- Email: [email protected]
Development of a portable heavy-water leak sensor based on laser absorption spectroscopy
Annals of Nuclear Energy 87 (2016) 350–355
Contents lists available at ScienceDirect
Annals of Nuclear Energy journal homepage: www.elsevier.com/locate/anucene
Development of a portable heavy-water leak sensor based on laser absorption spectroscopy Lim Lee ⇑, Hyunmin Park, Taek-Soo Kim, Minho Kim, Do-Young Jeong Quantum Optics Division, Korea Atomic Energy Research Institute, 1045 Daedeok-daero, Yuseong, Daejeon 305-600, Republic of Korea
a r t i c l e
i n f o
Article history: Received 4 May 2015 Received in revised form 14 September 2015 Accepted 15 September 2015
Keywords: Portable sensor Leak sensor Laser sensor Heavy water detection Coolant leak detection Nuclear reactor
a b s t r a c t A compact and portable leak sensor based on cavity enhanced absorption spectroscopy has been newly developed for a detection of heavy water leakage which may happen in the facilities using heavy water such as pressurized heavy water reactor (PHWR). The developed portable sensor is suitable as an individual instrument for the measuring leak rate and finding the leak location because it is sufficiently compact in size and weight and operated by using an internal battery. In the performance test, the minimum detectable leak rate was estimated as 0.05 g/day from the calibration curve. This new sensor is expected to be a reliable and promising device for the detection of heavy water leakage since it has advantages on real-time monitoring and early detection for nuclear safety. Ó 2015 Published by Elsevier Ltd.
1. Introduction Heavy water (D2O) is used as a neutron moderator and coolant in a pressurized heavy water reactor (PHWR). Although many improved technologies for safety have been developed and implemented in those facilities, the possible risk of heavy water leakages still exists. Since the leaked heavy water contains tritium (3H), the heavy water leaks can cause the release of the radioactive tritium into the environment as well as sudden shutdowns of the facilities. Actually, many leakage events have been reported at the PHWR plants and research reactors all over the world. For these reasons, the facilities using heavy water strongly requires a reliable system to continuously monitor heavy water leaks during normal operation of reactors (King, 2005; U.S. Nuclear Regulatory Commission, 2007). Currently, the PHWR plants uses a tritium detection method for heavy water leaks because leaked heavy water contains tritium, which is produced by the neutron activation of deuterium. There are two conventional tritium detection methods. One is Liquid Scintillation Counter (LSC) and the other is Fixed Tritium Air Monitor (FTAM). However, those two methods have some drawbacks. In case of LSC, they collect air samples over several hours in many rooms and measure the tritium concentration every 4–8 h. So, it is
⇑ Corresponding author. Tel.: +82 42 868 2843. E-mail address: [email protected] (L. Lee). http://dx.doi.org/10.1016/j.anucene.2015.09.017 0306-4549/Ó 2015 Published by Elsevier Ltd.
difficult to detect leaks in real time and also impossible to find the exact leak locations. For real time monitoring, they are using FTAM which contains an ion chamber. However, this method lacks of both sensitivity and linearity, and it has some limitation to detect a low rated leak and find the exact leak location. Conventional methods are less effective in PHWR with Tritium Removal facility (TRF) because these conventional methods basically detect radioactive tritium instead of leaked heavy water. As a specific method for the detection of heavy water, Fourier Transform infrared (FT-IR) spectroscopy is used. Although the FTIR spectroscopy is a good analytical method to measure heavy water without using any chemical reagents, the sensitivity is not enough to detect a low-rated leak. In order to overcome the limitation of the conventional techniques, we have recently developed a laser-based heavy water leak sensor (Lee et al., 2012). Our portable sensor is based on the OAICOS technique. It is one of the advanced techniques of cavity enhanced absorption spectroscopy (CEAS). It is considered to be a simple and robust spectroscopic tool because it doesn’t need additional complex setup for cavity length modulation or stabilization. Moreover, it is less susceptible to vibrations or perturbations compared to other long-path techniques because it is less sensitive to the beam alignment. Using this sensor, leak events could be monitored by detecting a small change in semi-heavy water (HDO) concentration induced by the exchange reaction of leaked heavy water (D2O) with light
L. Lee et al. / Annals of Nuclear Energy 87 (2016) 350–355
water (H2O). From the feasibility test, we found out that the sensor has an outstanding capability in terms of sensitivity for detecting a very low rated leak from reactor components and in its ability to find exact leak locations. However, the developed apparatus has some limitations caused by its weight and size. It should be placed at a fixed location of the PHWR plant because the movement of the detector was not easy (but possible). The collected air sample was sent to the laser leak detector by using a long hose up to 100 m and analyzed for detecting heavy water leaks. In this case, if the distance between the detector and the sampling position becomes longer, the sample transit time through a hose also becomes longer and, consequently, a time delay about 1 min occurred to determine whether the leak is or not. Moreover, it has had some constraints to find leaks occurring at pipelines and pumps located some narrow spaces. In view of these points, a smaller equipment as portable for individual use would enable a plant personnel to find leaks at the pipes, valves, and pumps more conveniently. In this paper, we reported about the more advanced leak monitoring device as a portable sensor for individual use in the facilities using heavy water. For the portability, we decreased its size and weight by improving the laser absorption cell, adopting a compact data analysis system, and using a miniature vacuum pump and internal battery. Finally, the developed portable device was assembled to be used as backpack type for individual use. Also, the sensitivity of the new device should be satisfied with the needed minimum sensitivity proposed by the general CANDU operational guidelines, 1 gallon per minute. After the feasibility test, we presume that this new portable sensor is expected to be a reliable and promising device for the detection of heavy water leakage since it has some advantages on real-time monitoring and early detection for nuclear safety.
