CONTROL SURFACE AND ACTUATOR DESIGN FOR A LOW DRAG,

LAMINAR FLOW AUV

S L Merry*,M J Large**,T J Whitten**,M R Wilkinson**and R J Ba.bb***
*Instituteof Sound and Vibration Research
University of Southampton
Southampton SO17 lBJ, UK
**Department of Aeronautics and Astronautics
University of Southampton
Southampton SO17 IBJ, UK
***Ocean Technology Division
Southampton OceanographyCentre
Empress Dock
Southampton SO14 3ZH, UK

Abstract oceanographic data collection [l]. The potential benefits of

this concept are such that the Na,tural Environmental
The design process for the control surfaces and actuators of Research Council (NERC) in the UK has funded a
the long distance AUV DOLPHIN (UK "Autosub" project) is Community Research Project for telchriological development
described. The main design criteria are low drag, reliability, of AUVs, named Autosub. The lpralject is hosted by the
robustness and energy-efficiency. The aerodynamic Ocean Technology Division of thle new Southampton
characteristics of four NACA foil sections and various Oceanography Centre (formerly pare of the Institute of
stabiliser/moveable fin geometries are compared, in order to Oceanographic Sciences, Deacon La.boratory).
reach the final design, which incorporates a NACA 66-009 wing
section with a partial span trailing edge flap. The actuator is One of the long-term goals of the project is to produce a
located in a wing-tip pod, which also houses the sensors for vehicle which will track across the: ocean basins, sampling
oceanographic data logging. Preliminary analysis of wind continuously. This Deep Ocean Long Path Hydrographic
tunnel tests on a full-scale model of the control surfaces confirms Instrument (DOLPHIN) will carry out profiles from the
that the lift force required to achieve the specified surface to the seabed, surfacing evlery 30 km to fix its
manoeuvrability for DOLPHIN is easily generated. Measured position by GPS satellite and to retulm data to shore stations.
drag data is lower than expected and this requires further
analysis. In order to meet the specified trims-oceanic range of
7000 km, within the limitations of currently available power
1. INTRODUCTION storage systems, an energy-efficient hull form is required for
the DOLPHIN. Research in the 1960s and 1970s identified
Widespread public concern about the increasing the inherently low drag charactenistics of natural laminar flow
greenhouse effect of the worlds atmosphere and consequent shapes [2, 31 and model tests on a laminar flow design by
global warming has led to a requirement for improved Huggins and Packwood [4] conffimied a drag coefficient of
climate forecasting. In order to meet that requirement, 0.0061 which is 75% lower than the drag of an equivalent
temporal and spatial monitoring of the worlds oceans on a torpedo-shaped hull-form at the same Reynolds No., in a
three-dimensional, global scale is needed, to validate controlled, low turbulence environment.
mathematical models of the ocean circulations. These
models can then be used at higher confidence levels to predict The final design of the DOLPHIN hull based on the
climate change. Huggins and Packwood shape is shown in Figure 1. The
overall length is 6.0 m, with a displaced volume of 4.4 m3.
Autonomous underwater vehicles (AUVs) has been
proposed as cost-effective and efficient platforms for

0-7803-3 185-0/96$5.0001996 IEEE

57
 The path profile described above, which necessitates only
gentle alterations from a straight-line course, defines the
48 m
specifications for the control surfaces of providing a
minimum turn radius of 100 m.

The main design consideration for the control system of

DOLPHIN is energy-efficiency. The selected foil geometry
must have low drag; the actuator mechanism and control
strategy must be designed to minimise energy consumption.
I I Other mechanical factors to be addressed include operation at
POIh1T OF
I 3333 ll IN F L E Y I Ill high pressure (maximum operating depth 6000 m), long term
reliability and resistance to mishandling during launch and
Figure 1. Hull Form of the AUV DOLPHIN recovery.

