COMPARISON OF EXPERIMENTAL AND COMPUTATIONAL SHIP AIR WAKES FOR YP CLASS PATROL CRAFT* Murray R. Snyder,† Joshua P. Shishkoff,‡ Franklin D. Roberson,§ Martin C. McDonald,§ Cody J. Brownell,** Luksa Luznik,†† David S. Miklosovic,‡‡ John S. Burks,§§ United States Naval Academy, Annapolis, MD 21402 Hyung Suk Kang,*** Johns Hopkins University, Baltimore, MD 21218 Colin H. Wilkinson Zenetex LLC, Patuxent River, MD 20670 Abstract This paper provides second year results from a multi-year research project that involves a systematic investigation of ship air wakes using an instrumented United States Naval Academy (USNA) YP (Patrol Craft, Training). The objective is to validate and improve Computational Fluid Dynamics (CFD) tools that will be useful in determining ship air wake impact on naval rotary wing vehicles. This project is funded by the Office of Naval Research and includes extensive coordination with Naval Air Systems Command. Currently, ship launch and recovery wind limits and envelopes for helicopters are primarily determined through at-sea in situ flight testing that is expensive and frequently difficult to schedule and complete. The time consuming and potentially risky flight testing is required, in part, because computational tools are not mature enough to adequately predict air flow and wake data in the lee of a ship with a complex superstructure. The top-side configuration of USNA YPs is similar to that of a destroyer or cruiser, and their size (length of 108 ft and above waterline height of 24 ft) allows for collection of air wake data that is in the same order of magnitude as that of modern naval warships, an important consideration in aerodynamic modeling. A dedicated YP has been modified to add a flight deck and hangar structure to produce an air wake similar to that on a modern destroyer. Three axis acoustic anemometers, fog generators and an Inertial Measurement Unit (IMU) have been installed. Repeated testing on the modified YP is being conducted in the Chesapeake Bay, which allows for the collection of data over a wide range of wind conditions. Additionally, a scale model of the modified YP has been constructed for testing in the 42×60×102 inch USNA wind tunnel. Significant wind tunnel measurements are scheduled for fall 2010. Comparison of YP in situ test data with wind tunnel data will be useful for validation of wind tunnel test methods and scale effects, as well as CFD models that could help predict ship air wake effects. The project involves USNA midshipmen who are participating in test planning, collecting and analyzing data, and in CFD modeling, providing the midshipmen with valuable professional and research experience. Additionally, the flight deck has been designed to allow operation of a 400-500 lb class rotary wing Unmanned Aerial Vehicle for direct measurement of the dynamic interface between the ship and helicopter. Summary of Important Interim Conclusions: 1. Good velocity component correlation exists between in situ and CFD data for the YP bow mounted reference anemometer. 2. Excellent repeatability has been observed for normalized in situ data collected for a given relative wind situation during different underway data collection periods. 3. Good comparison is shown between normalized in situ and CFD data over the flight deck for a headwind and for wind 15° off the starboard bow. * Presented at the American Society of Naval Engineers, Launch and Recovery Symposium 2010, Arlington, VA, December 8-9, 2010. † Permanent Military Professor, Mechanical Engineering Department, corresponding author ([email protected]). ‡ Ensign, US Navy. § Midshipman, US Navy. ** Assistant Professor, Mechanical Engineering Department. †† Assistant Professor, Mechanical Engineering Department. ‡‡ Associate Professor, Aerospace Engineering Department. §§ Visiting Professor, Aerospace Engineering Department. *** Associate Research Scientist, Mechanical Engineering Department. ASNE Launch & Recovery Symposium Report Documentation Page Form Approved OMB No. 0704-0188 Public reporting burden for the collection of information is estimated to average 1 hour per response, including the time for reviewing instructions, searching existing data sources, gathering and maintaining the data needed, and completing and reviewing the collection of information. Send comments regarding this burden estimate or any other aspect of this collection of information, including suggestions for reducing this burden, to Washington Headquarters Services, Directorate for Information Operations and Reports, 1215 Jefferson Davis Highway, Suite 1204, Arlington VA 22202-4302. Respondents should be aware that notwithstanding any other provision of law, no person shall be subject to a penalty for failing to comply with a collection of information if it does not display a currently valid OMB control number. 