2. Sensor description The assembled sensor was made easy to carry. The weight of the sensor is only 6.1 kg, the total weight including control PC, sampling probe, and the replacement battery is less than 8 kg. It is wearable and also easy to use to search for the leak point as shown in Fig. 1. A simplified architecture of the developed portable leak sensor is given in Fig. 2. It contains an optical module which consists with a laser diode as a light source, an optical cavity and a photodetector with a lens array. An air inlet and an outlet port were provided for the sample gas flow at the side of the absorption cell. A vacuum
351
pump for the sample gas flow is attached to the module, and the other units such as an oscilloscope, function generator and power supply are included in an aluminum box. Finally the box is packed in a backpack. After wearing the backpack, an air sampling tube and a control PC can be coupled through the port ① and port ③, respectively. A single-longitudinal-mode tunable distributed feedback (DFB) laser diode was used as a light source. By adjusting the temperature and driving the current of the laser diode, the wavelength could be tuned to near 1390 nm where several absorption peaks of H2O and HDO molecules exist. The driving current was directly controlled and modulated by an external function generator via an LD driver ④ (Thorlabs, CLD1015). In order for the sensor to be a portable device, the LD driver was selected by considering its size and the LD was installed in this driver. The DFB laser beam was sent to an absorption cell through a single mode fiber. Before the beam was inserted into the absorption cell, it was collimated using a fiberport containing an aspheric antireflection-coated lens. The fiberport ⑤ (Thorlabs, PAF-X-5-C) allowed an aspheric lens element to be micro-positioned with respect to the fixed position of the fiber tip, which enabled the lens to be linearly translated along all three axes and enabled the pitch and yaw of the lens to be adjusted. With the help of the fiberport, the incident angle and the divergence of the laser beam was conveniently adjusted for the appropriate alignment in the cell. We found that the slightly focused beam was good for minimizing the interference noise which might occur by the overlapping of the reflected beams and enhanced by the characteristics of the cavity. A vacuum-tight acetal (engineering plastic) absorption cell ⑥ was designed for a sensitive laser spectroscopy. The cavity enhanced absorption spectroscopy (CEAS) technique was adopted in our devices. It was designed as an optical cavity which was consisted of a 91 mm long tube and two highly reflective (R = 0.9995 at 1390 nm) plano-concave mirrors with 1 inch diameter. In the cavity, the light absorption by water molecules is enhanced by the lengthened absorption path due to the successive reflections between mirrors. The laser beam was aligned to be incident on the first mirror surface at a distance from the center of the mirror and it was also tilted with an angle with respect to the cavity axis. The optimized incident position and angle can be evaluated geometrically by using the results from Harriott Cell (Herriot et al., 1964). In this condition, the reflections on the mirror formed a circled spot pattern and the neighboring spots were separated with an angle. Then, the light becomes re-entrant after several round trips. As
Fig. 1. The developed portable leak sensor body (left) and picture of searching for the leak point with the sensor (right).
352
L. Lee et al. / Annals of Nuclear Energy 87 (2016) 350–355
Fig. 2. A brief architecture of the portable leak sensor developed for monitoring heavy water leaks. ① Air sample inlet, ② pump out, ③ USB connector, ④ laser diode driver, ⑤ fiberport, ⑥ absorption cell, ⑦ photo detector, ⑧ diaphragm pump, ⑨ oscilloscope and function generator, ⑩ battery pack, ⑪ regulator, ⑫ DC-DC convertor and ⑬ switchbox.