Laminar flow will be maintained for 70% of the length at the The control surface geometry should also incorporate a
cruise velocity of 2.5 m/s. Free ascent trials of a half-scale location for oceanographic sensors, such as a CTD
model in the open ocean indicated a 50% reduction in drag as measurement device.
compared with a torpedo-shaped hull of the same internal
volume, under realistic operating conditions [5]. 3. CONTROL SURFACE DESIGN

Various sub-systems of the vehicle, such as power storage, 3.1 Selection of the Aerofoil Section
buoyancy control, mission management and propulsion, are
currently under development. The present paper describes The aerodynamic characteristics of four symmetrical
the design and testing of control surfaces and actuators for NACA foils, the basic 0006 and 0009 and the laminar flow
DOLPHIN. 66-006 and 66-009, were compared in order to select the
most efficient section for the control surfaces [6]. A
2. DESIGNBRIEF graphical comparison is presented in Figure 3 (stall angle
versus Reynolds No.), Figure 4 (minimum drag coefficient
For its mission of oceanographic data logging through the versus Reynolds No.), and Figure 5 (maximum lift coefficient
water column, DOLPHIN will follow a saw-tooth path profile versus Reynolds No.).
as it traverses the ocean [Figure 21. The submersible will
descend from the surface at an angle of about 45", 14 - ~

continuously monitoring data such as temperature and

salinity as a function of depth. As it nears the ocean floor, 12 -
the vehicle attitude will change, in order to ascend vertically -
al
to the surface, where the data will be off-loaded via satellite. E 10-
Before repeating the cycle, a GPS position fix will be E
E.
obtained. g 8-
2
-m 6 -
Ci
C
.-c
U 4-
v)

2-

0-
0 xxxxxM40000006000000800000010M)o
0
Reynolds No.

Figure 2. Trans-Oceanic Path Profile of the AUV DOLPHIN Figure 3. Stall Angle versus Reynolds No. for Candidate Foil
Sections

58
 ~I Cdmin NACA ooo6
Cdmin NACA ooo9
--t Cdmin NACA 66-006
++Cdmin NACA 66-009
~
design specifications of long-term reliability and resistance to
mishandling. The final digit of a NACA designation gives
the maximum foil thickness as a percentage of chord. 6%
chord thickness was considered structurally fragile for the
envisaged application and therefore:the 0006 and 66-006 foils
appear unlikely candidates. Haiwevei: they were retained for
subsequent stages of the design process and analysis.

3.2 Control Surface Configuration

The vehicle's fins should incorporate both fixed stabilisers

and moveable control surfaces. The two configurations
shown in Figure 6 were considered: an all-moving tip
outboard of the fixed stabiliser and a trailing edge flap.
------c
0.002
200000040000006000000800000010000000
Reynolds No.

Figure 4. Minimum Drag Coefficient versus Reynolds No.

1 1 w
for Candidate Foil Sections
_--- -U---
--t Clmax NACA 0006 TPAILING EDGE FLAP ALL-MOVIING TIP
2 + Clmax NACA ooo9

1.8
-AF- Clmax NACA 66-CK)6 Figure 6. Candidate Fin Configurations
--Jt Clmax NACA f35-03
I C
Using similar areas for the moveable control surfaces, the lift:
drag ratio for zero incidence on thle stabiliser is plotted as a
function of tip angle for the moving tip configuration in
Figure 7 and as a function of flap angle for the trailing edge
flap configuration in Figure 8. The results for all four
candidate foil sections are presented.

Figure 5. Maximum Lift Coefficient versus Reynolds No. for

Candidate Foil Sections

The 0009 foil is clearly superior in terms of stall angle and

maximum lift coefficient: however, it also exhibits the
highest minimum drag coefficient. O - " = t ' D W D " ~

all-moving tip angle (degrees)

Although low drag is an important criterion, structural
robustness must also be considered, in order to meet the
Figure 7. Lift : Drag Ratio as a Funlction of Tip Angle for
Moving Tip Configuration

59
 18
16
14
;12
10
i
5 j -4-( LlD) NACA 66-006
Z 8 1 -+-(LID) NACA 66 0@9 1
6
4 I -+(LID) NACA 0006 I
2 j -X-_(LID)
_ NACA 0009
.--.__ - t
~

