See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/269567826

Ship-Helicopter Operating Limits Prediction Using Piloted Flight Simulation

and Time-Accurate Airwakes

Article in Journal of Aircraft · July 2012

DOI: 10.2514/1.C031525

CITATIONS READS

90 4,753

4 authors, including:

James Forrest Ieuan Owen

Prism Defence University of Liverpool
22 PUBLICATIONS 605 CITATIONS 137 PUBLICATIONS 2,596 CITATIONS

SEE PROFILE SEE PROFILE

Gareth D Padfield
University of Liverpool
188 PUBLICATIONS 2,988 CITATIONS

SEE PROFILE

All content following this page was uploaded by Ieuan Owen on 08 June 2015.

The user has requested enhancement of the downloaded file.

JOURNAL OF AIRCRAFT
Vol. 49, No. 4, July–August 2012

Ship–Helicopter Operating Limits Prediction Using Piloted

Flight Simulation and Time-Accurate Airwakes

James S. Forrest∗
Prism Defence Pty., Ltd., North Adelaide 5006, South Australia, Australia
Ieuan Owen†
University of Lincoln, Lincoln LN6 7TS, England, United Kingdom
Gareth D. Padfield‡
University of Liverpool, Liverpool L69 3BX, England, United Kingdom
and
Steven J. Hodge§
BAE Systems, Preston PR4 1AX, England, United Kingdom
DOI: 10.2514/1.C031525
This paper gives an overview of the ship–helicopter dynamic interface simulation facility at the University of
Liverpool, with an emphasis on recent improvements made through the inclusion of unsteady computational fluid
dynamics (CFD) ship airwake data. A FLIGHTLAB model of an SH-60B Seahawk helicopter has been flown in a full
motion base simulator to the deck of a Type 23 frigate and a Wave class auxiliary oiler, under the influence of
unsteady airwakes derived from CFD. Pilot workload ratings have been obtained for the deck landing task, using
both the Bedford workload rating scale and the deck interface pilot effort scale, from which fully simulated
ship–helicopter operating limits have been derived. Analysis of pilot ratings, comments, and control inputs has also
enabled both subjective and objective assessments of workloads at various wind-over-deck conditions, highlighting
the dominant aerodynamic airwake features which contribute to the difficulty of the landing task. Having access to
the underlying CFD data allows the aircraft handling qualities and pilot workload to be correlated with the
aerodynamic characteristics of the airwake and identification of the geometric features of the ship that cause them.

I. Introduction repeated landings. For example, at a 40
WOD angle the maximum
ANDING a helicopter on to the flight deck of a ship can be a allowable WOD speed would be 35 kt. Landings at conditions
L formidable task for even the most experienced of pilots. The
difficulties associated with the landing task arise due to several
outside the boundary are not normally permitted, except in extreme
circumstances. Furthermore, during operations (for example if the
environmental factors which are unique to the maritime environment. ship is part of a flotilla or is patrolling near the coast) it is not always
Sea swell leads to movement of the ship about its principal degrees of possible for a ship to turn to give the incoming pilot a favorable WOD
freedom (pitch, roll and heave), effectively making the landing spot a condition, so it is always operationally advantageous to maximize the
moving target; at the same time, air passing over the ship’s SHOL envelope.
superstructure forms large-scale turbulent eddies which pass over the In the United Kingdom, the Royal Navy (RN) requires SHOL
landing deck and perturb the aircraft during approach. This region of boundaries for each in-service ship–helicopter combination, with
disturbed flow is known as the ship’s airwake, and its severity is additional charts needed for day/night operations and different aircraft
dependent on the atmospheric wind speed, the ship’s forward speed weights [1]. The first-of-class flight trials (FOCFT) which are used to
and the relative wind-over-deck (WOD) angle. The close proximity of determine the SHOL boundaries are performed over a limited time
the helicopter to the ship during landing makes this a high-risk period, typically several weeks, and are at the mercy of the weather; as
maneuver and as both the ship motion and the ship airwake are a result, it is usually impossible to obtain test points at every desired
responsible for increasing pilot workload, it is necessary to develop combination of WOD speed and angle. This often leads to overly
ship–helicopter operating limits (SHOL) to minimize the risk of conservative SHOLs that are limited by scheduling and meteo-
accidents. rological constraints, rather than by aircraft or pilot limits. Further-
Figure 1 shows a typical SHOL diagram, with the relative more, at-sea SHOL testing is inherently hazardous due to the fact that
WOD direction (where the wind is coming from) around the pilots are operating close to their own limits, as well as those of the
circumference and the WOD speed on the radial axis. In accordance aircraft. Finally, the dedicated use of naval hardware during SHOL
with naval terminology, winds from the starboard side are termed testing ties up helicopters, ships, and personnel for significant periods
“green” and those from port termed “red.” The SHOL boundary is the of time, diverting resources from their primary operational roles.
thick black line which encloses all points that are deemed safe for For the reasons described previously, it has been suggested that
modeling and simulation of the ship–helicopter dynamic interface
(DI) may be used to augment the SHOL definition process [2].
Received 27 May 2011; revision received 8 September 2011; accepted for Potential benefits offered by DI simulation include 1) identification
publication 10 October 2011. Copyright © 2011 by the American Institute of of WOD “hot spots” before at-sea testing which can be used to inform
Aeronautics and Astronautics, Inc. All rights reserved. Copies of this paper the flight-test program; 2) the ability to assess particular WOD
may be made for personal or internal use, on condition that the copier pay the conditions which may have been missed during at-sea testing in order
$10.00 per-copy fee to the Copyright Clearance Center, Inc., 222 Rosewood to maximize the operational envelope; 3) investigation of flight deck
Drive, Danvers, MA 01923; include the code 0021-8669/12 and $10.00 in aerodynamics while new ships are still at the design stage to identify
correspondence with the CCC. potential improvements to superstructure design, landing spot
∗
Senior Fluids Engineer; james.forrest@prismdefence.com (Correspond-
ing Author).
locations and placement of equipment; 4) a greater understanding of
†
Professor of Mechanical Engineering, School of Engineering. ship airwake turbulence and the mechanisms which cause it; and 5) a
‡
Professor of Aerospace Engineering, Department of Engineering. realistic simulation environment in which to conduct pilot training
§
Senior Simulation Engineer, Flight Simulation Department. exercises.
1020
 FORREST ET AL. 1021

