- Email: [email protected]
Distributed temperature sensing inside a 19-rod bundle
Nuclear Engineering and Design 319 (2017) 201–209
Contents lists available at ScienceDirect
Nuclear Engineering and Design journal homepage: www.elsevier.com/locate/nucengdes
Distributed temperature sensing inside a 19-rod bundle S. Lomperski ⇑, N. Bremer, C. Gerardi Argonne National Laboratory, 9700 S. Cass Avenue, Lemont, IL, USA
h i g h l i g h t s
g r a p h i c a l a b s t r a c t
Bundle temperatures were measured
with two fiber optic distributed temperature sensors. One sensor generated 3300 readings along rod wire wraps. A second sensor generated 9300 readings in the gas space between the rods. Demonstrated feasibility of obtaining 3D temperature fields within heated rod bundles.
a r t i c l e
i n f o
Article history: Received 30 January 2017 Received in revised form 5 May 2017 Accepted 9 May 2017 Available online 23 May 2017 Keywords: Fiber optic distributed temperature sensor Nuclear fuel rod bundle
a b s t r a c t The temperature field within a model of a sodium-cooled fast reactor fuel rod bundle was measured using Ø155 lm fiber optic distributed temperature sensors (DTS). The bundle consists of 19 electrically-heated rods Ø6.3 mm and 865 mm long. Working fluids were argon and air at atmospheric pressure and Reynolds numbers up to 300. A 20 m-long DTS was threaded through Ø1 mm capillaries wound around rods as wire-wraps. The sensor generated 173 measurements along each rod at 5 mm resolution for a total of 3300 data locations. A second DTS, 58 m long, was suspended between rods to provide 9300 fluid temperature measurements at 20 mm resolution. Such data density makes it possible to construct 3D maps of the temperature field that are beyond the reach of traditional sensors such as thermocouples. This is illustrated through a series of steady-state and transient tests. The work demonstrates the feasibility of mapping temperature within the close confines of a rod bundle at resolutions suitable for validation of computational fluid dynamics codes. Ó 2017 Published by Elsevier B.V.
1. Introduction Computational fluid dynamics (CFD) can provide insights into the thermal hydraulics of reactor fuel bundles that are unavailable through traditional system codes. CFD tools generate detailed depictions of flow and temperature fields that can be applied to improving both fuel performance and reactor safety. CFD is increasingly employed in the design of advanced reactor cores (e.g., Podila and Rao, 2014) as well as optimization of conventional LWR cores (Ikeda, 2014) and troubleshooting (Cinosi and Walker, ⇑ Corresponding author. E-mail address: [email protected] (S. Lomperski). http://dx.doi.org/10.1016/j.nucengdes.2017.05.008 0029-5493/Ó 2017 Published by Elsevier B.V.
2016; Naveen Raj et al., 2016). Still, the thermal hydraulic behavior of reactor cores is complex and remains challenging to simulate in detail with CFD. Confidence in simulation accuracy can be obtained through code validation, but it is generally difficult to map velocity and temperature fields of real flows with spatial resolution adequate for rigorous CFD benchmarking. Fuel bundles are especially challenging because geometries are both intricate and compact with little room for instrumentation. Bundle heat transfer is of special interest because it plays fundamental roles in governing reactor efficiency and setting safety margins. Thermal hydraulic test data for heated bundles is generally limited to integral measurements such as pressure drop and flow rate
202
S. Lomperski et al. / Nuclear Engineering and Design 319 (2017) 201–209
along with point measurements of temperature from thermocouples. Recent work includes mixing vane and spacer grid studies for LWRs (In et al., 2015; Moon et al., 2014; Shin and Chang, 2009), thermal stratification tests in a 25-rod bundle (Kodama et al., 2014), and emergency cooling system testing with a 54-rod bundle (Patil et al., 2012). Also of interest are rod bundles for advanced reactors, such as those cooled by lead-bismuth eutectic (Martelli et al., 2015; Pacio et al., 2014) and supercritical water (Wang et al., 2016a). The abundance of recent studies attests to continued strong interest in rod bundle heat transfer. Yet these experiments do not generate the kind of whole-field data required for rigorous CFD code validation. Whole-field mapping is typically accomplished with optical techniques such as particle image velocimetry (PIV) and laserinduced fluorescence (LIF). Both rely upon imaging regions of a flow field illuminated by a laser sheet. But optical access to the interior of a bundle constructed of metals and ceramics is severely restricted and therefore largely inaccessible to these instruments. Some investigators have mapped velocity fields within bundles by employing transparent structural materials in so-called MIR (matched index of refraction) setups (Nishimura et al., 2012; Dominguez-Ontiveros and Hassan, 2009; Lee et al., 2013). A MIR setup was also used by Wang et al. (2016b) with a 3 3 bundle and LIF to characterize mixing. These experiments generate high resolution velocity field data, but lack the heat transfer and thermal mixing elements that are of ultimate interest in evaluating fuel bundle performance. MIR setups are generally isothermal since refractive indices of fluid and structure, which vary with temperature, must remain matched to avoid image distortions that lead to measurement errors. A solution lies in fiber optic distributed temperature sensors (DTS), which function without imaging. A single DTS can map temperature fields at resolutions far exceeding anything practicable with conventional sensors such as thermocouples. They function in opaque flows and can be woven through complex geometries to generate data in regions lacking optical access for techniques like LIF. DTSs are, however, infrequently applied to multidimensional-mapping of temperature or velocity across flow fields. Thomas et al. (2012) used them to map thermal structures in atmospheric flows across a region 8 m 8 m and Sayde et al. (2015) mapped wind speed along a 230 m gully using actively heated optical fibers. DTSs have also been used at smaller scales in laboratory settings for 2D mapping of subsurface water flow (Bersan et al., 2015), air jet flow (Chen et al., 2012), and air jet mixing in the context of thermal striping (Lomperski et al., 2015). Rod bundles are notoriously difficult to instrument. Thermocouples can be inserted into rods, but only in very limited numbers since they displace structural material such as cladding and insulation. Instrumenting the flow field between the rods is equally problematic since space is again too limited to accommodate numerous sensors. Sensing technologies based on optical fiber offer high data density since a single narrow fiber generates data at many measurement points. De Pauw et al. (2016) used fiber Bragg grating sensors in a bundle cooled by lead-bismuth eutectic, integrating 60 sensors into the cladding of a 7-rod bundle. These are essentially point sensors, however, and so it is difficult to deploy them in numbers sufficient for mapping flow fields. This paper demonstrates the feasibility of using DTSs for 3D mapping of rod bundle temperature fields. We begin with characterization of the bundle and test section followed by a description of the measurement technique and sensors. Selected data is then presented to illustrate the capability of this measurement technique and its potential utility in fluid dynamics testing for nuclear engineering applications.
2. Experiment 2.1. Rod bundle An acrylic test section houses 19 wire-wrapped heater rods in a configuration characteristic of many sodium fast reactor designs. Fig. 1 shows the bundle layout along with key dimensions. The rods are Ø6.3 mm and the wire-wraps Ø1 mm stainless steel capillaries with an inner diameter of 0.5 mm. The rod pitch/diameter ratio is 1.36 while the wire-wrap pitch is 215 mm. Total rod length is 865 mm with a heated length of 220 mm centered 130 mm above the middle of the rod. The heaters predate this study and their asymmetric heating layout was a fixed design parameter. Bundle support plates are 10 mm-thick polypropylene with Ø1 mm holes for passage of the wire-wrap capillaries and gas space DTS. The bundle is encased in transparent acrylic plates to form a hexagonal test section with a flow area of 1088 mm2, wetted perimeter 527 mm, and hydraulic diameter 8.2 mm. Working fluids were either argon or air delivered to the test section at atmospheric pressure and ambient temperature. Flow enters the top of the test section through a 6-inlet manifold and exits through six holes 30 mm above the base plate. Flow direction is arbitrary for this study. The downward direction was chosen for a longer flow path after the heated section, providing more measurements of thermal mixing before flow exits the test section. A port at the test section midpoint is reserved for tests involving jet injection perpendicular to the bundle. 2.2. Sensing systems The sensing systems used in this study are based on Rayleigh scattering. They use a tunable diode laser to launch a narrow-band infrared signal into a fiber followed by detection and analysis of light scattered back towards the source (Gifford et al., 2007). The spectrum of the backscattered light shifts with changes in a fiber’s physical dimensions or refractive index. Strain and temperature changes generate such shifts and so optical fibers can serve as sensors through measurements of the backscatter spectrum. The physical basis for the spectrum shift can be understood as that of a continuous, weak fiber Bragg grating (Kreger et al., 2006). A fiber Bragg grating consists of periodic variations in the core refractive index along a section of the fiber optic waveguide (Bhatia, 2001). Straining a grating alters its periodicity and shifts the reflected spectrum while temperature changes alter both periodicity and refractive indices, also shifting the spectrum. An analogous process occurs in a DTS as strain or changes in temperature alter the relative positions of scattering centers along the fiber optic cable. Therefore DTS response can be described by that of a } ksel et al., 2011): fiber Bragg grating (Gifford et al., 2007; Yu
Dk ¼ K T DT þ K e e k
ð1Þ
where Dk is the wavelength shift while KT and Ke are the temperature and strain coefficients, respectively. KT includes coefficients for thermal expansion and the index of refraction. The measured wavelength shift is a composite of strain and temperature signals, but sensing system electronics cannot distinguish between the two. One must be kept constant or known in some way in order to measure the other. The fibers are used here as temperature sensors and so strain is controlled in our setup. Alternative sensing methods utilizing Raman or Brillouin scattering do not share this difficulty in distinguishing strain from temperature but their spatial resolution and bandwidth are insufficient for the rod bundle application. The
S. Lomperski et al. / Nuclear Engineering and Design 319 (2017) 201–209
203
Fig. 1. a) Cross section of bundle support plate, and b) test section flow paths (dimensions in mm).
physics and capabilities of all these techniques are considered in reviews by Bao and Chen (2012) and Palmieri and Schenato (2013). Note that measurements are inherently differential in nature. The sensor measures DT from a ‘‘baseline”, a reference measurement performed under controlled conditions that is analogous to zeroing an instrument: new readings are identically zero as long as the sensor remains at the baseline temperature. Baselining for this study was performed at ambient temperature under isothermal conditions so that DT = 0 would be associated with a single temperature for the entire sensor. A link to absolute temperature is made through supplementary sensors such as a thermocouple or resistance temperature detector (RTD), which register absolute temperature while the baseline reading is recorded. The adjusted signal after baseline can then be written as:
TðxÞabs ¼ DTðxÞ þ TðxÞbase
ð2Þ
where T(x)abs is the absolute temperature along the DTS and T(x)base is the absolute temperature during baseline recorded by one or more TCs or RTDs. Baselining under isothermal conditions is the simplest case since conversion to absolute temperature is achieved with a single offset that applies to the entire sensor. Measurement accuracy for our bundle setup is estimated to be better than ±1 °C. Accuracy is not assessed here in detail since our purpose is to demonstrate unique sensing capabilities in the context of the complex geometry of a rod bundle. DTS accuracy and repeatability have been considered elsewhere under stagnant conditions (e.g., Rahim et al., 2015, Lomperski et al., 2013) and for sensors suspended within a flow field (Lomperski and Gerardi, 2014; Lomperski et al., 2016). Calibration issues and challenges are considered in the discussion section.