the re-entrance condition lengthens, the mode spacing continues to collapse and the Fabry–Perot theory breaks down (Paul et al., 2001; Engel and Moyer, 2007). Under the optimal condition, this cell has a feature of a long cavity with narrow free spectral range (FSR). When the FSR is significantly narrower than laser bandwidth, the resonances commonly associated with optical cavities are eliminated systematically while the absorption signal amplifying properties are preserved. In other words, this cavity becomes a cavity enhanced absorption cell with continuous mode spacing. Therefore, the cavity provides advantageous features of pathlength enhancement without introducing intensity noise in the transmitted spectrum. The OA-ICOS cell used in the sensor was designed to increase the effective optical path-length in the cavity. Since the cavity length is 91 mm, the FSR is about 1.7 GHz in the on-axis cavity configuration. The effective cavity length could be longer by the offaxis configuration, and effective cavity FSR should be narrower than the laser bandwidth or detector bandwidth. This means that the enhanced absorption signal could be observed as a continuous spectrum. Noises occurred by the interference between the partially overlapped beams in the cavity are also enhanced because of the same reason. Thus, in order to minimize the noise, the beam alignment and the wavelength scanning rate was optimized by considering the characteristics of the cavity. We found the optimal condition for accurate alignments of the incident beam by using a simulation of the ray tracing method, and we developed a unique method for the beam alignment according to the simulated results. We found that slightly focused beam with a divergent angle of about 103 radian was good for minimizing the interference noise because the overlapped area between reflected beams was minimized in our absorption cell. The scanning speed throughout the spectral range is concerned with the enhancement of the noise. With a consideration of detector bandwidth and spectral range, the scanning speed was selected as 100 Hz. The optical signal having passed through the absorption cell is detected with a photodiode ⑦ (Thorlabs, PDA50B) with a diameter of 5 mm. This is a germanium (Ge) detector the designed for detection of light signals with wavelength of 800 nm to 1800 mm. It has
a gain of 2.4 104 VA1 and a noise equivalent power (NEP) of 61.5 1011 W Hz1/2. The transmitted beams through the absorption cell were collected by the lens array, and they formed a pattern as a circle with a small enough diameter to be focused on this photodetector. A vacuum pump must be an oil-free type because the humid air in the cell flows out through the pump. A diaphragm pump ⑧ (Thomson, 1420 VP) was selected by considering the pumping capacity, power consumption, size and weight. With a pumping capacity of 11 l/min, the pump makes the pressure of the absorption cell sufficiently low in order to collect an air sample where the heavy water leaks may occur. The air pressure in the absorption cell was maintained as low as about 200 torr with the chosen pump. In order to scan the wavelength and acquire the data within the spectral region of interest, we used an arbitrary wave function generator and an oscilloscope integrated data acquisition board ⑨. The wavelength of the laser light was modulated by the modulated currents which were supplied from the LD driver. The currents were modulated by the consecutive triangular wave generated from the function generator. The power for the sensor is provided from a rechargeable lithium polymer 3 cell battery-packs ⑩ with a capacity of 5000 mAh (11.2 V). The stable supply at +12.0 V is generated by a regulator ⑪, and voltage of +12 V, 0 V, 12 V needed for the photodetector are generated by a DC–DC converter ⑫. It is important for the stable operation of the electronic elements, especially for the LD driver. Because of the conversion efficiency of the regulator and DC converter, we could use the capacity of about 3700 mAh at 12 V. The current that flows through the sensor at normal work was measured by 1200 mA, the capacity was sufficient to use the sensor for 3 h without exchanging the battery. 3. Measurement and analysis If the heavy water (D2O) is leaked into the moist air, the D2O molecules are converted to the HDO molecules by the isotope exchange reaction.
L. Lee et al. / Annals of Nuclear Energy 87 (2016) 350–355 k1
D2 O þ H2 O $ 2HDO
:
k1 ¼ 1:10 102 Pa1 s1
353
ð1Þ
For a detection of the heavy water leaks with a low rate, it is more effective to measure HDO molecules rather than D2O. As shown in Fig. 3, there are several absorption transition lines corresponding to the H2O and HDO molecules in the region of interest (http://cfa-www.Harvard.edu/HITRAN). We chose an absorption transition line of the HDO molecules at 7191.04 cm1 corresponding to the laser wavelength of about 1390 nm. In choosing the absorption line, we considered the adequate line intensity and minimum interference from nearby absorption lines of the H2O molecule. We could detect heavy water leaks by measuring the increments of the absorption peak. The amplitudes of the absorption peaks at 7190.74 cm1 corresponding to the H2O molecules are much larger than the amplitudes of the HDO in natural air. The overlapped absorption spectra caused by the large absorption of the H2O might lead to difficulties in the analysis on the heavy water leaks. But, the size of the H2O peak should be smaller because the humidity of the facilities where our sensor could be applied is very low and the concentration of the H2O molecules would be reduced by the exchange reaction (Eq. (1)). The absorption spectra for the dry air (gray line) without heavy water leaks and for the air with heavy water leaks (black line) were shown in Fig. 4. The increment of the absorption corresponding to the HDO molecules and the decrement of the absorption corresponding to the H2O molecules because of the heavy water are shown in the figure. Software for the sensor control, data acquisition and data analysis was developed using the National Instruments LabVIEW graphical programming environment. It was installed on the tablet PC, and it provides a graphical user interface for monitoring the status of the instrument and performing data analysis as shown in Fig. 5. The operation sequence of the sensor was optimized and automated in order to minimize the labor required for the sensor operation. Once the software was executed, and after the status of the function generator and the oscilloscope were checked, the temperature stabilization and currents modulation of the LD were sequentially started by touching button ⑤. The indicator light from alpha to gamma shows each condition of the LD temperature, LD currents, and currents modulation. When the heavy water is leaked out of the reactor components and the sampled air is sucked into the sensor, peak ⓒ corresponding with a HDO absorption will be increased and the integrated
Fig. 4. The absorption spectra for the dry air (gray line) in a nuclear reactor and for the vaporized heavy water contained dry water (black line).
