The NCAR-UWyo Aerosol Inlet

Transcription

1 The NCAR-UWyo Aerosol Inlet Jeff Snider Markus Petters Hiroshi Takagi Peter Liu Don Lukens Brent Glover Terry Deshler Derek Montague Perry Wechsler Gabor Vali Dave Rogers Bob Kelly This report summarizes characterizations of an inlet used for sampling aerosol from the University of Wyoming King Air. The inlet (Figure 1) was fabricated at NCAR and has be used on the UWyo King Air during three projects: 1) WAVEICE ( 2) WYICE-2000 ( and HICU ( The inlet will be referred to as the NCAR-UWyo inlet. Here both airborne and laboratory testing are reported. 1. General Characteristics In contrast to many aircraft inlet systems, utilizing a pump to draw sample into the aircraft, air is forced through the NCAR-UWyo inlet by ram pressure. Suction produced where sample air is exhausted from the plane, via a scarf tube, also assists flow through the NCAR- UWyo inlet. The inlet design is shown schematically in Figure 2. Air enters the diffuser (forward diameter 9.5 mm), decelerates through a section of 22 mm ID stainless steel tubing (radius of curvature = 200 to 300 mm) and enters the manifold. A section of electricallyconductive plastic tubing (15 mm ID) can be used to connect the forward section of the inlet (i.e., the diffuser and the curved section) to the manifold. A pitot tube located downstream of the manifold is located within a third section of 22 mm ID stainless steel tubing. After the pitot tube the airstream passes through ~2 m of metal-reinforced plastic tubing and is exhausted via the scarf tube. C:\jeff\keck\keck_inlet\inlet_report_v02.docx Last printed 5/24/2018 2:06:00 PM 1

2 Figure 2 shows a pump and a mass flow meter that were used during the laboratory studies but not during flight testing. The smallest velocity (i.e., at the front of the diffuser) tested during our laboratory studies was 36 m/s. After accounting for the diameter change between the front end of the diffuser and the manifold, 36 m/s translates to 7 m/s within the manifold. The corresponding Reynolds numbers are 22,000 and 9,000, respectively. Clearly, the inlet and the manifold are turbulent at this minimum pump flow setting and also at the larger settings typical of aircraft flight. The NCAR-UWyo inlet can be deiced. This option was employed during the WAVEICE experiments but not during WYICE-2000 or HICU. 2. Flight Characteristics Air velocities at the front of the NCAR-UWyo inlet were inferred using measurements made by the pitot tube. The formula used to calculate air velocity at the front end of the diffuser is V 2 p 1/ 2 D D man in 2 (1) Here p mm), and is the pitot pressure, is air density, D in D man is the diameter of the pitot manifold (22 is the diameter of the front end of the diffuser (9.5 mm). Regardless of date the pitot sensor data analyzed in this report were converted to differential pressure via the following relationship p c0 c1 pitot _ cnt (2) Here pitot _ cnt is the digital representation of the pitot output signal, and c0 and c1 are calibration constants (-1.78 and 5.43x10-3, respectively) obtained from the WAVEICE NetCDF header. In Figures 3a-3d the inlet velocities are plotted versus true air speed (i.e. the speed of the King Air relative to the wind). With the exception of the WYICE-2000 flight on , the inlet velocities are ~20 m/s larger than TAS. A velocity mismatch should result in a particle C:\jeff\keck\keck_inlet\inlet_report_v02.docx Last printed 5/24/2018 2:06:00 PM 2

