rpd/nci249.
REFERENCES [10]. Eugene Shwageraus, Pavel Hejzlar, Mujid S. Kazimi, A combined nonfertile and UO2 PWR [1]. OECD/NEA, Minor Actinide Burning in fuel assembly for actinide waste minimization, Thermal Reactors, Nuclear Energy Agency, Nuclear Technology, Vol. 149, March 2005. NEA No. 6997, 2013. [11]. T. A. Taiwo, T. K. Kim, J. A. Stillman, R. N.
[2]. Robert Jubin, Spent Fuel Reprocessing, Hill, M. Salvatores, P. J. Finck, Assessment of Introduction to Nuclear Chemistry and Fuel a heterogeneous PWR assembly for plutonium Cycle Separations Course, Consortium for Risk and minor actinide recycle, Nuclear Evaluation with Stakeholder Participation, Technology, Vol. 155, July 2006. http://www.cresp.org/education/courses/shortc [12]. Michael A. Pope, R. Sonat Sen, Abderrafi M. ourse/, 2008. Ougouag, Gilles Youinou, Brian Boer, Neutronic [3]. C.H.M. Broeders, E. Kiefhaber, H.W. Wiese, analysis of the burning of transuranics in fully Burning transuranium isotopes in thermal and ceramic micro-encapsulated tri-isotropic particle- fast reactors, Nuclear Engineering and Design fuel in a PWR, Nuclear Engineering and Design 202, 157–172, 2000. 252, 215– 225, 2012,
[4]. Z. Perkó, J. L. Kloosterman, S. Fehér, Minor http://dx.doi.org/10.1016/j.nucengdes.2012.07.013.
actinide transmutation in GFR600, Nuclear [13]. Bin Liu, Kai Wang, Jing Tu, Fang Liu, Liming Technology, Vol. 177, No. 1, pp. 83-97, Huang, Wenchao Hua, Transmutation of minor January 2012. actinide in the pressurized water reactors, Annals
[5]. Timothée Kooyman, Laurent Buiron, Gérald of Nuclear Energy 64 (2014) 86–92, Rimpault, A comparison of curium, neptunium http://dx.doi.org/10.1016/j.anucene.2013.09.042.
and americium transmutation feasibility, Annals [14]. Bin Liu, Rendong Jia, Ran Han, Xuefeng Lyu, of Nuclear Energy 112 (2018) 748–758, Jinsheng Han, Wenqiang Li, Minor actinide https://doi.org/10.1016/j.anucene.2017.09.041. transmutation characteristics in AP1000, Annals
[6]. H. N. Tran, Y. Kato, New 237Np burning strategy of Nuclear Energy 115 (2018) 116–125, https://doi.org/10.1016/j.anucene.2018.01.031. in a supercritical CO2-cooled fast reactor core attaining zero burnup reactivity loss, Nuclear [15]. Wenchao Hu, Bin Liu, Xiaoping Ouyang, Jing Science Engineering 159, 83-93, 2008. Tu, Fang Liu, Liming Huang, Juan Fu, Haiyan [7]. H. N. Tran, Y. Kato, P. H. Liem, V. K. Hoang, Meng, Minor actinide transmutation on PWR and S. M. T. Hoang, “Minor actinide burnable poison rods, Annals of Nuclear transmutation in supercritical-CO2-cooled and Energy 77 (2015) 74–82, sodium-cooled fast reactors with low burnup http://dx.doi.org/10.1016/j.anucene.2014.10.036.
25 STUDY ON TRANSMUTATION EFFICIENCY OF THE VVER-1000 FUEL ASSEMBLY…
[16]. Wenchao Hu, Jianping Jing, Jinsheng Bi, [19]. OECD/NEA, A VVER-1000 LEU and MOX Chuanqi Zhao, Bin Liu, Xiaoping Ouyang, Assembly Computational Benchmark, Nuclear Minor actinide transmutation on pressurized Energy Agency, NEA/NSC/DOC 10, 2002. water reactor burnable poison rods, Annals of [20]. A. Kotchetkov, I. Krivitskiy, N. Rabotnov, A. Nuclear Energy 110 (2017) 222–229, Tsiboulia, S. Iougai, Calculation and http://dx.doi.org/10.1016/j.anucene.2017.06.039. experimental studies on minor actinide reactor [17]. V. T. Tran et al., Study on Transmutation of transmutation, Proceedings of Fifth Minor Actinides as Burnable Poison in VVER- OECD/NEA Information Exchange Meeting 1000 Fuel, Science and Technology of Nuclear on Actinide and Fission Product Partitioning Installations, Volume 2019, Article ID and Transmutation, pp. 289-303, Mol, 5769147, 2019. Belgium, 25-27 November 1998.
[18]. OECD/NEA, A VVER-1000 LEU and [21]. K. Okumura, T. Kugo, K. Kaneko, and K. MOX Assembly Computational Tsuchihashi, SRAC2006: A Comprehensive Benchmark, Nuclear Energy Agency, Neutronics Calculation Code System, JAEA- NEA/NSC/DOC 10, 2002. Data/Code 2007-004, 2007.
26 Nuclear Science and Technology, Vol.9, No. 4 (2019), pp. 27-33 Determination of in situ detection efficiency for IM-NAA of non-standard geometrical samples
Nguyen Duy Quang1, Trinh Van Cuong1, Tran Tuan Anh1, Ho Van Doanh1, Nguyen Thi Tho1, Ho Manh Dung2 1 Dalat Nuclear Research Institute, 01 Nguyen Tu Luc Street., Da Lat, Lam Dong. 2 Center For Nuclear Techniques, 217 Nguyen Trai street, District 1, HCM City. Email: [email protected] (Received 29 November 2019, accepted 30 December 2019)
Abstract: The k0-based internal mono-standard (IM) method was first proposed for the concentration analysis of samples of non-standard geometry in the 2000s. The method has demonstrated several advantages such as the elimination of gamma-ray self-attenuation and geometrical effects. On the other hand, the accuracy of the method principally depends on the in situ relative detection efficiency, which requires to be obtained in each measurement. Therefore, the relative detection efficiency is always under consideration for the improvement of the analysis results. The present paper describes a simple and automatic procedure for the determination of the relative efficiency using one or more activation products emitting gamma rays over a considered range of the spectrum. The procedure can be applied for INAA and PGNAA analysis. Keywords: INAA, PGNAA, internal mono-standard method, relative efficiency, non-standard geometry samples.
I. INTRODUCTION and then used for elemental concentration analysis. The in situ relative detection Neutron activation analysis (NAA) is a efficiency plays a key role in the analysis as it sensitive multi-element analytical technique is valuable to the correction of sample used for both qualitative and quantitative geometry effects [6]. In this study, a computer analysis of elements in a vast amount of code for the determination of the in situ materials. The technique has applications in relative efficiency has been developed. Using chemistry, geology, archeology, medicine, Prompt Gamma NAA (PGNAA) and environmental monitoring and even in forensic Instrumental NAA (INAA) nuclear data, the science [1-4]. However, there are limitations software requires only a peak area report file due to sample size for application in bulk for calculation of relative efficiency and analysis, in particular for archaeology and perform further analysis. Results from cultural art artifacts, forensic materials as well measurements of standard reference materials as geological studies . For example, large and have been found to be in good agreement with non-standard geometry samples give rise to certified values. problems of neutron self-shielding, gamma rays attenuation and certain geometrical II. CONTENT effects. Therefore, the internal mono-standard analysis method has been proposed to A. Subjects and methods overcome the mentioned difficulties [5]. In this In INAA using k0 approach, consider method, the in situ relative detection efficiency two arbitrary elements x and y presented in an is required to be obtained in each measurement activated sample which emits two series of
©2019 Vietnam Atomic Energy Society and Vietnam Atomic Energy Institute DETERMINATION OF IN SITU DETECTION EFFICIENCY FOR IM-NAA OF NON-STANDARD... characteristic gamma rays Ex,i and Ey,j (i, j = 1, As clearly seen in eq. (3), the relative 2,...), respectively. The mass ratio of element x concentration of element x to y can be to y can be expressed as follow [5]: determined by the ratio of full peak detection efficiencies. This leads to the need for using m (SDC ( f Q ( ))) PkEEE 0, x 0 y x,,, i y j y j (1) relative detection efficiency which can be m( SDC ( f Q ( ))) P k y0 x Ey,,, j E x i 0, E x i determined straightforwardly from eq. 3 using Where S is the saturation factor, D is a fitting procedure. The relation of detection the decay factor, C is the measurement efficiencies is as follow: cE m factor, f is the ratio of the thermal to xi, x (6) EEy,, jcm x i epithermal neutron fluxes, Q0() is the ratio Eyyj, of the resonance integral-to-thermal neutron Hence, cross-section corrected for the non-ideal c epithermal neutron flux distribution (α), P is m Exi, ln( ) lnx ln ln( ) (7) the peak area and ε is the full energy peak EEy,, jmc x i yEyj, detection efficiency. In case of high f, the The efficiencies at different gamma value of in eq. (f Q00 ( )))yx / ( f Q ( ))) energies of each element are derived from (1) tends to unity and therefore, eq. (1) can eq. (6) be simplified as: c ln( ) lnEx,1 ln( ),i 2,3,... (8) ()SDC Pk EEx, i x ,1 mx y EEEx,,, i y j0, y j c (2) Exi, m() SDC P k y x Ey,,, j E x i0, E x i cE ln( ) lny,1 ln( ),j 2,3,... (9) EEy, jc y ,1 The situation is very simple in case of Eyj, PGNAA where the correction factors S, D and Thus, relative efficiency curves in C can be eliminated. Though, it should be logarithmic scale constructed individually noted that the k databases for two different 0 from each element are expected to be differed techniques are different. Having it in mind, the by constant factors, say t. Because of using mass ratio in both INAA and PGNAA can be relative efficiency, an arbitrary positive value rewritten as: can be firstly assigned to one detection
m cE E efficiency of each element, e.g. Arb E 10% x xi, yj, (3) k ,1 mc y Ey,, j E x i for any k-th element where “Arb” indicates the first choice of detection efficiency. The Where c is the coefficient calculated relative detection efficiencies are then using k database and experimental data 0 corrected by t-factors: P c Exi, (4) Exi, ln(Rel ) ln(Arb ) t (10) (SDC ) ( f Q ( ))) k Ek,, i E k i k x0 x 0, Exi, and in some cases can be simplified as In general, the expression for the relative efficiency curve is P Exi, n c (5) i Exi, ln(Rel (E )) a (ln E ) (11) ()SDCxE k0, i xi, i0