value will also be increased. The integrated value is proportional to the amount of leaked heavy water, and it is displayed as a digit at ③ and a form of an accumulated graph in the window ②. For the performance test of the developed portable sensor, we artificially controlled the leak rate from a leak source by changing the evaporation rate of heavy water by using a wetted paper on a heating plate. The paper was wetted by a certain amount of heavy water should be regarded as a leak source in PHWR. We could evaluate the measured HDO signals versus the leak rate because we knew the amount of heavy water in the paper and we could measure the time for total evaporation. The area of measured absorption curve in spectra was calculated and recorded in the window ②, and the total evaporation can be calculated by the area of the graph in the window ②. Assuming that the heavy water in the paper totally evaporated and the evaporated water vapor is fully captured by the sensor, the total evaporation amount is the same as the total leak amount. Then we could evaluate the leak rate by using the conventional linear regression method. The evaporation rate could be controlled by temperature and the amount of evaporation also could be controlled by the area of the paper. After several measurements, we finally acquired the
Fig. 3. The absorption spectra corresponding to the H2O and HDO molecules in the region of interest for developing the sensor.
354
L. Lee et al. / Annals of Nuclear Energy 87 (2016) 350–355
Fig. 5. The user interface for controlling the sensor and for the analysis of the leak rate. ① Absorption spectrum window, ② measuring history window, ③ measured HDO signal ④ peak position adjustors, ⑤ measuring start and stop, ⑥ reset of history window, ⑦ saving the accumulated data in history window, ⑧ saving the temporal data in spectrum window, ⑨ connection terminator, ⓐ position of H2O peak, ⓒ position of HDO peak, ⓑ measured value integration starting point, ⓓ integration ending point.
not maintained when the leak rate became higher than the range of the calibration curve. In such a high leak rate, the spectral linewidth of HDO absorption become broader than the integration range, and the integrated value would be saturated. From these preliminary experiments in the laboratory, we revealed that the minimum detectable leak rate was estimated to be lower than 0.05 g/day. From the calibration curve, the leak rate can be evaluated directly according to the measuring value. The sensitivity of 0.05 g/day (9 109 gpm) is much more precise than the proposed minimum sensitivity by NRC Guideline, 5500 g/day (1 gpm). It is sensitive enough to find invisible very tiny leaks.
4. Conclusion
Fig. 6. The calibration curve showing the relations between the measured HDO signals and leak rate.
calibration curve shown in Fig. 6. Keeping the similar condition in the real test at the nuclear plant, such as the humidity and temperature, the measurements were done in a closed cubic room with side lengths of 1.5 m. In order to refresh air in the room with minimal convection, dry air was supplied to the room at a constantly low pressure. Sensitivity or minimum detectable concentration (MDC) of a sensor is defined as the smallest change in the quantity that can be unambiguously discriminated from noise. The noises in this sensor are mainly caused by optical interference and ambient condition. The noise caused by interference is smaller than the noise caused by the fluctuation of the integration values due to the ambient condition, such as the air fluctuations. The fluctuation of the integration values was measured as about ±2, and the errors of the measured HDO signals shown in Fig. 6 are mainly caused by this. The calibration curve in Fig. 6 shows a linear correlation between the measured value and the leak rate. The linearity was
A new compact and portable leak sensor based on laser absorption spectroscopy for the detection of heavy water leakages has been developed. This sensor is suitable for individual and portable instruments for measuring leak rates and finding leak locations because it weighs only 6.1 kg and operates from an internal battery. With the help of a rechargeable lithium polymer battery, the detector can be operated continuously for 3 h without external power supply. The minimum detectable leak rate was estimated as lower than 0.05 g/day from the calibration curve. The leak rate can be directly evaluated at every second by measuring the HDO signals. This sensor has several advantages on real-time monitoring and early leak detection over conventional techniques. We expect that this sensor would be useful for a reliable and convenient heavy water leak detector in the industrial PHWR plants, research reactor or any facilities using heavy water.
Acknowledgements This work has been supported by the Nuclear Energy Research Program operated by the Ministry of Science, ICT and Future Planning in the Republic of Korea.
L. Lee et al. / Annals of Nuclear Energy 87 (2016) 350–355
References Engel, G.S., Moyer, E.J., 2007. Appl. Opt. 32, 704–706. Herriot, D., Kogelnik, H., Kompfner, R., 1964. Appl. Opt. 3, 523–526.. King C., 2005. In: Materials Reliability Program: Survey of On-Line PWR Primary Coolant Leak Detection Technologies (MRP-187), EPRI, Palo Alto, CA 10129, vol. 47.
355
Lee, L., Park, H., Kim, T.-S., Ko, K., Jeong, D.-Y., 2012. Nucl. Instrum. Methods Phys. Res. A 678, 8. Paul, J.B., Lapson, L., Anderson, J.G., 2001. Appl. Opt. 40, 4904–4910. U.S. Nuclear Regulatory Commission, Raft Regulatory Guide DG-1173, 2007. (Proposed Revision 1 of Regulatory Guide 1.45, dated May 1973).