Vertical Motion Simulator, located at the NASA Ames Research

Center, which was developed as part of the U.S. Joint Shipboard
Helicopter Integration Process program. The DI Modeling and
Simulation System was configured such that the fidelity levels of the
various subsystems could be altered to give an overall fidelity
configuration between level A and D, with level A corresponding to
full motion base, seat shaker, and high-performance image generator
with high-fidelity visual models. A series of simulated UH-60/LHA
deck landings were performed by several pilots, with results
compared to flight-test data which had been recorded during at-sea
landings. Using the five-point deck interface pilot effort scale
(DIPES) to rate the difficulty of the deck landings it was found that,
compared with the at-sea tests, mean DIPES ratings in the simulator
were within 1 point of the corresponding ratings awarded at sea.
However, the simulated SHOL was greatly expanded in comparison
with the real SHOL, largely due to the fact that insufficient high-
workload WOD conditions were encountered during sea trials due to
benign environmental conditions. This was highlighted as further
evidence of the need for high-fidelity piloted simulation capability.
The most recent example of a piloted ship–helicopter DI
simulation environment was presented by Cox and Duncan [8], who
described the United Kingdom’s Ship–Air Interface Framework
project. Using a networked “high-level architecture” simulation with
Fig. 1 A typical SHOL diagram. time-accurate ship airwake data (some of which was contributed by
the current authors), piloted simulation flight trials were conducted
for a Merlin helicopter to RN ships including the Type 23 frigate,
A great deal of DI modeling and simulation effort has focused on Wave class auxiliary oiler (AO) and Type 45 destroyer. It was found
improving the fidelity of piloted flight simulators such that the results that moving from a steady-state CFD-based airwake database with
from simulated SHOL trials are comparable to those from at-sea statistical turbulence modeling to a time-accurate database provided
flight trials. Indeed, significant progress towards this goal has been more realistic turbulent fluctuations, with an improved match
made by naval operators in the United Kingdom, United States, and between the simulated and at-sea flight-test ratings.
Australia in recent years [3–8]. One aspect of DI modeling which has To date, the common approach for CFD-based ship–helicopter
been identified as particularly important with regards to improving flight simulators has been for the ship airwake computations to be
fidelity is the ship airwake [9]. Much of the pilot workload performed elsewhere; either in a separate department or by
experienced during landing is a direct result of disturbances caused contracting out to other organizations. There is very little evidence in
by the airwake, so it follows that accurate modeling of the airwake is the literature to suggest that DI simulation researchers have
a key step in replicating appropriate levels of workload in any DI examined simulated flight trial results in the context of the underlying
simulation. aerodynamic airwake data. Given the wealth of information held
Blackwell et al. [3] and, later, Erm [4] from the Australian Defence within the CFD datasets, it is possible that researchers are missing
Science and Technology Organisation presented one early example opportunities to gain real insight into the nature of ship airwake
of an SH60B/FFG-7 frigate simulation capability, which was based turbulence and its impact on helicopter flight dynamics. A better
on aerodynamic ship airwake data obtained from wind-tunnel tests understanding of airwake turbulence and its role in driving pilot
on a model frigate. Significant differences were found between workload during ship–helicopter operations presents opportunities
airwake velocities predicted by the model and those measured during for improving the design of ship superstructures and augmented
at-sea tests. The discrepancies were attributed to the use of a wind- flight control systems, both of which should lead to improved safety
tunnel model whose superstructure was not sufficiently similar to the and expanded operational capability.
FFG-7. In addition, simplifications within the helicopter model such This paper presents the results of a series of piloted flight
as the use of an actuator disk rotor model and the assumption that the simulation trials in which an SH-60B Sea Hawk helicopter has been
airwake velocity at the aircraft center of gravity could be applied over flown to the deck of several different ships, under the influence of
the entire aircraft were identified as deficiencies which required unsteady CFD-based ship airwakes. Both the CFD computations
attention. It was recommended that a blade element rotor model [25,26] and the flight trials [24,27] have been conducted by the
which could detect velocity gradients across the rotor would improve current authors, whereby a key part of the analysis has been returning
the effectiveness of the simulation. to the CFD data to understand and explain various phenomena
As part of a review of collaborative DI modeling activities, observed during deck landings. Pilot workload ratings have been
Wilkinson et al. [5] described the development of a ship–helicopter used to derive, as far as the authors are aware, the first fully simulated
simulation facility based at the United Kingdom’s Defence SHOL diagrams published in the literature.
Evaluation and Research Agency. The airwake module was based on The first part of the paper describes the simulator facility, before
the superposition of basic flow patterns, with turbulent fluctuations details of the CFD airwake generation and integration are given.
provided by scaled random velocity time histories. Because of the Next, results from the flight trials are presented, in terms of control
empirical nature of this airwake database, the three-dimensional activity, pilot workload ratings and SHOL diagrams. Finally, some of
components of turbulence were not correlated. the underlying CFD airwake data are shown in order to explain
From the late 1990s the improvement in computational fluid certain results from the flight trials.
dynamics (CFD) codes and availability of high-performance
computing facilities meant that ship airwake modeling activities
increasingly moved from the wind tunnel to computer simulations. II. Ship–Helicopter Simulation Approach
Several researchers have published computational studies on ship A. HELIFLIGHT-R Flight Simulation Facility
airwake aerodynamics [10–25], with the results from Polsky [17], Lee Piloted flight trials were conducted in the University of
et al. [20], Roper et al. [23] and Forrest and Owen [25] being used to Liverpool’s HELIFLIGHT-R flight simulation facility, shown in
populate look-up tables for shipboard flight simulator investigations. Fig. 2. The facility consists of a six-degree-of-freedom, full motion
Bunnell [6] and Roscoe and Thompson [7] presented details of a base simulator, driven by several Linux-based PCs running
CFD-based shipboard helicopter flight simulation facility in the FLIGHTLAB aircraft models through the PilotStation software
1022 FORREST ET AL.

Fig. 5 Image of the Wave class AO (left) and Type 23 frigate (right) side
by side.

B. Computational Fluid Dynamics Ship Airwake Model

Fig. 2 The HELIFLIGHT-R flight simulator: a) external view and Time-accurate CFD computations of the simple frigate shape
b) internal view. (SFS2) research geometry, Type 23 frigate (T23), and Wave class
AO were performed using the FLUENT finite volume solver. The
detached-eddy simulation (DES) turbulence modeling approach was
package. The simulator itself is electrically actuated and capable of used [31], enabling the explicit capture of medium to large-scale
peak accelerations up to 1:0 g in heave and 0:7 g in surge and turbulent structures in refined regions of the mesh. The CAD
sway. HELIFLIGHT-R has been used successfully in a number geometries of the ships are shown in Fig. 4, with images of the T23
of rotorcraft and fixed-wing simulation research projects [28,29]. and AO also shown for reference in Fig. 5 (it is worth noting the
During the ship–helicopter trials the simulator was configured in a relative size of the ships, which will be shown later to play a factor in
side-by-side, two-seat helicopter arrangement, with visuals provided the difficulty of the simulated deck landings). A brief summary of the
by three LCD projectors giving a 220 by 65 deg field of view. computational method is given next; a more detailed description and
Because of the 12 ft projection dome, visuals are projected on to a validation studies can be found in Forrest and Owen [25].
region close to the pilot’s feet, however, there are no discrete chin Each ship geometry was placed within an oval-shaped subdomain,
windows. in the center of a squat cylindrical domain, as shown in Fig. 6. This
A FLIGHTLAB model of a UH-60 helicopter was used during the topology allowed the WOD angle to be changed simply by
current study, with the location of its rear tail-wheel modified to make specifying the x and y components of velocity at the outer boundary,
it representative of an SH-60B Seahawk (Fig. 3). Forces and avoiding the need for remeshing. The cylinder was given a radius of
moments on the four-bladed main rotor were calculated using a blade approximately r  4:5ls and a depth of approximately d  0:75ls,
element model, with a finite-state dynamic inflow model used to where ls is ship length; this was sufficient to prevent the domain
account for distortion of air flow into the rotor disk. The tail rotor was boundaries from affecting flow in the vicinity of the ship. Hybrid
modeled as a Bailey rotor disk, described in more detail in [30]. unstructured grids were created using a combination of the Gambit
Forces on the fuselage and empennage were calculated from look-up and T-grid meshing tools: using prism elements to capture the
tables of lift, drag and moment coefficients based on local flow viscous boundary layer, tetrahedra within the oval subdomain, and
velocities.