28 NGUYEN DUY QUANG et al.
Where ai is the coefficient and n is the of experimental relative detection efficiency at order of the polynomial that can be chosen different energies of all elements, i.e. depending on the energy range of interest. calculation of all characteristic factors t in eq. After the relative efficiency calibration curve 10 (see Fig. 2). In the next step, least-square (11) is constructed, the relative concentrations fitting is performed to construct a new can be calculated from eq. (3) and converted to efficiency curve. The Goodness Of Fit (GOF) absolute concentration using a well-know mass in the fitting step is used for stop condition. fraction of an element presented in the sample. The loop is forced to stop whenever the GOF If the concentration of m elements are required starts to increase. On the other hand, it will to be analysed there will be m+n+1 parameters stop if the number of the loop is large enough needed to be optimized in fitting procedure, and the GOF is almost saturated. A typical curve corresponding to the detector used for including ai, i=0,1,..n and tk, k=1,2,...m. The spectrum acquisition may be chosen as the iteration loop for determination of all original reference efficiency curve. It has been mentioned parameters is presented in Fig. 1. found that the employment of different original The loop starts with a reference efficiency reference efficiency curves gives a very small curve which is then used for the correction divergence on final analysis results.
Fig.1. Iteration for optimization of fitting parameters
Fig. 2. Illustration of relative efficiency correction for iron by tFe-factor. (a) before correction, (b) after correction
29 DETERMINATION OF IN SITU DETECTION EFFICIENCY FOR IM-NAA OF NON-STANDARD...
For evaluation of the method The detector solution is 1.90 keV for 1332.5 60 performance uscore test was implemented. uscore keV ( Co). factor was calculated as follow:
mmxy uscore (12) 22 mmxy
In this study, the limiting value for uscore has been set to 2.58 for a level of probability at 99% to determine if a result passes the evaluation. The uscore values less than 1.96 mean that the result probably does not differ significantly from the certified value while uscore less than 1.64 that means the result does not differ significantly from the certified value. B. Results and discussion In experiments, standard reference materials, BIR-1 and SMELS-III, were used for quality verification of the element concentration for PGNAA and INAA, respectively. BIR-1 sample had been irradiated and analyzed by KFKI lab using k0 approach. The acquired spectrum was re- used to construct a relative efficiency curve for internal monostandard analysis. In case of INAA, the SMELS III sample was sealed in a polyethylene bag and irradiated for 09 hours at the Rotary Rack channel of Dalat research reactor. The ratio f between thermal and epithermal neutron flux is 37.3 and epithermal neutron spectrum factor α is
0.073 [7]. The measurement was carried out for about ~18 hours after ~5 days of decay. Fig. 3. Construction of in situ relative efficiencies. To assess the feasibility of the method, (A) BIR-1 sample, original curve 1, analysis of large samples has been attempted. (B1) SMELS-III sample, original curve 1, (B2) SMELS-III sample, original curve 2. Two NIST-679 samples of different sizes and weights were prepared and irradiated. The In situ relative efficiency for each small one (103.25mg) was analyzed by k0 sample has been constructed using the approach while the large (1.365g) were mentioned fitting procedure. Fig. 3 indicates studied by IM-method using both optimized the improvement of efficiency curves before and non-optimized efficiency curves. Gamma and after fitting steps. As for the BIR-1 sample, spectrum was acquired by an HPGe detector. the efficiency curve demonstrates a small
30 NGUYEN DUY QUANG et al. change at high energy region while in the low the BIR-1 sample. A majority of IM’s energy region the divergence becomes large. element sconcentration has been found in To assess the feasibility of the procedure, good agreement with certified values, two different original efficiency curves were excluding results for Cr, Mn and Co. used in the study of the SMELS-III sample. However, k0 approach shows a very similar As clearly seen, the obtained relative pattern in case of Cr and Co when the efficiency curves are very similar, showing results are about 2 times greater than the the differ from each other by nearly a certified values. constant factor of about 1.3 in the logarithm The situation becomes better when using of relative efficiency. IM’s method in the INAA study of the Analysis of element concentration in SMELS-III sample (see Table. II). Despite the samples has been implemented using the using different original efficiency curves, the orresponding efficiency curve. Table I results are convergent and very close to shows the mass fraction of 15 elements in assigned values. Table I. Concentration found in BIR-1 sample (unit: Oxide form – wt%, Element – ppm)
Certified values k0-approach (KFKI) IM-approach Note No. El Conc. (1) +/- (2) Conc. Rel. Unc. (3) Conc. +/- u-score
1 Na 1.82 0.045 1.82 1.9 1.82 0.06 0.00 Oxide form 2 Mg 9.7 0.079 10 5. 9.4 0.6 0.50 Oxide form 3 Al* 15.5 0.15 15.0 2.6 15.5 0.9 0.00 Oxide form 4 Si 47.96 0.19 48 1.4 45.37 1.47 1.75 Oxide form 5 Ca 13.3 0.12 12.8 3.0 12.5 0.4 1.92 Oxide form 6 Sc 44 1 56 2.7 49.7 3.4 1.61 Element 7 Ti 0.96 0.01 1.01 2.6 0.98 0.03 0.63 Oxide form 8 V 310 11 401 3.5 336 40 0.63 Element 9 Cr 370 8 516 5. 626 47 5.37 Element 10 Mn 0.175 0.003 0.175 2.4 0.203 0.010 2.68 Oxide form 11 Fe 11.3 0.12 11.2 2.4 11.5 0.4 0.48 Oxide form 12 Co 52 2 104 4.0 116 8 7.76 Element 13 Ni 170 6 180 7. 200 37 0.80 Element 14 Sm 1.1 - 0.80 3.6 0.78 0.04 - Element 15 Gd 1.8 0.4 1.6 5. 1.82 0.11 0.05 Element Table II. Concentration found in SMELS III sample (unit: ppm) IM-approach, original IM-approach, original Assigned values k -approach No. El. 0 curve 2 curve 1 Conc. +/- Conc. +/- Conc. +/- u-score Conc. +/- u-score 1 Sc 1.140 0.031 1.21 0.01 1.136 0.039 0.08 1.15 0.039 0.2 2 Cr 86.7 2.6 90.01 3.79 83.5 2.9 0.82 88.8 3 0.53 3 Fe* 8200 190 8655 357 8200 190 - 8200 190 - 4 Co 24.3 0.33 25.45 1.04 23.9 0.8 0.46 23.9 0.8 0.46 5 Zn 618 11 660 27 608 21 0.42 610 21 0.34 6 Se 131 6 144 6 133.5 4.9 0.32 143 5 1.54 7 Sr 8150 200 8891 374 7767 272 1.13 8132 286 0.05 8 Cs 20.80 0.34 22.53 0.92 19.5 0.7 1.67 20.1 0.8 0.81 9 Tm 23.3 0.7 25 1 24.6 1.2 0.94 26.2 1.3 1.96 10 Yb 20.7 0.5 22.5 0.9 22.5 0.8 1.91 24.1 0.8 3.6 11 Au 0.901 0.016 - - 0.832 0.028 2.14 0.879 0.03 0.65