1. REPORT DATE 3. DATES COVERED 2010 2. REPORT TYPE 00-00-2010 to 00-00-2010 4. TITLE AND SUBTITLE 5a. CONTRACT NUMBER Comparison of Experimental and Computational Ship Air Wakes for YP 5b. GRANT NUMBER Class Patrol Craft 5c. PROGRAM ELEMENT NUMBER 6. AUTHOR(S) 5d. PROJECT NUMBER 5e. TASK NUMBER 5f. WORK UNIT NUMBER 7. PERFORMING ORGANIZATION NAME(S) AND ADDRESS(ES) 8. PERFORMING ORGANIZATION United States Naval Academy,Annapolis,MD,21402 REPORT NUMBER 9. SPONSORING/MONITORING AGENCY NAME(S) AND ADDRESS(ES) 10. SPONSOR/MONITOR’S ACRONYM(S) 11. SPONSOR/MONITOR’S REPORT NUMBER(S) 12. DISTRIBUTION/AVAILABILITY STATEMENT Approved for public release; distribution unlimited 13. SUPPLEMENTARY NOTES Launch & Recovery Symposium 2010, Dec 8-9, 2010, Arlington, VA. 14. ABSTRACT This paper provides second year results from a multi-year research project that involves a systematic investigation of ship air wakes using an instrumented United States Naval Academy (USNA) YP (Patrol Craft, Training). The objective is to validate and improve Computational Fluid Dynamics (CFD) tools that will be useful in determining ship air wake impact on naval rotary wing vehicles. This project is funded by the Office of Naval Research and includes extensive coordination with Naval Air Systems Command. Currently, ship launch and recovery wind limits and envelopes for helicopters are primarily determined through at-sea in situ flight testing that is expensive and frequently difficult to schedule and complete. The time consuming and potentially risky flight testing is required, in part, because computational tools are not mature enough to adequately predict air flow and wake data in the lee of a ship with a complex superstructure. The top-side configuration of USNA YPs is similar to that of a destroyer or cruiser, and their size (length of 108 ft and above waterline height of 24 ft) allows for collection of air wake data that is in the same order of magnitude as that of modern naval warships, an important consideration in aerodynamic modeling. A dedicated YP has been modified to add a flight deck and hangar structure to produce an air wake similar to that on a modern destroyer. Three axis acoustic anemometers, fog generators and an Inertial Measurement Unit (IMU) have been installed. Repeated testing on the modified YP is being conducted in the Chesapeake Bay, which allows for the collection of data over a wide range of wind conditions. Additionally, a scale model of the modified YP has been constructed for testing in the 42?0?02 inch USNA wind tunnel. Significant wind tunnel measurements are scheduled for fall 2010. Comparison of YP in situ test data with wind tunnel data will be useful for validation of wind tunnel test methods and scale effects, as well as CFD models that could help predict ship air wake effects. The project involves USNA midshipmen who are participating in test planning, collecting and analyzing data, and in CFD modeling, providing the midshipmen with valuable professional and research experience. Additionally, the flight deck has been designed to allow operation of a 400-500 lb class rotary wing Unmanned Aerial Vehicle for direct measurement of the dynamic interface between the ship and helicopter. 15. SUBJECT TERMS 16. SECURITY CLASSIFICATION OF: 17. LIMITATION OF 18. NUMBER 19a. NAME OF ABSTRACT OF PAGES RESPONSIBLE PERSON a. REPORT b. ABSTRACT c. THIS PAGE Same as 10 unclassified unclassified unclassified Report (SAR) Standard Form 298 (Rev. 8-98) Prescribed by ANSI Std Z39-18 Notation CAD = Computer Aided Design CFD = Computational Fluid Dynamics IMU = Inertial Measurement Unit LCS = Littoral Combat Ship USNA = United States Naval Academy YP = Patrol Craft, Training α = Angle of Attack β = Relative Wind Angle Introduction Launch and recovery of rotary wing aircraft from naval vessels can be very challenging and potentially hazardous. Ship motion combined with the air turbulence that is created as the wind flows over the ship’s superstructure can result in rapidly changing flow conditions for the rotary wing aircraft. Additionally, dynamic interface effects between the vessel air wake and the rotor wake are also problematic. To ensure aircraft and vessel safety, launch and recovery envelopes are prescribed for specific aircraft types on different ship classes (Figure 1).1 Permissible launch and recovery envelopes are often very restrictive because of limited flight envelope expansion upon initial certification. Flight testing required to expand the envelopes is frequently difficult