Contents lists available at ScienceDirect
Annals of Nuclear Energy journal homepage: www.elsevier.com/locate/anucene
Development of a portable heavy-water leak sensor based on laser absorption spectroscopy Lim Lee ⇑, Hyunmin Park, Taek-Soo Kim, Minho Kim, Do-Young Jeong Quantum Optics Division, Korea Atomic Energy Research Institute, 1045 Daedeok-daero, Yuseong, Daejeon 305-600, Republic of Korea
a r t i c l e
i n f o
Article history: Received 4 May 2015 Received in revised form 14 September 2015 Accepted 15 September 2015
Keywords: Portable sensor Leak sensor Laser sensor Heavy water detection Coolant leak detection Nuclear reactor
a b s t r a c t A compact and portable leak sensor based on cavity enhanced absorption spectroscopy has been newly developed for a detection of heavy water leakage which may happen in the facilities using heavy water such as pressurized heavy water reactor (PHWR). The developed portable sensor is suitable as an individual instrument for the measuring leak rate and finding the leak location because it is sufficiently compact in size and weight and operated by using an internal battery. In the performance test, the minimum detectable leak rate was estimated as 0.05 g/day from the calibration curve. This new sensor is expected to be a reliable and promising device for the detection of heavy water leakage since it has advantages on real-time monitoring and early detection for nuclear safety. Ó 2015 Published by Elsevier Ltd.
1. Introduction Heavy water (D2O) is used as a neutron moderator and coolant in a pressurized heavy water reactor (PHWR). Although many improved technologies for safety have been developed and implemented in those facilities, the possible risk of heavy water leakages still exists. Since the leaked heavy water contains tritium (3H), the heavy water leaks can cause the release of the radioactive tritium into the environment as well as sudden shutdowns of the facilities. Actually, many leakage events have been reported at the PHWR plants and research reactors all over the world. For these reasons, the facilities using heavy water strongly requires a reliable system to continuously monitor heavy water leaks during normal operation of reactors (King, 2005; U.S. Nuclear Regulatory Commission, 2007). Currently, the PHWR plants uses a tritium detection method for heavy water leaks because leaked heavy water contains tritium, which is produced by the neutron activation of deuterium. There are two conventional tritium detection methods. One is Liquid Scintillation Counter (LSC) and the other is Fixed Tritium Air Monitor (FTAM). However, those two methods have some drawbacks. In case of LSC, they collect air samples over several hours in many rooms and measure the tritium concentration every 4–8 h. So, it is
⇑ Corresponding author. Tel.: +82 42 868 2843. E-mail address: [email protected] (L. Lee). http://dx.doi.org/10.1016/j.anucene.2015.09.017 0306-4549/Ó 2015 Published by Elsevier Ltd.
difficult to detect leaks in real time and also impossible to find the exact leak locations. For real time monitoring, they are using FTAM which contains an ion chamber. However, this method lacks of both sensitivity and linearity, and it has some limitation to detect a low rated leak and find the exact leak location. Conventional methods are less effective in PHWR with Tritium Removal facility (TRF) because these conventional methods basically detect radioactive tritium instead of leaked heavy water. As a specific method for the detection of heavy water, Fourier Transform infrared (FT-IR) spectroscopy is used. Although the FTIR spectroscopy is a good analytical method to measure heavy water without using any chemical reagents, the sensitivity is not enough to detect a low-rated leak. In order to overcome the limitation of the conventional techniques, we have recently developed a laser-based heavy water leak sensor (Lee et al., 2012). Our portable sensor is based on the OAICOS technique. It is one of the advanced techniques of cavity enhanced absorption spectroscopy (CEAS). It is considered to be a simple and robust spectroscopic tool because it doesn’t need additional complex setup for cavity length modulation or stabilization. Moreover, it is less susceptible to vibrations or perturbations compared to other long-path techniques because it is less sensitive to the beam alignment. Using this sensor, leak events could be monitored by detecting a small change in semi-heavy water (HDO) concentration induced by the exchange reaction of leaked heavy water (D2O) with light
L. Lee et al. / Annals of Nuclear Energy 87 (2016) 350–355
water (H2O). From the feasibility test, we found out that the sensor has an outstanding capability in terms of sensitivity for detecting a very low rated leak from reactor components and in its ability to find exact leak locations. However, the developed apparatus has some limitations caused by its weight and size. It should be placed at a fixed location of the PHWR plant because the movement of the detector was not easy (but possible). The collected air sample was sent to the laser leak detector by using a long hose up to 100 m and analyzed for detecting heavy water leaks. In this case, if the distance between the detector and the sampling position becomes longer, the sample transit time through a hose also becomes longer and, consequently, a time delay about 1 min occurred to determine whether the leak is or not. Moreover, it has had some constraints to find leaks occurring at pipelines and pumps located some narrow spaces. In view of these points, a smaller equipment as portable for individual use would enable a plant personnel to find leaks at the pipes, valves, and pumps more conveniently. In this paper, we reported about the more advanced leak monitoring device as a portable sensor for individual use in the facilities using heavy water. For the portability, we decreased its size and weight by improving the laser absorption cell, adopting a compact data analysis system, and using a miniature vacuum pump and internal battery. Finally, the developed portable device was assembled to be used as backpack type for individual use. Also, the sensitivity of the new device should be satisfied with the needed minimum sensitivity proposed by the general CANDU operational guidelines, 1 gallon per minute. After the feasibility test, we presume that this new portable sensor is expected to be a reliable and promising device for the detection of heavy water leakage since it has some advantages on real-time monitoring and early detection for nuclear safety.