Two sensing systems were used in this study, an ODiSI-A for the gas space DTS and an ODiSI-B for the wire-wrap, both from Luna, Inc. The latter has higher bandwidth that could be better utilized with the gas space sensor, but assignments were based on system limits for sensor length. The former system accommodates sensors of more than 50 m while the latter is limited to 20 m, specifications that nearly match the two DTSs installed in the bundle. 2.3. Sensor configuration Both DTSs were made from commercial Ø155 mm polyimidecoated single-mode telecom fiber (SMB-G131OH, Specialty Photonics). The first is a 20 m sensor inserted through all 19 wirewraps while the second is 58 m long and suspended within the gas space as 54 vertical segments. A segment is defined as a single length of fiber spanning the distance between the top plate and baseplate. Segment length matches that of the bundle, 865 mm. Fig. 2 shows a close-up view of the bundle near the baseplate to highlight sensor positioning and physical differences between the wire-wrap sensor and the bare gas space sensor. Note that the wire-wrap sensor will be used to represent rod temperatures though it is actually in the capillaries. The wire-wraps are tightly wound around the rods for good thermal contact while the heat transfer coefficient between wire-wrap and argon is low. Consequently, the wire-wrap temperature is an adequate surrogate for rod temperature for the purposes of this demonstration study. Both DTSs pass repeatedly through the support plates and are looped outside the test section to return back into the bundle for subsequent segments. Fig. 3 shows the looping arrangement with fiber glued to the top plate. Loops are kept small since they are outside the bundle and effectively lost for sensing. Each consumes 170 mm of fiber, roughly 12 m for all 72 loops and 15% of total
204
S. Lomperski et al. / Nuclear Engineering and Design 319 (2017) 201–209
flow. This was accomplished here by hanging a weight from each of the loops below the base plate. Weights maintained constant segment tension since fibers pass freely through the /1 mm holes in the baseplate. Tension also positions the gas space segments, ensuring they remain parallel to the rods without touching them. The weights are inconsequential for the wire-wrap sensor housed in capillaries.
3. Results 3.1. 3D temperature field visualization
Fig. 2. Photo of rods and sensors above baseplate.
Fig. 3. Photo of fiber loops above top plate.
available sensor length. The fiber manufacturer specifies a ‘‘shortterm” bend radius limit of 10 mm and ‘‘long term” limit of 17 mm. The loop radius here, 10–15 mm, has been sustained for over a year without breakage or noticeable signal degradation. DTS mounting arrangements must preclude changes in sensor strain levels after the baseline. Sensors should not be strained by test section distortions accompanying pressurization, heating, or
A steady flow test was run at 28 L/min of argon entering the manifold 5 °C below ambient temperature and Reynolds number of approximately 160. The bundle was heated at 160 W until rod temperatures neared 100 °C, then power was cut and DTS data logged at 1 Hz. Fig. 4 presents raw DTS data in the form of DT from the baseline temperature, which was recorded shortly before the test while the setup was isothermal at room temperature with no flow. Data is raw in the sense that it is neither converted to absolute temperature nor mapped to the physical space of the test section. These curves include the loops outside the test section, which are later excluded when mapping data to the bundle. Fig. 4 shows a single scan of the wire-wrap DTS consisting of 4000 measurement points at 5 mm gage and 5 mm spacing. Gage length is the effective spatial resolution, the span over which Dk is analyzed to generate a single DT data point. The plot exhibits distinct peaks where the sensor passes along the heated sections of the rods. DT remains near zero, i.e., ambient temperature, along the cold ends of the rods and the loops outside the test section. Peak temperatures are not equal for two reasons: 1) cooling rates differ for interior and exterior rods, and 2) input power varied slightly among rods since it was distributed through three variable transformers, two wired to six rods each and the third to seven. Power to individual rods was not measured. The saw tooth pattern in the rod data is also seen in the gas space data. Sharp peaks occur where the sensor passes through the heated region of the bundle. Like the wire-wrap DTS, DT0 along external loops as seen in the plot inset. DT is slightly below zero upstream of the heated zone since argon enters the test section below ambient temperature. Gas space data was generated with a 20 mm gauge and sampled at 5 mm spacing for 11600 points along the entire sensor. Such oversampling is routinely applied to PIV data to maximize spatial resolution. The sensing system is capable of 10 mm gage, but signal noise associated with flow-induced vibration limited resolution to 20 mm. Both gage and spacing can be adjusted in post processing software to optimize tradeoffs between spatial resolution and signal noise. Such flexibility contrasts with the immobility of thermocouples, which are seldom repositioned in response to test findings to optimize a setup. While Fig. 4 illustrates the large amount of data available from each sensor, the curves give only a vague impression of the bundle temperature field. Sensor data is of limited utility until mapped onto the geometry of a test section. Complex patterns and trends are more readily revealed after decomposing 1D data and translating it into two or three-dimensional space. Such mapping was performed for the wire-wrap DTS data to generate the images in Fig. 5. With the data presented in this fashion, the extent of the heated zone is distinct and easily recognized. The magnified cutaway view reveals temperatures within the bundle interior, illustrating that data is generated where there is no access for optical instruments. Fig. 5 is based on 3300 total data points, 173 points/rod at 5 mm spacing. Though the DTS spirals around rods within the wire-wraps, circumferential variations in rod temperature are not
S. Lomperski et al. / Nuclear Engineering and Design 319 (2017) 201–209
205
Fig. 4. Raw DTS data for rods (left) and gas (right).
Fig. 5. Wire-wrap DTS data mapped onto rods; (a) full bundle, (b) magnified ¼bundle cutaway view.
considered here, only axial. The wire-wraps are 75 mm longer than the bundle, but data was clipped at the bundle length of 865 mm. As a result, rod temperature profiles are stretched 9%, but the distortion is irrelevant for the purpose of this feasibility study. For CFD code validation, however, DTS data would be more precisely mapped onto the rods to preclude such distortions. Gas space data is presented in Fig. 6 through two different plot styles. The first is a conventional 3D rendering with a single marker for each data point, 173 points/segment at 5 mm spacing, 9300 total for all 54 segments. As with the rod data, color indicates tem-
Fig. 6. Bundle gas temperatures represented with a) one marker per data point, b) interpolated solid, and c) overlay of gas and rod data.
perature level. The rendering provides a sense of the high data density available from a single sensor. The second image in Fig. 6 is a volumetric rendering of the data based on data point interpolation. Easily recognized is a warm plume centered where the rod bundle was seen to be hottest in Fig. 5. Downstream of the hot zone, gas quickly returns to ambient temperature as it flows along the cold
206
S. Lomperski et al. / Nuclear Engineering and Design 319 (2017) 201–209
end of the bundle. Such a detailed depiction of a bundle temperature field is unobtainable through conventional sensors such as thermocouples. The third image combines both the gas and rod data to emphasize that one could pair them to investigate correlations or patterns linking the two. For example, examining axial development of the thermal field as fluid flows along cold and hot regions of the rods. Data density here is sufficient for section views to examine the bundle interior and to scan the entire domain for unexpected phenomena, much as one would with simulation data. In addition, detailed field data can help confirm that boundary conditions are well-behaved and acceptable for meaningful CFD code validation, a shortcoming of some legacy experiments being used for such work. 3.2. 1D temperature field analysis Volumetric rendering of whole data sets is an efficient means of obtaining a general impression of an entire flow field. However, code validation and quantitative analysis are often better suited to manipulation of small data subsets. This is illustrated here with two test cases. The first involves step changes in flow rate to show variations in rod and gas temperature with Reynolds number. The second presents gas temperatures around the center rod for a single flow rate to highlight the ability to capture temperature field asymmetries. The first case began at a bundle power of 24 W and airflow rate of 14 L/min (Re = 100), followed by step increases of 14 L/min in 60 s intervals. All rods were heated, but each end of the bundle remained near ambient temperature. Fig. 7 shows temperature profiles along a selected rod and neighboring gas space. A perimeter rod was chosen rather than interior because it exhibits the largest temperature changes with flow rate for a more compelling demonstration case. The plot shows temperature profiles peaking near the center of the heated zone and flattening at higher flow rates. At Reynolds 100, gas temperature remains below rod temperature along the entire length of the bundle. At Reynolds 200, rod and gas temperatures are nearly equal between 0.4 and 0.6 m while for Reynolds 300 rod temperature falls below gas temperature between 0.5 and 0.6 m. This is not typical of actual reactor cores, but here the flow is not fully developed since the test section is short while rods are not all heated at the same power. Cross flows from hotter neighboring rods are likely skewing the temperature profiles. Regardless, the data demonstrates an ability
Fig. 7. Temperatures along a selected rod and gas segment versus Reynolds number. Measurement uncertainties ±1 °C and ±10 mm.
