Modification and Tests of Particle Probe Tips to Mitigate Effects of Ice Shattering (original) (raw)

1. Introduction

Understanding the formation and evolution of small ice particles in clouds has been a long-standing problem in cloud physics. Debates around this problem span back well over three decades and began once optical particle size spectrometers became accepted instruments for airborne cloud particle sampling. Early airborne measurements have shown that in glaciated clouds, the number concentration of ice particles is dominated by small particles with sizes less than 100 _μ_m in diameter. Small ice particles may play a significant role in radiation transfer and precipitation formation, and their associated parameterizations have been included in many numerical climate and weather prediction models.

One possible explanation for high concentrations of small ice particles is instrument-induced particle shattering. Prior to entering the instrument sample volume, an ice particle may impact the probe’s upstream tips or inlet and shatter into smaller fragments, which may cause multiple artificial counts of small particles of ice (Fig. 1a). The hypothesis of shattering assumes that after their impact with the surface of the probe’s arms or inlets, the ice particles travel several centimeters across the airflow in order to reach the sample volume. Even though the significance of the effect of shattering on measurements was pointed out as early as in the late 1970s (e.g., Cooper 1977), the shattering hypothesis was not commonly accepted in the cloud physics community for many years. It seemed counterintuitive to accept a possibility of particles traveling several centimeters across the airflow at an aircraft speed, ~100 m s−1. Many researchers argued that after bouncing, ice particles cannot travel that far away from the point of impact and proposed that shattered particle debris is instead shed over the surface of the arms and inlets, resulting in only a minimal effect on measurements. Despite numerous studies investigating the intercomparisons of measurements of different particle probes, which provide strong evidence in support of the effect of shattering (e.g., Gardiner and Hallett 1985; Gayet et al. 1996; Field et al. 2003, 2006; Korolev and Isaac 2005; Heymsfield 2007; McFarquhar et al. 2007; Vidaurre and Hallett 2009; Fugal and Shaw 2009; Jensen et al. 2009), the hypothesis of shattering has continued to be a topic of debate in the community, until recently.

Fig. 1.

Fig. 1.

Fig. 1.

(a) A conceptual diagram of the effect of ice particle shattering and bouncing on the measurements of ice particles with conventional probe tips. (b) The redesigned arm tips deflect ice particles away of the sample volume, resulting in the mitigation of the shattering.

Citation: Journal of Atmospheric and Oceanic Technology 30, 4; 10.1175/JTECH-D-12-00142.1

Direct experimental support for the shattering hypothesis was obtained in 2003 with a high-speed video collected during dedicated tests by the National Aeronautics and Space Administration’s (NASA) Glenn Research Center (GRC) at the Cox and Company wind tunnel facility. Several of these results will be presented in more detail in section 4. The video technique provides only a qualitative characterization of the possible flow of shattered fragments from a larger particle breakup. However, for the first time, it has been documented that at aircraft speed, ice particles bouncing off forward surfaces can travel several centimeters across the airflow before passing through the sample volume.

The results obtained in the Cox and Company wind tunnel initiated the process of reassessing the performance of the inlets and arm tips of existing particle probes and provided the impetus for researchers to begin considering new conceptions of inlet design that would improve the microphysical characterization of the ice cloud environment.

This work presents an overview of the efforts undertaken by Environment Canada in collaboration with NASA on the design of particle probe arm tips that are intended to mitigate the effect of ice shattering. This work was focused on the modifications of three probes: Particle Measuring Systems Inc. (PMS) Forward Scattering Spectrometer Probe (FSSP) and optical array probe (OAP-2DC), and Droplet Measurement Technologies (DMT) Cloud Imaging Probe (CIP). It is important to note that the results obtained here have a much broader impact beyond the parameters of these particular instruments and can be applied to most airborne particle probes. The main concepts and sets of requirements, which the arm tips should satisfy, are described in section 2. Section 3 presents the results of a particle bouncing simulation along with airflow modeling around the housing of probes with their originally designed (standard) tips as well as the modified tips. The outcomes of the wind tunnel tests of the modified and originally designed particle probe inlets are discussed in section 4. The conclusions and design recommendations for the probe tips are presented in section 5.

2. Principles of design for the arm tips

There are several basic requirements in the probes’ inlet design to mitigate the effect of potential artifacts related to particle sampling.

The first requirement consists of minimizing the inlet surface area, which may deflect shattered ice particles toward the sample area. Satisfying this condition will reduce the amount of artifacts related to shattering and bouncing.

The second requirement of the inlet design is related to minimizing the disturbance of the airflow on approach to the sample volume and obtaining the most uniform airflow across the sample volume. The airflow on approach to the sample volume experiences disturbances caused by the probes’ housing and inlet. The air disturbances may cause a number of unwanted effects, such as the spatial sorting of cloud particles before they pass through the sample volume (King 1984), the changing of the ice particle’s natural orientation (King 1986; Krämer et al. 2013), the elongation and breakdown of large drops (Pilch and Erdman 1987; Wierzba 1990), as well as the fragmentation of aggregates of ice particles (Korolev and Isaac 2005).

The third requirement involves minimizing water shedding toward the optical windows and protecting the optics from contamination by cloud water. Water shedding over optical windows may cause biases in particle sizing and concentration in scattering probes. For the case of imaging probes, the shedding water may be a source of various image artifacts. Accumulation of water on the optical windows may also cause data loss.

The fourth requirement consists of providing a proper arrangement of deicing heating for the inlets and probes housing. Failure to provide proper deicing may cause an accumulation of ice on the front surfaces of the probe and result in significant disturbances of airflow in the sample volume. In many cases, ice buildup on the inlets may cause partial or complete blockage of the sample volume and may result in data loss.