31 DETERMINATION OF IN SITU DETECTION EFFICIENCY FOR IM-NAA OF NON-STANDARD...
Table III. Concentration found in NIST-679 samples (unit: Fe-wt%, others-ppm) IM approach IM-approach Datasheet k -approach, u- 0 Non-optimized efficiency, Optimized efficiency, Value's Small sample score No. El. Large sample Large sample Conc. +/- Conc. +/- Conc. +/- Conc. +/-
1 Sc 22.5 - 22.4 0.5 23.3 0.6 23.4 0.7 - 2 Cr 109.7 4.9 120.2 5.5 107.5 2.7 106.7 3.3 0.51 3 Fe* 9.05 0.21 8.95 2.2 9.05 0.21 9.05 0.21 - 4 Co 26 - 24.9 0.8 26.3 0.6 26.1 0.7 - 5 Zn 150 - 163 15 130.4 3.5 130.4 3.8 - 6 Rb 190 - 201 17 211.7 6.1 211.8 6.6 - 7 Cs 9.6 - 9.4 0.5 9.6 0.3 9.54 0.36 - 8 Ce 105 - 118 4 98.1 2.6 105.4 3.7 - 9 Eu 1.9 - 1.6 0.1 1.62 0.04 1.65 0.06 - 10 Hf 4.6 - 4.3 0.2 4.54 0.13 4.53 0.16 - (1) Concentration, (2) Absolute Uncertainty, (3) Relative Uncertainty (%). *Reference Element
The study of samples of different sizes various geometry is required to verify and and weights has been attempted. Table III optimize the method. compares mass fractions of 10 elements presented in NIST-679 samples with ACKNOWLEDGMENT certified and informative values. As clearly The authors are thankful to the Ministry seen, the difference in sample size and of Science and Technology of Vietnamese weight gives rise to some variation in government for financial support through the obtained relative concentrations of Cr, Zn, ministerial-level project DTCB-01/18- Rb, Ce to Fe. If the optimized efficiency VNCHN. curve is employed it can help to correct the results for Ce. However, the difference in REFERENCES sample size and weight gives an [1]. Nguyen Canh Hai, Nguyen Nhi Dien, Vuong insignificant change in the relative Huu Tan, Tran Tuan Anh, Pham Ngoc Son, Ho efficiency curve. Therefore, it is desired for Huu Thang, “Determination of elemental further verification of the method using concentrations in biological and geological large samples of various shape and size. samples using PGNAA facility at the Dalat research reactor”, J Radioanal Nucl Chem, 319 III. CONCLUSIONS (3), 1165-1171, 2019.
[2]. Rahat Khan, Shayantani Ghosal, Debashish A procedure for the determination of in Sengupta, Umma Tamim, Syed Mohammod situ relative detection efficiencies for internal Hossain, Sudha Agrahari, “Studies on heavy monostandard neutron activation analysis has mineral placers from eastern coast of Odisha, been proposed. The element concentration India by instrumental neutron activation found in some standard samples were in good analysis”, J Radioanal Nucl Chem, 319 (1), agreement with certified values. The method 471-484, 2018. is promising as it has been successfully [3]. Jan Kamenik, Filipa R. F. Simoes, Pedro M. F. applied to a nonstandard geometrical sample. J. Costa, Jan Kucera, Vladimir Havranek, However, further analysis of large samples of “INAA and ion-beam analysis of elemental
32 NGUYEN DUY QUANG et al.
admixtures in carbon-based nanomaterials for [6]. A. G. C. Nair, R. Acharya, K. Sudarshan, S. battery electrodes”, J Radioanal Nucl Chem, Gangotra, A. V. R. Reddy, S. B. Manohar and 318 (3), 2463-2472, 2018. A. Goswami. “Development of an internal monostandard instrumental neutron activation [4]. Pasquale Avino, Geraldo Capannesi, Luigina analysis method based on in situ detection Renzi, Alberto Rosada, “Physiological efficiency for analysis of large and nonstandard parameters affecting the hair element content of geometry samples”, Anal. Chem. 75, 4868- young Italian population”, J Radioanal Nucl 4874, 2003. Chem, 306 (3), 737-743, 2015. [7]. Manh-Dung Ho, Quang-Thien Tran, Van- [5]. Sudarshan, K.; Nair, A. G. C.; Goswami, A. J. Doanh Ho, Dong-Vu Cao, Thi-Sy Nguyen,
Radioanal. “A proposed k0 based methodology “Quality evaluation of the k0-standardized for neutron activation analysis of samples of neutron activation analysis at the Dalat research non-standard geometry”, J Radioanal Nucl reactor”, J Radioanal Nucl Chem, 309 (1), 135- Chem, 256 (1), 93-98, 2003. 143, 2016.
33 Nuclear Science and Technology, Vol.9, No. 4 (2019), pp. 34-40
Evaluating uncertainty of some radiation measurand using Monte Carlo method
Bui Duc Ky*, Nguyen Ngoc Quynh, Duong Duc Thang, Le Ngoc Thiem, Ho Quang Tuan, Tran Thanh Ha, Bui Thi Anh Duong, Nguyen Huu Quyet, Duong Van Trieu Institute for Nuclear Science and Technology, Hanoi, Vietnam Email: [email protected] (Received 07 November 2019, accepted 26 December 2019)
Abstract: Evaluating measurement uncertainty of a physical quantity is a mandatory requirement for laboratories within the recognition ISO/IEC 17025 certification to access reliability of measured results. In this work, the uncertainty of ionizing radiation measurements such as air-kerma, personal dose equivalent was evaluated based on GUM method and Monte Carlo method. An uncertainty propagation software has been developed for evaluation of the measurement uncertainty more convenient. Key words: Uncertainty measurement, Monte Carlo method.
I. INTRODUCTION In this work, uncertainty of air-kerma and personal dose equivalent The measurement uncertainty is a quantities were evaluated by both methods. characteristic for the dispersion of measurable These two quantities are the fundamental values of a quantity to be measured [1, 2, 3]. quantities in radiation protection field. All Because without the measurement uncertainty, experimental data published in this work the results of the measurements cannot be were measured at the Secondary Standard compared to each other, nor can be compared Dosimetry Laboratory belongs to Institute for to conventional true values. Nuclear Science and Technology. In the field of measurement of ionizing II. METHODS radiation ISO/IEC and IAEA has provided guidance on measurement uncertainty for The relationship between a single real different measurement quantities. These output quantity y and a number of real input documents are primarily based on the quantities has the following equation (1). evaluation methods provided by the
International Commission on Measurement Guidelines (JCGM). The uncertainty A. GUM method propagation method described in the JCGM Uncertainty of the input quantities 100:2008 “Guide to the expression of are divided into two categories, based on how uncertainty in measurement” is often referred its values were evaluated. If it was evaluated to as the GUM method. The Monte Carlo based on statistical means, they are called type method was described in its supplement 1 A. Otherwise, they are nominated type B. “Guide to the expression of uncertainty in However, it is worth mentioning that this measurement – Propagation of distributions classification does not affect the uncertainty using a Monte Carlo method”. propagation law.
©2019 Vietnam Atomic Energy Society and Vietnam Atomic Energy Institute BUI DUC KY et al.
The uncertainty of output quantity Y is approximation. This makes calculated calculated as [2, 3, 4]: uncertainty in many cases inaccurate.