to characterize subtle shifts in the temperature field with changes in boundary conditions. Fig. 8 presents temperatures along the center rod with data from all six adjacent gas segments. The rod was heated at 9 W with airflow at 14 L/min and Re = 100. Data was logged about 1 h after heating began, brief enough that both rod ends remained near ambient temperature. Cool air 5 °C below ambient temperature was injected into the manifold, which is reflected in the gas segment readings. All are below the rod temperature in the inlet region upstream of the heated zone. Air warmed as it flowed along the center rod, but once past the heated zone it returned nearly to room temperature as it mixed with cold air from the rest of the bundle that made up the majority of the flow. The plot shows radial variation in the temperature field despite radial symmetry of the test section. As noted above, flow is not fully developed. Asymmetries in the velocity field are again seen to be reflected in the temperature field. Signal noise is low at Re = 100 and so data was processed at 10 mm gage instead of 20 mm. The extra resolution reveals sharp peaks at 120°, 180° and, to a lesser extent, 60°. These peaks coincide with proximity to the wire-wrap as it spirals up through the heated zone. Distance between the wire-wrap and bare fiber is <1 mm at the point of closest approach (see Fig. 1). The sequence of peaks progresses with the wire-wrap: the first at 0.26 m followed by the main peak at 0.3 m and the last at 0.34 m. The peak at 0.3 m (180°) is at the center of the heated zone and is the most distinct of the three. There are no peaks at 0°, 240°, and 300° because the wire-wrap is outside the heated zone at these positions. Figs. 7 & 8 demonstrate an ability to measure both rod and bulk fluid temperatures continuously along the bundle, allowing determination of local heat transfer coefficients. With fluid temperature measured at six positions around every interior rod, it is possible to examine circumferential as well as axial variations in the heat transfer coefficient. Such information could be used to identify crossflows or assess mixing vane efficiency. 3.3. Hot jet injection The volumetric renderings of the gas phase data seen earlier suggest data density is high enough to visualize complex flow structures within the bundle. This can be of particular interest in CFD code validation since much of the appeal of such codes lies
Fig. 8. Temperatures along the central rod and six neighboring gas segments at Re = 100. Data processed at 10 mm gage. Measurement uncertainties ±1 °C and ±10 mm.
S. Lomperski et al. / Nuclear Engineering and Design 319 (2017) 201–209
in their superior spatial resolution compared to traditional system codes, which rely upon lumped parameter models. Code proficiency in simulating multidimensional flow structures can be checked against high-resolution experimental data on complex flows. We chose jet injection as an example of such a flow. A stream of argon was injected into the bundle at its midpoint through a 5 mm diameter port, flow rate 50 L/min and temperature of 75 °C with no other flow into the system. The upper manifold was sealed so that gas entered the bundle perpendicular to the rods and flowed downward, mixing and eventually exiting through the six outlets near the base. Rods were unpowered and left at room temperature to achieve better contrast in visualizing the jet and impingement heating of the rods. Though this configuration does not mimic any specific flow phenomena characteristic of
207
sodium-cooled fast reactors, it produces a complex flow structure that is small enough to challenge the spatial resolution of the sensing network. Fig. 9 presents a series of snapshots portraying the transient. The thermal field develops both in and around the rods and so three different views are presented to convey the structure of the flow. The first shows gas temperature overlaid onto the rods, the second depicts only rods, and the third is a top view to track the advance of a ‘‘warm front” into the bundle. The first frame corresponds to stagnant conditions at ambient temperature while the second marks the start of flow. Gas space warming is clearly visible at this early stage, though there is little change in rod temperature. Advance of the warm front into the second row of rods can be seen from the top view. Note that images show gas only if its
Fig. 9. Bundle temperatures during jet injection; first of each bundle pair shows gas with rods, the second only rods. Gas at T < 21 °C was rendered transparent to emphasize jet structure.
208
S. Lomperski et al. / Nuclear Engineering and Design 319 (2017) 201–209
temperature is >21 °C, i.e., ambient-temperature gas is invisible, to isolate the structure of the warm zone created by the jet. The third frame at t = 21 s shows advance of the warm front and general increase in gas temperature. Warming of rods around the impingement zone is now apparent. Data for the rods must be viewed with some caution since mapping is 1D even though the DTS spirals around rods via the wire-wraps. In some places the jet impinges directly on wire-wraps while in other spots the wire-wraps are on the lee side of a rod. This is certain to generate artifacts that are not considered here. The final frame at t = 43 s shows a broad band of cyan-colored gas around and below the injection point. Cyan corresponds to very slight warming above ambient temperature, only 1 °C, and indicates that the thermal effects of the jet have advanced beyond the immediate vicinity of the injection point. Note that this warm zone does not extend above the injection point and recall that the top of the test section is sealed. The thermal map is seen to give an indirect indication of the flow pattern: argon flows downward from the injection point and mixes in the bundle while a stable, stagnant layer persists above the injection point. This snapshot series demonstrates that the sensor network provides data density high enough to characterize small thermal structures inside and around the bundle. An animated version of the transient is available as Supplementary data. 4. Discussion Mapping velocity and temperature fields in rod bundles has proven exceptionally challenging even in low-temperature, ambient-pressure models like the one used for this study. Here we demonstrated the feasibility of using fiber optic distributed temperature sensors to map temperature fields within a heated bundle. The DTSs generated 3300 measurements along rod wirewraps and 9300 measurements in gas spaces between rods. Such data density could not have been achieved with traditional physical probes like thermocouples. Optical methods such as LIF can match this data density and even achieve superior bandwidth and spatial resolution, but cannot access the interior of a bundle as the DTSs have here. Furthermore, DTSs perform equally well in transparent and opaque fluids while optical instruments do not. With distributed sensors, high-resolution temperature mapping across sizable expanses of a rod bundle becomes conceivable. Sensor networks spanning an entire test section can be of great value in fluid dynamics experiments, especially when locations of interesting phenomena are not known a priori. It can be difficult to detect small flow structures like the hot jet presented earlier using only thermocouples, even in exceptionally instrumented setups. Interesting and potentially important phenomena are effectively nonexistent until detected by instrumentation. Distributed sensors can increase the likelihood of observing the unexpected. Despite the potential value of DTSs for bundle mapping, several issues remain in implementing this relatively young technology into fluid dynamics experiments. Two are predominant: signal noise associated with flow-induced vibration and errors from strain changes after the baseline. Though both are essentially uncontrolled variations in strain, the former dynamic and the latter largely static, they differ in how they disturb the temperature signal and in practice are dealt with separately. A third and distinctly different issue is the special difficulty of calibrating these sensors. All three are briefly considered below. 4.1. Dynamic errors Signal noise associated with flow-induced vibration is, in our experience, the most troublesome issue. It can give rise to unpre-
dictable and sometimes debilitating noise levels, especially with long (>20 m) sensors. This is a serious limitation since the full potential of distributed sensing is realized with long sensors. We have investigated noise behavior as the dominant factor limiting effective spatial resolution in long sensors used in a jet mixing experiment (Lomperski and Gerardi, 2014; Lomperski et al., 2016). Indeed, such noise limited testing here to laminar flows. Though laminar flow is relevant to some natural circulation and severe accident states, turbulent flows are of more general interest in nuclear engineering. In practice, noise can be reduced by decreasing sensor length, increasing gage length, and through sensor support arrangements that reduce vibration. These adjustments will increase effective limits on flow rates and Reynolds numbers. For example, vibration of the gas space DTS might have been reduced with heavier weights on the loops, though we observed no remarkable improvement after increasing weight from 2 g to 9 g. However, this fiber can support considerably more (proof strength 0.7 GPa) and, in separate testing, withstood a 250 g load for weeks without failing. Noise is far less troublesome for sensors housed in rigid capillaries. There were no signal noise difficulties with the wire-wrap DTS, but its thermal response is much slower than that of the bare fiber. The extent of the tradeoff is illustrated with an example based on a simple lumped-parameter thermal analysis: response time in a cross flow of argon at 1 m/s is roughly 0.2 s for the bare fiber used here, increasing to 9 s when housed in a /1 mm steel capillary. Response times in water are faster, of the order of milliseconds for bare fiber and 0.05 s with the capillary. Others have measured response times of bare fibers (Van Wyk et al., 2006; Pan et al., 2016) and fibers housed in capillaries (Rinaudo et al., 2016). In the future, signal noise is expected to be less troublesome as optical fiber interrogators evolve. In addition, others have demonstrated the ability to distinguish between stress and strain using paired fibers with differing strain and temperature sensitivities (Ding et al., 2016) and single, polarization-maintaining fibers (Li et al., 2013). 4.2. Static errors Control of static strain, the second issue, is essential but can be problematic even for sensors housed in capillaries. For example, measurement errors were observed in 7-m long fibers around mounting points for steel capillaries (Lisowski et al., 2013), a consequence of thermal expansion upon heating the test loop. Deformation of sensor supports or the test section itself can strain fibers via the capillaries. Strain control typically involves simply ensuring a complete lack of fiber tension. We accomplished this in other setups involving 2D arrays by maintaining slack on all fiber segments. Taut segments will experience strain shifts with even minute deformations of supports, generating measurement errors. For the bundle setup, the weights maintained constant segment tension, but at the cost of giving up hermetic seals between sensor and baseplate. This is a clear disadvantage since liquids tend to be of more interest than gases in nuclear engineering applications. Capillaries facilitate sealing and can be an alternative for applications where slower thermal response time is acceptable. 4.3. Accuracy The third concern involves sensor calibration and accuracy. Calibration, outlined earlier, is simple in principle and can be accomplished with a single thermocouple. But the thermocouple, a point sensor, must serve as a standard for many meters of DTS embodying thousands of sensing points. Adding thermocouples could improve the situation, but the incongruity of calibrating