to schedule, expensive and potentially hazardous. Currently, the launch and recovery wind limits and air operation envelopes are primarily determined via the subjective analysis of test pilots (e.g., excessive flight control inputs are required to safely land on the flight deck), using a time consuming and potentially risky iterative flight test Figure 1. Launch and recovery envelopes for build-up approach. The risk and cost could be MH-60S helicopters on USS Ticonderoga (CG 47) mitigated through the use of computational tools, but class cruiser (Ref 1). current methods are insufficiently validated for ships with a complex superstructure, such as a destroyer, In Situ Measurements of Ship Air Wakes. cruiser or Littoral Combat Ship (LCS).2-9 This paper presents an update of a multi- USNA operates a fleet of YP (Patrol Craft, Training) year project to develop and validate Computational vessels for midshipman training. The USNA YPs, Fluid Dynamics (CFD) tools to reduce the amount of (Figure 2) are relatively large vessels (length of 108 at-sea in situ flight testing required and make rotary ft (32.9 m) and an above waterline height of 24 ft wing launch and recovery envelope expansion safer, (7.3m)) with a superstructure and deck configuration more efficient, and affordable. that resembles that of a destroyer or a cruiser. The size of the YPs is such that air wake data can be Project Description collected with Reynolds numbers in the same order of magnitude as those for modern naval warships, an The Ship Air Wake project leverages unique important consideration in aerodynamic modeling. resources available at the United States Naval (Reynolds number is the ratio of inertia forces to Academy (USNA) that allow for a systematic viscous forces.) As shown in Figures 3 and 4, YP676 analysis of ship air wakes. has been modified to add a representative flight deck and hangar structure that model those on modern US Navy ships. ASNE Launch & Recovery Symposium Ultrasonic anemometers have been installed to allow for direct measurement of wind velocity and direction over the flight deck (Figure 5). The anemometers are the Applied Technology Inc. “A” style three-velocity component model with a 5.91 inch path length and a measurement accuracy of ± 1.18 inch/sec. The anemometers are connected to a data packer unit that allows up to eight different anemometers to be sampled concurrently at up to 20 Hz, which is the sample rate for the current measurements. Figure 2. Unmodified USNA YP (Patrol Craft, Figure 5. Ultrasonic anemometers installed on Training). YP676 flight deck. To estimate the reference or “true” wind (i.e. the wind minimally impacted by flow over the ship), one anemometer is mounted 3.5 feet forward and 7.0 feet above the ship’s bow (Figure 6). Placement of the reference anemometer was based upon CFD flow simulations that showed this is an accessible location which was expected to have minimal disturbance to the incoming wind due to the ship hull and superstructure. Currently six anemometers are installed on YP676; eventually eight anemometers will be installed. Other supporting data collected simultaneously with velocity measurements include Figure 3. YP676 with added flight deck. ship pitch and roll (Crossbow Inertial Measurement Unit model VG600), temperature and atmospheric pressure. In addition, real time meteorological data from the Thomas Point meteorological station (NOAA—National Data Buoy Center—Station TPLM2), which is in close proximity to the underway testing area, is also collected. Figure 4. YP676 added hangar-like superstructure modification. ASNE Launch & Recovery Symposium Figure 8. Detail being added to YP model. Computational Fluid Dynamics Simulations Numerical simulations have been performed by USNA midshipmen with parallel processing using both Cobalt,10 a commercial CFD code, and the NASA USM3Dns11,12 code. Both codes use an unstructured tetrahedral grid. As shown in Figure 9, the unstructured grid allows for finer resolution Figure 6. Bow mounted reference anemometer. where greater variation in air flow is expected. The tetrahedral grid is divided into Wind Tunnel Measurements partitions with communication between these partitions accomplished through the use of Message Wind tunnel tests of a 4% scale model of the YP will Passing Interface. Such partitioning speeds up the commence in USNA’s large wind tunnel (42 × 60 × solution generation by allowing an individual 120 inch test section) in October 2010. Figure 7 processor to solve the flow field in a limited number shows the hull and basic superstructure while Figure of tetrahedrons. 