B. Monte Carlo method ∑ The Monte Carlo method simulates input
∑ ∑ ( ) quantities based on initial probability Where, and are distribution. The distribution of the input sensitivity coefficients, is the quantities will affect the output quantity estimated covariance associated with and . according to the model in the equation 1. As result, the distribution function of the output The GUM uncertainty framework quantity was obtained. Therefore, not only the requires [4]: standard deviation but other characteristics of a) The non-linearity of the measurement output quantity can be determined (i.e. function to be insignificant. skewness, coverage interval). The process of b) The central limit theorem to apply, uncertainty evaluation using Monte Carlo implying the representativeness of the method was presented in fig.1 [2, 4]. Probability density function (PDF) for the The advantages of the Monte Carlo output quantity by a Gaussian distribution or a method are that it doesn’t make any t-distribution. assumption about linearity of measurement c) The adequacy of the Welch- function nor PDF of the output quantity. Satterthwaite formula for calculating the Therefore, the Monte Carlo method is valid for effective degrees of freedom. wider range of problem compare to GUM method. Its result can be used to validate the In practice, the GUM method is result of GUM method. frequently used in violation of the requirements listed above or without knowing whether these The disadvantage of the Monte Carlo requirements hold (with an unquantified degree method is that it is impossible using hand of approximation). Furthermore, the equation 2 calculation. This method must be implemented is only the first order Taylor series in a computer software.
Fig. 1. Uncertainty measurement using a Monte Carlo method for a univariate, real measurement function.[4]
35 EVALUATING UNCERTAINTY OF SOME RADIATION MEASURAND USING…
C. INST-MC software measurement are included: Gaussian A software program, namely INST-MC distribution, t-distribution, Poisson was developed to facilitate the uncertainty distribution, uniform distribution, triangular evaluation. The interface of the software is distribution, etc. The program was validated shown in fig.2. Both methods of uncertainty by comparing with the NIST uncertainty evaluation discussed above were machine, an uncertainty software has been implemented. In the first version of INST- developed by National Institute of Standard MC, most common distributions in radiation and Technology/ USA.
INST-MC
Fig. 2. Interface of INST-MC uncertainty software.
III. RADIATION MEASURAND for the deviation of the actual air pressure P from the reference temperature 137 A. Air-kerma of Cs source. mbar, corrects for the unstable of The air-kerma is obtained from Eq.3 ionization chamber, corrects for the possible deviation of the actual distance of the
reference source to the measuring instrument from the nominal calibration distance. Where: is calibration factor of ionization chamber, is correction factor of Using the equation 2, uncertainty of air- the difference between the reference beam kerma U(K) and corrected reading U(Mcorr) is quality, , and the actual quality, , during the given by: measurement and is corrected reading of ionization chamber: √
is reading of ionization chamber,
( ) ( ) ( ) corrects for the deviation of the actual
air temperature T from the reference ( ) ( ) temperature K, corrects √
36 BUI DUC KY et al.
B. Personal dose equivalent using corrects for the inhomogeneity response of
TLD dosimeter. dosimeter, corrects for others affect. Personal dose equivalent Hp(d) is The uncertainty of personal dose obtained from equation (7): equivalent ) is obtained from equation 10.
Where: M is reading of exposed dosimeter, is reading of background dosimeter. ( ) ( )
ECC is elements correction coefficients. ( ) ( )
̅
( ) ( ) ( ) ( ) √ ̅ is average reading of n dosimeters and is reading of ith dosimeter (i= ̅̅̅ ̅̅̅ ̅ ). IV. RESULTS AND DICUSSION
RCF is reader calibration factor: A. Uncertainty of air-kerma with the 137 gamma rays of Cs.
The measurement uncertainty of air- C is reading of calibration set, is reading of kerma using approximation method background calibration set, is conventional estimated around 1.26%. The uncertainty true value (exposed dose), corrects for corresponds to a coverage factor k = 1 and a energy dependence of dosimeter, corrects level of confidence factor of approximately p = for non-linearity of dosimeter, corrects for 68%. The details of uncertainty of components the loss signal before reading dosimeter, showed in Table I. Table I. Uncertainty budget of air-kerma
Relative standard Type of Degree of Source of uncertainty deviation (%) uncertainty freedom Calibration factor of ionization 0.41 B - chamber Reading of ionization chamber 0.10 A 9 Air pressure 0.11 B - Air temperature 0.10 B - Distance 0.13 B - Stability of ionization chamber 0.60 A, B - Others 1.00 A, B -
1.26
37 EVALUATING UNCERTAINTY OF SOME RADIATION MEASURAND USING…
Calculation model of air-kerma Based on data in table III, Monte Carlo using INST-MC software is given by method estimated measurement uncertainty of equation (11): air-kerma approximately 1.21% correspond a coverage factor k = 1 and a level of confidence
factor of approximately p = 68%.
Table II. Distribution of input quantities of air- kerma
Average value Standard Degree of Input quantities, Distribution of deviation freedom Reading of ionization chamber, 4.175 0.035 Student 3
Calibration factor of ionization 50.23 0.27 Student 3 chamber,
0.997 0.007 Rectangular -
1.008 0.0006 Rectangular -
1.001 - Constant -
0.999 - Rectangular -
Figure 3 show the detail of the close. That means GUM method is valid, uncertainty result of air-kerma using and uncertainty of air-kerma can be Monte Carlo method. The uncertainty calculated by either GUM method or evaluation results of two method are very Monte Carlo method.
Fig. 3. Measurement uncertainty result of air kerma using INST-MC software
B. Uncertainty of dose equivalent uncertainty corresponds to a coverage factor using TLD dosimeters. k = 1 and a level of confidence factor of The measurement uncertainty of dose approximately p = 68%. The details of equivalent using approximation uncertainty of components are showed in method estimated around 18.1%. The Table III.
38 BUI DUC KY et al.
Table III. Uncertainty budget of dose equivalent
Relative standard Type of Degree of Source of uncertainty deviation (%) uncertainty freedom Conventional true value (exposed dose), Hc 2.36 B -
Reading of dosimeters, M 2.7 A 4 Elements Correction Coefficients, ECC 2.1 A 99 Reader Calibration Factor, RCF 3.5 A, B 9
12.6 B -
6.5 B -
9.1 B -
1.9 B -
others 3.0 - -
18.1
Calculation model of dose equivalent equivalent approximately 18.7% correspond a is given by equation (7). Based on coverage factor k = 1 and a level of confidence distribution of input quantities of personal dose factor of approximately p = 68%. Fig.4 show equivalent in table IV, INST-MC the detail of uncertainty of personal dose estimated measurement uncertainty of dose equivalent Hp(d) using Monte Carlo method.
Table IV. Distribution of input quantities of personal dose equivalent
Standard Degree of Input quantities, Value of Distribution deviation freedom
Reading of dosimeter, 4817 90.48 Student 5
Elements Correction Coefficients, ECC 0.88 0.07 Student 99
Reading of calibration set, 15443 545 Student 5
Conventional true value, 6.8 0.082 Student 7
0.615 0.08 Rectangular -
1.02 0.034 Rectangular -
1.02 0.091 Rectangular -
0.98 0.019 Rectangular -
39 EVALUATING UNCERTAINTY OF SOME RADIATION MEASURAND USING…
Fig. 4. Measurement uncertainty result of ) using INST-MC software
IV. CONCLUSIONS [4]. NPL report MS6 - Software support for The measurement uncertainty of air- metrology – Uncertainty evaluation, M G Cox and P M Harris, National Physical kerma and the personal dose equivalent Laboratory, march 2010. were evaluated by the GUM method and Monte Carlo method, which were [5]. On a Monte Carlo method for measurement implemented in the INST-MC software uncertainty evaluation and its program. The results showed that deviations implementation, P M Harris and M G Cox, of air-kerma and personal dose equivalent Metrologia 51, 2014. calculated by two methods are 3.9% [6]. Uncertainty analysis of phase and amplitude of and 3.3%, respectively. Compared with the harmonic components of bearing inner ring approximation method, INST-MC is more four-point roundness measurement, Raine convenient to calculate and it also shows the Viitala et al, Precision Engineering, 2018. probability distribution of the obtained results. [7]. Estimation of uncertainty of effective area of a In further research, uncertainty pneumatic pressure reference standard using evaluation of other quantities in SSDL will be Monte Carlo method, Singh et al, Indian estimated by Monte Carlo method. Journal of Pure & Applied Physics, 2016. [8]. A Guide on Measurement Uncertainty in RERERENCES Medical Testing, Technical Guide 4, the [1]. ISO 17025:2005, General requirements for Singapore Accreditation Council, 2013. the competence of testing and calibration [9]. Calibration of radiation protection and laboratories. monitoring instrument, IAEA safety report [2]. Evaluation of measurement data – Guide to the series No.16, 2000. expression of uncertainty in measurement, [10]. Evaluation of measurement data – JCGM 100:2008, 2008. Supplement 1 to the "Guide to the expression [3]. Revision of the Guide to Expression of of uncertainty in measurement" – Propagation Uncertainty in Measurement, Walter Bich et of distributions using a Monte Carlo method, al, Metrologia, 49, 2012. JCGM 101:2008, 2008.