S. Lomperski et al. / Nuclear Engineering and Design 319 (2017) 201–209
distributed sensors with point sensors remains. Still, the situation is not unprecedented. It parallels that of PIV, which is often benchmarked against point sensors such as hot-wire anemometers or laser Doppler velocimeters. Accuracy assessments are similarly challenging since strain sensitivity effectively rules out the common practice of sensor calibration at the point of origin. In-situ calibration is mandatory, making it difficult to achieve traceability to recognized standards such as NIST. The nature of in-situ calibrations will depend upon the application and so sensor accuracy and repeatability will vary accordingly. This again parallels PIV, which enjoys widespread use and acceptance despite similar issues. We have performed rudimentary assessments of DTS accuracy and stability in a jet mixing setup (Lomperski and Gerardi, 2014; Lomperski et al., 2016) that can serve as examples of sensor performance in a fluid dynamics experiment. Acknowledgment Data visualizations by VisIt; Bethel, E.W., Childs, H., Hansen, C., High performance visualization: enabling extreme-scale scientific insight, Chapman & Hall, 2012. The submitted manuscript has been created by UChicago Argonne, LLC, Operator of Argonne National Laboratory (‘‘Argonne”). Argonne, a U.S. Department of Energy Office of Science laboratory, is operated under Contract No. DE-AC0206CH11357. The U.S. Government retains for itself, and others acting on its behalf, a paid-up nonexclusive, irrevocable worldwide license in said article to reproduce, prepare derivative works, distribute copies to the public, and perform publicly and display publicly, by or on behalf of the Government. The Department of Energy will provide public access to these results of federally sponsored research in accordance with the DOE Public Access Plan. http://energy.gov/downloads/doe-public-access-plan. Appendix A. Supplementary data Supplementary data associated with this article can be found, in the online version, at http://dx.doi.org/10.1016/j.nucengdes.2017. 05.008. References Bao, X., Chen, L., 2012. Recent progress in distributed fiber optic sensors. Sensors 12, 8601–8639. http://dx.doi.org/10.3390/s120708601. Bersan, S., Schenato, L., Rajendran, A., Palmieri, L., Cola, S., Pasuto, A., Simonini, P., 2015. Application of a high resolution distributed temperature sensor in a physical model reproducing subsurface water flow. Measurement. http://dx.doi. org/10.1016/j.measurement.2015.09.018. Bhatia, V., 2001. Properties and applications of fiber gratings. Proc. Int. Conf. Fiber Optics Photonics SPIE 4417:154160. Chen, T., Wang, Q., Zhang, B., Chen, R., Chen, K.P., 2012. Distributed flow sensing using optical hot-wire grid. Opt. Express 20 (8), 10. Cinosi, N., Walker, S.P., 2016. CFD analysis of localized crud effects on the flow of coolant in nuclear rod bundles. Nucl. Eng. Des. 305, 28–38. De Pauw, B., Lamberti, A., Ertveldt, J., Rezayat, A., Vanlanduit, S., Van Tichelen, K., Berghmans, F., 2016. Temperature monitoring using fibre optic sensors in a lead-bismuth eutectic cooled nuclear fuel assembly. Nucl. Eng. Des. 297, 54–59. Ding, Z., Yang, D., Du, Y., Liu, K., Zhou, Y., Zhang, R., Xu, Z., Jiang, J., Liu, T., 2016. Distributed strain and temperature discrimination using two types of fiber in OFDR. IEEE Phot. J. 8 (5), 8. http://dx.doi.org/10.1109/JPHOT.2016.2605011. Dominguez-Ontiveros, E.E., Hassan, Y.A., 2009. Non-intrusive experimental investigation of flow behavior inside a 5 5 rod bundle with spacer grids using PIV and MIR. Nucl. Eng. Des. 239, 888–898. Gifford, D.K., Kreger, S.T., Sang, A.K., Froggatt, M.E., Duncan, R.G., Wolfe, M.S., Soller, B.J. 2007 Swept-wavelength interferometric interrogation of fiber Rayleigh scatter for distributed sensing applications. In: Proc SPIE 6770 Fiber Optic Sens Appl V 67700F. Ikeda, K., 2014. CFD application to advanced design for high efficiency spacer grid. Nucl. Eng. Des. 279, 73–82. In, W.K., Shin, C.H., Lee, C.Y., 2015. Convective heat transfer experiment of rod bundle flow with twist-vane spacer grid. Nucl. Eng. Des. 295, 173–181.
209
Kodama, S., Yoshida, K., Kataoka, I., 2014. Thermal stratification and mixing under single- and two-phase flow in rod bundle with axially distributed heat flux in vessel without forced convection. Nucl. Eng. Des. 268, 164–172. Kreger, S.T., Gifford, D.K., Froggatt, M.E., Soller, B.J., Wolfe, M.S., 2006. High resolution distributed strain or temperature measurements in single- and multi-mode fiber using swept-wavelength interferometry. Opt. Fiber Sens. OSA/ OFS 2006, 4. Lee, C.Y., Shin, C.H., Park, J.Y., In, W.K., 2013. An experimental investigation on turbulent flow mixing in a simulated 3 x 3 dual-cooled annular fuel bundle using particle image velocimetry. Nucl. Eng. Des. 260, 134–144. Li, W., Chen, L., Bao, X., 2013. Compensation of temperature and strain coefficients due to local birefringence using optical frequency domain reflectometry. Optics Communications 311, 26–32. http://dx.doi.org/10.1016/j.optcom.2013.08.022. Lisowski, D., Lomperski, S., Farmer, M.T., Bremer, N., Kilsdonk, D.J., Aeschlimann, R. W., Vik, A., Engel, D., 2013. NSTF demonstration test report, Argonne National Laboratory technical report ANL-NSTF-000000-TECH-009-NC. Lomperski, S., Gerardi, C., 2014. Assessment of distributed fiber optic sensors for flow field temperature mapping. In: Proc. ASME 2014 4th Joint US-European Fluids Engineering Division Summer Meeting and 12th International Conference on Nanochannels, Microchannels, and Minichannels FEDSM2014 August 3–7, Chicago. Lomperski, S., Gerardi, C., Pointer, D.W., 2013. Distributed fiber optic temperature sensing for CFD code validation. In: Proceedings of 15th International Topical Meeting on Nuclear Reactor Thermal Hydraulics (NURETH-15) Pisa, Italy. Lomperski, S., Gerardi, C., Pointer, W.D., 2015. Distributed fiber optic temperature sensor mapping of a jet-mixing flow field. Exp. Fluids 56, 55. http://dx.doi.org/ 10.1007/s00348-015-1918-6. Lomperski, S., Bremer, N., Gerardi, C., Lisowski, D., 2016. Performance assessment of a 50 m-long fiber optic distributed temperature sensor in a fluid dynamics experiment. In: Proc. Specialist Workshop on Advanced Instrumentation Measurement Techniques for Nuclear Reactor Thermal Hydraulics, Livorno, Italy, June 15–17, paper 33. Martelli, D., Forgione, N., Di Piazza, I., Tarantino, M., 2015. HLM fuel pin bundle experiments in the CIRCE pool facility. Nucl. Eng. Des. 292, 76–86. Moon, S., Kim, J., Cho, S., Kim, B.J., Park, J.K., Youn, Y.J., Song, C., 2014. Single-phase convective heat transfer enhancement by spacer grids in a rod bundle. J. Nucl. Sci. Tech. 51 (4), 543–557. Naveen Raj, M., Velusamy, K., Maity, R.K., 2016. Thermal hydraulic investigations on porous blockage in a prototype sodium cooled fast reactor fuel pin bundle. Nucl. Eng. Des. 303, 88–108. Nishimura, M., Sato, H., Kamide, H., Ohshima, H., Nagasawa, K., Imai, Y., 2012. Investigation on velocity distribution around the wrapping wire in an inner subchannel of fuel pin bundle. In: Proc. 20th Int. Conf. Nucl. Eng. July 30-August 3 Anaheim, CA, USA. pp. 11. Pacio, J., Daubner, M., Fellmoser, F., Liftin, K., Marocco, L., Stieglitz, R., Taufall, S., Wetzel, Th., 2014. Heavy-liquid metal heat transfer experiment in a 19-rod bundle with grid spacers. Nucl. Eng. Des. 273, 33–46. Palmieri, L., Schenato, L., 2013. Distributed optical fiber sensing based on Rayleigh scattering. Open Opt. J. 7 (Suppl-1, M7), 104–127. Pan, Y., Jiang, J., Liu, K., Wang, S., Liu, T., 2016. Note: response time characterization of fiber Brag grating temperature sensor in water medium. Review Sci. Instr. 87, 116102. http://dx.doi.org/10.1063/1.4967720. Patil, N.D., Das, P.K., Bhattacharyya, S., Sahu, S.K., 2012. An experimental assessment of cooling of a 54-rod bundle by in-bundle injection. Nucl. Eng. Des. 250, 500– 511. Podila, K., Rao, Y.F., 2014. Assessment of CFD for the Canadian SCWR bundle with wire wraps. Prog. Nucl. Energy 77, 373–380. Rahim N.A.A., Davis, M.A., Routhier, L., Chang, B., Chevalier, J., Bos J.J., Kreger S.T., Sanborn E.E., 2015. Accuracy and survivability of distributed fiber optic temperature sensors. In: 53rd AIAA Aerospace Sciences Meeting, AIAA SciTech Forum, (AIAA 2015–1920). doi: 10.2514/6.2015-1920. Rinaudo, P., Paya-Zaforteza, I., Calderón, P., Salvador, S., 2016. Experimental and analytical evaluation of the response time of high temperature fiber optic sensors. Sensors Actuators A 243, 167–174. http://dx.doi.org/10.1016/j. sna.2016.03.022. Sayde, C., Thomas, C.K., Wagner, J., Selker, J., 2015. High-resolution wind speed measurements using actively heated fiber optics. Geophys. Res. Lett. 42, 10064– 10073. http://dx.doi.org/10.1002/2015GL066729. Shin, B.S., Chang, S.H., 2009. CHF experiment and CFD analysis in a 2 3 rod bundle with mixing vane. Nucl. Eng. Des. 239, 899–912. Thomas, C.K., Kennedy, A.M., Selker, J.S., Moretti, A., Schroth, M.H., Smoot, A.R., Tufillaro, N.B., Zeeman, M.J., 2012. High-resolution fibre-optic temperature sensing: a new tool to study the two-dimensional structure of atmospheric surface-layer flow. Boundary Layer Metorol. 142, 177–192. http://dx.doi.org/ 10.1007/s10546-011-9672-7. Van Wyk, A.J., Swart, P.L., Chtcherbakov, A.A., 2006. Fibre Bragg grating gas temperature sensor with fast response. Meas. Sci. Tech. 17, 1113–1117. http:// dx.doi.org/10.1088/0957-0233/17/5/S29. Wang, H., Bi, Q., Leung, L.K.H., 2016a. Heat transfer from a 2 2 wire-wrapped rod bundle to supercritical pressure water. Int. J. Heat Mass Transfer 97, 486–501. Wang, X., Wang, R., Du, S., Chen, J., Tan, S., 2016b. Flow visualization and mixing quantification in a rod bundle using laser induced fluorescence. Nucl. Eng. Des. 305, 1–8. } ksel, K., Mégret, P., Wuilpart, M., 2011. A quasi-distributed temperature sensor Yu interrogated by optical frequency-domain reflectometer. Meas. Sci. Technol. 22, 115204. http://dx.doi.org/10.1088/0957-0233/22/11/115204.