3 size-dependent sampling bias. How large is the inlet sampling bias? To answer that question we rely on the aerosol sampling theory discussed in Reist (1984). ( V o Aerosol concentrations are perturbed relative to their ambient values if the free-stream ) and aerosol inlet velocities (V ) are different. This concentration perturbation is C Co 2 Stk V o Stk V (3) where C is the perturbed concentration, C o is the concentration corresponding to the freestream state and Stk is the Stokes number defined as Stk Vo R (4) In Equation 4 is the particle relaxation time and R is the radius of the inlet. The calculations are shown in Figure 4a. Note that a 20 m/s velocity mismatch has a negligible effect submicron particle concentrations. Larger absolute biases, approaching a factor of two at 1 um, occur because of the velocity mismatch within the sampling manifold. This sampling bias can also be analyzed with equations 3 and 4. Here the free stream velocity becomes the velocity measured with the pitot tube 1/ 2 2 p V o (5) and the aerosol inlet velocity is the value dictated by the flow rate into instrumentation attached to the manifold (4 actual liter per minute (ALPM) during WYICE-2000 and 8 ALPM during HICU) and by the diameter of the sample tubes (5 mm ID during WYICE-2000 and 4 mm ID during HICU). Figure 4b shows that the manifold sampling bias enhances the concentration of particles greater than 0.2 um but only modestly affects the CN concentration (particles greater than 0.01 um). Equations 3 and 4 approximate the enhancement (or attenuation) of particle concentration due to anisokinetic sampling. Figure 4 shows that the phenomenon is occurring at relatively large sizes (D>0.2 um) since it is here that the Stokes number (a measure of the particle inertia) C:\jeff\keck\keck_inlet\inlet_report_v02.docx Last printed 5/24/2018 2:06:00 PM 3

4 becomes significant (Equation 3). There are other mechanisms for altering particle concentration during aerosol sampling. Turbulence within the inlet is an issue since it decreases the depth of the laminar boundary layer and thus enhances the rate of particle deposition via the process of Brownian diffusion. In the UWyo-Keck laboratory this phenomenon has been addressed by examining particle transmission through tubes flowing at relatively small volumetric flow rates ( ). Typical volumetric rates which are smaller than 5 ALPM. Because of the requirement of laminar flow in the 3071 DMA used to produce the test particles these tests cannot be scaled up to volumetric flow rates needed for the NCAR-UWyo inlet (volumetric rates ~500 ALPM). To address the problem of characterizing sampling biases introduced by the NCAR-UWyo inlet we conduced two different types of studies: 1) comparisons of ground-based and airborne aerosol measurements, and 2) comparisons of ground-based aerosol measurements made when sampling either with or without the NCAR-UWyo inlet. 3. Comparison to Ground-based and Airborne Aerosol Measurements The flight conducted on was dedicated to the comparison of the King Air and ground-based measurements of CN and CCN. The ground-based measurements were made at the UWyo Balloon Launch Facility located ~2 km SE of the main runway at the Laramie, WY Airport. Data collected during three low-altitude flybys (flight altitude m agl) and during two 60 km legs (altitude 300 m agl) conducted upwind of the ground site were compared. The CCN comparison, based on concurrent ground- and airborne-data (i.e., the upwind legs) showed that the ground-based values were ~30% larger than the airborne values. This was true at all applied supersaturations (S=0.8, 1.2 and 1.6%). The 30% difference suggests a sampling bias, either due to particle loss in the NCAR-UWyo inlet or due to spatial non-uniformity of the aerosol. Even larger differences (+60%) were noted in the comparison of the ground-based and upwind leg CN data and in the comparison of CN data collected during the flybys. Note that these concentration biases are opposite the direction of the sampling bias predicted for (WYICE-2000) due to the velocity mismatch in the NCAR-UWyo manifold (Section 2). Figure 5 shows data (1 Hz airborne measurements of CN and potential temperature) collect during a flyby of the Balloon Launch Facility and during landing (~15 min after the flyby). Observe that the lowest 15 m of the sounding is stabilized by a 1.5 K inversion and that C:\jeff\keck\keck_inlet\inlet_report_v02.docx Last printed 5/24/2018 2:06:00 PM 4