The design work was divided into three stages. The first stage of the design work consisted of a sequence of iterations of modifications of mechanical design followed by the airflow and particle bouncing simulations (section 3). During the second stage, the tips were tested at the Cox and Company wind tunnel facility in ice particle and liquid conditions. These tests were focused on the following tasks: (i) identifying the frequency of ice particles bouncing toward the probe’s sample area, (ii) observing whether water would shed into the optical windows, and (iii) testing the quality of the deicing heating system (section 4). During the third stage, the probes with the modified tips were installed on the National Research Council (NRC) Convair-580 aircraft side by side with the probes with the originally designed inlets and tested during in situ measurements. The preliminary results of these tests were described in Korolev et al. (2011).

The conceptual design of the arm tips is shown in Fig. 1b. The tips have sharp leading edges having a wedge- or spear-type shape. The tips’ surfaces facing the sample volume are parallel to the airflow. The opposite surface of the tips is tilted with respect to the airflow, and it deflects bouncing particles away from the airflow (Korolev 2009). Such tips minimize the probability of particles bouncing toward the sample volume. Ice particles may still shatter into the sample volume if they impact the sharp edge of the tips. Despite the fact that such events were documented on high-speed video during the wind tunnel tests (section 4), the probability of these occurrences is actually lower than the probability of contamination of the sample volume with ice particles bounced from the spherical (e.g., OAP-2DC) or saucer [e.g., CIP, two-dimensional stereo (2D-S) probe] type of tips.

The following modifications were applied to the OAP-2DC, FSSP, and CIP in order to improve particle sampling.

a. OAP-2DC

The experimental antishattering OAP-2DC tips had both circular and flat surfaces facing the sample volume. To adapt the new tips, the OAP-2DC arms were modified and remachined. To keep the arm surface in front of the sample volume flat, the antisplashing guards were removed as a source of additional shattering. Some examples of the antishattering tips designed for the OAP-2DC are shown in Fig. 2. The tip facets opposite the sample volume had three types of surfaces: flat (Figs. 2a,c) and carved in two different directions (Figs. 2b,d). The arm and tip surfaces facing the sample volume were designed to be fully flat (Fig. 2a), partially flat (Fig. 2c), or circular (Figs. 2b,d).

b. FSSP

The tubular inlets have a higher probability of shattering than the open concept inlets, for example, arm-type inlets. Therefore, the original FSSP inlet tube was uninstalled, and the spherical tips were replaced by the modified ones, similar to those shown in Fig. 2.

c. CIP

Besides the mitigation of the effect of shattering, the modification of the CIP arms is also intended to reduce the sample volume length in order to decrease the amount of out-of-focus donut-looking images. The out-of-focus images result in particle oversizing and information about particle shape is lost (e.g., Korolev et al. 1998). The modified CIP tips protrude from the arms toward each other and block some fraction of the original sample volume, which shortens the sample volume length.

During the course of the modification of the probe’s arms and tips, the position of the optical elements was preserved in the same locations and did not affect the optical magnification and angular apertures of the receiving optics. In total, 29 modified tips were tested in the frame of this work for the three probes: OAP-2DC (18), FSSP (4), and CIP (7).

3. Flow and bouncing simulation

Computational fluid dynamics (CFD) simulations have been intensively used in the mechanical design and modification of probe arms and tips as a form of feedback regarding the quality of the airflow in the vicinity of the sample area of the probes. One of the main objectives was to design a set of probe tips with maximum uniformity and minimal disturbance to the airflow prior to entering the sample volume. The airflow disturbances related to convergence or divergence may result in the spatial sorting of particles with different aerodynamic sizes and therefore cause errors in measurements of particle concentration in different size bins. Acceleration and deceleration of the airflow in front of the sample area may cause changes to the natural orientation of ice particles or a deformation of raindrops, which may be misleading in the process of particle habit recognition that is done from the 2D imagery measured by the probes.

The probe-tip-shape optimization was a highly iterative process. After generating an initial 3D model of the tips based on conceptual design, the CFD simulation around the model was performed. The results of the modeling were analyzed and modifications to the 3D model were applied, if found to be necessary. Then, the CFD simulation around the modified model was performed again followed by one more repeated cycle of “design–simulation.” The CFD simulations were conducted at different angles of attack, yaw, airspeeds, temperatures, and pressures. The CFD analysis was performed with the help of the flow simulation tool in SolidWorks software (Solidworks 2012). The adaptive grid changed progressively to finer cells toward the smaller features and fluid interfaces with the smallest cells approaching 0.2 mm. Most cases had over 80 000 fluid cells plus 50 000 partial cells at the solid/fluid boundaries. The air was considered to be “dry” and having zero turbulence as an initial condition. The probes’ surfaces are considered to be perfectly smooth, and both heat transfer and gravity were ignored. The transition of laminar to turbulent flow was taken into consideration.

Below, we show some results of the CFD and particle bouncing simulations for the particle probes with standard and modified arms.

a. Flow simulation

Figures 3a–c show the CFD simulations of the airflow at an airspeed of 100 m s−1, a pressure of 700 mb, and a temperature of 0°C around the front part of the FSSP, OAP-2DC, and CIP with standard configurations of the arms and tips, as they were built by manufacturers. The diagrams in the central column depict the following changes: airspeed along the sample volume (Figs. 3d–f), and along the line parallel to the airflow and passing through the center of the sample area (Figs. 3g–i). Particles passing through the FSSP inlet tube experience deceleration and then acceleration before crossing the sample area (Figs. 3g). The flow inside the tube has asymmetry caused by the knob mounted on one of the optical apertures (Figs. 3g). The airspeed gradually decreases during its approach to the sample area of both the OAP-2DC and the CIP (Figs. 3h,i). However, the airspeed along the sample volume of these probes increases from the center toward the arms. The ice particles passing close to the arms may experience aerodynamic stresses caused by the wind shear. This may cause aggregated ice particles with weak bonds such as aggregates of dendrites or needles to break apart.

Fig. 3.

Fig. 3.

Fig. 3.