Contents lists available at ScienceDirect
Nuclear Engineering and Design journal homepage: www.elsevier.com/locate/nucengdes
Distributed temperature sensing inside a 19-rod bundle S. Lomperski ⇑, N. Bremer, C. Gerardi Argonne National Laboratory, 9700 S. Cass Avenue, Lemont, IL, USA
h i g h l i g h t s
g r a p h i c a l a b s t r a c t
Bundle temperatures were measured
with two fiber optic distributed temperature sensors. One sensor generated 3300 readings along rod wire wraps. A second sensor generated 9300 readings in the gas space between the rods. Demonstrated feasibility of obtaining 3D temperature fields within heated rod bundles.
a r t i c l e
i n f o
Article history: Received 30 January 2017 Received in revised form 5 May 2017 Accepted 9 May 2017 Available online 23 May 2017 Keywords: Fiber optic distributed temperature sensor Nuclear fuel rod bundle
a b s t r a c t The temperature field within a model of a sodium-cooled fast reactor fuel rod bundle was measured using Ø155 lm fiber optic distributed temperature sensors (DTS). The bundle consists of 19 electrically-heated rods Ø6.3 mm and 865 mm long. Working fluids were argon and air at atmospheric pressure and Reynolds numbers up to 300. A 20 m-long DTS was threaded through Ø1 mm capillaries wound around rods as wire-wraps. The sensor generated 173 measurements along each rod at 5 mm resolution for a total of 3300 data locations. A second DTS, 58 m long, was suspended between rods to provide 9300 fluid temperature measurements at 20 mm resolution. Such data density makes it possible to construct 3D maps of the temperature field that are beyond the reach of traditional sensors such as thermocouples. This is illustrated through a series of steady-state and transient tests. The work demonstrates the feasibility of mapping temperature within the close confines of a rod bundle at resolutions suitable for validation of computational fluid dynamics codes. Ó 2017 Published by Elsevier B.V.
1. Introduction Computational fluid dynamics (CFD) can provide insights into the thermal hydraulics of reactor fuel bundles that are unavailable through traditional system codes. CFD tools generate detailed depictions of flow and temperature fields that can be applied to improving both fuel performance and reactor safety. CFD is increasingly employed in the design of advanced reactor cores (e.g., Podila and Rao, 2014) as well as optimization of conventional LWR cores (Ikeda, 2014) and troubleshooting (Cinosi and Walker, ⇑ Corresponding author. E-mail address: [email protected] (S. Lomperski). http://dx.doi.org/10.1016/j.nucengdes.2017.05.008 0029-5493/Ó 2017 Published by Elsevier B.V.
2016; Naveen Raj et al., 2016). Still, the thermal hydraulic behavior of reactor cores is complex and remains challenging to simulate in detail with CFD. Confidence in simulation accuracy can be obtained through code validation, but it is generally difficult to map velocity and temperature fields of real flows with spatial resolution adequate for rigorous CFD benchmarking. Fuel bundles are especially challenging because geometries are both intricate and compact with little room for instrumentation. Bundle heat transfer is of special interest because it plays fundamental roles in governing reactor efficiency and setting safety margins. Thermal hydraulic test data for heated bundles is generally limited to integral measurements such as pressure drop and flow rate
202
S. Lomperski et al. / Nuclear Engineering and Design 319 (2017) 201–209
along with point measurements of temperature from thermocouples. Recent work includes mixing vane and spacer grid studies for LWRs (In et al., 2015; Moon et al., 2014; Shin and Chang, 2009), thermal stratification tests in a 25-rod bundle (Kodama et al., 2014), and emergency cooling system testing with a 54-rod bundle (Patil et al., 2012). Also of interest are rod bundles for advanced reactors, such as those cooled by lead-bismuth eutectic (Martelli et al., 2015; Pacio et al., 2014) and supercritical water (Wang et al., 2016a). The abundance of recent studies attests to continued strong interest in rod bundle heat transfer. Yet these experiments do not generate the kind of whole-field data required for rigorous CFD code validation. Whole-field mapping is typically accomplished with optical techniques such as particle image velocimetry (PIV) and laserinduced fluorescence (LIF). Both rely upon imaging regions of a flow field illuminated by a laser sheet. But optical access to the interior of a bundle constructed of metals and ceramics is severely restricted and therefore largely inaccessible to these instruments. Some investigators have mapped velocity fields within bundles by employing transparent structural materials in so-called MIR (matched index of refraction) setups (Nishimura et al., 2012; Dominguez-Ontiveros and Hassan, 2009; Lee et al., 2013). A MIR setup was also used by Wang et al. (2016b) with a 3 3 bundle and LIF to characterize mixing. These experiments generate high resolution velocity field data, but lack the heat transfer and thermal mixing elements that are of ultimate interest in evaluating fuel bundle performance. MIR setups are generally isothermal since refractive indices of fluid and structure, which vary with temperature, must remain matched to avoid image distortions that lead to measurement errors. A solution lies in fiber optic distributed temperature sensors (DTS), which function without imaging. A single DTS can map temperature fields at resolutions far exceeding anything practicable with conventional sensors such as thermocouples. They function in opaque flows and can be woven through complex geometries to generate data in regions lacking optical access for techniques like LIF. DTSs are, however, infrequently applied to multidimensional-mapping of temperature or velocity across flow fields. Thomas et al. (2012) used them to map thermal structures in atmospheric flows across a region 8 m 8 m and Sayde et al. (2015) mapped wind speed along a 230 m gully using actively heated optical fibers. DTSs have also been used at smaller scales in laboratory settings for 2D mapping of subsurface water flow (Bersan et al., 2015), air jet flow (Chen et al., 2012), and air jet mixing in the context of thermal striping (Lomperski et al., 2015). Rod bundles are notoriously difficult to instrument. Thermocouples can be inserted into rods, but only in very limited numbers since they displace structural material such as cladding and insulation. Instrumenting the flow field between the rods is equally problematic since space is again too limited to accommodate numerous sensors. Sensing technologies based on optical fiber offer high data density since a single narrow fiber generates data at many measurement points. De Pauw et al. (2016) used fiber Bragg grating sensors in a bundle cooled by lead-bismuth eutectic, integrating 60 sensors into the cladding of a 7-rod bundle. These are essentially point sensors, however, and so it is difficult to deploy them in numbers sufficient for mapping flow fields. This paper demonstrates the feasibility of using DTSs for 3D mapping of rod bundle temperature fields. We begin with characterization of the bundle and test section followed by a description of the measurement technique and sensors. Selected data is then presented to illustrate the capability of this measurement technique and its potential utility in fluid dynamics testing for nuclear engineering applications.
2. Experiment 2.1. Rod bundle An acrylic test section houses 19 wire-wrapped heater rods in a configuration characteristic of many sodium fast reactor designs. Fig. 1 shows the bundle layout along with key dimensions. The rods are Ø6.3 mm and the wire-wraps Ø1 mm stainless steel capillaries with an inner diameter of 0.5 mm. The rod pitch/diameter ratio is 1.36 while the wire-wrap pitch is 215 mm. Total rod length is 865 mm with a heated length of 220 mm centered 130 mm above the middle of the rod. The heaters predate this study and their asymmetric heating layout was a fixed design parameter. Bundle support plates are 10 mm-thick polypropylene with Ø1 mm holes for passage of the wire-wrap capillaries and gas space DTS. The bundle is encased in transparent acrylic plates to form a hexagonal test section with a flow area of 1088 mm2, wetted perimeter 527 mm, and hydraulic diameter 8.2 mm. Working fluids were either argon or air delivered to the test section at atmospheric pressure and ambient temperature. Flow enters the top of the test section through a 6-inlet manifold and exits through six holes 30 mm above the base plate. Flow direction is arbitrary for this study. The downward direction was chosen for a longer flow path after the heated section, providing more measurements of thermal mixing before flow exits the test section. A port at the test section midpoint is reserved for tests involving jet injection perpendicular to the bundle. 2.2. Sensing systems The sensing systems used in this study are based on Rayleigh scattering. They use a tunable diode laser to launch a narrow-band infrared signal into a fiber followed by detection and analysis of light scattered back towards the source (Gifford et al., 2007). The spectrum of the backscattered light shifts with changes in a fiber’s physical dimensions or refractive index. Strain and temperature changes generate such shifts and so optical fibers can serve as sensors through measurements of the backscatter spectrum. The physical basis for the spectrum shift can be understood as that of a continuous, weak fiber Bragg grating (Kreger et al., 2006). A fiber Bragg grating consists of periodic variations in the core refractive index along a section of the fiber optic waveguide (Bhatia, 2001). Straining a grating alters its periodicity and shifts the reflected spectrum while temperature changes alter both periodicity and refractive indices, also shifting the spectrum. An analogous process occurs in a DTS as strain or changes in temperature alter the relative positions of scattering centers along the fiber optic cable. Therefore DTS response can be described by that of a } ksel et al., 2011): fiber Bragg grating (Gifford et al., 2007; Yu
Dk ¼ K T DT þ K e e k
ð1Þ
where Dk is the wavelength shift while KT and Ke are the temperature and strain coefficients, respectively. KT includes coefficients for thermal expansion and the index of refraction. The measured wavelength shift is a composite of strain and temperature signals, but sensing system electronics cannot distinguish between the two. One must be kept constant or known in some way in order to measure the other. The fibers are used here as temperature sensors and so strain is controlled in our setup. Alternative sensing methods utilizing Raman or Brillouin scattering do not share this difficulty in distinguishing strain from temperature but their spatial resolution and bandwidth are insufficient for the rod bundle application. The
S. Lomperski et al. / Nuclear Engineering and Design 319 (2017) 201–209
203
Fig. 1. a) Cross section of bundle support plate, and b) test section flow paths (dimensions in mm).
physics and capabilities of all these techniques are considered in reviews by Bao and Chen (2012) and Palmieri and Schenato (2013). Note that measurements are inherently differential in nature. The sensor measures DT from a ‘‘baseline”, a reference measurement performed under controlled conditions that is analogous to zeroing an instrument: new readings are identically zero as long as the sensor remains at the baseline temperature. Baselining for this study was performed at ambient temperature under isothermal conditions so that DT = 0 would be associated with a single temperature for the entire sensor. A link to absolute temperature is made through supplementary sensors such as a thermocouple or resistance temperature detector (RTD), which register absolute temperature while the baseline reading is recorded. The adjusted signal after baseline can then be written as:
TðxÞabs ¼ DTðxÞ þ TðxÞbase
ð2Þ
where T(x)abs is the absolute temperature along the DTS and T(x)base is the absolute temperature during baseline recorded by one or more TCs or RTDs. Baselining under isothermal conditions is the simplest case since conversion to absolute temperature is achieved with a single offset that applies to the entire sensor. Measurement accuracy for our bundle setup is estimated to be better than ±1 °C. Accuracy is not assessed here in detail since our purpose is to demonstrate unique sensing capabilities in the context of the complex geometry of a rod bundle. DTS accuracy and repeatability have been considered elsewhere under stagnant conditions (e.g., Rahim et al., 2015, Lomperski et al., 2013) and for sensors suspended within a flow field (Lomperski and Gerardi, 2014; Lomperski et al., 2016). Calibration issues and challenges are considered in the discussion section.