Fig. 6 The cylindrical mesh topology used for the CFD computations.
Fig. 3 An SH-60B Seahawk hovering over a frigate landing deck. The oval subdomain is indicated by an arrow.

Fig. 4 CAD geometry of the ships used for the CFD airwake computations: a) SFS2, b) Type 23 frigate, and c) Wave class AO.
 FORREST ET AL. 1023

Table 1 List of CFD airwakes generated for the airwake database. Indicated WOD speeds
are defined at the ship anemometer height
Ship WOD angles, deg WOD speed Velocity profile
SFS2 0, 15, 30, 45, 60, 75, 90 40 kt Uniform
Type 23 frigate 0, 15, 30, 45, 60, 90, 135, 180 40, 37.5 kt Uniform, ABL
Wave class AO 0, 15, 30, 45, 60, 90, 135, 180 39.8 kt ABL

hexahedral cells throughout the rest of the domain. Size functions true wind speeds at the deck are expected to be higher for the AO than
were employed to ensure gradual growth of cells away from the ship the T23 for any given WOD condition.
and also to create a refined region over the flight deck, in order to
capture as wide a range of length scales as possible in this critical
area. Cell counts were approximately 5:8  106, 7:4  106 and 8:4  C. Airwake Integration
106 for the SFS2, T23 and AO, respectively, reflecting the increasing The outcome of each of the CFD computations was a large number
geometric complexity and physical size of the ship geometries. of data files consisting of velocity components at each grid point,
The curved cylindrical boundary was given a far-field boundary with each data file corresponding to a different time within the time
condition, with velocity specified in terms of x, y and z components. history. The data were not suitable for direct implementation into the
Runs were performed with uniform velocity profiles (as commonly HELIFLIGHT environment; partly due to formatting issues, but
seen in wind tunnels with boundary-layer removal) and also using primarily due to the fact that the CFD output was unstructured.
power law profiles to model the effects of the Earth’s atmospheric Therefore, a structured airwake grid was generated for each of the
boundary layer (ABL) [32]. For the uniform flow cases the sea SFS2, T23 and AO ships. When designing these grids it was
surface and the upper domain boundaries were given slip conditions; necessary to take into account the computational resources available
for the cases with ABL flow profiles the sea was modeled as a rough in the simulator. Because of the FLIGHTLAB implementation, it was
surface with wall functions and the upper domain boundary was necessary to hold the full airwake time history in main memory
modeled as a moving smooth wall with velocity equal to the wind at (RAM). Through testing it was found that the maximum size of
the boundary height. In all cases the ship surfaces were given a no- airwake data file which could be used without crashing the software
slip condition. was approximately 300 MB. The size of the airwake data file was a
Each case was initiated using a steady-state solver, with the function of the update frequency, airwake length (in terms of time)
resulting velocity and turbulence fields used to initialize the time- and number of data points in the airwake grid; it was therefore
accurate DES runs. After starting the unsteady solver, the solution necessary to strike a balance between these parameters to meet the
was run for a period of 15 s before any airwake data or flow statistics 300 MB constraint.
were recorded; this was found to be sufficient time to initialize the With the preceding in mind, the airwake grids shown in Fig. 7 were
solution. During the main computation phase the solver was run for a constructed for the ships; the T23 grid shown in Fig. 7a was also used
further 90 s, with the complete velocity field written to disk at each for the SFS2 flight tests as the two ships are very similar in size. Each
time-step, corresponding to a sampling rate of 80 Hz. grid used uniform 1 m spacing in each direction, giving cell counts of
Given the high cell counts and the need for time-accurate DES 18,000 and 32,062 for the T23 and AO, respectively. The AO
computations, it was necessary to perform the CFD runs in parallel required a larger domain than the T23 because it is a much bigger
on the University of Liverpool’s high-performance computing ship. The 1 m spacing was chosen as this allowed a large enough
cluster. Each run was partitioned over 32 processing cores, allowing domain to conduct the deck landing task, while still providing a
results to be obtained in approximately 8–10 days for each of the reasonable cell count. Each cell was similar in size to the blade
various cases. Typically several jobs were run concurrently to speed section length on the main rotor blades and was small enough to
up the process. provide a cell density which would allow large rotational airwake
To ensure that the airwake database contained sufficient flight structures to be resolved on the grid.
conditions to develop a full SHOL envelope, computations were The final airwake grids allowed airwake time histories of 30 and
performed at a number of WOD conditions for each ship; these are 22 s (with an update rate of 20 Hz) for the T23 and AO, respectively. It
shown in Table 1 for reference. The WOD angle is defined relative to should be noted that the AO airwakes have not been validated against
the ship’s bow and the WOD speed is defined at the ship anemometer experimental data, because no suitable data were available at the time
height. Although the ABL computations for the T23 and the AO used of these tests; however the computational method used to produce
the same velocity profile, the AO’s anemometer is higher above the them was shown in Forrest and Owen [25] to give good agreement
sea, leading to the higher WOD speed for the AO. Incidentally, the with wind-tunnel and full-scale data for the SFS2 and T23. The
AO’s flight deck is also higher above the sea than that of the T23, so fact that the CFD method had been validated for two ships gave

Fig. 7 Airwake output grids for the a) Type 23 frigate and b) Wave class AO.
1024 FORREST ET AL.