Simulations of the airflow for _V_∞ = 100 m s−1, P = 700 mb, and T = 0°C around the particle probes (a) FSSP, (b) OAP-2DC, and (c) CIP having standard housing. The FSSP inlet tube in (a) is cut in half. The sample areas of the probes are indicated by a star in (a) and by a dashed line in (b) and (c). Normalized air velocity V/_V_∞ for _V_∞ = 100, 150, and 200 m s−1 along the sample volume of (d) FSSP, (e) OAP-2DC, and (f) CIP. Normalized air velocity V/_V_∞ upstream and downstream from the sample area along the line passing through the center of the sample volume of (g) FSSP, (h) OAP-2DC, and (i) CIP. Red dots indicate the location of the sample volume. For the OAP-2DC and the CIP, the sample volume in (e) and (f) is presented by the entire length of the beam between the arms.

Citation: Journal of Atmospheric and Oceanic Technology 30, 4; 10.1175/JTECH-D-12-00142.1

The velocity changes in the diagrams in Figs. 3d–i are presented in terms of normalized velocity _V/V_∞, where _V_∞ is undisturbed velocity at a large distance from the probe and V is the local air velocity. The calculations were performed for _V_∞ = 100, 150, and 200 m s−1. Presenting the flow calculations in terms of the ratio _V/V_∞ allows for the estimation of the effect of the airspeed on the flow. As seen from Figs. 3d–i, the changes of the ratio _V/V_∞ remain nearly the same for the aircraft speed range. This is an important outcome that facilitates the flow analysis.

Figure 4 shows a similar set of diagrams for the same conditions and the same type of instruments as in Fig. 3 but with modified tips. The modified tips improved the uniformity of the airflow along the sample volume (Figs. 4d–f). Unlike the standard tips, in the modified version the airflow becomes more uniform upstream from the sample volume for the FSSP and OAP-2DC, and it varies less than 5%. The airflow upstream from the CIP experiences deceleration and then acceleration within ±5%. The acceleration of the airspeed between the CIP arms is caused by the depression behind the tips, resulting in the Venturi effect.

One of the tasks related to the modification of the tips was the protection of the optical windows from water shedding over the surface. Water might shed from the upstream lip onto an optical aperture, contaminate the optics, and eventually degrade the measurements. As a result, a groove with angled edges around the open aperture was considered as one of the potential solutions to this problem. The shed water was expected to be trapped inside the groove and dumped out to the sides.

Figure 5 shows the flow trajectories over the surface of the tip upstream from the aperture and groove. The results of the CFD simulation suggest that the leading-edge vortex sweeps the majority of the flow down and around the tip. The remainder moves along the top surface of the tip and enters into the groove. Then it is diverted by the groove and exits out through the groove’s side corners. The wind tunnel tests described in section 4c confirmed the success of this design in protecting the optics from water contamination.

Fig. 5.

Fig. 5.

Fig. 5.

An illustration of the flow trajectories along the modified tip surface of OAP-2DC (as in Figs. 2b, 12a), calculated from the CFD analysis. The flow analysis was performed for _V_∞ = 150 m s−1, P = 800 mb, and T = 0°C.

Citation: Journal of Atmospheric and Oceanic Technology 30, 4; 10.1175/JTECH-D-12-00142.1

It is anticipated that the flow pattern around the tip and groove remains largely unchanged over a fairly wide speed range. However, local compressibility effects, heat conduction, and icing susceptibility may all contribute to a change in the local flow around the optics. More study is required for optimizing this issue, also with a performance test across a wide variety of flight conditions.

b. Bouncing simulation

A CFD simulation of bounced ice particles was performed to estimate the trajectory of deflected particles and to ascertain which regions of the sample area could potentially be contaminated by deflected ice particles. In the simulation, the ice particles were assumed to be solid spheres having the density of ice that bounce from the probe’s surface without shattering. The flow simulation of two-phase flow (fluid + particle) calculated the motion of particles in a steady flow (Solid Works 2012). The interaction of particles with the surfaces was taken into account by specifying the particle velocity restitution (reflection) coefficients, and , where _V_2,n and _V_2,τ are the velocities’ surface normal and tangential components after reflection, _V_1,n and _V_1,τ, the ones before the impingement. The restitution coefficient was estimated from the trajectories of ice particles deflected from the solid surface, which were obtained from the analysis of high-speed videos recorded in the Cox and Company wind tunnel facility.

It was estimated that the normal coefficient of restitution (ɛ_n_) of ice particles generated in the wind tunnel varies from 0.55 to 0.85. Variation in this value is probably due to variations in individual particle shape, off-axis camera viewing angles, and is further compounded by an unknown tangential coefficient of restitution (ɛ_t_) between the probe tip and the geometry of wind tunnel ice particles.

It should be noted that the restitution coefficient of ice particles grown through the water vapor diffusion process may be quite different from those generated in the wind tunnel by means of shaving solid ice blocks. The diffusion-grown ice particles usually have lower bulk density due to their developed branches, and they are more fragile than those produced in the wind tunnel. Upon impact, such particles most likely shatter first and then bounce away. This leads to a redistribution of the kinetic energy of the original particle into energy required for the particle fragmentation, heating, partial melting, and kinetic energy of the bounced fragments. Therefore, the restitution coefficient of the individual fragments of shattered particles is expected to be lower than that for the case of bouncing without shattering. The analysis of the high-speed video obtained in the wind tunnel tests showed that the interaction of the tunnel-generated ice with a solid surface in many cases occurred without shattering (section 4a), although shattering events were also observed.

