Development and Application of Molecular Tagging Velocimetry for Gas Flows in Thermal Hydraulics

Abstract Molecular tagging velocimetry (MTV) is a nonintrusive velocimetry technique based on laser spectroscopy. It is particularly effective in challenging gas flow conditions encountered in thermal hydraulics where particle-based methods such as particle image (or tracking) velocimetry do not perform well. The main principles for designing and operating this diagnostic are presented as well as a set of gases that have been identified as potential seeds. Two gases [H2O and nitrous oxide (N2O)] have been characterized extensively for thermodynamic conditions ranging from standard temperature and pressure to environments encountered in integral effects test (IET) facilities for high-temperature gas reactors. A flexible, modular, and transportable laser system has been designed and demonstrated with H2O and N2O seed gases. The laser system enables determining the optimum excitation wavelength, tracer concentration, and timing parameters. Velocity precision and thermodynamic domain of applicability are discussed for both tracers. The spectroscopic nature of the diagnostics enables one to perform first-principle uncertainty analysis, which makes it attractive for validating numerical models. Molecular tagging velocimetry is demonstrated for two flows. First, in blowdown tests with H2O seed, the unique laser system enables one of the largest dynamic ranges reported to date for velocimetry: 5000:1 (74 dB). N2O-MTV is then deployed in situ in an IET facility, i.e., the High-Temperature Test Facility at Oregon State University, during a depressurized conduction cooldown (DCC) event. Data enable researchers to gain insights into flow instabilities present during DCC. Thus, MTV shows a strong potential to gain a fundamental understanding of gas flows in nuclear thermal hydraulics and to provide validation data for numerical solvers.


I. INTRODUCTION
Gas flows are commonly encountered in nuclear thermal hydraulics and can be challenging to instrument, particularly for measuring velocity. Examples of gas flows include containment studies, 1 steam flow during reflood (liquid cooling after a loss-of-coolant accident) of pressurized water reactor (PWR) fuel bundles, 2 or hightemperature gas-cooled reactors 3 (HTGRs). Velocimetry methods that rely on Mie scattering of solid particles to seed the flow [such as particle image velocimetry (PIV), particle tracking velocimetry, or laser Doppler velocimetry] have been demonstrated successfully for some instances. 1 However, conditions such as accident scenarios in HTGR *E-mail: matandre@gwu.edu NUCLEAR TECHNOLOGY · VOLUME 205 · 262-271 · JANUARY-FEBRUARY 2019 © 2019 American Nuclear Society DOI: https://doi.org/10.1080/00295450.2018.1516954 can involve very high-speed flows (blowdown), very high temperature (pressurized conduction cooldown), or very slow transients [depressurized conduction cooldown (DCC)]. The inertia and buoyancy of particles will preclude them from accurately seeding the flow. Additionally, in PWR reflood, liquid water flashes to steam; the steam flow cools the fuel bundles in a process that needs to be better understood. Seeding the steam with particles could be challenging. One possibility would be to add particles to the liquid water and hope that sufficient particles are entrained in the steam following the phase change. Also, for some of these flows, the speed is low, and the particles may settle out. Nonparticle-based laser spectroscopy techniques such as molecular tagging velocimetry (MTV) could address some of these limitations.
Molecular tagging velocimetry is a time-of-flight velocimetry method that relies on locally creating and tracking molecular tracers. 4 This nonintrusive technique is applicable to a wide variety of gas and liquid flows ranging from stagnant to hypersonic, from cryogenic to flame temperature, and over a large range of pressures. The performance of MTV was assessed against that of PIV (Ref. 5) and showed uncertainties of the same order of magnitude as PIV.
The instrument most commonly uses two sets of lasers. A first ("write") laser creates tracers with a predetermined spatial pattern, and then a second ("read") laser illuminates a cross section of the flow within a controlled time interval dt and excites the fluorescence of tracers "written" in the gas. The location of the displaced tracers is recorded for each read pulse with a camera, ultimately leading to velocity fields. In gases, tracers are typically created through photodissociation of specific seed molecules and are tracked with planar laser-induced fluorescence. MTV has been demonstrated in gas with a variety of tracers. Examples of seed gas include nitrous oxide (N 2 O) (Ref. 6), NO 2