5 CN concentrations increase by a factor of 2.5 in this layer. Also note that the minimum altitude of the flybys (50 m agl) was above the cool layer containing enhanced concentrations of CN. Surface CN concentrations (not shown) varied between 1400 and 1800 cm -3 during the flybys (only data from one of three flybys is shown in Figure 5). Clearly the conclusion that the airborne values of CN were substantially smaller than the surface measurements is valid, but since the CN/Theta sounding shows a strong stratification the aerosol comparisons from cannot be viewed as proof of a sampling bias. 4. Laboratory Testing Laboratory testing the NCAR-UWyo inlet was motivated by the disparities seen between ground-based and airborne aerosol concentration measurements on Both an indoor aerosol (outside air filtered by the Engineering building air-handling system) and an outdoor aerosol were tested. Two sensors, referred to as the internal and external sensors (Figure 2), were used in the tests. Flow rates through the sensors were controlled by critical orifices - 1 ALPM and 3 ALPM for the CN and CCN instruments, respectively. Figure 6a shows a comparison of the two CN sensors when sampling the indoor aerosol via a common inlet tube (not the NCAR-UWyo inlet). Both instruments reported nearly the same average concentration and the correlation coefficient is large (r=0.97). Figures 6b-6d show the internal/external CN comparison as a function of inlet velocity (i.e., the velocity inferred via Equation 3). Increasing velocity is associated with a decrease in the internal/external concentration ratio but only a 7% departure from unit slope is observed at the largest velocity (140 m/s; Figure 6d). The observed decrease in the internal/external CN ratio may result from increased turbulence, and therefore increased deposition to the wall of the NCAR-UWyo inlet, associated with increasing inlet velocity. Also, a comparison of the data shown in Figure 6a and in Figures 6b-6d shows that there is a degradation of the correlation between the CPC1 and CPC3 measurements when CPC1 is sampling within the NCAR-UWyo inlet. This phenomenon may be due to the ~100 mm separation of the CPC3 and the NCAR-UWyo inlets (Figure 2). More analysis is needed to examine both of these hypotheses. In contrast with CN, which are measured with a response time of ~10 s, CCN measurements are only made every 30 s. Therefore data derived from the internal-to-external CCN comparisons are presented as frequency distributions. Furthermore, the CCN testing focused on low supersaturation CCN measurements (S=0.2 and S=0.4%). To insure large C:\jeff\keck\keck_inlet\inlet_report_v02.docx Last printed 5/24/2018 2:06:00 PM 5

6 concentrations, and thus favorable CCN counting statistics, we conducted these inlet tests on the roof of the Engineering building. During sampling the aerosol in Laramie was contaminated by urban aerosol advected from east of the Colorado Front Range. All of the CCN tests were conducted at the maximum inlet speed, 140 m/s. The results of an internal-to-internal CCN tests (control experiments) are shown in Figure 7a (S=0.2%) and Figure 7c (S=0.4%). The frequency distributions have similar shape and are characterized by nearly the same mode value. Surprisingly, the external CCN concentrations at S=0.2% were shifted to smaller values (mode at 225 cm -3 ) compared to the larger mode corresponding to the internal measurements (mode at 325 cm -3 ). These results are explained by the low precision of CCN measurements made at such small supersaturations (Snider et al., 2003). The internal-toexternal CCN comparison at S=0.4% did however reveal good agreement between the internal/external CCN measurements (Figures 7c and 7d). As in the ground-based CN testing, the CCN comparisons shown in Figure 7 do not uncover a sampling bias attributable to the NCAR-UWyo inlet. 5. Conclusions and Recommendations 1) For CCN the internal-to-external laboratory-based testing did not reveal any significant loss within the NCAR-UWyo inlet. This conclusion is based on a limited set of measurements and should be supplemented with data acquired at larger applied supersaturations. 2) For CN the laboratory testing did reveal a small velocity-dependent bias. If we are correct in our speculation that this is related to turbulence then it should vary with particle size. At present we cannot test that hypothesis with size-classified aerosol becasue we do not have the capability of producing classified particles at the relevant volumetric flow rates. Continued ground-to-aircraft and laboratory-based internal/external comparisons, complimented with diffusion batteries to limit the minimum sampled size, should be used to probe this issue. 3) Unbiased sampling of particles in the 0.1 to 1 um range will require closer matching of velocities within the NCAR-UWyo manifold. Consideration should also be given to the particle loss via wall deposition, in the diffuser, and where sample streams are split and lead into instrumentation and particle samplers. This topic has been the focus of recent work at NCAR ( C:\jeff\keck\keck_inlet\inlet_report_v02.docx Last printed 5/24/2018 2:06:00 PM 6