confidence that the process could be used successfully for other

geometries.
Once the airwake grids had been defined, the unstructured data
were interpolated onto them using a linear interpolation routine.
Finally, the resulting airwake data were converted into a format
suitable for export to FLIGHTLAB using MATLAB scripts. Figure 8
shows a comparison of longitudinal velocity contours before and
after interpolation on to the airwake grid, indicating that although
some spatial resolution has been lost, the overall flow pattern remains
the same.
During simulation the airwake database interacted with the
aerodynamic surfaces (e.g., blade element, fuselage) by applying a
time-varying velocity perturbation, based on the surface location and
simulation time. The nature of the airwake/helicopter coupling was
one-way; that is, the airwake perturbations affected the aircraft
aerodynamics, but the aerodynamic forces (e.g., rotor inflow,
downwash) did not, in turn, affect the airwake. Clearly, the true
aerodynamic situation is a highly complex, fully coupled flowfield. Fig. 9 Final stages of the recovery of a RN helicopter to a single-spot
frigate.
However, until CFD computations of sufficient fidelity can be
performed in real time, one-way coupling is the only feasible method
for running piloted simulations with unsteady airwakes. deck landings. The helicopter was given a mass of approximately
For each ship, the flowfield for each WOD angle was computed at 7400 kg, which is equivalent to a lightly loaded aircraft. All runs were
only one WOD speed. Polsky [17] showed that, for typical WOD performed at day, in good visibility, with ship motion equivalent to a
speeds, ship airwake data can be scaled linearly in terms of velocity sea state 4. The ship motion consisted of 5 min time histories of six-
magnitude, as the flow over bluff bodies at high Reynolds numbers is degree-of-freedom motion as recorded on board a RN aircraft carrier;
insensitive to moderate changes in Reynolds number (this was also this data had subsequently been scaled to make it more representative
tested and verified independently by the current authors using CFD). of a smaller frigate-sized vessel. The airwake grid was fixed relative
A value of approximately 40 kt WOD was chosen for all of the CFD to the ship, such that when the ship moved in pitch, roll or heave the
runs as this WOD speed was expected to be in the middle of the airwake moved with it. This behavior was deemed preferable to
SHOL envelope, meaning that the data would only have to be scaled keeping the airwake fixed relative to sea, as it was anticipated to be a
moderately in both directions. The ability to scale airwake velocities more realistic representation of the real at-sea situation. However, the
resulted in a significant reduction in the required number of CFD effect of ship motion on the airwake is not fully understood, and
computations, and provided a great deal of flexibility during certainly warrants further investigation.
simulator trials by allowing any given WOD speed to be specified. It Each deck landing sortie followed the standard RN deck landing
should be noted that the WOD conditions tested represent a approach [33] as illustrated in Fig. 9. This consists of an initial
stationary ship, with all WOD generated by the freestream wind. At approach down the red 165 deg glide slope to a stabilized hover
other combinations of wind/ship speeds, the incident wind profile alongside the ship, followed by a lateral translation to hover over the
changes due to a combination of the uniform “ship only” wind and landing spot, before finally making a vertical descent to touchdown.
the power law profile generated by the ABL. To simulate this The full evolution was split into discrete mission task elements
correctly would have required a large number of additional CFD runs (MTEs), with an additional station keeping MTE added between the
which was outside the scope of this project, although the translation and landing MTEs. The station keeping MTE required the
phenomenon warrants further research. pilot to maintain the aircraft in a hover over the landing spot for a
minimum of 10 s. This maneuver is not routinely performed during
D. Flight Trial Methodology at-sea SHOL testing, but presented a valuable opportunity in which
The deck landing trials were designed to mimic, as far as possible, to gather data for analysis of pilot control activity in airwake
the real SHOL derivation process used during FOCFT testing, with a turbulence. As the pilot was started off in a trimmed hover alongside
highly experienced former RN test pilot employed to carry out the the ship at the start of each sortie, the approach phase was not

Fig. 8 Comparison of contours of longitudinal velocity component between raw CFD data (top) and interpolated airwake grid (bottom) for the Wave
class AO at red 135 deg.
 FORREST ET AL. 1025

Fig. 10 The DIPES.

required; the three MTEs were therefore defined as 1) lateral ratings scales used were the DIPES and the Bedford workload rating
translation across the deck to a hover over the landing spot; 2) a scale (based on the Cooper–Harper handling qualities scale [34]);
minimum of 10 s station keeping, in line with the hangar roof, over these scales are shown in Figs. 10 and 11 for reference. The pilot also
the landing spot; and 3) vertical descent to touchdown. provided verbal feedback on the difficulty of the task, the realism of
Following each deck landing the pilot gave subjective ratings, the airwake turbulence and any other significant factors affecting
indicating the level of workload required during the task and the performance; these comments were recorded alongside the ratings.
ability of an average fleet pilot to repeat the maneuver safely. The two In addition to the qualitative pilot workload ratings, activity from

Fig. 11 The Bedford workload rating scale.

1026 FORREST ET AL.

Table 2 WOD conditions tested for each ship during the piloted simulation flight trials
Ship WOD angles, deg WOD speeds Velocity profile Test points
SFS2 15, 0, 15, 30, 45, 60, 90 20–50 kt Uniform 19
Type 23 frigate 15, 0, 15, 30, 45, 60, 90 25–50 kt Uniform 21
Type 23 frigate 0, 15, 30, 45, 60, 90 30–50 kt ABL 34
Wave class AO 0, 15, 30, 45, 60, 90 25–45 kt ABL 28