II.A. Tracer Selection
The selection of appropriate tracers is the first and most important step in designing an MTV diagnostic. For a particular test condition, the tracers should have adequate chemistry and lifetimes, seeding properties, and chemical compatibilities. For instance, some seed gases such as H 2 O are naturally present in many environments and would be appropriate for some tests in air. However, condensation could be detrimental and in high-temperature applications, H 2 O could lead to oxidation. 15 Furthermore, OH radicals are highly reactive and will quickly recombine with O 2 , H, or itself. 12 Nitric oxide (NO) is stable in inert environments such as helium or nitrogen, which is useful for increasing the probe delay time to improve velocity precision for low-speed flows.
Finally, the tracer concentration (if not naturally present in the flow) should be low enough so that it does not significantly affect the relevant thermophysical properties of the flow under study. For instance, the change in gas density should be small. Additionally, the change in heat capacity and speed of sound should be small for compressible and nonisothermal flows. Tracers typically used for MTV are at a concentration of a few percent or less, which minimizes the changes of thermophysical properties. These can be calculated, as done hereafter, to assess if the relative changes are acceptable for a given application.

II.A.1. Potential Tracers
The gases in Table I have been identified as potential  tracers for MTV tests in thermal hydraulics. A 193-nm beam is produced with an ArF excimer laser, while 355 nm is obtained from a frequency tripled Nd:YAG laser. The other wavelengths are obtained with tunable lasers, such as tunable dye lasers or optical parametric oscillators. This is covered in Sec. II.B. Note that other read laser wavelengths, corresponding to other vibrational transitions, can be used for these tracers. Those reported in Table I are the most efficient and widely used. Of the five seed gases listed in Table I

II.A.2. Precision Limits
Temperature, pressure, tracer concentration, probe time, and working gas affect the precision that can be MOLECULAR TAGGING VELOCIMETRY FOR GAS FLOWS · ANDRÉ et al. 263 obtained with MTV. For tracers that are chemically inert, molecular diffusion of the tracers limits the probe time by spreading the tag line. Very diffuse lines will degrade the pattern matching between the initial and displaced lines. The diffusion of the tracers increases with dt: where w 0 is the initial width of the tracers created by the write pulse and D is the binary diffusion coefficient of the tracer and working gas. Based on kinetic theory of gases, D scales as T 3=2 =P. Higher temperature will result in worse precision due to faster diffusion of the tracers, while high-pressure environments will reduce the effect of diffusion. In particular, tracers at the conditions of scaled HTGR, such as the HTTF (1100 K and 8 atm) will diffuse less than at standard temperature and pressure (STP). The working gas will also significantly affect the diffusivity; for example, the diffusivity of NO in He is nearly seven times larger than in N 2 (Ref. 18). Additionally, the signal-to-noise ratio (SNR) strongly affects the precision of the technique. SNR depends on the ability of the write beam to effectively photodissociate tracers. This is a function of the laser wavelength, fluence, and seed gas concentration and absorption cross section σ. The latter depends on pressure and temperature. For example, the photodissociation cross section of water vapor at 193 nm is nearly 40 times larger at 1000 K than at 300 K.