40 Nuclear Science and Technology, Vol.9, No. 4 (2019), pp. 41-47 Evaluation of image reconstruction algorithms in cone-beam computed tomography technique
Tran Thuy Duong, Bui Ngoc Ha Hanoi University of Science and Technology Email: [email protected] (Received 07 November 2019, accepted 15 December 2019)
Abstract Cone-beam computed tomography (CBCT) technique is largely used in medical diagnostic imaging and nondestructive materials testing, especially in cases which require fast times and high accuracy level. In this paper, the pros and cons of Feldkamp-Davis-Kress (FDK) and simultaneous iterative reconstruction technique (SIRT) algorithms used in CBCT technique is studied. The method of simulating CBCT systems is also used to provide richer projection data, which helps the research to evaluate many aspects of algorithms. Keywords: CT, cone beam, reconstruction algorithm, FDK, SIRT.
I. INTRODUCTION Together with the development of hardware, CBCT reconstruction algorithms are Computed Tomography (CT) was first also researched and improved. Practical CBCT employed in Medical Imaging Diagnosis in the system employs two major algorithms to early 1970s. It has been improved with seven reconstruct the image from projections. They commercial generations in the past 50 years. are Filtered Back Projection (FBP) so called a Today, CT became popular not only in medical convolution method and series expansion applications but also in material analysis and method which includes Algebraic non-destructive testing (NDT). Especially, the Reconstruction Technique (ART), latest CT generation so-called Cone Beam CT Simultaneous Iterative Reconstruction (CBCT) can produce three-dimensional (3D) Technique (SIRT), Iterative Least-Squares imaging with high-resolution to makes CT now Technique (ILST). Series expansion method is more suitable for observing the inner structure the most accurate reconstruction method, but it of materials and detecting material defects. The requires a very high level of hardware. This most advantages of using cone-beam geometry method only can be implemented when all are reducing data acquisition time and projections are collected so that it slows the increasing spatial resolution. By employing reconstruction period. Although the cone-beam geometry, whole 3D information reconstructed image has poorer accuracy, inside the sample could be gathered in a short convolution method is still widely used time and could be used in reconstruction because of their flexibility and having higher process to obtain 3D imaging or cross-section processing speed. FBP is one of the important of any sample’s part. Recent, development of algorithms for practical CBCT due to their radiation detector technology especially Flat simplicity and parallel computing capability, Panel Detector (FPD) allowed CBCT is being FBP may produce a high-quality image if step widely studied and utilized in diversified angle between two adjacent projections is applications [1-3]. small enough [1, 3-6].
©2019 Vietnam Atomic Energy Society and Vietnam Atomic Energy Institute EVALUATION OF IMAGE RECONSTRUCTION ALGORITHMS IN CONE-BEAM COMPUTED…
Fig. 1. CBCT configuration
In this paper, standard Back projection source in this simulation is cone-beam X-ray method is Feldkamp-Davis-Kress algorithm tube with cone angle is 30o, the focal spot size is (FDK) and standard series expansion method is 4μm and maximum tube voltage is 240kV. Two SIRT will be evaluated . This paper will show samples (phantoms) are used to generate dataset results and evaluations of image quality when of projections in this research have rectangular applying difference filter mask in and cylindrical shape with dimensions of reconstruction process. Thereby it can show 2.5×2.5×6.0 cm (Length x Width x Height) and that FDK algorithm is the most appropriate 10×8 cm (Diameter x Height) respectively. algorithm for industrial CBCT in Vietnam. They are made by plastic and aluminum. F4 Tally combine with Fmesh Card are used in II. CONTENT MCNP to get the result (average radiation flux in a cell), this allows a result has statistical error A. Subjects and methods at single cell <3% and can meet requirement for Today, the study of CBCT technique is good imaging quality in radiography. A two- still new in Vietnam. There is a national project dimensional matrix is a result of each simulation (KC.05.18/16-20) which is being implemented process in which contained flux value of cells, in Hanoi University of Science and Technology. equivalent to the gray level of pixels. Data But hardware of system in this project has not processing and imaging reconstruction are finished yet. Hence, this research employed the implemented by using Python language. Monte Carlo simulation to simulate CBCT Projection data obtained from simulation will be system which has configuration is shown in used to reconstruct 3D image of sample through Figure 1. In this configuration, detector and X- FDK and SIRT algorithms, several filter masks ray source have a fix position. The sample is are also applied to evaluate result. In this rotated around the axis which is perpendicular research, ASTRA Toolbox – an open-source to the line between X-ray source and center of tool will be used to integrate into MATLAB or detector array. With the configuration shown in PYTHON language to facilitate developing figure 1, X-ray detector is Flat Panel Detector tomography system [7-9], this tool can well has 43x43.9 cm2 effective area and support imaging reconstruction process with 143μm×143μm pixel size, detector uses CsI(Tl) reduction of coding work by using intuitive as scintillator with thickness is 0.3 mm. X-ray integrated library. The advantages and
42 TRAN THUY DUONG, BUI NGOC HA disadvantages of two algorithms will be database. The gray value distribution of pixels analyzed in detail below. on a line of reconstructed image is shown in the figure 2. Reconstruction process is B. Results implemented on Workstation computer with the First, the quality of reconstructed images configuration: Intel® Xeon® CPU E5-2630 v4 by using the FDK and SIRT algorithm via @ 2.20GHz. For SIRT algorithm, all images changing number of projections will be are reconstructed with the number of iterations evaluated. Figure 2 displays reconstructed is 150. Reconstruction time performed by two images of center slide of rectangular object algorithms for similar projection data set is (200×200 pixels) with difference projection recorded and shown in Figure 3.
Fig. 2. Reconstructed images by FDK and SIRT with difference number of projections
In figure 4, a phantom with a two- 400 iterations). Computational cost to dimensional image size of 500x500 pixels is reconstruct by the FDK algorithm is 43.18 used to investigate the quality of reproduced seconds, and by the SIRT algorithm (with 400 images by the SIRT algorithm with difference iterations) is 1439.11 seconds (~23.9 minutes). numbers of iteration. Figure 5a represents In order to evaluate quality of an image, reconstructed images with 180 projections by following criterial is concerned as gray-scale using the FDK and the SIRT algorithms (with value and image noise. From figure 5b, one
43 EVALUATION OF IMAGE RECONSTRUCTION ALGORITHMS IN CONE-BEAM COMPUTED… could realize that SIRT’s image has higher which require short time for reconstructing gray value at interested peak and lower gray image (less than 10 minutes). So, in this paper, value in low-frequency noise range in FDK algorithm is recommended to reconstruct comparison with FDK’s image. It proves that images of cone-beam computed tomography SIRT algorithm allows to achieve higher systems. In addition, three-dimensional image quality of reconstructed image than FDK of the FDK algorithm also was performed. The algorithm. With respect to images which results were shown in Figure 6, the dataset in require a larger number of pixels (higher this study consists of 720 projections in which resolution), image reconstruction process of the intensity of radiation was increased gradually SIRT spend much more time than FDK. from left to right and from top to bottom with a Therefore, SIRT is not suitable for applications ratio of 1: 4: 8: 10.
Fig. 3. Changing of reconstruction time according to number of projections for FDK and SIRT algorithm
50 iterations 150 iterations
300 iterations 600 iterations Fig. 4. Reconstructed images by the SIRT algorithm when increasing the number of iterations
44 TRAN THUY DUONG, BUI NGOC HA
Finally, reconstructed images of a data set With a small number of projections, it can still were filtered with different filter functions as reconstruct images with required quality. The Ram-Lak, Shepp-Logan, Cosine, Hamming and reconstruction time of both algorithms linearly Hann. These images were shown in Figure 7. increases according to the number of projections but the execution time of the SIRT C. Discussion algorithm increases more strongly than the one From the obtained results, the quality of of FDK algorithm. the image reconstructed by both two Figure 6 shows that when the projection algorithms increases and its noise decreases as intensity was increased, the 3D reconstructed the number of projected images increases. image by using the FDK algorithm is more However, with the same number of projections, clearly. Thus, the quality of reconstructed contrast of reconstructed images using the image depends not only on projection data but SIRT are better than FDK (FDK) algorithm. also on intensity of X-ray source.