each of the pilot’s control inceptors was recorded for offline analysis. fidelity CFD airwakes are able to reproduce such realistic effects is
As well as giving an objective measure of how hard the pilot was encouraging. Indeed, it is hoped that simulated flight trials may lead
working, these data were also used to determine whether the pilot was to an improved understanding of these effects and provide valuable
regularly exceeding control margins during the landing task. In input to future ship superstructure design studies. Complimenting
accordance with commonly used DI test procedures (as described by this, a further research study at the University of Liverpool is using
Roscoe and Thompson in [7]), WOD conditions where the pilot experimental methods to investigate rotor response in airwake
consistently had less than 10% control margin remaining in any axis turbulence [35]; this is providing additional insight into the nature of
were assigned a DIPES rating of five. ship airwake turbulence.
As time with the test pilot was limited, the flight tests were planned
such that test points where the airwake was expected to have an
important effect were given priority. WOD conditions from red B. Fully Simulated Ship–Helicopter Operating Limits Envelopes
15 deg to green 90 deg were tested first for each ship, with red 30 to One of the key outcomes of this research project was the
90 deg completed later if time allowed. Although winds from astern development of a simulation environment in which SHOL envelopes
would normally be tested during at-sea SHOL trials, these are not could be predicted before at-sea flight trials. Therefore, SHOL
usually limited by ship airwake turbulence, so were omitted from the diagrams have been produced based on DIPES ratings given during
simulated flight trials due to a lack of pilot time. the simulated landing tasks. In converting DIPES ratings to SHOL
At each WOD angle, the pilot was given a 30 kt WOD speed as a boundaries, ratings of 1–3 are deemed acceptable, while 4–5 are
first test point; this was then scaled up or down by either 5 or 10 kt for outside the SHOL (see Fig. 10). An important consideration for the
subsequent sorties, depending on the rating given at 30 kt. Once a DIPES scale is that ratings are given based on the perceived ability of
DIPES rating of four was obtained for a given WOD angle, or if the an average fleet pilot, so although a highly capable test pilot may be
pilot deemed it was not worth adding more WOD speed due to able to safely land for a given WOD condition, a rating may be
exceedance of control margins, a limit was found and the next WOD awarded which excludes that point from the envelope if it is deemed
angle was loaded. The full test matrix achieved during the flight trials too difficult for a fleet pilot to perform on a regular basis.
is shown in Table 2. Figure 13 presents the DIPES scores and the corresponding SHOL
boundary for the two real ships. In construction of the SHOL
boundaries a lateral velocity limit of 30 kt is assumed for the
III. Piloted Flight Trial Results
helicopter. This assumes that the critical flight condition for tail rotor
A. General Observations and Subjective Pilot Comments authority is 35 kt sideways, allowing 5 kt side-slip into wind in case
Figure 12 shows typical traces of the aircraft main rotor hub of overshoot. For both ships the lateral velocity limit results in a
position plotted on a y-z plane, for several WOD angles which were contraction of the SHOL on the red side. On the green side it was not
obtained during several deck landing tasks. It can be seen that for the necessary to artificially impose such a limit as the pilot either
three cases presented the approach profiles are similar, and these are exceeded pedal control margins or deemed the task outside the
typical of the great majority of the landing tasks. The green 45 deg SHOL due to the need for excessive amounts of right pedal.
case shows slight deviation at the start of the maneuver, which is most Some major differences can be observed between the two ships, in
likely due to the aircraft being subject to strong airwake turbulence in terms of DIPES ratings and the resultant SHOL boundaries. All AO
this off-deck location for that particular WOD condition. Figure 12 ratings are at least one DIPES point higher than the corresponding
also illustrates a phenomenon which is observed at many of the red T23 ratings, and in many cases the difference is two. On a five-point
test points. For the red 15 deg case, as the aircraft approaches the port rating scale these variances are significant. Both SHOL diagrams
deck edge it gains several metres in height. This is due to an updraft exhibit a similar shape; in general the SHOL boundary contracts on
caused as air passes up and over the port side of the flight deck. Pilot
comments confirmed that this is realistic aircraft behavior for red
winds.
Pilot comments were, generally, very favorable for the unsteady
airwakes. In the majority of cases, where the aircraft was expected to
encounter turbulence, the pilot commented that he experienced
multi-axis disturbances. For example, at green 45 deg conditions the
pilot observed strong turbulence alongside the deck which reduced in
intensity as he traversed to hover over the landing spot. In contrast,
for red winds very little turbulence was noticed alongside, with an
increase in disturbances during station keeping. These observations
were in good agreement with the pilot’s experiences of at-sea deck
landings.
Certain ship airwake phenomena are commonly experienced on
approach to frigates for winds close to the bow. The so-called
pressure wall effect and the effect of the aircraft being pulled towards
the hangar have both been documented [9] and were also observed by
the test pilot during these trials. The pressure wall requires the pilot to
increase the lateral cyclic input as the aircraft approaches the deck
edge in order to break through an invisible “wall”; as the aircraft
punches through the wall, the aircraft accelerates laterally and reverse Fig. 12 Traces showing the position of the aircraft main rotor hub
lateral cyclic is required to prevent overshoot. Although the cause of during the deck landing task for the Type 23 frigate at several WOD
these phenomena is still not well understood, the fact that high- angles for 40 kt airwakes.
 FORREST ET AL. 1027

Fig. 13 DIPES ratings and SHOL diagrams for the a) Type 23 frigate and b) Wave class AO. The dashed line indicates the SHOL boundary purely from
DIPES scores; the main boundary assumes a 30 kt lateral velocity envelope limit for the SH-60B.

either side of headwind, to a minimum at beam winds. However, the been missed. This technique has been used to explain the cause of
AO boundary is 5–15 kt lower than the T23 boundary at all locations some specific phenomena observed during the simulated flight trials.
except red 60 to 90 deg; this area is fixed due to the lateral velocity
constraint.
An interesting feature is seen on the AO SHOL, where the 1. Effect of Ship Geometry
boundary is seen to expand between red 45 to 60 deg. At these WOD As discussed in the previous section, a difference in workload
conditions lower workload is experienced due to the fact that the ratings was observed between the T23 and AO during the flight trials.
aircraft is out of airwake turbulence for the majority of the deck At the 30 kt green 30 deg WOD condition, a DIPES rating of one was
landing. The windward hangar edge shear layer and turbulence shed given for the T23 and three was awarded for the AO. Bedford
from the superstructure are both swept away over the flight deck at an workload ratings of four and three, and seven and five were given to
angle that has less impact on the helicopter than for winds closer to the lateral translation and station keeping MTEs for the T23 and AO,
the bow. At these relatively low WOD speeds the tail rotor power respectively. Figure 14 shows the time history of airwake velocity
requirements are not limiting, so the points are within the SHOL. A components recorded at a location on the starboard side of the rotor
similar feature is seen at red 45 deg for the T23, but the point is disk (defined at 75% span, azimuth  90 deg) during the complete
prevented from being included due to the fact that the 45 kt WOD deck landing maneuver for the 30 kt green 30 deg test point, for both
speed is outside the aircraft’s lateral velocity envelope. ships. The diagram is split into segments, each representing the
Although a methodology for deriving SHOL diagrams from CFD- approximate start and end time of each of the MTEs, with an extra
based simulation has been presented, it has not been possible to MTE defined at the start of the sortie as an initial trim and
compare the SHOL diagrams in Fig. 13 to any real SHOL diagrams stabilization phase before translation. It can be seen that the majority
for validation. Besides the fact that no SHOLs exist for the ship– of airwake turbulence is encountered during the translation MTE,
helicopter combinations examined, SHOL charts for similar aircraft which is as expected, due to the aircraft passing through areas of
interfacing to these ships are working military documents and, as highest turbulence during this phase. Figure 15 shows contours of
such, are restricted due to their sensitive nature. Although the authors turbulence intensity plotted at hangar height for both ships to
are aware of a United Kingdom Ministry of Defence research illustrate this point.
programme comparing simulated SHOLs with at-sea data [8], there Although the velocity traces shown in Fig. 14 are somewhat
exists a need for more generic ship–helicopter flight-test data for the chaotic, the clear trend is that the AO disturbances are larger in
wider research community. magnitude than those seen from the T23. This applies to all three
velocity components, with only the lateral component of the T23
airwake exhibiting perturbations close in magnitude to the AO
airwake. It is also evident that the AO velocity fluctuations are more
C. Ship Aerodynamics and Aircraft Handling Qualities persistent during translation and station keeping than the T23. The
A valuable benefit of having team members both computing the CFD results shown in Fig. 15 suggest that this is most likely due to
CFD airwakes and conducting the flight trials is that in-depth the fact that the wider superstructure of the AO prevents the aircraft
knowledge of the ship aerodynamics can be used to gain insight into from encountering clear, freestream air until much later in the
the aircraft handling qualities. Furthermore, the ability to return to the maneuver. It can be seen that, in each axis, levels of turbulence
CFD results and interrogate the flow data using postprocessing tools intensity are higher and areas of high turbulence are larger for the AO
allows the identification of flow features which may otherwise have compared to the T23.
1028 FORREST ET AL.