II.A.3. Laser Beam Absorption
Another factor affecting SNR is the absorption of the laser beams by gases in the test section and in the laboratory. This can be particularly problematic if the write pulse loses too much energy to photodissociate enough tracers. The number density of photodissociated molecules n d over a distance L is where h is Planck's constant and A is beam cross-sectional area. The absorbed beam energy E abs is given by the Beer-Lambert law: where E 0 is the initial beam energy and n i is the seed gas number density. The fluorescent signal can be improved by increasing the seed gas concentration; however, this will also result in less energy being transmitted to the test region. Thus, the optimal concentration of seed gas is a trade-off involving the beam path length in the test section L and is given by 1=ðσLÞ. For distances longer than 0.5 m at STP, the N 2 O concentration should be less than 1% (σ ¼ 9 Â 10 À20 cm 2 ). H 2 O is relatively unaffected by this constraint as its absorption cross section is two orders of magnitude lower than that of N 2 O at STP (σ ¼ 8 Â 10 À22 cm 2 ). For instance, beam absorption by 3% H 2 O would become an issue only for path lengths exceeding 10 m. Note that since σ depends on pressure and temperature, this constraint will change accordingly with the test conditions. In addition to the gases in the test section, gases outside of the test section can significantly absorb laser light. For example, at 193 nm, oxygen absorbs and the write beam is significantly attenuated in the laboratory. This can be remedied by enclosing the beam in a tube and purging the air from the tube with an inert gas like N 2 .
Finally, ultraviolet light is absorbed by common transparent glass materials used in viewports or prisms, which results in reduced energy being delivered to the test section. Optical components must be selected carefully.

II.A.4. Chemical Compatibility
A nonnegligible factor in selecting tracers is the chemical compatibility with the materials of the test section. For example, NO 2 is very aggressive with corroding most gasket materials, or acetone could irreparably damage plastic components. Additionally, some seed gases are toxic, and appropriate procedures must be deployed to protect the operators. Finally, some seed gases can be prohibitively expensive for large facilities.

II.B. MTV System
The instrument used here relies on several lasers and a camera synchronized together at 10 Hz. The seed gas is initially dispersed in the flow being probed. The first laser beam (the write pulse) is from an excimer laser (GAM Laser EX5) at 193 nm and creates the tracer molecules (OH or NO) nearly instantaneously ( %10 ns) along its path. Two laser pulses shaped in a 3-mm-thick sheet (read pulses) are then emitted from a tunable dye laser (Sirah Cobra-Stretch pumped by two Nd:YAG lasers) to illuminate the tracers and induce fluorescence. An intensified camera (QImaging QIClick charge-coupled-device camera coupled with a LaVision IRO intensifier) records the fluorescent signal and rejects the laser light (282 or 226 nm, depending on tracer selection) with a suitable long-pass filter, measuring on the location of the tracers at two different instants in time. Figure 1 shows the spatial arrangement of the lasers. The velocity is then obtained by measuring the displacement using cross-correlation techniques and knowing the delay between the two images. Since only a single tag line was used in this study, only a single component of velocity can be resolved. Finally, the system is transportable and can be deployed in laboratories of collaborators in a time-and cost-effective manner.

II.C. Test Sections
A stainless steel pressure vessel was built to investigate in a laboratory the performance of the diagnostics in conditions as encountered in a HTGR, namely, high pressure and high temperature. The test section, pictured in Fig. 2a and shown schematically in Fig. 3a, is fitted with two ultraviolet-transparent fused-silica viewports to allow the laser beams to enter and the fluorescence signal to be captured. A 9.0-mm-diameter jet discharges vertically in the section to generate a well-controlled flow of nitrogen or helium containing a small fraction of seed gas that can be probed with MTV. A computerized data acquisition system controls and monitors pressure and temperature in the chamber as well as jet flow rate and seed gas injection rate.
A second test section, pictured in Fig. 2b and shown schematically in Fig. 3b, was assembled to investigate large velocity ranges. It consists of a 5.7-L vessel that can be pressurized and then vented to the atmosphere to create a choked flow at the jet exit. The velocity can range from sonic speed to zero.