0 Y--t--'--+1--tti--t-tri-c-l
o r d q a m
-o 2 :
0 0? 04 06 08 1
Flap Angle (degrees)
Flap Chord Ratio fcdc)
Figure 8. Lift : Drag Ratio as a Function of Flap Angle for Figure 9. Effectiveness Ratio versus Flap : Chord Ratio for
Trailing Edge Flap Configuration Different Trailing Edge Flap Configurations

In each case, the trailing edge flap gives a more favourable Figure 9 shows the relationship between effectiveness ratio
maximum lift: drag ratio, which is sustained over a broader and flap chord: wing chord ratio ( c j /c) for three geometric
range of flap angles. This is the configuration which was configurations: (a) full-span trailing edge flaps, (b) half-span
selected for the final design. trailing edge flaps, centrally located on the wing span,
(c) half-span trailing edge flaps located at the outboard sector
The geometry of the trailing edge flap was specified after of the wing span.
consideration of the data presented in Figure 9 [7]. The
effectiveness ratio of a wing flap is defined as the lift According to Figure 9 full-span flaps are the most
differential per degree angle of attack of the flap, in effective, but in the DOLPHIN layout, this would disturb the
comparison to that of the surface as a whole, i.e. propeller approach flow [see Figure 11. Therefore partial-
span flaps were selected, to be located outboard of the
propeller inflow region.

where a = wing angle of attack, 6 = flap deflection and

Cl = lift coefficient.

L = 6.0
I--
I

Figure 10. General Arrangement and Dimensions of the Control Surfaces on DOLPHIN

60
3.3 Final Design of the Control Surfaces 4. ACTUATOR DIESIGN

A parametric study of the effect of tail area and tail fin 4.1 Design Considerations
location on the stability of the DOLPHIN vehicle during
steady cruise, using the experimental data published in The design process for the control surface actuators
Reference 4, was used to determine the final size and position focused on compactness and energy-efficiency. The limited
of the tail fins. The fins were designed with a swept leading power and space availability in a long distance AUV drives
edge, to reduce the likelihood of entanglement in weed or the requirement to minimise both thle energy losses in the
debris. servo-mechanism and the space occupied within the hull.
The main decisions therefore concerned the location of the
The tail force required to turn a vertical circle of radius servo-motor and the linkage between the motor and the shaft
lOOm was calculated to be 200N, or lOON per fin. The of the control flap.
dimensions of the trailing edge flap, as presented in Figure
10, were calculated from this value; the flap:wing span ratio The initial concept of locating the servo-motor inside the
is 0.7 and the average flap:wing chord ratio is 0.38. The DOLPHIN hull was compared with the idea of fitting a pod
general arrangement and location of the fins on the prototype on the wing tip, to house both the motor and the CTD sensor.
vehicle are also presented in Figure 10.

The final choice of wing section was limited to the laminar

flow foils (66-006 and 66-009),because of their superior drag
characteristics (Figure 4). Absolute values of the drag of the
control surfaces, operating at 22 m/s in air (which gives a
Reynolds No equivalent to the full size vehicle operating at
2.5 m/s in seawater) is presented in Figure 11 for these foils.

INT ERNAL A C T IJ A TOR

C L E A N AERRD i N A M I C S + -
POD
m
MOUNTEri A C T U A T O R

ADDITIONAL PUD DRAG

I N C L O S E D SERVOMECHANICM + ZEPVOMECHANISM EXPEYED

~

FREE FROM SNAGGING + 3 0 S L I B I L I T Y OF SNAGGING

LIMITED INTERNAL VOLUME t CAN BE SIZED AS REQUIRED

>RUBLEM OF C T I i L O C A T I O N + ‘,PALE I N POPS FOR C T D

GPEATEP C O M P L E X I T Y t I I M A L L I1’1 K GG I b1G

M IN

DIFFiCaJLT ACCESS + E A S E OF MhINTENANCE

Figure 12. Comparison of Actuator Locations in the Hull

0 4- 4 -
- +-i--c-i-+t--i (Internal) and Pod Mounted (External)
o w a c o m g ~ g