In commenting on the differences between airwakes for the two

ships, the pilot remarked that the AO airwake felt more “wallowy”
than the T23, with lower frequency disturbances that were larger in
magnitude. This required less frequent, but more severe corrections
in order to maintain control of the aircraft. To investigate this further,
the CFD airwake data were postprocessed by deriving power spectral
density (PSD) plots from velocity time histories which were sampled
at a point directly over the port side deck edge of both ships. The
resulting data are plotted in Fig. 16, showing the frequency content of
the airwakes. In each axis, and over the whole frequency range, the
AO airwake contains more power than the T23 airwake, with
significantly more in the vertical axis. This further confirms the
discussion in the preceding paragraphs regarding the relative
magnitude of airwake disturbances. Where peaks in the PSD plots
exist for the two airwakes, the AO peak occurs at a lower frequency
than the T23. For example, in the longitudinal axis a peak in the AO
airwake is seen at approximately 0.3 Hz, with the T23 peak closer to
0.5 Hz. This is consistent with the pilot’s comments regarding lower
frequency disturbances during AO deck landings.
The relative size of the ships as shown in Fig. 5 can be related to the
Fig. 14 Airwake velocity components recorded at a location on the frequency shift observed in Fig. 16. It can be seen from Fig. 5 that the
starboard side of the rotor disk (defined at 75% span, azimuth AO is significantly larger than the T23 and will therefore shed

90 deg) during the deck landing task for the Type 23 frigate and vortical structures which are larger than those shed from the T23 and,
Wave class AO (green 30 deg, 30 kt). Time histories are split into four consistent with Strouhal scaling, will be shed less frequently.
MTEs: a) stabilized hover alongside ship, b) lateral translation across Similarly, for a given strength, larger vortices will induce higher
deck, c) station keeping over landing spot, and d) vertical landing.
velocities at their extremities. Both of these factors combine to
suggest that, for ships with a conventional configuration (landing
deck to stern behind a vertical hangar door), the airwake becomes
more difficult to operate in with increasing size; although clearly

Fig. 15 Contours of turbulence intensity plotted at hangar height for the Type 23 frigate (left) and the Wave class AO (right) at green 30 deg.
Longitudinal (a), lateral (b), and vertical (c) components are plotted, normalized by velocity magnitude at the anemometer height. The approximate
location of the rotor disk during the station keeping task is denoted by the circle.
 FORREST ET AL. 1029

Fig. 16 PSD plots of Type 23 frigate and Wave class AO airwake velocity components at a location over the port side deck edge, in line with the landing
spot, at hangar height (green 30 deg). Longitudinal (a), lateral (b), and vertical (c) components are shown.

small ships have their own operational limitations in terms of landing turbulent airwake features drive pilot workload and control strategy
deck size and ship motion. at different WOD angles. For the purposes of this study it was
A further consideration in comparing workload ratings for convenient to choose two WOD conditions which were given
different ships is the difference in visual cues available to the pilot. It significantly different workload scores by the pilot, as a contrast in
was observed during the trials that the test pilot relied heavily on the workload rating should also indicate a dissimilar airwake structure.
natural horizon to maintain the aircraft in a level hover; the technique Therefore, it was decided that the T23 headwind and green 45 deg
was described as “looking through the ship.” Behind the T23 hangar WOD angles would be analyzed, both for WOD speeds of 40 kt. For
the pilot still had a reasonable view of the horizon, despite some the station keeping task the test pilot gave the headwind case a
superstructure partially blocking his view. However, the larger AO Bedford workload rating of 4, classified as “Insufficient spare
hangar and upwind superstructure almost totally blocked out the capacity for easy attention to additional tasks.” The green 45 deg case
forward view of the horizon, leaving only a small region to the was given a rating of 6, classified as “Little spare capacity; level of
extreme port and starboard sides of his field of vision. It is possible effort allows little attention to additional tasks.” It is instructive to
that this disparity between visual cues may have contributed to the look at a visualization of the flowfields at the chosen WOD angles in
higher workload experienced during AO deck landings. This serves order to determine what may be causing the differences in workload.
to highlight the complex interaction between cue environment, Figure 17 shows contours of turbulence intensity on a plane at 6 m
aircraft disturbances and pilot response which contribute to the above the flight deck. A marked difference in flowfields can be seen.
difficulty of the shipboard landing task. For the headwind, turbulence is moderate over the whole of the flight
deck, whereas the green 45 deg case shows a flight deck bisected
by a region of high turbulence which is shed from the windward
2. Effect of Wind-over-Deck Angle hangar edge. Although at green 45 deg much of the velocity field
It is clear from the DIPES ratings diagrams shown in Fig. 13 that experienced by the rotor disk is at freestream conditions, the region
winds either side of the headwind tend to cause increased workload. between 180 and 330 deg azimuth (counter-clockwise from the rear
This is mainly due to the effects of shear layers and vortices which of the disk) is subjected to levels of airwake turbulence with peak
strengthen as the wind angle becomes more oblique; these flow vertical velocity magnitudes up to 75% higher than the headwind
features being the primary mechanisms for airwake turbulence case.
generation. It was anticipated that results from the deck landing trials Figure 18 largely confirms the pilot’s assertion that he was
would be able to provide a better understanding of how these working harder during the green 45 deg landing, showing PSD plots

Fig. 17 Contours of turbulence intensity plotted at a height of 6 m above the flight deck. The approximate location of the rotor disk during the station
keeping task is denoted by the circle: a) headwind and b) green 45 deg.
1030 FORREST ET AL.

Fig. 18 PSD plots of pilot control activity during the deck landing task for the 40 kt headwind and green 45 deg cases (PSD values have been normalized
by 1  103 for clarity). Diagrams show, from left to right, lateral cyclic, longitudinal cyclic, collective, and pedal.