Two sensing systems were used in this study, an ODiSI-A for the gas space DTS and an ODiSI-B for the wire-wrap, both from Luna, Inc. The latter has higher bandwidth that could be better utilized with the gas space sensor, but assignments were based on system limits for sensor length. The former system accommodates sensors of more than 50 m while the latter is limited to 20 m, specifications that nearly match the two DTSs installed in the bundle. 2.3. Sensor configuration Both DTSs were made from commercial Ø155 mm polyimidecoated single-mode telecom fiber (SMB-G131OH, Specialty Photonics). The first is a 20 m sensor inserted through all 19 wirewraps while the second is 58 m long and suspended within the gas space as 54 vertical segments. A segment is defined as a single length of fiber spanning the distance between the top plate and baseplate. Segment length matches that of the bundle, 865 mm. Fig. 2 shows a close-up view of the bundle near the baseplate to highlight sensor positioning and physical differences between the wire-wrap sensor and the bare gas space sensor. Note that the wire-wrap sensor will be used to represent rod temperatures though it is actually in the capillaries. The wire-wraps are tightly wound around the rods for good thermal contact while the heat transfer coefficient between wire-wrap and argon is low. Consequently, the wire-wrap temperature is an adequate surrogate for rod temperature for the purposes of this demonstration study. Both DTSs pass repeatedly through the support plates and are looped outside the test section to return back into the bundle for subsequent segments. Fig. 3 shows the looping arrangement with fiber glued to the top plate. Loops are kept small since they are outside the bundle and effectively lost for sensing. Each consumes 170 mm of fiber, roughly 12 m for all 72 loops and 15% of total
204
S. Lomperski et al. / Nuclear Engineering and Design 319 (2017) 201–209
flow. This was accomplished here by hanging a weight from each of the loops below the base plate. Weights maintained constant segment tension since fibers pass freely through the /1 mm holes in the baseplate. Tension also positions the gas space segments, ensuring they remain parallel to the rods without touching them. The weights are inconsequential for the wire-wrap sensor housed in capillaries.
3. Results 3.1. 3D temperature field visualization
Fig. 2. Photo of rods and sensors above baseplate.
Fig. 3. Photo of fiber loops above top plate.
available sensor length. The fiber manufacturer specifies a ‘‘shortterm” bend radius limit of 10 mm and ‘‘long term” limit of 17 mm. The loop radius here, 10–15 mm, has been sustained for over a year without breakage or noticeable signal degradation. DTS mounting arrangements must preclude changes in sensor strain levels after the baseline. Sensors should not be strained by test section distortions accompanying pressurization, heating, or
A steady flow test was run at 28 L/min of argon entering the manifold 5 °C below ambient temperature and Reynolds number of approximately 160. The bundle was heated at 160 W until rod temperatures neared 100 °C, then power was cut and DTS data logged at 1 Hz. Fig. 4 presents raw DTS data in the form of DT from the baseline temperature, which was recorded shortly before the test while the setup was isothermal at room temperature with no flow. Data is raw in the sense that it is neither converted to absolute temperature nor mapped to the physical space of the test section. These curves include the loops outside the test section, which are later excluded when mapping data to the bundle. Fig. 4 shows a single scan of the wire-wrap DTS consisting of 4000 measurement points at 5 mm gage and 5 mm spacing. Gage length is the effective spatial resolution, the span over which Dk is analyzed to generate a single DT data point. The plot exhibits distinct peaks where the sensor passes along the heated sections of the rods. DT remains near zero, i.e., ambient temperature, along the cold ends of the rods and the loops outside the test section. Peak temperatures are not equal for two reasons: 1) cooling rates differ for interior and exterior rods, and 2) input power varied slightly among rods since it was distributed through three variable transformers, two wired to six rods each and the third to seven. Power to individual rods was not measured. The saw tooth pattern in the rod data is also seen in the gas space data. Sharp peaks occur where the sensor passes through the heated region of the bundle. Like the wire-wrap DTS, DT0 along external loops as seen in the plot inset. DT is slightly below zero upstream of the heated zone since argon enters the test section below ambient temperature. Gas space data was generated with a 20 mm gauge and sampled at 5 mm spacing for 11600 points along the entire sensor. Such oversampling is routinely applied to PIV data to maximize spatial resolution. The sensing system is capable of 10 mm gage, but signal noise associated with flow-induced vibration limited resolution to 20 mm. Both gage and spacing can be adjusted in post processing software to optimize tradeoffs between spatial resolution and signal noise. Such flexibility contrasts with the immobility of thermocouples, which are seldom repositioned in response to test findings to optimize a setup. While Fig. 4 illustrates the large amount of data available from each sensor, the curves give only a vague impression of the bundle temperature field. Sensor data is of limited utility until mapped onto the geometry of a test section. Complex patterns and trends are more readily revealed after decomposing 1D data and translating it into two or three-dimensional space. Such mapping was performed for the wire-wrap DTS data to generate the images in Fig. 5. With the data presented in this fashion, the extent of the heated zone is distinct and easily recognized. The magnified cutaway view reveals temperatures within the bundle interior, illustrating that data is generated where there is no access for optical instruments. Fig. 5 is based on 3300 total data points, 173 points/rod at 5 mm spacing. Though the DTS spirals around rods within the wire-wraps, circumferential variations in rod temperature are not
S. Lomperski et al. / Nuclear Engineering and Design 319 (2017) 201–209
205
Fig. 4. Raw DTS data for rods (left) and gas (right).
Fig. 5. Wire-wrap DTS data mapped onto rods; (a) full bundle, (b) magnified ¼bundle cutaway view.
considered here, only axial. The wire-wraps are 75 mm longer than the bundle, but data was clipped at the bundle length of 865 mm. As a result, rod temperature profiles are stretched 9%, but the distortion is irrelevant for the purpose of this feasibility study. For CFD code validation, however, DTS data would be more precisely mapped onto the rods to preclude such distortions. Gas space data is presented in Fig. 6 through two different plot styles. The first is a conventional 3D rendering with a single marker for each data point, 173 points/segment at 5 mm spacing, 9300 total for all 54 segments. As with the rod data, color indicates tem-
Fig. 6. Bundle gas temperatures represented with a) one marker per data point, b) interpolated solid, and c) overlay of gas and rod data.
perature level. The rendering provides a sense of the high data density available from a single sensor. The second image in Fig. 6 is a volumetric rendering of the data based on data point interpolation. Easily recognized is a warm plume centered where the rod bundle was seen to be hottest in Fig. 5. Downstream of the hot zone, gas quickly returns to ambient temperature as it flows along the cold
206
S. Lomperski et al. / Nuclear Engineering and Design 319 (2017) 201–209
end of the bundle. Such a detailed depiction of a bundle temperature field is unobtainable through conventional sensors such as thermocouples. The third image combines both the gas and rod data to emphasize that one could pair them to investigate correlations or patterns linking the two. For example, examining axial development of the thermal field as fluid flows along cold and hot regions of the rods. Data density here is sufficient for section views to examine the bundle interior and to scan the entire domain for unexpected phenomena, much as one would with simulation data. In addition, detailed field data can help confirm that boundary conditions are well-behaved and acceptable for meaningful CFD code validation, a shortcoming of some legacy experiments being used for such work. 3.2. 1D temperature field analysis Volumetric rendering of whole data sets is an efficient means of obtaining a general impression of an entire flow field. However, code validation and quantitative analysis are often better suited to manipulation of small data subsets. This is illustrated here with two test cases. The first involves step changes in flow rate to show variations in rod and gas temperature with Reynolds number. The second presents gas temperatures around the center rod for a single flow rate to highlight the ability to capture temperature field asymmetries. The first case began at a bundle power of 24 W and airflow rate of 14 L/min (Re = 100), followed by step increases of 14 L/min in 60 s intervals. All rods were heated, but each end of the bundle remained near ambient temperature. Fig. 7 shows temperature profiles along a selected rod and neighboring gas space. A perimeter rod was chosen rather than interior because it exhibits the largest temperature changes with flow rate for a more compelling demonstration case. The plot shows temperature profiles peaking near the center of the heated zone and flattening at higher flow rates. At Reynolds 100, gas temperature remains below rod temperature along the entire length of the bundle. At Reynolds 200, rod and gas temperatures are nearly equal between 0.4 and 0.6 m while for Reynolds 300 rod temperature falls below gas temperature between 0.5 and 0.6 m. This is not typical of actual reactor cores, but here the flow is not fully developed since the test section is short while rods are not all heated at the same power. Cross flows from hotter neighboring rods are likely skewing the temperature profiles. Regardless, the data demonstrates an ability
Fig. 7. Temperatures along a selected rod and gas segment versus Reynolds number. Measurement uncertainties ±1 °C and ±10 mm.