8 shows detail being added to the model. Wind tunnel tests of the detailed model will primarily be conducted at a wind tunnel velocity of 300 ft/sec. This velocity allows matching of Reynolds number between the 4% scale model and the actual YP with a seven knot (nm/hr) wind over deck. An important feature of the present research is the ability to match Reynolds number between the wind tunnel, the in situ measurements and numerical simulations. Figure 9. Unstructured YP surface grid. Figure 7. YP hull and basic superstructure in USNA large wind tunnel. ASNE Launch & Recovery Symposium Interim Results A similar match is observed for β=15° (wind from 15 degrees off the starboard bow). Figure 11 As of the submission of this paper, 14 underway test shows this underway data which indicates α=5.95°; periods in the Chesapeake Bay have been completed. corresponding time-averaged CFD solutions show α Additionally, midshipmen interns have performed of 5.61° and 5.30°, respectively, for 7 and 20 knots CFD analysis for both 7 and 20 knots of wind over incident winds. deck. CFD analysis, using a grid size of The close agreement in flow direction approximately 20 million tetrahedrons, was between in situ and CFD simulations indicates that performed for a head wind and for crosswind from the CFD provides an accurate flow solution near the the starboard bow from 15, 30, 45, 60, 75 and 90°. front of the YP. This is graphically shown in Figures 12 to 14, wherein the black vectors show the in situ Underway Measurements data and the white vectors show the 7 knot time- averaged CFD solution (note that the white vectors Using up to six anemometers simultaneously, in situ have a 7 knots magnitude while the black vector data has been collected for a head wind, and a exhibit the average magnitude seen in the 34 data crosswind 15 or 30° off the starboard bow. Data has sets). been collected for the bow reference anemometer and for anemometers mounted directly above the flight deck at various heights. During data collection real time data output is continuously monitored to ensure desired relative wind is approximately maintained and that data quality is satisfactory. This information is displayed on the YP’s bridge such that the ship’s helmsman can take real time corrective action to adjust ship heading. Analysis of Reference Anemometer Data As mentioned previously, reference measurements are taken above and forward of the YP bow. The Figure 10. Variation in relative wind β and angle reference anemometer data is analyzed to ensure data of attack α for β=0° data from multiple underway sampling above the stern flight deck on the YP is test periods. conducted under a known, specified flow scenario. The relative wind angle β is measured from the bow, increasing to 90° when off the starboard beam. Due to the nature of field data collected while underway in the Chesapeake Bay, there are inevitable fluctuations in the air flow incident upon the ship. Additionally, the reference measurements, which are taken approximately seven feet above and three and half feet forward of the uppermost portion of the YP’s bow, see effects from the ship and do not accurately represent the undisturbed free stream flow. Figure 10 shows in situ reference anemometer data for a headwind (β=0°) which includes 34 data sets collected over multiple underway test periods with mean wind velocity in the range of 8-23 knots. The vertical bars represent 95% Figure 11. Variation in relative wind β and angle confidence bands defined as ± 2 standard deviations of attack α for β=15° data from multiple from the mean. While the mean flow is primarily underway test periods. horizontal, a small but consistent vertical flow component is observed with the incident wind averaging an angle of attack α of 5.10°. The positive α indicates the vertical flow is in the upward direction. Time-averaged CFD solutions computed at 7 and 20 knots, similarly, show α of 4.97° and 3.89°, respectively. ASNE Launch & Recovery Symposium Repeatability and Scaling of Flight Deck In Situ Data Collection Incoming flow conditions significantly vary between different sampling periods depending on the daily weather conditions (wind speed and direction, atmospheric pressure and temperature, and wavelength and height). The YP676 operating conditions (engine rpm, ship’s speed and craft master's skill level at maintaining a given relative wind component) also vary. Figure 12. β=0° bow reference anemometer data Data repeatability is tested qualitatively in (black vector) vs. 7 knot CFD flow visualization two points of view: the first is the mean velocities (white vectors; 7 knots = 141 inches/sec). above the flight deck; with the other being turbulence quantities including Reynolds shear stresses. To