III.A. Tracer Spectra and Wavelength Selection
The fluorescence excitation spectrum of each tracer is first measured and analyzed to identify the optimum wavelength for the dye laser (read pulse). The MTV signal MOLECULAR TAGGING VELOCIMETRY FOR GAS FLOWS · ANDRÉ et al. 265 is recorded while the dye laser scans the wavelength and all other parameters (laser power, seed gas concentration, and probe time) are held constant. Spectra for OH and NO are presented in Fig. 4 at ambient temperature and pressure. The respective spectra computed with the LIFBASE software 19 are also plotted in Fig. 4 and show good agreement with the experimental data. The strongest signal (highest peak) is obtained at wavelengths of 281.905 and 226.186 nm for OH and NO, respectively. Tests were also performed at higher pressure (up to 3 atm) and temperature (up to 800 K). While pressure and temperature affect the peak height and width through collisional quenching and broadening, the best signal was still obtained at the same wavelength, which ensures the dye laser would not need to be tuned for changing pressure or temperature.

III.B. Effect of Temperature, Pressure, Probe Time, and Seed Gas Concentration
For the data described in this section, the fluorescence signal is measured by keeping the wavelength of the read pulse constant (where the signal is maximum) while the temperature, pressure, and seed gas concentration are varied. The tracers are probed at various times after being created by the write pulse. Detailed results for OH and NO tracers can be found in Refs. 5 and 18, respectively. The main findings are summarized here.

III.B.1. Hydroxide Tracers
The signal from OH tracers improves with increasing temperature (higher H 2 O photodissociation efficiency) and deteriorates with an increase in pressure due to collisional quenching of fluorescence. Hydroxide recombination also increases with pressure.
The seed gas concentration is based on equilibrium with vapor pressure of water at a given pressure and at ambient temperature. This corresponds to 2.7% and 0.9% molar at 1 and 3 atm, respectively.
Probe time measurements ranged from 5 μs to 6.4 ms. At short dt (<100 μs), the signal decreases with an increase in dt, approximately following a 1= ffiffiffiffi dt p trend. This decrease in signal is attributed to the recombination of OH with itself and other photoproducts. Chemical reaction simulations conducted with the Cantera code 20 confirmed the observed decay rate of OH molecules. Above 3 ms, the SNR becomes too low to precisely measure the tracer location. Therefore, this limits the precision that can be achieved with this tracer in these conditions, as discussed in Sec. III.C.

III.B.2. Nitric Oxide Tracers
The effects of pressure and temperature on the signal of NO tracers are qualitatively similar to those of OH, for similar reasons.
The seed gas concentration (N 2 O) is varied from 0.1% to 6% molar. The signal increases with concentration for low concentration (<2%). The SNR then plateaus or decreases between 2% and 4%. This effect is caused by the attenuation of the write beam on its path to the measurement region. For high seed gas concentration, the write beam will be strongly absorbed before it reaches that region.
The probe time of NO can be extended to tens of milliseconds due to its low reactivity in inert gases. The limiting factor to increasing dt is tracer diffusion, which

III.C. Velocity Measurement Precision
Molecular tagging velocimetry is a time-of-flight technique, and the velocity is obtained by measuring a displacement over a period of time according to the equation V ¼ MΔx=dt with M the image magnification and Δx the measured tracer displacement in pixels. Since the measurement resolution of the displacement is limited by the camera resolution (discretized in pixels), larger dt is required to probe slower flows for a given magnification (0.137 mm/pixel for the present experiments).
The uncertainty on V is a combination of the contributing factors and is calculated as follows: . With proper image calibration and accurate timing, the main contribution to the error is the measured displacement in pixels σ x . The precision of the velocity measurement can then be approximated by σ v % Mσ x =dt. Using image cross correlation followed by a curve fit of the correlation peak allows for subpixel accuracy in the displacement measurement. The exact value of σ x depends on the SNR (Ref. 16) and ranges from 0.1 to 1 pixel for a SNR of 6 to 1.3, respectively. 18 The main factors affecting the SNR are camera noise, optics aperture, signal brightness, and background illumination.
Comparison of the precision of the measurements obtained with OH and NO is presented in Fig. 5 at ambient temperature and pressure. At short dt, OH has better precision (smaller standard deviation); however, this is a consequence of a higher SNR resulting from a higher seed gas concentration (2.7% for H 2 O versus 0.5% for N 2 O) as well as higher read pulse energy (10 versus 4 mJ).
As dt is increased, the precision improves (i.e., the standard deviation decreases) for both tracers, following the theoretical 1=dt trend (dashed line). The precision with OH tracers then plateaus at about 0.1 m/s around dt ¼ 1 ms due to recombination of the tracers. The signal is then completely lost above 3 ms. Nitric oxide tracers are more stable and can be probed over longer dt, thus improving the precision. The best precision with NO tracers in the present measurements is 0.004 m/s at dt ¼ 40 ms. At longer dt, the gain in precision is only marginal due to the molecular diffusion of the tracers spreading the tag line.