2. Sensor description The assembled sensor was made easy to carry. The weight of the sensor is only 6.1 kg, the total weight including control PC, sampling probe, and the replacement battery is less than 8 kg. It is wearable and also easy to use to search for the leak point as shown in Fig. 1. A simplified architecture of the developed portable leak sensor is given in Fig. 2. It contains an optical module which consists with a laser diode as a light source, an optical cavity and a photodetector with a lens array. An air inlet and an outlet port were provided for the sample gas flow at the side of the absorption cell. A vacuum
351
pump for the sample gas flow is attached to the module, and the other units such as an oscilloscope, function generator and power supply are included in an aluminum box. Finally the box is packed in a backpack. After wearing the backpack, an air sampling tube and a control PC can be coupled through the port ① and port ③, respectively. A single-longitudinal-mode tunable distributed feedback (DFB) laser diode was used as a light source. By adjusting the temperature and driving the current of the laser diode, the wavelength could be tuned to near 1390 nm where several absorption peaks of H2O and HDO molecules exist. The driving current was directly controlled and modulated by an external function generator via an LD driver ④ (Thorlabs, CLD1015). In order for the sensor to be a portable device, the LD driver was selected by considering its size and the LD was installed in this driver. The DFB laser beam was sent to an absorption cell through a single mode fiber. Before the beam was inserted into the absorption cell, it was collimated using a fiberport containing an aspheric antireflection-coated lens. The fiberport ⑤ (Thorlabs, PAF-X-5-C) allowed an aspheric lens element to be micro-positioned with respect to the fixed position of the fiber tip, which enabled the lens to be linearly translated along all three axes and enabled the pitch and yaw of the lens to be adjusted. With the help of the fiberport, the incident angle and the divergence of the laser beam was conveniently adjusted for the appropriate alignment in the cell. We found that the slightly focused beam was good for minimizing the interference noise which might occur by the overlapping of the reflected beams and enhanced by the characteristics of the cavity. A vacuum-tight acetal (engineering plastic) absorption cell ⑥ was designed for a sensitive laser spectroscopy. The cavity enhanced absorption spectroscopy (CEAS) technique was adopted in our devices. It was designed as an optical cavity which was consisted of a 91 mm long tube and two highly reflective (R = 0.9995 at 1390 nm) plano-concave mirrors with 1 inch diameter. In the cavity, the light absorption by water molecules is enhanced by the lengthened absorption path due to the successive reflections between mirrors. The laser beam was aligned to be incident on the first mirror surface at a distance from the center of the mirror and it was also tilted with an angle with respect to the cavity axis. The optimized incident position and angle can be evaluated geometrically by using the results from Harriott Cell (Herriot et al., 1964). In this condition, the reflections on the mirror formed a circled spot pattern and the neighboring spots were separated with an angle. Then, the light becomes re-entrant after several round trips. As
Fig. 1. The developed portable leak sensor body (left) and picture of searching for the leak point with the sensor (right).
352
L. Lee et al. / Annals of Nuclear Energy 87 (2016) 350–355
Fig. 2. A brief architecture of the portable leak sensor developed for monitoring heavy water leaks. ① Air sample inlet, ② pump out, ③ USB connector, ④ laser diode driver, ⑤ fiberport, ⑥ absorption cell, ⑦ photo detector, ⑧ diaphragm pump, ⑨ oscilloscope and function generator, ⑩ battery pack, ⑪ regulator, ⑫ DC-DC convertor and ⑬ switchbox.