Flap Angle (degrees) The two arrangements are presented schematically in Figure
12, with the relative advantages for each system. The elegant
Figure 11. Drag Comparison for the Candidate Laminar solution of a combined location for instrumentation and the
Flow Aerofoil Sections [Full scale, 22 m . d in air] servo-motor in the tip pod, which permits a simple linkage
mechanism and ease of access for maintenance, while
releasing valuable space within the vehicle hull, was finally
Although the drag of the thicker, 66-009 section is selected. Reference 8 indicates that the additional drag will
marginally higher, the drag increment amounts to less than -
be small (cd 0.05, based on pod diameter) and that there
1% of the total vehicle drag under steady cruise conditions. may be a beneficial effect on the lift generated by the wing
It was agreed that the mechanical advantages offered by a section, similar to end plates.
thicker section, such as robustness and ease of fabrication,
more than offset the slight drag penalty and therefore the 66-
009 section was selected for the final control surface design.

61
 A design trade-off was conducted between the use of a connected to the control surface shaft. Nylon gears are
pushrod or rotating shaft mechanism for the actuator linkage. particularly suitable in applications such as this, where zero
Although the pushrod system is simpler and more efficient, backlash is required. The motor is housed in a pressure-
this is offset by the disadvantages of a variable torque-arm balanced, oil-filled vessel, to accommodate operation at
and the possibility of buckling. Therefore a purely rotational 6000 m depth in the ocean.
system was the mechanism of choice.
Figure 13 also illustrates the plate which provides access to
Since the servo-motor shaft will be aligned with the axis of a recess at the end of the control surface shaft, where the
the tip pod (i.e. at right angles to the shaft of the control flap) potentiometer which provides feedback on flap deflection for
a bevel gearbox or worm gear mechanism must be the vehicle controller will be located. Controller design will
incorporated in the linkage. A bevel gearbox is smaller and be the subject of a future publication.
more efficient, offering minimal backlash (which is
advantageous for the controller design). The main benefit of 5. MODEL TESTING
using a worm gear is the unidirectional transmission, which
can lock the fins in a given position, thereby reducing the 5.1 Experimental Procedure
energy required by the controller. However, the
consequences of failure of the control system with fins locked Testing of a full-scale model of a single control-surface and
could be catastrophic and the worm gear option was actuator was performed in a low speed wind tunnel at the
discarded. University of Southampton. The tunnel has a working
section of 7' x 5' and a maximum air speed of 44 m/s. The
4.2 Final Actuator Design tests were conducted at 22 m/s, providing a Reynolds No
corresponding to full-scale vehicle operation at 2.5 m/s in
The final layout of the servo-motor and linkage is seawater and also to the half-scale model wind tunnel tests
presented in Figure 13. A 12 volt D.C. motor is located in a described in Reference 4.
pod at the tip of the control surface, behind the CTD sensor
housing. The motor drive is transmitted via a 800: 1 reduction The fin was vertically mounted above a false floor,
planetary gearbox, through 1:l nylon bevel gears which are extending 2.5 m in front and 0.6 m to the rear, which ensured

EULVHEAD
\ ;ERVOMOTOF

/
/
FIN _~___ +-
i

ACCESS PLQTE T O
POSITIDN F C t D B A C k
POTENTIOMETER i

I"

Figure 13. Final Design of the Control Surface Actuator

62
the same boundary layer thickness as that at the tail fin
location on the DOLPHIN hull, under normal operating
conditions. The fin was supported by a 5-component
dynamometer beneath the false floor, giving measurementsof I --- i
axial and side force, and bending moments about three 4-Drag (Theory) (N)
orthogonal axes. An automated data acquisition system logs
the average and standard deviation of 4 sample voltages from
10 /
r'
each of the dynamometer strain gauges for a given test
condition.
- 5
Trials were conducted for fin angles of attack between -16" E
0
and +16" to the tunnel axis, in steps of 2". For each fin angle,
the angle of the trailing edge flap was varied between -14" to 10 15
+14" to the fii, again in steps of 2".
Test Reyviolds INo. = 503500
Smoke injection and fibre tufts (attached to the fin and
flap) were used to visualise the flow. -10 1 W
Flap Defllection (degrees)
5.2 Experimental Results