45 EVALUATION OF IMAGE RECONSTRUCTION ALGORITHMS IN CONE-BEAM COMPUTED…
Fig. 6. A three-dimensional reconstructed image by the FDK algorithm as the dose level
Original Ram-Lak Shepp-Logan
Cosine Hamming Hann Fig. 7. Image before and after using the filtering function of the reconstructed image
46 TRAN THUY DUONG, BUI NGOC HA
Figure 7 shows the differences between REFERENCES images with different filter functions. Because [1]. Wang, G., Lin, T. H., Cheng, P. C., & each filter function has different frequency Shinozaki, D. M. A general cone-beam domain responses, it affects different parts of reconstruction algorithm. IEEE Transactions the image. For this image, the Ram-Lak filter on Medical Imaging, 12(3), 486-496, 1993. function improves the image best. Depending [2]. Xing, Y., & Zhang, L. A free-geometry cone on the specific case, an appropriate filter can beam CT and its FDK-type be chosen to achieve the best quality of reconstruction. Journal of X-Ray Science and Technology, 15(3), 157-167, 2007. reconstructed image. [3]. Jia, X., Dong, B., Lou, Y., & Jiang, S. B. III. CONCLUSIONS GPU-based iterative cone-beam CT reconstruction using tight frame In this paper, some properties of the regularization. Physics in Medicine & FDK and SIRT algorithm used for cone- Biology, 56(13), 3787, 2011. beam computed tomography systems have [4]. Hsieh, J., Nett, B., Yu, Z., Sauer, K., Thibault, investigated. The results show that the J. B., & Bouman, C. A. Recent advances in CT reconstructed image quality of the SIRT image reconstruction. Current Radiology Reports, 1(1), 39-51, 2013. algorithm is better than the FDK algorithm. However, when the number of pixels [5]. Pack, J. D., Noo, F., & Clackdoyle, R. Cone- beam reconstruction using the backprojection of increases, to achieve the same image locally filtered projections. IEEE Transactions on quality, the SIRT algorithm takes a longer Medical Imaging, 24(1), 70-85, 2005. time than FDK algorithm. Therefore, FDK [6]. Scherl, H., Koerner, M., Hofmann, H., Eckert, W., algorithm is recommended to reconstruct Kowarschik, M., & Hornegger, J. Implementation images in industrial-used CBCT systems of the FDK algorithm for cone-beam CT on the with fast speed requirements. In addition, cell broadband engine architecture. In Medical the quality of reconstructed image by FDK Imaging 2007: Physics of Medical Imaging (Vol. 6510, p. 651058). International Society for Optics algorithm was investigated when changing and Photonics, March 2007. intensity of X-ray source and when using additional image filtering functions. Our [7]. Van Aarle, W., Palenstijn, W. J., Cant, J., Janssens, E., Bleichrodt, F., Dabravolski, A., ... results are consistent with other studies in & Sijbers, J. Fast and flexible X-ray the world [1, 3-6]. tomography using the ASTRA toolbox. Optics express, 24(22), 25129-25147, 2016. ACKNOWLEDGMENTS [8]. Palenstijn, W. J., Batenburg, K. J., & Sijbers, This research is supported by J. The ASTRA tomography toolbox. In 13th International Conference on Computational KC.05.18/16-20 Project of Ministry of Science and Mathematical Methods in Science and and Technology and Mitsubishi Heavy Engineering, CMMSE, Vol. 2013, pp. 1139- Industries (MHI) Group of Japan. 1145), June 2013.
47 Nuclear Science and Technology, Vol.9, No. 4 (2019), pp. 48-55 Relative output factors of different collimation systems in truebeam STx medical linear accelerator
Do Duc Chi1, Tran Ngoc Toan2, Robin Hill3,4, Nguyen Do Kien1 1108 Military Central Hospital, Hanoi, Vietnam 2Vietnam Atomic Energy Institute, Hanoi, Vietnam 3Chris O’Brien Lifehouse, NSW, Australia 4School of Physics, The University of Sydney, NSW Australia Email: [email protected] (Received 05 September 2019, accepted 25 September 2019)
Abstract: The IAEA TRS483 and TRS398 Code of Practices (CoP) were used to calculate relative output factors for small photon beams of 6X, 6XFFF energies shaped by High Definition Multileaf Collimator (HDMLC), jaws and cones mounted on TrueBeam STx medical linear accelerator (Varian Medical Systems), respectively. A comparison between these results were made. The results show a large discrepancy in relative output factor curves found among different collimation systems of the same equivalent field sizes and between the CoPs. Therefore, the specific beam modelling in treatment planning system for each type of the collimation system to be used for small fields maybe required for better computational accuracy. Keywords: TRS483 code of practice, small field dosimetry, relative output factors.
I. INTRODUCTION Practical issues encountered are: photon beam data for treatment planning Modern radiotherapy techniques such system (TPS) are usually collected for jaw- as Intensity-Modulated Radiation Therapy shaped beams while we use these data for (IMRT), Volumetric Modulated Arc computation of Multileaf Collimator Therapy (VMAT), Stereotactic (MLC) - shaped beams. Furthermore, Radiosurgery (SRS) and Stereotactic HDMLC-shaped beams are constituted Radiation Therapy (SRT) make use of small from very tiny beamlets, much smaller than photon beams in order to deliver complex smallest collected beam data of field size radiation treatments. However, there are 2 still many physical and technical aspects of 3 × 3 cm (at isocenter), which may which need to be considered in order to affect the computation accuracy of TPS, commission small photon beams safely and especially for small tumors. In Eclipse efficiently in clinical practice such as: v.13.6 (Varian Medical Systems), warning changing in photon fluence spectrum message “inaccuracy” was often seen when making beam quality changing by field size, making treatment plans for tumors less lateral disequilibrium of charged particles than 3 cm diameter. Radiation oncologists may leading to wrong estimation of tend to use HDMLC for small tumor absorbed dose as well as detector size radiosurgery because of its small thickness compared to field size [1–4]. (2.5 mm at isocenter) and convenience.
©2019 Vietnam Atomic Energy Society and Vietnam Atomic Energy Institute DO DUC CHI et al.
Fig. 1. Occlusion of photon source in the case of narrow collimation. Left: the full, extended source can be “viewed” by an observer on the central axis. Right: only partial view of the source is possible by an observer on the central axis [13].
Conical collimators are dedicated for very small field collimations [2], [5–9]. In radiosurgery of small tumors. With cone- this study, we made a comparison of shaped beams, field size diameters are of relative output factors of different 17.5 mm down to 4 mm cone, but they are collimation systems (jaws, HDMLC and previously measured using TRS 398 CoP cones) for further estimation of the published by the IAEA [14]. It has been computational accuracy of TrueBeamSTx shown that the beam quality of photon TPS using newly published TRS 483 CoP beam changes significantly due to these by the IAEA [2].
Fig. 2. TrueBeam STx treatment head with collimation systems: a) Jaws (highest), HDMLC (midlle) [10] and b) Cone (lowest)
49 RELATIVE OUTPUT FACTORS OF DIFFERENT COLLIMATION SYSTEMS IN TRUEBEAM…
II. MATERIAL AND METHOD has an over-response in large fields because of the significant amount of phantom scatter TrueBeamSTx medical linear accelerator component of low energy photons. The with integrated HDMLC with 20 central leaf consequence is an underestimation of field pairs of 2.5 mm thickness and 40 peripheral output factors when they are normalized to a leaf pairs of 5.0 mm thickness at isocenter. large field size (e.g. the conventional 10 cm × Beam shaping using HDMLC (and also jaws) 2 10 cm reference field) [2]. were of field sizes 0.5 × 0.5, 1 × 1, 2 × 2, 3 × 3, 4 × 4, 5 × 5, 7 × 7, 10 × 10 cm2. MLC-shaped According to TRS398 CoP, the output fields were created when the jaws were factor may be determined as the ratio of “optimized” and at “recommended positions” corrected dosimeter readings measured under by software. Inversely, jaw-shaped fields were a given set of non-reference conditions to created when MLC are fully retracted. Beam that measured under reference conditions. However, in TRS483 CoP, the field output shapinga using the cones were with diameter of 4.0, 5.0, 7.5, 10.0, 12.5, 15.0 and 17.5 mm. factor, , relative to is defined
Photon beam energies of 6X (with Flatterning by the following equation : Filter), 6XFFF (Flatterning Filter-Free), 10X,
10XFFF were used for measurements. The (2) linac was calibrated for all photon energies at 10 × 10 cm2 jaw-shaped field to be used for all Where and are the other collimation systems. The dose measurements were readings of the detector (corrected for performed in Blue Phantom 2 (IBA) using a influence quantities) in the clinical field (fclin) Razor chamber (IBA) and Razor diode (IBA) and the machine specific reference field (fmsr), respectively. is a beam quality under Source-to-Axis Distance setup (100 cm SAD, 5 cm depth). The TRS398 and TRS483 correction factor which changed by field size. CoP are both applied to determine relative The intermediate field ( ) method output factors. Relative output factor curves was used by applying formula (2) for field were compared for 3 different collimation sizes bigger than measured by Razor systems and in both CoPs. All data were ionization chamber, and field sizes smaller normalized to 10× 10 cm2 field size. The than measured by Razor unshielded diode equivalent square of the cone defined fields for using formula (2): were calculated using formula [2] :
√ √ (1)
[ ] [ ] (3)
Razor diode (unshielded, p-type silicon diode chip, active detector diameter of 0.6 mm) with high spatial resolution and high sensitivity where “det” refers to the small field is superior to Razor chamber (total active detector (Razor diode) and “IC” to the length of 3.6 mm) in relative dosimetry of ionization chamber (Razor chamber). The small photon beams. However, Razor diode output correction factor [ ] is
50 DO DUC CHI et al. obtained from the tabulated output correction III. RESULTS and DISCUSSION factors with respect to the machine specific A. Output factors of collimation systems reference field as below: using TRS398 CoP
[ ] Using the conventional formula from [ ] (4)
[ ] TRS398 CoP, we got the result as Table I.