III.D. Velocity Dynamic Range
Molecular tracers are well suited for probing low-speed flows due to the absence of settling. They also perform very well for high-speed and acceleration flows thanks to their perfect frequency response (no inertia lag). However, time-of-flight methods such as PIV or MTV are usually limited in terms of the velocity dynamic range by the measurement accuracy of the tracer displacement (usually 0.2 to 0.5 pixels for MTV, depending on the SNR). To maintain a reasonably good spatial resolution and resolve local (instead of spatially averaged) velocity, a maximum acceptable tracer displacement must be set, typically less than 100 pixels. This results in a velocity dynamic range typically on the order of 10 2 . There exist various methods to increase this range with PIV such as multgrid interrogation and multiple exposure recording. For MTV with a single tag line, only the latter is possible, but it requires several read laser pulses.
The majority of MTV studies to date have used only one read pulse. The initial location of the tracers is recorded in a separate step by firing the read pulse within a few nanoseconds of the write pulse. The assumption that the initial location of the tracers is constant is reasonable when doing measurements in tabletop experiments, where the environment is well controlled. Factors potentially affecting the position of the write beam are vibrations, beam pointing stability, beam steering through turbulent flow, and thermal expansion. Therefore, it is preferable to rely on two probe beams (read pulses) when doing field measurements where such issues may occur, as done in the present work.
In more controlled experiments where the initial tracer location is constant, the two probe pulses can be taken advantage of to increase the velocity dynamic range. For instance, a first pulse can be fired after a short delay, allowing measurement of high-speed flows, and then a MOLECULAR TAGGING VELOCIMETRY FOR GAS FLOWS · ANDRÉ et al. 267 second pulse is fired after a longer time, enabling the measurement of lower velocities. The dynamic range of this MTV scheme was investigated with OH tracers. Based on the above results, NO tracers would work as well. The tests were performed in the test section shown in Fig. 3b, pressurized with helium at 3.5 atm and with a small amount of liquid water to seed the gas flow with vapor in an equilibrium condition (0.77% at 3.5 bars). The effect of the seed gas on the fluid properties is calculated to be −1.3%, +0.5%, and +2.6% for the speed of sound, heat capacity, and density, respectively. The variation of the speed of sound is on the order of the measurement precision and thus is acceptable. At the beginning of the test, a valve is quickly opened to vent the chamber to the atmosphere. Such flow would be similar to a blowdown, as is experienced during the depressurization phase of a DCC. Velocity is measured with MTV at the valve exit. The first read pulse is 3 μs after the write pulse and is used for resolving high-speed flow [Oð100Þ m/s]. For lowspeed flow [Oð1Þ m/s], the displacement after 3 μs is very small, and the relative precision is poor. The lower-speed phase of the flow is more precisely resolved with a longer dt of 253 μs. The initial tracer location is obtained from measurements before the opening of the valve, when the flow is quiescent and the line is not convected.
The time history of the jet centerline velocity is plotted in Fig. 6 using a logarithmic scale for the velocity axis. Before the valve opens at t ¼ 1:5 s, the quiescent flow is measured with the second dt ¼ 253 μs pulse. At the valve opening, the flow is initially choked (at Mach 1, with the speed of sound c ¼ 1005 m/s for helium at 293 K and c ¼ 992 m/s for the helium-vapor mixture) and is correctly captured by the first dt ¼ 3 μs pulse. Data for the second dt ¼ 253 μs pulse are not shown because the signal is outside of the camera field of view (FOV) (and would probably be too smeared by the turbulent flow to be resolved). When the flow is slow enough (less than 20 m/s), the second pulse becomes visible and is then used to measure the velocity. The image of the first pulse is still visible (○ symbols in Fig. 6), but the very small displacement makes the measured velocity very imprecise (on the order of AE10 m/s). The precision with the second pulse measurement is about 0.2 m/s. Overall, the flow is resolved from 1000 m/s down to 0.2 m/s, which corresponds to a dynamic range of 5000:1 versus about 100:1 using a single pulse. To our knowledge, this is the highest reported dynamic range for MTV. Theoretically, the dynamic range could be increased to 100 2 ; however, it was chosen to have some overlap in the range 10 to 20 m/s to ensure good continuity of the velocity time history.
Uncertainties due to calibration and timing are small compared to the displacement precision; thus, the aforementioned measurement precision provides an estimation of the overall uncertainty, i.e., 10 m/s for dt ¼ 3 μs and 0.2 m/s for dt ¼ 253 μs.
It was attempted to compare these results to one-dimensional isentropic flow theory. 21 The model captured the first few tenths of a second (five data points) of the jet but then diverged from the observed flow; the most likely reason for the discrepancy is incorrect adiabatic and reversible flow assumptions.