to characterize subtle shifts in the temperature field with changes in boundary conditions. Fig. 8 presents temperatures along the center rod with data from all six adjacent gas segments. The rod was heated at 9 W with airflow at 14 L/min and Re = 100. Data was logged about 1 h after heating began, brief enough that both rod ends remained near ambient temperature. Cool air 5 °C below ambient temperature was injected into the manifold, which is reflected in the gas segment readings. All are below the rod temperature in the inlet region upstream of the heated zone. Air warmed as it flowed along the center rod, but once past the heated zone it returned nearly to room temperature as it mixed with cold air from the rest of the bundle that made up the majority of the flow. The plot shows radial variation in the temperature field despite radial symmetry of the test section. As noted above, flow is not fully developed. Asymmetries in the velocity field are again seen to be reflected in the temperature field. Signal noise is low at Re = 100 and so data was processed at 10 mm gage instead of 20 mm. The extra resolution reveals sharp peaks at 120°, 180° and, to a lesser extent, 60°. These peaks coincide with proximity to the wire-wrap as it spirals up through the heated zone. Distance between the wire-wrap and bare fiber is <1 mm at the point of closest approach (see Fig. 1). The sequence of peaks progresses with the wire-wrap: the first at 0.26 m followed by the main peak at 0.3 m and the last at 0.34 m. The peak at 0.3 m (180°) is at the center of the heated zone and is the most distinct of the three. There are no peaks at 0°, 240°, and 300° because the wire-wrap is outside the heated zone at these positions. Figs. 7 & 8 demonstrate an ability to measure both rod and bulk fluid temperatures continuously along the bundle, allowing determination of local heat transfer coefficients. With fluid temperature measured at six positions around every interior rod, it is possible to examine circumferential as well as axial variations in the heat transfer coefficient. Such information could be used to identify crossflows or assess mixing vane efficiency. 3.3. Hot jet injection The volumetric renderings of the gas phase data seen earlier suggest data density is high enough to visualize complex flow structures within the bundle. This can be of particular interest in CFD code validation since much of the appeal of such codes lies
Fig. 8. Temperatures along the central rod and six neighboring gas segments at Re = 100. Data processed at 10 mm gage. Measurement uncertainties ±1 °C and ±10 mm.
S. Lomperski et al. / Nuclear Engineering and Design 319 (2017) 201–209
in their superior spatial resolution compared to traditional system codes, which rely upon lumped parameter models. Code proficiency in simulating multidimensional flow structures can be checked against high-resolution experimental data on complex flows. We chose jet injection as an example of such a flow. A stream of argon was injected into the bundle at its midpoint through a 5 mm diameter port, flow rate 50 L/min and temperature of 75 °C with no other flow into the system. The upper manifold was sealed so that gas entered the bundle perpendicular to the rods and flowed downward, mixing and eventually exiting through the six outlets near the base. Rods were unpowered and left at room temperature to achieve better contrast in visualizing the jet and impingement heating of the rods. Though this configuration does not mimic any specific flow phenomena characteristic of
207
sodium-cooled fast reactors, it produces a complex flow structure that is small enough to challenge the spatial resolution of the sensing network. Fig. 9 presents a series of snapshots portraying the transient. The thermal field develops both in and around the rods and so three different views are presented to convey the structure of the flow. The first shows gas temperature overlaid onto the rods, the second depicts only rods, and the third is a top view to track the advance of a ‘‘warm front” into the bundle. The first frame corresponds to stagnant conditions at ambient temperature while the second marks the start of flow. Gas space warming is clearly visible at this early stage, though there is little change in rod temperature. Advance of the warm front into the second row of rods can be seen from the top view. Note that images show gas only if its
Fig. 9. Bundle temperatures during jet injection; first of each bundle pair shows gas with rods, the second only rods. Gas at T < 21 °C was rendered transparent to emphasize jet structure.
208
S. Lomperski et al. / Nuclear Engineering and Design 319 (2017) 201–209
temperature is >21 °C, i.e., ambient-temperature gas is invisible, to isolate the structure of the warm zone created by the jet. The third frame at t = 21 s shows advance of the warm front and general increase in gas temperature. Warming of rods around the impingement zone is now apparent. Data for the rods must be viewed with some caution since mapping is 1D even though the DTS spirals around rods via the wire-wraps. In some places the jet impinges directly on wire-wraps while in other spots the wire-wraps are on the lee side of a rod. This is certain to generate artifacts that are not considered here. The final frame at t = 43 s shows a broad band of cyan-colored gas around and below the injection point. Cyan corresponds to very slight warming above ambient temperature, only 1 °C, and indicates that the thermal effects of the jet have advanced beyond the immediate vicinity of the injection point. Note that this warm zone does not extend above the injection point and recall that the top of the test section is sealed. The thermal map is seen to give an indirect indication of the flow pattern: argon flows downward from the injection point and mixes in the bundle while a stable, stagnant layer persists above the injection point. This snapshot series demonstrates that the sensor network provides data density high enough to characterize small thermal structures inside and around the bundle. An animated version of the transient is available as Supplementary data. 4. Discussion Mapping velocity and temperature fields in rod bundles has proven exceptionally challenging even in low-temperature, ambient-pressure models like the one used for this study. Here we demonstrated the feasibility of using fiber optic distributed temperature sensors to map temperature fields within a heated bundle. The DTSs generated 3300 measurements along rod wirewraps and 9300 measurements in gas spaces between rods. Such data density could not have been achieved with traditional physical probes like thermocouples. Optical methods such as LIF can match this data density and even achieve superior bandwidth and spatial resolution, but cannot access the interior of a bundle as the DTSs have here. Furthermore, DTSs perform equally well in transparent and opaque fluids while optical instruments do not. With distributed sensors, high-resolution temperature mapping across sizable expanses of a rod bundle becomes conceivable. Sensor networks spanning an entire test section can be of great value in fluid dynamics experiments, especially when locations of interesting phenomena are not known a priori. It can be difficult to detect small flow structures like the hot jet presented earlier using only thermocouples, even in exceptionally instrumented setups. Interesting and potentially important phenomena are effectively nonexistent until detected by instrumentation. Distributed sensors can increase the likelihood of observing the unexpected. Despite the potential value of DTSs for bundle mapping, several issues remain in implementing this relatively young technology into fluid dynamics experiments. Two are predominant: signal noise associated with flow-induced vibration and errors from strain changes after the baseline. Though both are essentially uncontrolled variations in strain, the former dynamic and the latter largely static, they differ in how they disturb the temperature signal and in practice are dealt with separately. A third and distinctly different issue is the special difficulty of calibrating these sensors. All three are briefly considered below. 4.1. Dynamic errors Signal noise associated with flow-induced vibration is, in our experience, the most troublesome issue. It can give rise to unpre-
dictable and sometimes debilitating noise levels, especially with long (>20 m) sensors. This is a serious limitation since the full potential of distributed sensing is realized with long sensors. We have investigated noise behavior as the dominant factor limiting effective spatial resolution in long sensors used in a jet mixing experiment (Lomperski and Gerardi, 2014; Lomperski et al., 2016). Indeed, such noise limited testing here to laminar flows. Though laminar flow is relevant to some natural circulation and severe accident states, turbulent flows are of more general interest in nuclear engineering. In practice, noise can be reduced by decreasing sensor length, increasing gage length, and through sensor support arrangements that reduce vibration. These adjustments will increase effective limits on flow rates and Reynolds numbers. For example, vibration of the gas space DTS might have been reduced with heavier weights on the loops, though we observed no remarkable improvement after increasing weight from 2 g to 9 g. However, this fiber can support considerably more (proof strength 0.7 GPa) and, in separate testing, withstood a 250 g load for weeks without failing. Noise is far less troublesome for sensors housed in rigid capillaries. There were no signal noise difficulties with the wire-wrap DTS, but its thermal response is much slower than that of the bare fiber. The extent of the tradeoff is illustrated with an example based on a simple lumped-parameter thermal analysis: response time in a cross flow of argon at 1 m/s is roughly 0.2 s for the bare fiber used here, increasing to 9 s when housed in a /1 mm steel capillary. Response times in water are faster, of the order of milliseconds for bare fiber and 0.05 s with the capillary. Others have measured response times of bare fibers (Van Wyk et al., 2006; Pan et al., 2016) and fibers housed in capillaries (Rinaudo et al., 2016). In the future, signal noise is expected to be less troublesome as optical fiber interrogators evolve. In addition, others have demonstrated the ability to distinguish between stress and strain using paired fibers with differing strain and temperature sensitivities (Ding et al., 2016) and single, polarization-maintaining fibers (Li et al., 2013). 4.2. Static errors Control of static strain, the second issue, is essential but can be problematic even for sensors housed in capillaries. For example, measurement errors were observed in 7-m long fibers around mounting points for steel capillaries (Lisowski et al., 2013), a consequence of thermal expansion upon heating the test loop. Deformation of sensor supports or the test section itself can strain fibers via the capillaries. Strain control typically involves simply ensuring a complete lack of fiber tension. We accomplished this in other setups involving 2D arrays by maintaining slack on all fiber segments. Taut segments will experience strain shifts with even minute deformations of supports, generating measurement errors. For the bundle setup, the weights maintained constant segment tension, but at the cost of giving up hermetic seals between sensor and baseplate. This is a clear disadvantage since liquids tend to be of more interest than gases in nuclear engineering applications. Capillaries facilitate sealing and can be an alternative for applications where slower thermal response time is acceptable. 4.3. Accuracy The third concern involves sensor calibration and accuracy. Calibration, outlined earlier, is simple in principle and can be accomplished with a single thermocouple. But the thermocouple, a point sensor, must serve as a standard for many meters of DTS embodying thousands of sensing points. Adding thermocouples could improve the situation, but the incongruity of calibrating