confirm the repeatability on different underway test periods, velocity measurements are performed at the same measurement locations above the deck, e.g., at x/H=2.15, y/H=0 (centerline of the deck) at three different heights of z/H=0.28, 0.67 and 1.07, where H is the hanger height (59 inches), x is the distance aft of the hanger, y is the distance to starboard of the centerline and z is the height above the deck. The un-scaled velocity vectors on the (x,z)- plane are shown in Figure 15 (a). The red and green vectors were sampled on different days with the corresponding mean stream wise bow velocities of 239 inch/sec and 298 inch/sec, respectively. Although there is good agreement in flow angle between the two data sets, the effect of the difference in incoming velocity magnitude is clear. However, when the velocity components are scaled with their corresponding mean stream wise bow velocities, the Figure 13. β=0° bow reference anemometer data two data sets agree well each other, as shown in (black vector) vs. 7 knot CFD flow visualization Figure 15(b). (white vectors; 7 knots = 141 inches/sec). Figure 14. β=15° bow reference anemometer data (black vector) vs. 7 knot CFD flow visualization (white vectors; 7 knots = 141 inches/sec). ASNE Launch & Recovery Symposium minutes are data are available for analysis. Figure 16 shows that in situ data convergence should occur within the sample window. Due to computational limitations, and based upon prior NAVAIR experience, 30 seconds of flow simulations are averaged to provide the time- averaged CFD results presented in this paper. Figure 16. In situ Reynolds shear stress running mean 〈u’w’〉 vs. time. Comparison Between In Situ Flight Deck Data with CFD Simulations Over 14 separate underway test periods, in situ velocity data has been collected over the flight deck at 22 different x,y plane positions and at various z Figure 15. Comparison of data collected on two heights above a given x,y position. different days (red and green). (a) Un-scaled Figures 17 to 19 compare the in situ data, velocity vectors; (b) Scaled velocity vectors. normalized and rescaled to the bow reference anemometer mean velocity, to the 7 knots time- Data repeatability can also be analyzed from averaged CFD data. The black vectors are the convergence of turbulence statistics, particularly the normalized and rescaled in situ data whilst the white Reynolds shear stress,13 which is the mean shear vectors and color back scale are the 7 knots CFD stress of the fluctuating components of velocity (e.g. simulation results. Multiple black vectors at a given 〈u’w’〉 where u’ is a fluctuating horizontal component sampling location show the repeatable of results and w’ is the fluctuating vertical component). Figure collected on different underway test periods. 16 shows the mean in situ Reynolds shear stress Figures 17 and 18 clearly show a large collected during numerous underway data collection recirculation zone that forms behind the aft end of the periods. Figure 16 shows that a majority of the hanger structure for both β=0° and β=15°. This sampled sets converge after approximately two recirculation zone is well captured by both the in situ minutes. A few sampled data sets, however, do not data and CFD simulation. Of note, flow vectors, show convergence until well past six minutes of black for in situ data and white for CFD, show very continuous sampling. These sets likely correspond to good correlation between in situ data and the measurements with large fluctuations in relative wind corresponding CFD simulation. Flow fields conditions and or ship motion. downstream of this large recirculation zone also A related concern for underway testing is correspond well between in situ data and CFD determination of the time necessary to collect data for simulation. a given crosswind condition. Typical data collection Figure 19 show both in situ data and CFD periods, for a given β, last approximately 25 minutes. simulation for a horizontal plane 17 inches above the After excluding data taken when the reference flight deck (z/H=0.289) for β=15°. Again, one can anemometer indicates relative wind is greater than ± observe good correlation between in situ data and 5° of the desired wind angle, typically five to 10 CFD simulation. ASNE Launch & Recovery Symposium Figure 17. β=0° centerline (y/H = 0) normalized in situ data (black vectors) vs. 7 knots time-averaged CFD data (white vectors and color scale; 7 knots = 141 inches/sec). Figure 18. β=15° starboard side (y/H = 0.6) normalized in situ data (black vectors) vs. 7 knots time-averaged CFD data (white vectors and color scale; 7 knots = 141 inches/sec). ASNE Launch & Recovery Symposium