IV.A.1. H 2 O
The lifetime of OH formed from H 2 O is about 1 ms in inert environments. The signal is the strongest at short dt (<500 μs), which limits the applicability of OH in terms of minimum resolvable flow velocity (for a given magnification).
The typical molar ratio of H 2 O seed gas in the literature is 1% (Ref. 22) and 2.8% (Ref. 17). In our experiments, it was 2.7% at 1 atm and 0.9% at 3 atm. For reference, 3% seeding at ambient conditions corresponds to 22 g/m 3 . Injection of seed gas can be done by adding a measured mass or volume of liquid water. Some facilities also have water vapor (or humidity) naturally occurring and thus may not require additional seeding.
As discussed in Sec. III.B.2 related to the write beam absorption, H 2 O-MTV (or HTV) can be deployed in

IV.A.2. N 2 O
Nitric oxide formed from N 2 O has a relatively long lifetime. A measurable signal can be obtained at dt up to tens of milliseconds, which makes N 2 O-MTV an ideal technique for probing low-speed flows. N 2 O has been used at a molar ratio of 4% in the literature. 6 Our experiments showed that 0.5% is sufficient to get an adequate signal. For 0.5% N 2 O at ambient P and T, this corresponds to 9 g/m 3 . N 2 O gas seeding can be controlled with a mass flowmeter or by monitoring the increase in pressure during injection in a closed vessel. For instance, a seeding of 0.5% can be achieved by injecting seed gas until the pressure increases by 0.5%. The vessel can then be mixed and vented to the required test pressure.
The high absorption of the write beam by N 2 O limits the distance the beam can travel inside the facility before reaching the test section. For 0.5% N 2 O, a maximum distance (resulting in 90% beam attenuation) is about 2 m. This limitation could still be circumvented by adding a purged beam tube inside the facility. Distances can also be increased with higher-power lasers resulting in more energy remaining for the measurement.

IV.A.3. Summary
As discussed in Sec. III.D, both tracers are suitable for high-speed flows since they perform well at short dt. Furthermore, over a short distance, the seed gas concentration can be increased to improve the signal strength. The photodissociation process requires a similar amount of energy from the write beam for a given test condition. Finally, both seed gases are safe for humans to handle and are readily available.  23 at a temperature of 550°C and partial pressure of 5 to 20 kPa (5% to 20% molar). By extrapolating to 500°C and 1% N 2 O, a rate on the order of 10 −2 mg/g·h is estimated. This is higher than for H 2 O, but again, the rate remains small and may be tolerable.