S. Lomperski et al. / Nuclear Engineering and Design 319 (2017) 201–209
distributed sensors with point sensors remains. Still, the situation is not unprecedented. It parallels that of PIV, which is often benchmarked against point sensors such as hot-wire anemometers or laser Doppler velocimeters. Accuracy assessments are similarly challenging since strain sensitivity effectively rules out the common practice of sensor calibration at the point of origin. In-situ calibration is mandatory, making it difficult to achieve traceability to recognized standards such as NIST. The nature of in-situ calibrations will depend upon the application and so sensor accuracy and repeatability will vary accordingly. This again parallels PIV, which enjoys widespread use and acceptance despite similar issues. We have performed rudimentary assessments of DTS accuracy and stability in a jet mixing setup (Lomperski and Gerardi, 2014; Lomperski et al., 2016) that can serve as examples of sensor performance in a fluid dynamics experiment. Acknowledgment Data visualizations by VisIt; Bethel, E.W., Childs, H., Hansen, C., High performance visualization: enabling extreme-scale scientific insight, Chapman & Hall, 2012. The submitted manuscript has been created by UChicago Argonne, LLC, Operator of Argonne National Laboratory (‘‘Argonne”). Argonne, a U.S. Department of Energy Office of Science laboratory, is operated under Contract No. DE-AC0206CH11357. The U.S. Government retains for itself, and others acting on its behalf, a paid-up nonexclusive, irrevocable worldwide license in said article to reproduce, prepare derivative works, distribute copies to the public, and perform publicly and display publicly, by or on behalf of the Government. The Department of Energy will provide public access to these results of federally sponsored research in accordance with the DOE Public Access Plan. http://energy.gov/downloads/doe-public-access-plan. Appendix A. Supplementary data Supplementary data associated with this article can be found, in the online version, at http://dx.doi.org/10.1016/j.nucengdes.2017. 05.008. References Bao, X., Chen, L., 2012. Recent progress in distributed fiber optic sensors. Sensors 12, 8601–8639. http://dx.doi.org/10.3390/s120708601. Bersan, S., Schenato, L., Rajendran, A., Palmieri, L., Cola, S., Pasuto, A., Simonini, P., 2015. Application of a high resolution distributed temperature sensor in a physical model reproducing subsurface water flow. Measurement. http://dx.doi. org/10.1016/j.measurement.2015.09.018. Bhatia, V., 2001. Properties and applications of fiber gratings. Proc. Int. Conf. Fiber Optics Photonics SPIE 4417:154160. Chen, T., Wang, Q., Zhang, B., Chen, R., Chen, K.P., 2012. Distributed flow sensing using optical hot-wire grid. Opt. Express 20 (8), 10. Cinosi, N., Walker, S.P., 2016. CFD analysis of localized crud effects on the flow of coolant in nuclear rod bundles. Nucl. Eng. Des. 305, 28–38. De Pauw, B., Lamberti, A., Ertveldt, J., Rezayat, A., Vanlanduit, S., Van Tichelen, K., Berghmans, F., 2016. Temperature monitoring using fibre optic sensors in a lead-bismuth eutectic cooled nuclear fuel assembly. Nucl. Eng. Des. 297, 54–59. Ding, Z., Yang, D., Du, Y., Liu, K., Zhou, Y., Zhang, R., Xu, Z., Jiang, J., Liu, T., 2016. Distributed strain and temperature discrimination using two types of fiber in OFDR. IEEE Phot. J. 8 (5), 8. http://dx.doi.org/10.1109/JPHOT.2016.2605011. Dominguez-Ontiveros, E.E., Hassan, Y.A., 2009. Non-intrusive experimental investigation of flow behavior inside a 5 5 rod bundle with spacer grids using PIV and MIR. Nucl. Eng. Des. 239, 888–898. Gifford, D.K., Kreger, S.T., Sang, A.K., Froggatt, M.E., Duncan, R.G., Wolfe, M.S., Soller, B.J. 2007 Swept-wavelength interferometric interrogation of fiber Rayleigh scatter for distributed sensing applications. In: Proc SPIE 6770 Fiber Optic Sens Appl V 67700F. Ikeda, K., 2014. CFD application to advanced design for high efficiency spacer grid. Nucl. Eng. Des. 279, 73–82. In, W.K., Shin, C.H., Lee, C.Y., 2015. Convective heat transfer experiment of rod bundle flow with twist-vane spacer grid. Nucl. Eng. Des. 295, 173–181.
209
Kodama, S., Yoshida, K., Kataoka, I., 2014. Thermal stratification and mixing under single- and two-phase flow in rod bundle with axially distributed heat flux in vessel without forced convection. Nucl. Eng. Des. 268, 164–172. Kreger, S.T., Gifford, D.K., Froggatt, M.E., Soller, B.J., Wolfe, M.S., 2006. High resolution distributed strain or temperature measurements in single- and multi-mode fiber using swept-wavelength interferometry. Opt. Fiber Sens. OSA/ OFS 2006, 4. Lee, C.Y., Shin, C.H., Park, J.Y., In, W.K., 2013. An experimental investigation on turbulent flow mixing in a simulated 3 x 3 dual-cooled annular fuel bundle using particle image velocimetry. Nucl. Eng. Des. 260, 134–144. Li, W., Chen, L., Bao, X., 2013. Compensation of temperature and strain coefficients due to local birefringence using optical frequency domain reflectometry. Optics Communications 311, 26–32. http://dx.doi.org/10.1016/j.optcom.2013.08.022. Lisowski, D., Lomperski, S., Farmer, M.T., Bremer, N., Kilsdonk, D.J., Aeschlimann, R. W., Vik, A., Engel, D., 2013. NSTF demonstration test report, Argonne National Laboratory technical report ANL-NSTF-000000-TECH-009-NC. Lomperski, S., Gerardi, C., 2014. Assessment of distributed fiber optic sensors for flow field temperature mapping. In: Proc. ASME 2014 4th Joint US-European Fluids Engineering Division Summer Meeting and 12th International Conference on Nanochannels, Microchannels, and Minichannels FEDSM2014 August 3–7, Chicago. Lomperski, S., Gerardi, C., Pointer, D.W., 2013. Distributed fiber optic temperature sensing for CFD code validation. In: Proceedings of 15th International Topical Meeting on Nuclear Reactor Thermal Hydraulics (NURETH-15) Pisa, Italy. Lomperski, S., Gerardi, C., Pointer, W.D., 2015. Distributed fiber optic temperature sensor mapping of a jet-mixing flow field. Exp. Fluids 56, 55. http://dx.doi.org/ 10.1007/s00348-015-1918-6. Lomperski, S., Bremer, N., Gerardi, C., Lisowski, D., 2016. Performance assessment of a 50 m-long fiber optic distributed temperature sensor in a fluid dynamics experiment. In: Proc. Specialist Workshop on Advanced Instrumentation Measurement Techniques for Nuclear Reactor Thermal Hydraulics, Livorno, Italy, June 15–17, paper 33. Martelli, D., Forgione, N., Di Piazza, I., Tarantino, M., 2015. HLM fuel pin bundle experiments in the CIRCE pool facility. Nucl. Eng. Des. 292, 76–86. Moon, S., Kim, J., Cho, S., Kim, B.J., Park, J.K., Youn, Y.J., Song, C., 2014. Single-phase convective heat transfer enhancement by spacer grids in a rod bundle. J. Nucl. Sci. Tech. 51 (4), 543–557. Naveen Raj, M., Velusamy, K., Maity, R.K., 2016. Thermal hydraulic investigations on porous blockage in a prototype sodium cooled fast reactor fuel pin bundle. Nucl. Eng. Des. 303, 88–108. Nishimura, M., Sato, H., Kamide, H., Ohshima, H., Nagasawa, K., Imai, Y., 2012. Investigation on velocity distribution around the wrapping wire in an inner subchannel of fuel pin bundle. In: Proc. 20th Int. Conf. Nucl. Eng. July 30-August 3 Anaheim, CA, USA. pp. 11. Pacio, J., Daubner, M., Fellmoser, F., Liftin, K., Marocco, L., Stieglitz, R., Taufall, S., Wetzel, Th., 2014. Heavy-liquid metal heat transfer experiment in a 19-rod bundle with grid spacers. Nucl. Eng. Des. 273, 33–46. Palmieri, L., Schenato, L., 2013. Distributed optical fiber sensing based on Rayleigh scattering. Open Opt. J. 7 (Suppl-1, M7), 104–127. Pan, Y., Jiang, J., Liu, K., Wang, S., Liu, T., 2016. Note: response time characterization of fiber Brag grating temperature sensor in water medium. Review Sci. Instr. 87, 116102. http://dx.doi.org/10.1063/1.4967720. Patil, N.D., Das, P.K., Bhattacharyya, S., Sahu, S.K., 2012. An experimental assessment of cooling of a 54-rod bundle by in-bundle injection. Nucl. Eng. Des. 250, 500– 511. Podila, K., Rao, Y.F., 2014. Assessment of CFD for the Canadian SCWR bundle with wire wraps. Prog. Nucl. Energy 77, 373–380. Rahim N.A.A., Davis, M.A., Routhier, L., Chang, B., Chevalier, J., Bos J.J., Kreger S.T., Sanborn E.E., 2015. Accuracy and survivability of distributed fiber optic temperature sensors. In: 53rd AIAA Aerospace Sciences Meeting, AIAA SciTech Forum, (AIAA 2015–1920). doi: 10.2514/6.2015-1920. Rinaudo, P., Paya-Zaforteza, I., Calderón, P., Salvador, S., 2016. Experimental and analytical evaluation of the response time of high temperature fiber optic sensors. Sensors Actuators A 243, 167–174. http://dx.doi.org/10.1016/j. sna.2016.03.022. Sayde, C., Thomas, C.K., Wagner, J., Selker, J., 2015. High-resolution wind speed measurements using actively heated fiber optics. Geophys. Res. Lett. 42, 10064– 10073. http://dx.doi.org/10.1002/2015GL066729. Shin, B.S., Chang, S.H., 2009. CHF experiment and CFD analysis in a 2 3 rod bundle with mixing vane. Nucl. Eng. Des. 239, 899–912. Thomas, C.K., Kennedy, A.M., Selker, J.S., Moretti, A., Schroth, M.H., Smoot, A.R., Tufillaro, N.B., Zeeman, M.J., 2012. High-resolution fibre-optic temperature sensing: a new tool to study the two-dimensional structure of atmospheric surface-layer flow. Boundary Layer Metorol. 142, 177–192. http://dx.doi.org/ 10.1007/s10546-011-9672-7. Van Wyk, A.J., Swart, P.L., Chtcherbakov, A.A., 2006. Fibre Bragg grating gas temperature sensor with fast response. Meas. Sci. Tech. 17, 1113–1117. http:// dx.doi.org/10.1088/0957-0233/17/5/S29. Wang, H., Bi, Q., Leung, L.K.H., 2016a. Heat transfer from a 2 2 wire-wrapped rod bundle to supercritical pressure water. Int. J. Heat Mass Transfer 97, 486–501. Wang, X., Wang, R., Du, S., Chen, J., Tan, S., 2016b. Flow visualization and mixing quantification in a rod bundle using laser induced fluorescence. Nucl. Eng. Des. 305, 1–8. } ksel, K., Mégret, P., Wuilpart, M., 2011. A quasi-distributed temperature sensor Yu interrogated by optical frequency-domain reflectometer. Meas. Sci. Technol. 22, 115204. http://dx.doi.org/10.1088/0957-0233/22/11/115204.