for each of the control axes derived from time histories of pilot contrast, the right side of Fig. 17 shows high vertical velocities,
control inputs. In all axes except the longitudinal cyclic the PSD both positive and negative, on the rotor disk between approximately
levels are significantly higher for the green 45 deg case than for the 210 and 290 deg azimuth; the most severe gradient occurs at
headwind. It can be seen that all activity is to the lower end of the approximately 270 deg. When animated, these large-amplitude
typical 0.2–2 Hz pilot closed-loop response range. The very large perturbations are seen to convect across the disk at a lower frequency
low-frequency peak seen for the pedal activity in the green 45 deg than the lower-amplitude headwind turbulence. It appears that the
case is likely to be exaggerated due to the gradual application of left frequency and magnitude of the green 45 deg turbulence shed from
pedal during the lateral translation into freestream conditions; when the starboard hangar edge are the main drivers for workload at this
data from just the station keeping MTE is isolated and plotted a condition. The pilot commented that it was easier to make constant,
smaller peak is observed, but this is still significantly higher in small adjustments to the controls than to be hit suddenly with large
magnitude than the headwind data. For the green 45 deg cases the disturbances that required more extreme control inputs, hence the
pilot consistently reported pedal activity as being a dominant factor higher ratings for the green 45 deg case; again, these comments
in terms of workload. For the cyclic, similar levels of activity are seen concur with the underlying CFD data.
in both axes during the green 45 deg test point, with slightly more in
the lateral channel. In contrast, for the headwind case there is very
little activity in the lateral axis, with a marked increase in longitudinal
activity, comparable in magnitude to the green 45 deg case. IV. Conclusions
Although the contours of turbulence intensity shown in Fig. 17 A ship–helicopter DI flight simulation facility based at the
give an indication of which regions are likely to experience large University of Liverpool has been described, which allows simulated
fluctuations, they give no indication about the instantaneous deck landings to be conducted under the influence of realistic
perturbations. Figure 19 shows contours of instantaneous vertical unsteady ship airwakes. The airwake perturbations are provided by
velocity at an estimated rotor disk location during the station keeping aerodynamic look-up tables which have been populated by offline
task, for both WOD conditions. The vertical axis was chosen as it is CFD computations of the air flow over several ships at various WOD
well known that the vertical velocity component is important in angles. Simulated flight trials have been conducted to mimic, as far as
causing rotor disk disturbances, by changing the induced velocity possible, at-sea FOCFT testing; this has so far been limited to the
distribution and hence the effective angle of attack of the blades [36]. deck landing part of the ship–helicopter interface. DIPES ratings
The headwind case on the left side shows that there are moderate taken from a highly experienced test pilot have been used to
spatial velocity variations and pilot comments suggest that some construct, the first known published, fully simulated SHOL
airwake turbulence at the headwind condition was felt, but it was at a diagrams.
magnitude which was manageable, hence the Bedford score of 4. In Pilot comments regarding the simulated flying experience were,
generally, very good. The pilot reported feeling the effects of
turbulence in locations where it was expected. For example, at a
green 45 deg WOD condition, turbulence was encountered to the port
side of the flight deck in the lee of the superstructure, but not to the
starboard side of the deck. Other phenomena, such as the so-called
pressure wall and updraft over the port side deck edge for red winds,
were also observed.
The ability to return to the CFD and use the underlying
aerodynamic data as a tool for understanding certain flight-test
results was shown to be a key aspect of this methodology. Differences
in airwake severity were observed between the T23 and AO, leading
to a more restrictive SHOL for the AO. Aerodynamic data suggested
that this was due to the relative size of the ships leading to airwakes
with differing frequency content. Similarly, the increase in reported
workload as winds move from ahead around to green 45 deg were
attributed to the difference in the structure of the airwake and the
increasing strength of turbulent structures as winds become more
oblique.
The simulation methodology described in this paper was
conducted using virtual models (ship CAD data, aircraft flight
Fig. 19 Contours of instantaneous vertical velocity plotted at the model) throughout, lending itself well to virtual prototyping. In
approximate location of the rotor disk during the station keeping task, particular, with ships still at the design stage, this type of DI
for the 40 kt headwind (left) and green 45 deg (right) cases. simulation approach could be a valuable tool to assess the structure
 FORREST ET AL. 1031