the re-entrance condition lengthens, the mode spacing continues to collapse and the Fabry–Perot theory breaks down (Paul et al., 2001; Engel and Moyer, 2007). Under the optimal condition, this cell has a feature of a long cavity with narrow free spectral range (FSR). When the FSR is significantly narrower than laser bandwidth, the resonances commonly associated with optical cavities are eliminated systematically while the absorption signal amplifying properties are preserved. In other words, this cavity becomes a cavity enhanced absorption cell with continuous mode spacing. Therefore, the cavity provides advantageous features of pathlength enhancement without introducing intensity noise in the transmitted spectrum. The OA-ICOS cell used in the sensor was designed to increase the effective optical path-length in the cavity. Since the cavity length is 91 mm, the FSR is about 1.7 GHz in the on-axis cavity configuration. The effective cavity length could be longer by the offaxis configuration, and effective cavity FSR should be narrower than the laser bandwidth or detector bandwidth. This means that the enhanced absorption signal could be observed as a continuous spectrum. Noises occurred by the interference between the partially overlapped beams in the cavity are also enhanced because of the same reason. Thus, in order to minimize the noise, the beam alignment and the wavelength scanning rate was optimized by considering the characteristics of the cavity. We found the optimal condition for accurate alignments of the incident beam by using a simulation of the ray tracing method, and we developed a unique method for the beam alignment according to the simulated results. We found that slightly focused beam with a divergent angle of about 103 radian was good for minimizing the interference noise because the overlapped area between reflected beams was minimized in our absorption cell. The scanning speed throughout the spectral range is concerned with the enhancement of the noise. With a consideration of detector bandwidth and spectral range, the scanning speed was selected as 100 Hz. The optical signal having passed through the absorption cell is detected with a photodiode ⑦ (Thorlabs, PDA50B) with a diameter of 5 mm. This is a germanium (Ge) detector the designed for detection of light signals with wavelength of 800 nm to 1800 mm. It has
a gain of 2.4 104 VA1 and a noise equivalent power (NEP) of 61.5 1011 W Hz1/2. The transmitted beams through the absorption cell were collected by the lens array, and they formed a pattern as a circle with a small enough diameter to be focused on this photodetector. A vacuum pump must be an oil-free type because the humid air in the cell flows out through the pump. A diaphragm pump ⑧ (Thomson, 1420 VP) was selected by considering the pumping capacity, power consumption, size and weight. With a pumping capacity of 11 l/min, the pump makes the pressure of the absorption cell sufficiently low in order to collect an air sample where the heavy water leaks may occur. The air pressure in the absorption cell was maintained as low as about 200 torr with the chosen pump. In order to scan the wavelength and acquire the data within the spectral region of interest, we used an arbitrary wave function generator and an oscilloscope integrated data acquisition board ⑨. The wavelength of the laser light was modulated by the modulated currents which were supplied from the LD driver. The currents were modulated by the consecutive triangular wave generated from the function generator. The power for the sensor is provided from a rechargeable lithium polymer 3 cell battery-packs ⑩ with a capacity of 5000 mAh (11.2 V). The stable supply at +12.0 V is generated by a regulator ⑪, and voltage of +12 V, 0 V, 12 V needed for the photodetector are generated by a DC–DC converter ⑫. It is important for the stable operation of the electronic elements, especially for the LD driver. Because of the conversion efficiency of the regulator and DC converter, we could use the capacity of about 3700 mAh at 12 V. The current that flows through the sensor at normal work was measured by 1200 mA, the capacity was sufficient to use the sensor for 3 h without exchanging the battery. 3. Measurement and analysis If the heavy water (D2O) is leaked into the moist air, the D2O molecules are converted to the HDO molecules by the isotope exchange reaction.
L. Lee et al. / Annals of Nuclear Energy 87 (2016) 350–355 k1
D2 O þ H2 O $ 2HDO
:
k1 ¼ 1:10 102 Pa1 s1
353
ð1Þ
For a detection of the heavy water leaks with a low rate, it is more effective to measure HDO molecules rather than D2O. As shown in Fig. 3, there are several absorption transition lines corresponding to the H2O and HDO molecules in the region of interest (http://cfa-www.Harvard.edu/HITRAN). We chose an absorption transition line of the HDO molecules at 7191.04 cm1 corresponding to the laser wavelength of about 1390 nm. In choosing the absorption line, we considered the adequate line intensity and minimum interference from nearby absorption lines of the H2O molecule. We could detect heavy water leaks by measuring the increments of the absorption peak. The amplitudes of the absorption peaks at 7190.74 cm1 corresponding to the H2O molecules are much larger than the amplitudes of the HDO in natural air. The overlapped absorption spectra caused by the large absorption of the H2O might lead to difficulties in the analysis on the heavy water leaks. But, the size of the H2O peak should be smaller because the humidity of the facilities where our sensor could be applied is very low and the concentration of the H2O molecules would be reduced by the exchange reaction (Eq. (1)). The absorption spectra for the dry air (gray line) without heavy water leaks and for the air with heavy water leaks (black line) were shown in Fig. 4. The increment of the absorption corresponding to the HDO molecules and the decrement of the absorption corresponding to the H2O molecules because of the heavy water are shown in the figure. Software for the sensor control, data acquisition and data analysis was developed using the National Instruments LabVIEW graphical programming environment. It was installed on the tablet PC, and it provides a graphical user interface for monitoring the status of the instrument and performing data analysis as shown in Fig. 5. The operation sequence of the sensor was optimized and automated in order to minimize the labor required for the sensor operation. Once the software was executed, and after the status of the function generator and the oscilloscope were checked, the temperature stabilization and currents modulation of the LD were sequentially started by touching button ⑤. The indicator light from alpha to gamma shows each condition of the LD temperature, LD currents, and currents modulation. When the heavy water is leaked out of the reactor components and the sampled air is sucked into the sensor, peak ⓒ corresponding with a HDO absorption will be increased and the integrated
Fig. 4. The absorption spectra for the dry air (gray line) in a nuclear reactor and for the vaporized heavy water contained dry water (black line).