Figure 6 shows the trajectories of ice spheres with diameters 7, 20, and 40 _μ_m bouncing from the FSSP inlet tube calculated for two different conditions: P = 800 mb, T = −5°C, V = 100 m s−1 (Figs. 6a–c) and P = 200 mb, T = −40°C, V = 200 m s−1 (Figs. 6d–f). The normal restitution coefficient during the bouncing simulation was assumed to be 0.85. The angle of attack (AoA) for this simulation was 0°. Some fraction of the front surface of the FSSP inlet tube is slanted toward the center, and therefore it deflects the particles inside the tube. As seen from Figs. 6a and 6d, after bouncing, ice particles with D = 7 _μ_m do not reach the sample volume located on the axis in the rear section of the inlet tube. However, larger ice particles (D = 20 _μ_m and D = 40 _μ_m) after bouncing have higher kinetic energy than the smaller ones. Therefore, large particles can travel farther across the flow and their trajectories cross the sample volume and contaminate the entire inlet tube (Figs. 6b,c,e,f). It is interesting to note the effect of focusing of the rebound particles in Figs. 6c and 6e. In the diagram in Fig. 6f, the secondary focusing of ice particles occurs after bouncing from the walls of the inlet tube. Comparisons of the trajectories of Figs. 6a–c and Figs. 6d–f demonstrate a well-pronounced effect of pressure on the particle trajectories: at lower pressure the rebounding particles experience lower drag force and they can travel farther distances across the airflow. It should be emphasized that Fig. 6 demonstrates the effect of bouncing with no shattering. Accounting for particle shattering will significantly increase the contamination of the FSSP measurements by the artifact counts.

Fig. 6.

Fig. 6.

Fig. 6.

Simulation of bouncing of ice spheres with diameters 7, 20, and 40 _μ_m from the FSSP inlet tube at 0° AoA. The simulation was performed for (a)–(c) P = 800 mb, T = −5°C, _V_∞ = 100 m s−1 and (d)–(f) P = 200 mb, T = −40°C, V_∞ = 200 m s−1. For both cases AoA = 0°, ɛ_n = 0.85. The diagrams show two views of the trajectories inside of the inlet tube: (left) across the flow and (right) along the flow.

Citation: Journal of Atmospheric and Oceanic Technology 30, 4; 10.1175/JTECH-D-12-00142.1

The bouncing simulation presented in Fig. 6 suggests that the FSSP measurements in ice clouds may be severely contaminated by particles rebounding from the front edge of the inlet tube. Sharpening the edge of the inlet tube may help to mitigate the effect of shattering. However, bouncing from the inner walls of the inlet tube at nonzero angles of attack will still contaminate the measurements. As demonstrated in Korolev et al. (2011) during in situ measurements, removing the FSSP inlet tube and installing modified tips have effectively reduced the amount of artifacts related to shattering and significantly improved the performance of the probe as compared to the FSSP with the standard inlet tube.

To estimate the characteristic travel distances of ice particles that rebound from the OAP-2DC tips, the simulation of particles bouncing from a hemispherical head has been performed. The dimensions of the hemispherical head are the same as that for the OAP-2DC arm (see Fig. 11), although the antisplashing guards installed on the OAP-2DC arms were not included in the simulation. The antisplashing guards were assumed to be far enough downstream from the initial impact and therefore to be of little importance. The flow blockage due to their presence should have negligible effect on the net range of resulting trajectories. The simulation assumed that particles rebound without shattering, and that the particles are spherical and have the density of ice. The calculations were performed for the restitution coefficient of 1.0, representing a perfect reflection and thus giving an estimate of the maximum limit of the travel distance.

Figure 7 shows the trajectories of 20-_μ_m spherical ice particles after bouncing off the surface of the hemispherical head. The simulation of trajectories along x direction was discontinued at the location of the OAP-2DC sample volume at x = 4.5 cm downstream from the tip of the head. As seen from Fig. 7, after bouncing, the ice particles travel across the airflow, forming an envelope of trajectories with no rebound particles in the vicinity of the arm surface. Particles traveling close to the edge of the tip are deflected by the airflow around the tip edge, forming a shadow zone with no particles of this specific size in this zone.

Fig. 7.

Fig. 7.

Fig. 7.

Simulation of the trajectories of ice spherical particles with D = 20 _μ_m after their impact with the hemispherical head. The diameter of the hemispherical head is 2.5 cm. The location of the OAP-2DC sample volume would be on the right side of the diagrams, where the particle trajectories are terminated. The calculations were performed for the restitution coefficient of 1.0, _V_∞ = 100 m s−1, and P = 1000 mb.

Citation: Journal of Atmospheric and Oceanic Technology 30, 4; 10.1175/JTECH-D-12-00142.1

Figure 8 shows the trajectories of spherical ice particles with different diameters after impact with the spherical tip head at two different airspeeds, 75 and 200 m s−1, and two air pressures, P = 1000 mb and P = 200 mb, respectively. Figure 8 indicates that the travel distance of the particles across the airflow increases with an increase in the particle size (mass) and a decrease in the air pressure P. As seen, that even at high pressures (P = 1000 mb, Fig. 8a), rebounded ice particles with a D ~ 50 _μ_m collectively covered the entire distance between the OAP-2DC arms, which is equal to 7.6 mm. Since the travel distance of the rebound particles increases with the increase of their size, it can be concluded that shattered and rebounded particles with bigger sizes will contaminate the entire area between the OAP-2DC arms with spherical tips.

Fig. 8.

Fig. 8.

Fig. 8.

As in Fig. 7, but the simulations were performed for ice particles with different diameters D =10, 20, and 50 _μ_m and two different conditions: (a) _V_∞ = 75 m s−1, P = 1000 mb and (b) _V_∞ = 200 m s−1, P = 200 mb.

Citation: Journal of Atmospheric and Oceanic Technology 30, 4; 10.1175/JTECH-D-12-00142.1

Figure 9 shows the maximum travel distance across the airflow _L_max at x = 4.5 cm versus the pressure, calculated for different velocities V and different particle sizes. As seen from Fig. 9 the curves _L_max(P) calculated from different airspeeds are grouped together, showing a weak dependence of _L_max versus V. This result is suggestive of the fact that _L_max primarily depends on particle size and air pressure, and that velocity has a minor effect on _L_max.