Table I. Output factors of collimation systems using TRS398 CoP and Razor chamber.
Cone (mm) 4 5 7.5 10 12.5 15 17.5 100 (square) 6X 0.420 0.523 0.662 0.746 0.797 0.834 0.859 1 6XFFF 0.473 0.574 0.702 0.772 0.816 0.846 0.866 1 MLC FS (mm) 0.5 1 2 3 4 5 7 10 6X 0.529 0.754 0.861 0.895 0.921 0.940 0.968 1 6XFFF 0.560 0.782 0.876 0.909 0.932 0.948 0.975 1 Jaw FS (mm) 0.5 1 2 3 4 5 7 10 6X 0.345 0.704 0.850 0.888 0.914 0.936 0.967 1 6XFFF 0.377 0.736 0.867 0.906 0.929 0.947 0.974 1
B. Output factor of collimation systems cone, intermediate field was 17.5 mm using TRS483 CoP: conical field because we need to
Intermediate field ( ) of 4 × 4 normalize these data to that of 10 × 10 cm2 was selected for calculation of jaw- cm2 field size. The results were obtained shaped fields and MLC-shaped fields. For as Table II.
Table II. Output factor of collimation systems using TRS483 CoP (Razor chamber and Razor diode)
Cone (mm) 4 5 7.5 10 12.5 15 17.5 100 (square) Square Equi. Field size (cm) 0.708 0.885 1.327 1.77 2.212 2.655 3.097 10 6X 0.522 0.599 0.713 0.773 0.814 0.841 0.864 1 6XFFF 0.578 0.648 0.745 0.797 0.830 0.856 0.871 1 MLC Field size(mm) 0.5 1 2 3 4 5 7 10 6X 0.608 0.774 0.871 0.905 0.927 0.945 0.971 1 6XFFF 0.643 0.792 0.881 0.918 0.938 0.954 0.978 1 Jaw Field size(mm) 0.5 1 2 3 4 5 7 10 6X 0.619 0.756 0.84 0.876 0.901 0.924 0.961 1 6XFFF 0.652 0.771 0.846 0.884 0.907 0.928 0.963 1
C. Comparison of results between TRS483 field size and for jaw-shaped field size less than and TRS398 CoP: 3 × 3 cm for both the 6X and 6XFFF beams. Based on these results, the difference The smallest difference was observed between ROF curves is significant between the with MLC-shaped fields while the biggest two different methods (CoPs) for MLC-shaped difference was observed with cone-shaped
51 RELATIVE OUTPUT FACTORS OF DIFFERENT COLLIMATION SYSTEMS IN TRUEBEAM… fields as seen in Fig.3 and Fig.4. At 0.5 × 0.5 and cone-shaped fields, respectively. TRS398 cm2 squared field and 4 mm conical field, the CoP gave underestimation of a relative output output factor difference of 6X and 6XFFF factor in comparison with TRS483 CoP. The beams were -44.2%/-42.2%, -13.0%/-13.0%, - large difference was always seen at field sizes 19.6%/-18.2% for jaw-shaped, MLC-shaped smaller than 4 × 4 cm2.
6X Jaw Output Factors 6XFFF Jaw Output Factors
1 1 0.9 0.9
0.8 0.8
0.7 0.7 ROF ROF 0.6 TRS483 0.6 TRS483 0.5 TRS398 0.5 TRS398 0.4 0.4 0.3 0.3 0 2 4 6 8 10 0 2 4 6 8 10 Jaw Field Size (cm x cm) Jaw Field Size (cm x cm)
6X MLC Output Factors 6XFFF MLC Output Factors 1 1
0.9 0.9
0.8 0.8
Razor Chamber - ROF ROF 0.7 Razor Chamber - 0.7 TRS398 0.6 TRS398 0.6 TRS483
0.5 0.5 0 2 4 6 8 10 0 2 4 6 8 10 MLC Field Size (cm x cm) MLC Field Size (cm x cm) Fig. 3. Difference of TRS398 and TRS483 CoP in relative output factor of MLC and Jaws collimations.
6X Cone Output Factors 6XFFF Cone Output Factors 1 1 0.9 0.9
0.8 0.8
0.7 0.7 ROF ROF TRS483 TRS483 0.6 0.6 TRS398 TRS398 0.5 0.5 0.4 0.4 0 10 20 30 40 50 60 70 80 90 100 0 10 20 30 40 50 60 70 80 90 100 Coned Field Size (mm) Coned Field Size (mm) Fig. 4. Difference of TRS398 and TRS483 CoP in relative output factor of cone collimations.
52 DO DUC CHI et al.
Noticingly, the Razor chamber’s D. Comparison of results between 6X, reading differences between 10 × 10 cm2 6XFFF (TRS483 CoP): MLC-shaped field and 10 × 10 cm2 jaw- For the same collimation system, output shaped field were just 0.61% and 0.25% for factor comparisons were also made for 6X and 6X and 6XFFF, respectively. Therefore, these 6XFFF beams after applying TRS483 CoP (Fig. 5). The biggest differences in output relative output factors could be used for direct factor were seen at 0.5 × 0.5 cm2 jaw-shaped comparison between jaw-shaped field and field, 0.5 × 0.5 cm2 MLC-shaped field and MLC-shaped field of “the same” nominal 4mm cone-shaped field with values of 5.3%, field size. 5.8% and 10.5%, respectively.
TRS483 Jaws ROF TRS483 MLC ROF TRS483 Cone ROF
1 1 1
ROF
ROF ROF 0.9 0.9 0.9
0.8 0.8 0.8
6X 6X 6X 0.7 0.7 0.7 6XFFF 6XFFF 6XFFF
0.6 0.6 0.6
0.5 0.5 0.5 0 2 4 6 8 10 0 2 4 6 8 10 2.5 5 7.5 10 12.5 15 17.5 Jaw Field Size (cm) Coned Field Size (mm) MLC Field Size (cm) Fig. 5. Difference in output factor of 6X and 6XFFF beams in each collimation system.
E. Comparison of Output Factor curves MLC’s but it is inverse for field size less than 1 between different collimation systems × 1 cm2. (TRS483 CoP): In a multi-centre analytical study of The relative output factor comparisons small field output factor calculations in were made between MLC, jaws and Cone radiotherapy reported by Krzysztof Chełmiński 2 systems for both 6X and 6XFFF beams. and Wojciech Bulski, for 2 × 2 cm MLC- shaped fields of Varian linacs, the differences Conical collimators are independent between the treatment planning system output from MLC and Jaws systems. The conical factors (based on collected beam data) often collimation system has smallest relative exceeded 5% and were below 10% [11]. In output factor in comparison with that of MLC our study, these differences were -1.1% (6X) and Jaws systems for both 6X and 6XFFF and -0.5% (6XFFF) for MLC-shaped fields, beams as Fig. 6. 1.3% (6X) and 2.6% (6XFFF) for jaw-shaped For field sizes bigger than 1 × 1 cm2, jaw fields, -7.0% (6X) and -5.7% (6XFFF) for system has lower relative output factor than cone-shaped fields. The smaller differences
53 RELATIVE OUTPUT FACTORS OF DIFFERENT COLLIMATION SYSTEMS IN TRUEBEAM… observed in our study for MLC-shaped field 31% of the calculated ROF of the 2 × 2 cm2 may came from our small field detector, the field exceeded the action limit of 3% for Razor chamber. nominal beam energies of 6 MV and for nominal beam energies higher than 6 MV, A multinational audit of small field respectively [12]. output factors calculated by treatment planning systems used in radiotherapy, the The discrepancy above may come ROF for small fields calculated by TPSs from accuracy of treatment planning were generally larger than measured algorithms on measured output factors, reference data. On a national level, 30% and especially for small fields.