value will also be increased. The integrated value is proportional to the amount of leaked heavy water, and it is displayed as a digit at ③ and a form of an accumulated graph in the window ②. For the performance test of the developed portable sensor, we artificially controlled the leak rate from a leak source by changing the evaporation rate of heavy water by using a wetted paper on a heating plate. The paper was wetted by a certain amount of heavy water should be regarded as a leak source in PHWR. We could evaluate the measured HDO signals versus the leak rate because we knew the amount of heavy water in the paper and we could measure the time for total evaporation. The area of measured absorption curve in spectra was calculated and recorded in the window ②, and the total evaporation can be calculated by the area of the graph in the window ②. Assuming that the heavy water in the paper totally evaporated and the evaporated water vapor is fully captured by the sensor, the total evaporation amount is the same as the total leak amount. Then we could evaluate the leak rate by using the conventional linear regression method. The evaporation rate could be controlled by temperature and the amount of evaporation also could be controlled by the area of the paper. After several measurements, we finally acquired the
Fig. 3. The absorption spectra corresponding to the H2O and HDO molecules in the region of interest for developing the sensor.
354
L. Lee et al. / Annals of Nuclear Energy 87 (2016) 350–355
Fig. 5. The user interface for controlling the sensor and for the analysis of the leak rate. ① Absorption spectrum window, ② measuring history window, ③ measured HDO signal ④ peak position adjustors, ⑤ measuring start and stop, ⑥ reset of history window, ⑦ saving the accumulated data in history window, ⑧ saving the temporal data in spectrum window, ⑨ connection terminator, ⓐ position of H2O peak, ⓒ position of HDO peak, ⓑ measured value integration starting point, ⓓ integration ending point.
not maintained when the leak rate became higher than the range of the calibration curve. In such a high leak rate, the spectral linewidth of HDO absorption become broader than the integration range, and the integrated value would be saturated. From these preliminary experiments in the laboratory, we revealed that the minimum detectable leak rate was estimated to be lower than 0.05 g/day. From the calibration curve, the leak rate can be evaluated directly according to the measuring value. The sensitivity of 0.05 g/day (9 109 gpm) is much more precise than the proposed minimum sensitivity by NRC Guideline, 5500 g/day (1 gpm). It is sensitive enough to find invisible very tiny leaks.
4. Conclusion
Fig. 6. The calibration curve showing the relations between the measured HDO signals and leak rate.
calibration curve shown in Fig. 6. Keeping the similar condition in the real test at the nuclear plant, such as the humidity and temperature, the measurements were done in a closed cubic room with side lengths of 1.5 m. In order to refresh air in the room with minimal convection, dry air was supplied to the room at a constantly low pressure. Sensitivity or minimum detectable concentration (MDC) of a sensor is defined as the smallest change in the quantity that can be unambiguously discriminated from noise. The noises in this sensor are mainly caused by optical interference and ambient condition. The noise caused by interference is smaller than the noise caused by the fluctuation of the integration values due to the ambient condition, such as the air fluctuations. The fluctuation of the integration values was measured as about ±2, and the errors of the measured HDO signals shown in Fig. 6 are mainly caused by this. The calibration curve in Fig. 6 shows a linear correlation between the measured value and the leak rate. The linearity was
A new compact and portable leak sensor based on laser absorption spectroscopy for the detection of heavy water leakages has been developed. This sensor is suitable for individual and portable instruments for measuring leak rates and finding leak locations because it weighs only 6.1 kg and operates from an internal battery. With the help of a rechargeable lithium polymer battery, the detector can be operated continuously for 3 h without external power supply. The minimum detectable leak rate was estimated as lower than 0.05 g/day from the calibration curve. The leak rate can be directly evaluated at every second by measuring the HDO signals. This sensor has several advantages on real-time monitoring and early leak detection over conventional techniques. We expect that this sensor would be useful for a reliable and convenient heavy water leak detector in the industrial PHWR plants, research reactor or any facilities using heavy water.
Acknowledgements This work has been supported by the Nuclear Energy Research Program operated by the Ministry of Science, ICT and Future Planning in the Republic of Korea.
L. Lee et al. / Annals of Nuclear Energy 87 (2016) 350–355
References Engel, G.S., Moyer, E.J., 2007. Appl. Opt. 32, 704–706. Herriot, D., Kogelnik, H., Kompfner, R., 1964. Appl. Opt. 3, 523–526.
355
Lee, L., Park, H., Kim, T.-S., Ko, K., Jeong, D.-Y., 2012. Nucl. Instrum. Methods Phys. Res. A 678, 8. Paul, J.B., Lapson, L., Anderson, J.G., 2001. Appl. Opt. 40, 4904–4910. U.S. Nuclear Regulatory Commission, Raft Regulatory Guide DG-1173, 2007. (Proposed Revision 1 of Regulatory Guide 1.45, dated May 1973).