and severity of the expected ship airwakes, and their potential impact Simulation of Ship Airwake,” Computers and Fluids, Vol. 29, No. 4,
on ship–helicopter operations. 2000, pp. 451–465.
doi:10.1016/S0045-7930(99)00033-X
[16] Sharma, A., and Long, L. N., “Airwake Simulations on an LPD-17
Acknowledgments Ship,” AIAA Paper 2001-2589, 2001.
The authors would like to acknowledge financial support from [17] Polsky, S. A., “A Computational Study of Unsteady Ship Airwake,”
40th AIAA Aerospace Sciences Meeting and Exhibit, Reno, NV, 14–
Westland Helicopters, Defence Science and Technology Laboratory, 17 Jan. 2002.
and the ongoing support from Engineering and Physical Sciences [18] Polsky, S. A., “CFD Prediction of Airwake Flowfields for Ships
Research Council. ANSYS, Inc., have also provided valuable Experiencing Beam Winds,” 21st AIAA Applied Aerodynamics
assistance. The first author is employed by the Virtual Engineering Conference and Exhibit, Orlando, FL, 23–26 June 2003.
Centre, which is a University of Liverpool project partially funded by [19] Syms, G. F., “Numerical Simulation of Frigate Airwakes,” Interna-
North-West Development Agency and European Regional Develop- tional Journal of Computational Fluid Dynamics, Vol. 18, No. 2, 2004,
ment Fund located at the Daresbury Laboratory of the Science and pp. 199–207.
Technology Facilities Council. The authors would like to thank the doi:10.1080/10618560310001634159
test pilot, Andy Berryman, for his professionalism and guidance. [20] Lee, D., Sezer-Uzol, N., Horn, J. F., and Long, L. N., “Simulation of
Helicopter Shipboard Launch and Recovery with Time-Accurate
Many hours of help and support have also been given by Philip Airwakes,” Journal of Aircraft, Vol. 42, No. 2, 2005, pp. 448–461.
Perfect, Neil Cameron, and Mark White from the University of doi:10.2514/1.6786
Liverpool’s Flight Science and Technology research group. Cliff [21] Lee, D., and Horn, J. F., “Simulation of Pilot Workload for a Helicopter
Addison helped immensely with the computing cluster setup and Operating in a Turbulent Ship Airwake,” Proceedings of the Institution
maintenance. of Mechanical Engineers, Part G: Journal of Aerospace Engineering,
Vol. 219, No. 5, 2005, pp. 445–458.
doi:10.1243/095441005X30298
References [22] Sezer-Uzol, N., Sharma, A., and Long, L. N., “Computational Fluid
[1] Carico, D., Fang, R., Finch, R. S., Geyer, W. P., Jr., Krijns, H. W., and Dynamics Simulations of Ship Airwake,” Proceedings of the Institution
Long, K., “Helicopter/Ship Qualification Testing,” RTO/NATO, of Mechanical Engineers, Part G: Journal of Aerospace Engineering,
TR RTO-AG-300-V22, 2003. Vol. 219, No. 5, 2005, pp. 369–392.
[2] Advani, S. K., and Wilkinson, C. H., “Dynamic Interface Modelling and doi:10.1243/095441005X30306
Simulation: A Unique Challenge,” Royal Aeronautical Society [23] Roper, D. M., Owen, I., Padfield, G. D., and Hodge, S. J., “Integrating
Conference on Helicopter Flight Simulation, London, Nov. 2001. CFD and Piloted Simulation to Quantify Ship-Helicopter Operating
[3] Blackwell, J., Arney, A. M., Gilbert, N. E., and Truong, T. T., Limits,” Aeronautical Journal, Vol. 110, No. 1109, 2006, pp. 419–428.
“Modelling the Helicopter/Ship Dynamic Interface for the Seahawk/ [24] Forrest, J. S., Hodge, S. J., Owen, I., and Padfield, G. D., “An
FFG-7,” Australian Aeronautical Conference, Proceedings No. 93/6, Investigation of Ship Airwake Phenomena Using Time-Accurate CFD
Institution of Engineers, Australia, Sept. 1993. and Piloted Helicopter Flight Simulation,” 34th European Rotorcraft
[4] Erm, L. P., “A Preliminary Study of the Airwake Used in an Existing Forum, Liverpool, England, U.K., 16–19 Sept. 2008.
SH-60B/FFG-7 Helicopter/Ship Simulation Program,” DSTO, [25] Forrest, J. S., and Owen, I., “An Investigation of Ship Airwakes Using
TR DSTO-TR-0015, 1994. Detached-Eddy Simulation,” Computers and Fluids, Vol. 39, No. 4,
[5] Wilkinson, C., Zan, S., Gilbert, N., and Funk, J., “Modelling and 2010, pp. 656–673.
Simulation of Ship Air Wakes for Helicopter Operations: A doi:10.1016/j.compfluid.2009.11.002
Collaborative Venture,” AGARD Symposium on Fluid Dynamics, [26] Forrest, J. S., Owen, I., Padfield, G. D., and Hodge, S. J., “Detached-
Problems of Vehicles Operating Near or in the Air–Sea Interface, Eddy Simulation of Ship Airwakes for Piloted Helicopter Flight
Amsterdam, 5–8 Oct. 1998. Simulation,” First International Aerospace CFD Conference, Paris, 18–
[6] Bunnell, J. W., “An Integrated Time-Varying Airwake in a UH-60 Black 19 June 2007.
Hawk Shipboard Landing Simulation,” AIAA Modeling and [27] Forrest, J. S., Hodge, S. J., Owen, I., and Padfield, G. D., “Towards Fully
Simulation Technologies Conference and Exhibit, Montreal, 6– Simulated Ship-Helicopter Operating Limits: The Importance of Ship
9 Aug. 2001. Airwake Fidelity,” American Helicopter Society 64th Annual Forum,
[7] Roscoe, M. F., and Thompson, J. H., “JSHIP’s Dynamic Interface Vol. 1, Montreal, 29 April–1 May 2008, pp. 339–351.
Modeling and Simulation System: A Simulation of the UH-60A [28] White, M., Perfect, P., Padfield, G., Gubbels, A., and Berryman, A.,
Helicopter/LHA Shipboard Environment Task,” American Helicopter “Acceptance Testing of a Rotorcraft Flight Simulator for Research and
Society 59th Annual Forum, Phoenix, AZ, 6–8 May 2003. Teaching: The Importance of Unified Metrics,” 35th European
[8] Cox, I. G., and Duncan, J. M., “Optimising the Ship/Air Vehicle Rotorcraft Forum, Hamburg, Germany, 21–25 Sept. 2009.
Interface via Simulation,” Royal Institution of Naval Architects [29] Perfect, P., White, M., Padfield, G., Gubbels, A., and Berryman, A.,
International Conference: Warship 2009: Airpower at Sea, London, “Predicting Simulator Fidelity through ADS-33 Metrics,” 36th
17–18 June 2009, pp. 53–62. European Rotorcraft Forum, Paris, 7–9 Sept. 2010.
[9] Zan, S. J., “On Aerodynamic Modelling and Simulation of the Dynamic [30] Manimala, B., Walker, D., and Padfield, G., “Rotorcraft Simulation
Interface,” Proceedings of the Institution of Mechanical Engineers, Modelling and Validation for Control Design and Load Prediction,”
Part G: Journal of Aerospace Engineering, Vol. 219, No. 5, 2005, 31st European Rotorcraft Forum, Florence, Italy, 13–15 Sept. 2005.
pp. 393–410. [31] Spalart, P. R., Jou, W. H., Strelets, M., and Allmaras, S. R., “Comments
doi:10.1243/095441005X30315 on the Feasibility of LES for Wings, and on a Hybrid RANS/LES
[10] Tai, T. C., and Carico, D., “Simulation of DD-963 Ship Airwake by Approach,” Advances in DNS/LES, Greyden, 1997, pp. 137–147.
Navier–Stokes Method,” Journal of Aircraft, Vol. 32, No. 6, 1995, [32] Counihan, J., “Adiabatic Atmospheric Boundary Layers: A Review and
pp. 1399–1401. Analysis of Data from the Period 1880–1972,” Atmospheric
doi:10.2514/3.46892 Environment, Vol. 9, No. 10, 1975, pp. 871–905.
[11] Landsberg, A. M., Young, T. R., Jr, and Boris, J. P., “Analysis of the doi:10.1016/0004-6981(75)90088-8
Nonlinear Coupling Effects of a Helicopter Downwash with an [33] Fang, R., Krijns, H. W., and Finch, R. S., “Dutch/British Clearance
Unsteady Ship Airwake,” 33rd AIAA Aerospace Sciences Meeting and Process,” RTO AGARDograph 300: Helicopter/Ship Qualification
Exhibit, Reno, NV, 9–12 Jan. 1995. Testing, edited by G. D. Carico, R. Fang, R. Finch, W. P. Geyer, Jr., H.
[12] Landsberg, A. M., Sandberg, W. C., Young, T. R., and Boris, J. P., W. Krijns, and K. Long, Vol. 22, Flight Test Techniques Series, NATO
“DDG-51 Flt-IIA Airwake Study Part 2: Hangar Interior Flow,” Naval Research and Technology Organization, 2003.
Research Lab., TR NRL/MR/6410-96-7898, 1996. [34] Roscoe, A. H., and Ellis, G. A., “A Subjective Ratings Scale for
[13] Liu, J., Long, L. N., and Modi, A., “Higher Order Accurate Solutions of Assessing Pilot Workload in Flight: A Decade of Practical Use,” RAE,
Ship Airwake Flow Fields Using Parallel Computer,” American TR TR90019, 1990.
Helicopter Society 54th Annual Forum, Washington, D.C., 20– [35] Kääriä, C. H., Wang, Y., Curran, J., Forrest, J., and Owen, I., “AirDyn:
22 May 1998, pp. 58–70. An Airwake Dynamometer for Measuring the Impact of Ship Geometry
[14] Polsky, S. A., and Bruner, C. W. S., “Time-Accurate Computational on Helicopter Operations,” 36th European Rotorcraft Forum, Paris, 7–
Simulations of an LHA Ship Airwake,” 18th AIAA Applied 9 Sept. 2010.
Aerodynamics Conference, Denver, CO, 14–17 Aug. 2000. [36] Padfield, G. D., Helicopter Flight Dynamics, 2nd ed., Blackwell
[15] Reddy, K. R., Toffoletto, R., and Jones, K. R. W., “Numerical Science, Boston, 2007.

View publication stats