Fig. 9.

Fig. 9.

Fig. 9.

Simulations of the maximum travel distance across the airflow at the location x = 4.5 cm after particle bouncing from the hemispherical head with a diameter of 2.5 cm. The distance x = 4.5 cm corresponds to the location of the OAP-2DC sample volume.

Citation: Journal of Atmospheric and Oceanic Technology 30, 4; 10.1175/JTECH-D-12-00142.1

The calculation of the trajectories of the rebound particles was performed for the restitution coefficient of 1.0, and therefore it gives the upper estimate of the travel distance _L_max. However, even assuming that the restitution coefficient is twice as low, the rebounded particles will still entirely contaminate the OAP-2DC sample volume.

The process of ice particle fragmentation as a result of impact with a solid surface is quite a complex and not well-understood process. Vidaurre and Hallett (2009) showed that the number of fragments is growing with an increase of particle size, and thus an ice crystal of approximately 300 _μ_m in size may shatter into ~103 fragments at 130 m s−1. Other than the particle size, the number of shattered fragments depends on several other parameters. At the moment, it is recognized that it depends on airspeed, air pressure and air temperature, the shape and temperature of the tips, ice particle habit, and particle orientation at the moment of impact. None of these dependences has been studied experimentally. The direction and speed of the fragments after shattering are two other unknown variables, which cannot be specified in numerical simulations. Therefore, we consider the modeling of bouncing of shattered ice particles as a premature effort and so we have limited our analysis to the simulation of bouncing without shattering. As mentioned above, the simulation of bouncing without shattering provides an upper estimate of the travel distance of the rebound ice particles.

4. Wind tunnel tests

During the course of the modification of the probes’ arms and tips, a series of wind tunnel tests were performed. The main objectives of these tests were as follows:

  1. to evaluate the general performance of the probes after their modification,
  2. to characterize the effectiveness of the modified tips in mitigating shattering and to compare them with the originally designed ones,
  3. to characterize the effect of shedding water on optics contamination, and
  4. to determine the efficiency of the deicing heaters and their capability to keep the arms’ surface in front of the sample area free from ice buildup.

The present section will be focused mainly on objectives (ii) and (iii).

The tests were conducted in the Cox and Company wind tunnel facility. The characterization of the performance of the probe tips was conducted in both ice and liquid sprays. Most of the tests were conducted at an airspeed of 80–85 m s−1. The performance of the modified probes was tested at three angles of attack: 0°, 5°, and 10°.

a. High-speed video analysis

The characterization of the effectiveness of the tips in mitigating ice particle bouncing toward the probe’s sample volume was conducted in ice sprays. The trajectories of the bouncing ice particles were recorded on high-speed digital video Vision Research Phantom V7.0 and V9.0 cameras.

Figure 10 shows high-speed video frames with the trajectories of ice particles bouncing from the originally designed arm tips of the OAP-2DC and the CIP and the FSSP inlet tube. The images of individual particles appear as tracks, whose length is determined by the camera’s shutter speed and the particle velocity. The high-speed videos associated with these snapshots can be downloaded (from ftp://64.176.164.83/pub/bouncing\_video).

Fig. 10.

Fig. 10.

Fig. 10.

High-speed video snapshots of the trajectories of ice particles bouncing from the standard (a) OAP-2DC tip, (b) FSSP inlet tube, and (c) CIP tip. The video recording was conducted in ice sprays in the Cox wind tunnel at V = 80 m s−1, P = 1000 mb, and T = −10°C. Grayish strips in (a) and (c) highlight the sample volumes of OAP-2DC and CIP, respectively.

Citation: Journal of Atmospheric and Oceanic Technology 30, 4; 10.1175/JTECH-D-12-00142.1

The analysis of the high-speed videos revealed that after bouncing from the original OAP-2DC arm tips, the ice particles may travel several centimeters across the airflow, which results in the contamination of the entire sample area by shattered and bounced ice particles. This observation is in qualitative agreement with the prediction obtained from the bouncing simulation in section 3b.

Figure 10b shows a high-speed video snapshot of the trajectories of ice particles bounced from the FSSP inlet tube. The illumination used for the high-speed photography does not allow the recording equipment to capture footage of particles entering the sample tube. However, the qualitative similarity between the trajectories of bouncing particles outside the inlet tube obtained from simulation (Fig. 6) and the particles observed in the wind tunnel (Fig. 10b) suggests that, on a qualitative level, the results of the simulation can be applied to the FSSP. The similarity between the wind tunnel observation and bouncing simulation suggests that ice particles bounce and pass through the sample volume of the probe in a qualitatively similar way as obtained in the bouncing simulation.

The CIP has saucer-shaped tips with a sharp leading edge (Fig. 10c). Such a shape results in a bouncing pattern different from that for the OAP-2DC (Fig. 10a). Since the incident angle for the CIP tips remains approximately the same for all particles, they bounce from the surface at approximately the same angle. As a result, the rebound particles form a relatively narrow cone with an angle of approximately 20°–30°, which contaminates approximately 2–3 cm of the sample area nearest to the tip (Fig. 10c). The high-speed videos documented frequent instances when trajectories originated on the cylindrical part of the tip and ended up in the sample area. In other words, ice particles may bounce over the rib of the saucer tip. Such particles may travel a much farther distance across the airflow than the particles that rebound from the tilted section of the leading edge of the tip. An example of such a jump over the CIP tip rib is shown in Fig. 11. Since the travel distance increases with the decrease of the air density (section 3b), it is anticipated that the contamination by particles coming from the cylindrical section of the probe tip will increase for the airborne measurement (P < 1000 mb), as compared to the wind tunnel tests conducted at P = 1000 mb.

Fig. 11.

Fig. 11.

Fig. 11.