Preliminary analysis indicates a fair correlation between Figure 15. Comparison of Drag Forces from Wind Tunnel
the theoretical lift data used in the design calculations and the Tests with Theoretical Calculations
measured values from the model tests (see Figure 14).
data points at 0" and 12" flap1 deflection which indicate
negative drag must be disregardld and the remaining values
are between 25% (for large angles of deflection) and 75%
(for small angles of deflection) lower than indicated by the
theoretical design calculations. The discrepancy may have
been caused by experimental enror; there is a suspicion that
I Y'? the dynamometer or model was fouling on the false floor of
the tunnel. This is the probable explanation for the two
'rogue' data points in Figure 15.

/ @
Full analysis of the test results, which is on-going at the
9 / Test Reynolds time of writing, will be the subject of a future publication.

"+ I /' J',f

/
No. = 503500
6. COWCILUSION
Ot-= - -7i- -__ i
-
The design of the control surfaces and actuators for the
0 5 10 15
Autosub DOLPHIN vehicle have been presented. Wind
Flap Deflection (degrees) tunnel tests confirm that the tail lift, force required to achieve
the specified vehicle manoeuirrability is generated with a flap
Figure 14. Comparison of Lift Forces from Wind Tunnel deflection of 7",which is well bdow the flap stall angle. The
Tests with Theoretical Calculations corresponding drag measurements were lower than expected
and this area requires further analysis.
Although the empirical lift values are lower than theory, the
specified lift force of lOON is generated during tests with a 7. ACKNOWLEDGEMENTS
flap deflection of 7 O and the foil has not stalled at 14"
deflection, when the lift force exceeds 200 N. The authors wish to thank 13r R Allen, Dept of Mechanical
Engineering and Miss H Jones, Department of Electronics
Comparison of the experimental drag data with theory is and Computer Science, University of Southampton, for their
less satisfactory,as shown in Figure 15. The two empirical contributions to this project. The work was carried out as a
Master of Engineering Group Design Project in the Faculty of
Engineering, University of Soutlhampton and generously

63
sponsored by the Ocean Technology Division of the
Southampton Oceanography Centre.

8. REFERENCES

B. S . McCartney and P. G. Collar, "Autonomous

Submersibles - Instrument Platforms of the Future,"
Underwater Technology, 15(4), pp.19-25, 1989.
B. H. Carmichael, "Underwater Vehicle Drag
Reduction through Choice of Shape," AIM Second
Propulsion Joint Specialist Conference, Colorado
Springs, Colorado, Paper No. 66-657, 1966.
J. S . Parsons, R. E. Goodson and F. R. Goldschmied,
"Shaping of Axisymmehic Bodies for Minimum Drag
in Incompressible Flow," Journal of Hydronautics,
8(3), pp.100-107, July 1974.
A. Huggins and A. R. Packwood, "Wind Tunnel
Experiments on Fully Appended Laminar Flow
Submersible for Oceanographic Survey," Ocean
Engineering, 22, pp. 207-221, Feb 1995.
R. J. Babb, "Instrumentation for a Low Drag
Hydrodynamic Test Vehicle," IEE International
Conference on Electronics in Oceanography,
Cambridge, UK, pp. 24-30, July 1994.
F. W. Riegels, "Aerofoil Sections - Results from Wind
Tunnel Investigations,"Theoretical Foundations, 1961.
S . F. Hoerner, "Fluid-Dynamic Lift" Publ.
L. A. Hoemer 1975.
S . F.Hoerner, "Fluid-Dynamic Drag" Publ.
S. F. Hoemer 1965.

SYMBOLS

angle of attack for wing

flap deflection angle
L
lift coefficient Cl =
WpGA
vl
Reynolds No. Re = -
V
D
drag coefficient c d =
WpgA
lift force (N)
drag force (N)
velocity (m.s-1)
length (m)
area (m2>
density (kg m-3)

64