IV.B. Compatibility with Facility Components
Nitric oxide has a higher reaction rate than N 2 O at similar conditions, but the NO partial pressure is only a small fraction of that of N 2 O in the experiment. For 0.1% NO (conservative estimate), the oxidation rate is also on the order of 10 −2 mg/g·h.

V. DEMONSTRATION OF N 2 O-MTV IN AN IET FACILITY
The present diagnostic was designed to be deployed in an IET of an HTGR: the HTTF at OSU to perform velocity measurements during a DCC. The implementation of this technique to the HTTF is reported in Ref. 18. Sample results are presented here to illustrate the capabilities of MTV for gas reactor investigations. Detailed analysis of the results obtained in the facility will be presented in a subsequent paper.
The HTTF replicates a DCC by opening both hot and cold legs of the primary loop into a large reactor cavity simulation tank, which is a surrogate for the ambient gas surrounding the reactor. Velocity profiles are measured at the exit of the hot leg (a 300-mm-diameter pipe) at a frequency of 10 Hz for 30 min. In a DCC scenario, simulations predict helium to flow out of the reactor and air (here nitrogen is used as a surrogate) to ingress into the reactor, in a buoyancydriven flow called a lock-exchange. For this experimental campaign, N 2 O (seeding at 0.5%) was chosen because of (a) lower risk of short circuit to the electric heaters, (b) better measurement precision for low-speed flows, and (c) easy control of the seeding with pressure readings. The effect of MOLECULAR TAGGING VELOCIMETRY FOR GAS FLOWS · ANDRÉ et al. 269 the seed gas on the fluid properties is calculated to be −2.4%, +0.4%, and +4.8% for the speed of sound, heat capacity, and density, respectively. While the density change is not negligible, it is not likely to have a significant effect on the measured flow, which is driven by density variation on the order of 700% between helium and nitrogen. Figure 7 shows an example of velocity time evolution during an isothermal DCC. These results are the first detailed velocity data obtained in an IET with a laser spectroscopic technique and give insights into the complex flows associated with these events. The high measurement precision (down to 0.006 m/s) was crucial in these experiments because of the low velocities experienced in the HTTF (<2 m/s). The laboratory investigation of the effect of temperature, pressure, and seed gas concentration and close integration between the teams enabled us to select the best tracer and laser parameters for this study and made this challenging in situ measurement campaign a success. The present data will help develop and validate models for describing such flows. Such models will be presented in a future publication.

VI. CONCLUSION
The performance and applicability of MTV to gas flows present in thermal hydraulics have been investigated. While the study focused primarily on HTGR, the demonstrated application domain overlaps with conditions expected for containment studies and PWR reflood tests. Two gases (H 2 O and N 2 O) were extensively characterized as seed gases to probe flows over a large range of pressures, temperatures, and velocities. Only small amounts (<3%) of those innocuous seed gases are necessary to be adequately photodissociated in tracers that provide sufficient SNR for the tests conducted here. HTV has been determined to operate satisfactorily at up to 100% steam concentration.
The applicability of both gases to gas reactors was discussed in terms of ease of implementation, limitations, and risk to the facility. HTV was found to be preferable for moderate-to high-speed flows, in very large facilities, or in facilities where water vapor is already present. Its precision also increases with temperature. N 2 O-MTV was found to be preferable for measuring low-speed flows and in facilities where condensation of water vapor could be problematic.
High-precision measurements (down to a standard deviation of 0.004 m/s, the lowest value ever reported for MTV in gas) and large velocity dynamic range (up to 5000:1, the largest known to us for laser velocimetry) were obtained in the laboratory. In situ measurements were successfully conducted in an IET demonstrating the applicability and benefits of the MTV technique to HTGRs and thermal hydraulics in general.