Snapshot of the trajectories of ice particles bouncing from the cylindrical surface of the CIP tip. The video recording was conducted in ice spays in the Cox wind tunnel at V = 80 m s−1, P = 1000 mb, and T = −10°C.

Citation: Journal of Atmospheric and Oceanic Technology 30, 4; 10.1175/JTECH-D-12-00142.1

Figure 11b also shows that ice particles at V = 80 m s−1 and P = 1000 mb may bounce upstream of the airflow more than 1 cm. This is an important observation, which should be taken into account during the design of particle probe inlets and hot-wire sensors. As follows from the results of the bouncing simulation (section 3b), the travel distance of the particles bouncing upstream will increase with the decrease of pressure and the increase of the particle mass.

Figures 12a and 12b show high-speed video snapshots of ice particle trajectories around the modified tips for the OAP-2DC and the CIP at a 0° angle of attack. As seen in Figs. 12a and 12b, trajectories associated with bouncing toward the sample volumes of these instruments are not observed here. The particles impacting the tips are deflected to the side opposite to the sensing area. Occasionally, ice particles hit the very end of the tip and may shatter into the sample volume. The probability of such events depends of the particle projection area and its number concentration. The analysis of the high-speed video shows that such events in the Cox and Company wind tunnel are rare. However, the number of such cases will increase with the increase of the particle size.

Fig. 12.

Fig. 12.

Fig. 12.

High speed video snapshots of the trajectories of ice particles bouncing from the modified arm tips of (a),(c) OAP-2DC and (b),(d) CIP at a (a),(b) 0° and (c),(d) 10° AoA. Frames are from high-speed videos taken in ice sprays in the Cox wind tunnel at V = 80 m s−1, P = 1000 mb, and T = −10°C. White strips in (b) and (d) highlight the sample volumes of CIP.

Citation: Journal of Atmospheric and Oceanic Technology 30, 4; 10.1175/JTECH-D-12-00142.1

For a nonzero angle of attack, particles experiencing impact with the inner surface of the tips are deflected toward the sample volumes. The snapshots of the cases with a 10° angle of attack are shown in Figs. 12c and 12d. As depicted in these snapshots, the airflow in this case is nonparallel to the tip surface facing the sample volume and it is exposed to the impact with particles. After impact, particles bounce off or shed along this surface toward the sample volume. It should be mentioned that most research airplanes do not fly with a 10° angle of attack. Such an angle was selected for research purposes in order to exaggerate the effect of particle bouncing.

Figure 13 demonstrates two different scenarios of behavior of ice particles impinging on an unheated tip surface. In the first case, a particle rebounds as one piece without shattering (Figs. 13a–d). In the second case, the ice particle shatters into a number of fragments that bounce off in different directions (Figs. 13e–h). The dotted tracks of the rebound particles in Figs. 13b,c,d,g,h are related to the modulation of light from the facets of rotating ice particles toward the camera. As mentioned in section 3b, the naturally grown cloud ice particles usually have developed branches and may contain hollow cavities inside their bodies, and therefore they are more fragile. For this reason, it is anticipated that cloud ice particles will break down into more fragments on impact with solid surfaces as compared to the ice particles generated in the wind tunnel by means of shaving ice blocks. It is anticipated that frozen cloud droplets, before they develop branches, will rebound without shattering, similar to that shown in Figs. 13a–d. However, ice particles grown through the vapor deposition (bullet rosettes, dendrites, columns, etc.) will shatter on impact with a solid surface at aircraft speed.

Fig. 13.

Fig. 13.

Fig. 13.

Sequence of snapshots of high-speed video showing two types of interactions of ice particles impacting the OAP-2DC arm tip: (a)–(d) the ice particle rebounds from the tip without shattering and (e)–(h) the ice particle shatters after impact with the tip and then bounces off. The arrows in (a)–(d) point to the tracks of the rebound particle. The video was recorded in the Cox wind tunnel in ice spray at V = 85 m s−1, P = 1000 mb, and T = −10°C.

Citation: Journal of Atmospheric and Oceanic Technology 30, 4; 10.1175/JTECH-D-12-00142.1

b. Still photography

Still photography of the illuminated sample volume exposed to the spray is another way to characterize the inhomogeneity of the particle flow across the probe’s sensing area. This method in many ways is similar to that used by the laser sheet technique. When particles pass through the illuminated sample volume, they scatter light toward the camera. The amount of light received by the camera (intensity times exposure time) from each individual volume of the beam depends on the size of the particles in this volume, their concentration, and their speed. For a uniform flow with a spatially homogeneous spray, the brightness of the beam on photograph should remain constant. The nonuniformity of the beam brightness means that one of the following reasons, or a combination, is currently present at the beam location: (i) inhomogeneity of particle sizes, (ii) inhomogeneity of particle concentration, and (iii) nonuniform velocity along the sensing area.

Since the intensity of the scattered light is proportional to the intensity of incident light, the probe’s condensing optics focusing the beam and thus creating inhomogeneities of the beam intensity were uninstalled for this experiment. During the still photography phases, all probes had collimated beams with a small divergence, which was insignificant for the purposes of this experiment.

Figure 14 shows still pictures of the OAP-2DC sample volume with the standard tips. The pictures were taken during one ice spray and three liquid sprays with the nominal mean volume diameters (MVDs) 22, 40, and 150 _μ_m. Depending on the spray, the time exposure of the camera varied from 2 to 10 s to get a statistically significant sample of particles to pass through the sample volume and to average the turbulent fluctuations of the airflow in the wind tunnel.

Fig. 14.

Fig. 14.

Fig. 14.

Still pictures of the OAP-2DC sample volume taken in three liquid and one ice spray at V = 80 m s−1 and P = 1000 mb.