7 4) Since Figure 3 suggests that the pitot calibration has changed with time, flow measurement uncertainties are indicated. The NCAR-UWyo inlet flow measurement needs to be improved, both for laboratory and airborne application. References Jaenicke, R., Tropospheric aerosols, in Aerosol-Cloud-Climate Interactions, edited by P.V. Hobbs. Academic Press, San Diego, CA Reist, P.C., Aerosol Science and Technology, second Edition, McGraw Hill, New York, 1984 Snider, J.R., and, S.Guibert, J.-L. Brenguier, and J.-P.Putaud, Aerosol activation in marine stratocumulus clouds: Part II Köhler and parcel theory closure studies, J. Geophy. Res., 108, 8629, doi: /2002jd002692, 2003 C:\jeff\keck\keck_inlet\inlet_report_v02.docx Last printed 5/24/2018 2:06:00 PM 7

8 Figure 1 - Photos of the NCAR-UWyo inlet. C:\jeff\keck\keck_inlet\inlet_report_v02.docx Last printed 5/24/2018 2:06:00 PM 8

9 9.5 mm to 22 mm ID diffuser Dwyer Pitot Tube 22 mm ID manifold and forward-facing sample tubes (stainless steel) External Sensor (CPC1 and/or CCN104) Fuji VFC404A Ring Compressor Internal sensor (CPC3 and/or CCN108) Sierra 640 Mass Flow Meter Figure 2 - The NCAR-UWyo inlet configured for laboratory testing. Measurements from the external sensors are compared to those from the internal sensors and used to assess particle loss during transmission through the inlet. C:\jeff\keck\keck_inlet\inlet_report_v02.docx Last printed 5/24/2018 2:06:00 PM 9

10 200 Inlet Speed, m/s WAVEICE y = 1.13x WYICE y = 1.01x Inlet Speed, m/s HICU y = 1.07x HICU y = 1.01x True Air Speed, m/s True Air Speed, m/s Figure 3 - One Hz true air speed and inlet velocity measurements. The latter correspond to the velocity at the front end of the diffuser. Data is from a WAVEICE flight, a WYICE flight and two HICU flights. C:\jeff\keck\keck_inlet\inlet_report_v02.docx Last printed 5/24/2018 2:06:00 PM 10

11 WYICE HICU V=Vo+20 (Vo=100 m/s) 1.2 C/Co Diameter, mm Figures 4a - Size-dependent concentration ratios corresponding to velocities both external to and within the NCAR-UWyo inlet dn/dlogd or Cumulative Conc., cm Free-tropospheric spectrum Spectrum with WYICE sampling bias Diameter, mm Figure 4b - Size spectra corresponding to a free-tropospheric climatology (Jaenicke, 1993) and the effect of the sampling bias introduced by the velocity mismatch in the manifold during WYICE-2000 C:\jeff\keck\keck_inlet\inlet_report_v02.docx Last printed 5/24/2018 2:06:00 PM 11

12 75 Potential Temperature, K Radar altimeter altitude, m ag CN, Sounding CN, Flyby Theta, Sounding Theta, Flyby CN Concentration, cm -3 Figure 5 - CN and potential temperature measurements from the UWyo King Air on C:\jeff\keck\keck_inlet\inlet_report_v02.docx Last printed 5/24/2018 2:06:00 PM 12

13 Figure 6 - Comparisons of internal and external CN measurements and external/external control measurements made during indoor sampling of room air. C:\jeff\keck\keck_inlet\inlet_report_v02.docx Last printed 5/24/2018 2:06:00 PM 13

14 0.6 7a 104/Internal/C/S= /Internal/C/S= b 104/External/I/S= /Internal/I/S= Relative Frequency c 104/Internal/C/S= /Internal/C/S= d 104/External/I/S= /Internal/I/S= Relative Frequency CCN Concentration, cm CCN Concentration, cm -3 Figure 7 - Comparisons of internal and external CCN measurements and internal/internal control measurements made during outdoor sampling C:\jeff\keck\keck_inlet\inlet_report_v02.docx Last printed 5/24/2018 2:06:00 PM 14