6X TRS483 Output Factors 6XFFF TRS483 Output Factors 1 1 0.95 0.95 0.9 0.9 0.85 0.85
0.8 0.8
0.75 0.75
ROF ROF 0.7 0.7 MLC 0.65 MLC 0.65 0.6 Jaws 0.6 Jaws 0.55 Cone 0.55 Cone 0.5 0.5 0 2 4 6 8 10 0 2 4 6 8 10 Field Size (cm, square equivalent) Field Size (cm, square equivalent)
Fig. 6. Difference in output factor of difference collimation system for 6X and 6XFFF beams.
CONCLUSIONS that jaw-based beam data itself may not suitable for MLC-based treatment planning. An international code of practice for the Additional measurement of small beam dosimetry of small static fields used in external percentage depth dose and profiles as well as beam radiotherapy (TRS483 CoP) was specific modelling of photon beam for MLC successfully applied to recalculate relative system may be required. output factors for cone system with correction. Relative output factors for jaw collimation REFERENCES system were extensively obtained for field size less than 3 × 3 cm2 for Eclipse v.13.6 for 6X [1]. P. Andreo, “The physics of small megavoltage and 6XFFF beams using TRS483 CoP. photon beam dosimetry,” Radiother. Oncol., vol. 126, no. 2, pp. 205–213, 2018. Relative output factors were also measured for
MLC collimation system to be compared with [2]. International Atomic Energy Agency, TRS483, that of the jaw collimation system. The “An International Code of Practice for discrepancy of output factor between jaw- Reference and Relative Dose Determination”, shaped fields and MLC-shaped fields suggests November, 2017.
54 DO DUC CHI et al.
[3]. I. J. Das, G. X. Ding, and A. Ahnesjö, “Small [10]. M. Constantin, H. Hsu, S. Mansfield, and M. fields: Nonequilibrium radiation dosimetry,” Svatos, “Mitigation Strategies to Reduce the Med. Phys., vol. 35, no. 1, pp. 206–215, 2008. Relative Extrafocal Dose ( RED ),” vol. 33, p.
[4]. I. J. Das, J. Morales, and P. Francescon, “Small 2525, 2006. field dosimetry: What have we learnt?,” in AIP Conference Proceedings, 2016. [11]. K. Chełmiński and W. Bulski, “A multi-centre analytical study of small field output factor [5]. K. Park, J. Bak, S. Park, W. Choi, and S. W. calculations in radiotherapy,” Phys. Imaging Park, “Determination of small-field correction factors for cylindrical ionization chambers Radiat. Oncol., vol. 6, no. September 2017, pp. using a semiempirical method,” Phys. Med. 1–4, 2018. Biol., vol. 61, no. 3, 2016. [12]. W. Lechner et al., “A multinational audit of [6]. H. Bouchard, J. Seuntjens, S. Duane, Y. Kamio, small field output factors calculated by and H. Palmans, “Detector dose response in treatment planning systems used in megavoltage small photon beams. I. Theoretical radiotherapy,” Phys. Imaging Radiat. concepts,” Med. Phys., vol. 42, no. 10, 2015. Oncol., vol. 5, no. September 2017, pp. 58– [7]. E. Mainegra-Hing and B. R. Muir, “On the 63, 2018. impact of ICRU report 90 recommendations on kQ factors for high-energy photon beams,” Med. [13]. Aspradakis, M. M. (2009). Small Field MV Phys., vol. 45, no. 8, pp. 3904–3908, 2018. Photon Dosimetry. World Congress on
[8]. H. Benmakhlouf, J. Sempau, and P. Andreo, Medical Physics and Biomedical Engineering, “Output correction factors for nine small field September 7 - 12, 2009, Munich, Germany, detectors in 6 MV radiation therapy photon 854–854. beams: A PENELOPE monte carlo study,” Med. Phys., vol. 41, no. 4, 2014. [14]. Absorbed Dose Determination in External Beam Radiotherapy: An International Code of [9]. I. J. Das, M. B. Downes, A. Kassaee, and Z. Practice for Dosimetry based on Standards of Tochner, “Choice of Radiation Detector in Dosimetry of Stereotactic Radiosurgery- Absorbed Dose to Water. Technical Reports Radiotherapy,” J. Radiosurgery, vol. 3, no. 4, Series No. 398, International Atomic Energy pp. 177–186, 2000. Agency, Vienna, 2000.
55 INSTRUCTIONS FOR AUTHORS
GENERAL INFORMATION MANUSCRIPT PREPARATION Nuclear Science and Technology (NST), an Manuscripts must be written in English with international journal of the Vietnam Atomic adequate margins and indented paragraph. All Energy Society (VAES) and Vietnam Atomic manuscript must use SI (metric) units in text, Energy Institute (VINATOM), quarterly publishes figures, and tables. Manuscripts should in general articles related to theory and application of nuclear be organized in the following order: title, names of science and technology. All papers and technical authors and their complete affiliation including zip notes will be refereed. code, abstract (not exceeding 200 words), It is understood that the paper has been neither keywords (up to 7), introduction, main body of a published nor currently submitted for publication paper, acknowledgments, references, appendices, elsewhere. The copyright of all published papers table & figure captions, tables and figures. and notes will be transferred in VAES. Unnecessary sections may be omitted.
DETAILED FIELDS Headings: Use I, II,… for major headings and A, B, … for secondary headings. NST coves all fields of nuclear science and technology for peaceful utilization of nuclear Mathematical formulas: All mathematical energy and radiation. Authors should choose one formulas should be clearly written, with special of the following fields at the time they submit consideration to distinctive legibility of sub-and their manuscript: 1) Nuclear Physics, 2) Nuclear superscripts. Equation (at least the principal ones) Data, 3) Reactor Physics, 4) Thermal Hydraulics, should be numbered consecutively using Arabic 5) Nuclear Safety, 6) Nuclear I&C, 7) Nuclear numerals in parentheses in the right hand margin. Fuel and Materials, 8) Radioactive Waste Tables and Figures: Tables should be numbered Management, 9) Radiation Protection, 10) with Roman numerals. Figures should be Radiation Technology, 11) Nuclear Techniques in numbered consecutively with Arabic numerals in Food and Agriculture, 12) Nuclear Medicine and order of their first appearance and have a complete Radiotherapy, 13) Nuclear Techniques in descriptive title. They should be typed on separate Industries, 14) Environment Radioactivity, 15) sheets. Tables should no repeat data which are Isotope Hydrology, 16) Nuclear Analytical available elsewhere in the paper. Figures should Methods, 17) Health Physics, 18) Fusion and be original ink drawing or computer drawn figures Laser Technology. in the original and of high quality, ready for direct MANUSCRIPT SUBMISSION reproduction. Figures should be referred to in the text as, for example, Fig. 1., or Fig. 2. . Manuscript for publication should be submitted to the Editorial Office in triplicate by postal mail. Reference: References should be listed at the end For electronical submission use of the text and presented as follows: [email protected]. [1] C. Y. Fu et al., Nuclear Data for Science and Technology, S. M. Qaim (Ed.), p. 587 (1991). Submission Address [2] C. Kalbach, Z. Phys, A283, 401 (1977). Department of Planning, R&D Management Vietnam Atomic Energy Institute, 59 Ly Thuong [3] S. Shibata, M. Imamura, T. Miyachi and M. Kiet Street, Hanoi, Vietnam Mutou, “Photonuclear spallation reactions in E-mail: [email protected]. Cu”, Phys. Rev. C 35, 254 (1987). KHOA HỌC VÀ CÔNG NGHỆ HẠT NHÂN
Chịu trách nhiệm xuất bản TRẦN HỮU PHÁT
Chịu trách nhiệm nội dung TRẦN HỮU PHÁT TRẦN CHÍ THÀNH
Trình bày LÊ THÚY MAI
In 200 cuốn, khổ 19x26,5cm tại Công ty TNHH Trần Công Địa chỉ: số 12 ngách 155/176 Đường Trường Chinh, Hà Nội Giấy đăng ký kế hoạch xuất bản số: 770/GP-BTTTT cấp ngày 20 tháng 5 năm 2011 In xong và nộp lưu chiểu Quý I năm 2020
25 000đ