Citation: Journal of Atmospheric and Oceanic Technology 30, 4; 10.1175/JTECH-D-12-00142.1

The distributions of the intensity of the scattered light along the sample volumes of the OAP-2DC, FSSP, and the CIP with standard and modified tips are shown in Fig. 15. These distributions were obtained from the digitization of the still pictures of the laser beams used for illumination of the sample volumes. The distributions of intensity were corrected for the light attenuation by the sprays. For the purposes of comparison, the central sections of the distributions were normalized on unity.

Fig. 15.

Fig. 15.

Fig. 15.

Comparisons of the intensity of the scattered light along the sample volume of the probes with (left) standard and (right) modified arm tips: (a),(b) OAP-2DC, (c),(d) FSSP, and (e),(f) CIP. The gray areas in (c) and (d) indicate the location of the FSSP sample volume along the beam axis. The intensity distributions were obtained from the digitization of the still pictures of the illuminated sample volumes exposed to the sprays with MVD = 22, 40, and 150 _μ_m, and ice spray. The measurements were conducted at V = 80 m s−1 and P = 1000 mb.

Citation: Journal of Atmospheric and Oceanic Technology 30, 4; 10.1175/JTECH-D-12-00142.1

For all sprays, the brightness of the beam has a tendency to increase toward the edges of the sample area. However, in some cases, after passing its maximum, the brightness may decrease near the walls of the arms. This observation is in general agreement with the analysis of the high-speed video (section 4a) and the CFD simulations (section 3a). For the liquid sprays, the inhomogeneity of intensity increases with the increase of the droplet MVDs. This increase of the intensity near the arms is most likely related to the splashing of large drops present in the sprays. Of all liquid sprays, the distribution of intensity turned out to have the smallest inhomogeneities for MVD = 22 _μ_m. Since small droplets follow the airstream better than the large ones and they do not splash, the spray with MVD = 22 _μ_m can be used as a reference for comparison with the sprays with larger droplets, when splashing becomes a dominant artifact.

As seen from Fig. 15, the inhomogeneities of intensity extend approximately 1 cm from the surface of the walls. The amplitude of the relative intensity of the inhomogeneities for the probes with the standard tips is approximately the same (Figs. 15a,c,e). For the FSSP inlet tube, the inhomogeneities of the light intensity reached the center of the tube, where the probe’s sample volume is located (Fig. 15c). This observation suggests that in the presence of large drops (drizzle, rain) the FSSP measurements can be contaminated by artifacts related to spurious splashing of the drops at the leading edge of the inlet tube.

For the probes with modified tips, the intensity of the beams appears to be essentially more uniform, although some inhomogeneities of the intensity are still present near the walls of the arms, indicating the disturbance of the flow and spatial sorting of particles caused by the upstream parts of the tips.

c. Shedding water

Water shedding over the surface of the tips and arms may be a source of different artifacts during measurements. Shedding water may form on the surface of the arms in warm clouds after impact with liquid droplets, or may result from the partial or complete melting of ice particles after impact with the heated surface of tips. Since the air velocity on the arm surface is close to zero, small droplets slowly move downstream, coagulating with other droplets and eventually forming larger pools of shedding water. If such a shedding drop passes over the optical aperture, it may contaminate optics and degrade the measurements.

The flow of shedding water on the surface of the tips and arms was studied with the help of the high-definition video camera Panasonic AJ-HDC27 HD running at 60 frames per second in progressive scan mode and high-speed Phantom V7.0 and V9.0 digital video cameras.

The shape of the tips plays an important role in determining the water flow over the arm’s surface. The following two examples demonstrate the effect of the shape of the tips on the distribution of the shed water, which moves along the airstream over the arm surfaces.

Figure 16a shows the trajectory of the water flow over the surface of the conical tips and arms of OAP-2DC formed in liquid sprays. The flow of the shedding water is indicated by the arrows. The trajectories of the shedding water were deduced from the analysis of the footage of the high-speed Phantom camera. Figures 16c and 16d show the shape of the ice formed after the water shedding downstream from the OAP-2DC and FSSP arms refroze. The shape of the ice is indicative of a nearly uniform distribution of the water flow over the arm surface. As seen from Figs. 16a and 16b, the shedding water streams pass over the optical aperture and therefore may get inside and contaminate the optics.

Fig. 16.

Fig. 16.

Fig. 16.

(a) Trajectories of the shedding water over the surface of the conical tips. (b),(c) The shape of the ice resulting from the refrozen water indicates that the layer of water shedding down the arms. The measurements were conducted in supercooled sprays at V = 80 m s−1 and P = 100 mb.

Citation: Journal of Atmospheric and Oceanic Technology 30, 4; 10.1175/JTECH-D-12-00142.1

The second example shows the streams of water shedding over the surfaces of the modified CIP (Fig. 17a) and the modified FSSP (Fig. 17b) tips obtained from the analysis of the high-definition Panasonic camera footage. The CIP tips have flat plane facets facing the airflow, and a flat surface on the sample area side, which is parallel to the airstream. The front section of the FSSP tips was carved, forming a concave surface. Figure 17a shows that the CIP’s front tip facets collect water in such a way that it only travels along a narrow stream that forms on the side of the rib of the tip, keeping the surface behind the rib completely dry. For the case of the FSSP tip, the spray water is collected in the concave depression, channeled downstream and then dumped into the airflow. For both versions of the modified CIP and FSSP tips shown in Fig. 17, no shedding water was observed behind the ribs separating the front and side sections of the tips, and this area remained dry during the experiments. In contrast to the case in Fig. 16, for the heated CIP and FSSP tips in supercooled sprays no ice was formed downstream from the sample volume. The analysis of the measurements of the CIP and FSSP conducted in liquid sprays did not reveal any artifacts (e.g., elongated images for CIP, anomalous changes in FSSP droplet size distributions during spray) typically observed when the shedding water contaminates optics. These observations suggest that the proposed design of the CIP and FSSP tips (Fig. 17) can be safely used for the sampling of liquid sprays when the airflow is parallel to the arm axis.

Fig. 17.

Fig. 17.

Fig. 17.

Trajectories of the shedding water over the surface of (a) the CIP and (b) FSSP modified tips in supercooled sprays with MVD = 150 _μ_m, at V = 80 m s−1 and P = 100 mb.

Citation: Journal of Atmospheric and Oceanic Technology 30, 4; 10.1175/JTECH-D-12-00142.1

During aircraft maneuvers or flying at nonzero angles of attack or yaw, the probe’s axis is misaligned with respect to the airflow, and cloud water may start accumulating on the arm surfaces facing the sample volume and eventually get into the optical windows. As it was mentioned in section 3a, in order to protect the optical apertures from contamination by shedding water, a special groove in front of the window was designed. The effectiveness of the antishedding groove was tested during the wind tunnel experiments. Figure 18 shows a high-speed video snapshot of the inner part of the modified tip with a V-shaped groove in front of the optical window. The probe was installed at a 10° angle of attack and exposed to a liquid spray with an air temperature above the freezing point. The streakers formed by shedding water appear in the section upstream from the optical window. The shedding water trapped by the angular opening can be seen exiting on the side of the groove. The analysis of the videos with shedding water showed that the designed groove collects the shedding water, thus minimizing contamination of the optics by water.

Fig. 18.

Fig. 18.

Fig. 18.

High-speed video snapshots of the water shedding over the inner surface of the modified OAP-2DC tip. The shedding water is collected by the groove in front of the aperture and dumped on the side. The measurements were conducted in a liquid spray with an MVD = 22 _μ_m at V = 80 m s−1 and P = 100 mb.

Citation: Journal of Atmospheric and Oceanic Technology 30, 4; 10.1175/JTECH-D-12-00142.1

d. Deicing heaters

Ice forming on the tips changes the aerodynamic performance of the arms, and this ice may act to shatter and deflect rebound ice particles toward the sample volume. Icing of the tips may also result in disturbing the airflow, thereby changing the airspeed and the direction of the airflow at the sample volume location, partially or completely blocking the sample volume. During the Cox wind tunnel experiments, the tip’s deicing heaters were tested to protect the tips and arms from ice buildup in supercooled sprays. The tips were loaded with cartridge heaters, whose electric power was controlled with the help of a transformer. It was found that in order to prevent the arm tips from icing in a supercooled spray with liquid water content 1 g m−3 at V = 85 m s−1 and T = −10°C, the deicing heater power for OAP-2DC, FSSP, and CIP should be approximately 60°–70°W, 80°–90°W, and 60°–70°W per tip, respectively.

A complementary way of mitigating the artifacts related to bouncing and shattering is by overheating the tips. Overheating the tips causes the small ice particles stick to the heated surface of the tips and melt, instead of bouncing. This effect was observed during the wind tunnel experiments with heated and unheated surfaces.

It should be noted that icing of the front domes of the probe’s canisters will move the velocity depression area forward, and it may change the local airspeed in the sample volume. Therefore, it is highly recommended to deice the front domes as well.

5. Concluding remarks

The present study describes some results of the tests performed during the modification and optimization of the shape of the tips in order to mitigate particle shattering and to improve the performance of airborne particle probes. It has been demonstrated during the wind tunnel experiments and the high-speed video recordings that the proposed shape of the tips can significantly mitigate shattering and particle bouncing toward the probes’ sample volume. The Airborne Icing Instrumentation Experiment (AIIE) flight campaign, dedicated to the study of the effect of shattering on the microphysical measurements in ice clouds, demonstrated the successfulness of the undertaken approach to lessen shattering through the modification of the shape of the tips (Korolev et al. 2011). The conducted tests suggest that the arms with circular cross sections and with tip shapes as in Fig. 2b have the best overall performance in terms of mitigating shattering, water shedding, and airflow uniformity. The arms with a flattened inner section (Figs. 2a,c) deflect more shattered particles toward the sample volume at a nonzero angle of attack. The airflow between such arms is also more sensitive to nonzero yaw and angle of attack in comparison to the arms with a circular cross section.

The numerical simulation of bouncing, analysis of high-speed videos from the wind tunnel tests, and the results of the AIIE flight campaign suggests that shattering, in a complex way, depends on air density, airspeed, angle of attack, the shape and temperature of the tips, ice particle size and habit, and particle orientation at the moment of impact. Taking all of these effects into account in numerical models of shattering is a challenging problem. It is hindered by the lack of information about the rebound coefficient of ice, number and size of shattered fragments, angle of bouncing of shattered fragments, and other factors.

It should be noted that the proposed modification of the tips minimizes the effect of shattering; however, it does not completely solve the problem of shattering. The modified tips may occasionally shatter particles or bounce them into the sample volume, if the airplane is flying at a nonzero yaw or angle of attack. Generally speaking, any part of the probes’ housing upstream from the sample volume one way or another under certain conditions may shatter and deflect ice particles toward the sample volume. This is a principal limitation of existing particle measuring techniques that keeps the sample volume inside the inlet tubes or between the arms. To address this important problem in cloud microphysics, more efforts should be invested in the design of new instruments and methods free of the shattering problem.

Acknowledgments

This work was funded by Environment Canada, Transport Canada, the Federal Aviation Administration, and NASA. The authors express their gratitude to Cox and Company personnel and, in particular, to Adam Lawrence for such a high level of cooperation and support in operating the Cox and Company wind tunnel facility. The NASA Glenn Research Center video group Vince Reich, Chris Lynch, and Quentin Schwinn did an excellent job capturing high-speed videos during the Cox and Company wind tunnel tests. It is hard to overestimate the role of Vladimir Torgashev, who did mechanical drawing and supervised manufacturing of the probe tips. Special thanks to Alex Shahshkov of Environment Canada for his help in data analysis. The authors thank Alain Protat and two anonymous reviewers for their thoughtful comments.

REFERENCES