Vous êtes sur la page 1sur 68

SNAME Transactions, Voi. 89, 1981, pp.

23-90

Rudder Torque Prediction


R. L. Harrington,1 Life Member
The various considerations to be evaluated when establishing rudder torque design values are discussed. Procedures for predicting rudder torque requiremements are analyzed, and examples are
given for rudders of the spade and horn types. Rudder torque calculations are made for six ships, and
the results are compared with data recorded during sea trials.

Background

The difficulties encountered in accurately predicting the


forces and moments acting on a rudder are self-evidently
considerable, but the establishment of the design torque capacity of the steering gear involves other onerous considerations
which must be dealt with. One is that large ships are generally
designed with balanced rudders in order to minimize the required torque rating of the steering gear. From Figs. 2 and 3

SOME YEARSAGO, Dr. K. E. Schoenherr was commissioned

by Panel H-10 (Controllability) of the SNAME Technical and


Research Program to undertake an investigation of the existing
uncertainties in estimating the torque and lateral force action
on an operating rudder. The results of that investigation were
published by SNAME in 1965 as reference [1], 2 and a review
of reference [1] indicates that neither the magnitude nor the
nature of the rudder-torque problem has changed to any significant degree since that investigation was conducted. Taking
into consideration the large amount of study that has gone into
the analysis of the flow around ships' hulls and the performance
of control surfaces, it would appear that an accurate prediction
of the torque required to exercise the rudder on a large ship
would be a simple undertaking; however, such is far from the
case. After studying the numerous factors that theoretically
must be dealt with when predicting rudder torques, some of
which were outlined by Schoenherr, it is apparent that the
achievement of precise rudder-torque predictions will remain
out of reach for many years to come. Such being the case, the
practicing naval architect must determine how to make best
use of the data and analysis techniques currently available.
Figure 1 shows the motion of a ship when turning in a simple
maneuver. With the initial throw of the rudder, the rudder
normal force causes a simultaneous outward movement of the
ship and establishment of a drift angle, fl (the angle between
the direction of the ship's instantaneous velocity and the centerline of the ship). As the ship assumes a drift angle, the speed
of the ship decreases until the ship settles into a turning circle
of essentially constant diameter. It is apparent that there is a
complex flow of water around the hull and across the rudder
even during the simple maneuver illustrated by Fig. 1; therefore, zigzag maneuvers, during which the rudder is exercised
to alternate hardover positions, would obviously introduce
variables which defy an accurate definition in a general sense.
For example, a hardover left rudder could be ordered during
any position of the continuum indicated on Fig. l before
equilibrium is reached, and each of these positions would entail
a unique set of transient conditions before a different equilibrium is established.

\'\,\
i

. o

/"

Fig. 1

Motion of a ship when turning

it can be seen that two rudders could be designed, each having


the same maximum resultant pressure force, R; however, the
maximum torque required to control the balanced rudder can
be seen to be considerably less than for the unbalanced rudder
in that the torque RL2 is less than the torque RL1.
The design of balanced rudders must be approached with
more caution than the design of the unbalanced type. Figure
4 shows the dimensions of the rudder used on a series of C4
cargo ships and also indicates the position of the estimated
center of the maximum resultant pressure force during ahead
steering. The center of this force is seen to be 6.3 in. aft of the

1 Newport News Shipbuilding, Newport News, Virginia.


2 Numbers in brackets designate References at end of paper.
Presented at the January 28, 1981 meeting of the Hampton Roads
Section of T H E SOCIETY OF NAVAl. ARCHITECTS AND MARINE ENGINEERS.
Author Harrington received the 1981 Vice Admiral E. L. Cochrane
Award for this paper, chosen as the best paper presented before a
Section of the Society.
23

RUDDERSTOCK
CENTERLINE- ~

RUDDERSTOCK I
CENTERLINE-~

RUDDERSTOCK
CENTERLINE

9.4'

CENTEROFMAXIMUM
RESULTANTPflESSURE
FORCE.fl

CENTEROFMAXIMUM

Fig. 2

--~[~--

ESTIMATEDPOSITIONOFMAXIMUM
RESULTANTPRESSUREFORCE
DURINGAHEADSTEERING

RESULTANTPRESSURE
FORCE,R

Unbalanced rudder

Fig. 3

Balanced rudder

centerline of the rudderstock. It is obvious that a very small


error in the estimated position of the maximum resultant
pressure force would have a serious impact on the accuracy of
the rudder-torque estimate. For example, an error of 4 in. in
the estimated position of the maximum resultant pressure force
on the rudder illustrated by Fig. 4 would appear reasonable to
expect; however, this is the magnitude of the error required to
explain the seemingly erratic ahead-steering data, shown by
Table 1, which was recorded during steering gear trials for the
series of sister ships illustrated by Fig 4. The astern steering
data in Table 1 is seen to be more consistent than the ahead data;
this is largely attributed to the fact that during astern steering,
the moment arm of the resultant pressure force about the
ruddersteek is approximately 50 in. Therefore, small variations
in the position of the resultant pressure force are of little consequence.
Figure 5 is a simplified illustration (with friction, etc. neglected) of torque curves characteristic of a balanced rudder.
The sense of the rudder torque is determined by the natural
tendency of the rudder's movement;'that is, if the rudder tends
to take charge and naturally move in the desired direction, the
torque is considered to be negative (restraint is required); if the
rudder must be forced in the desired direction, the torque required is considered to be positive. It therefore follows that
Fig. 5 applies only when the rudder is moving to increasing
angles; if the rudder were moving to decreasing angles, then
the applicable torque curves would be the opposite of those
shown. It may be seen that if the rudder illustrated by Fig 5
were released while going in the ahead direction, the rudder
would have an initial tendency to become stable at an angle
somewhat greater than 20 deg.
For small rudder angles during ahead steering, the position
of the resultant pressure force is forward of the rudderstock of
balanced rudders, thereby producing a negative torque such
as shown by Fig. 5. For progressively larger rudder angles, the
magnitude of the resultant pressure force increases and moves
aft, causing a reversal in the sense of the rudder torque as the

25.5'

!3

14 3'

Fig. 4 Estimated position of maximum resultant pressure force during


ahead steering for a typical C4 cargo ship [Ship E as defined by Figs.
26 and 28(b) and Table 4]

rudderstock centerline is crossed. Thereafter, until the rudder


stalls, the rudder torque increases rapidly since both the resultant pressure force and the lever arm about the rudder stock
increase.
During astern steering, the lever arm of the resultant pressure
force decreases with increasing rudder angles and the rudder
stalls at a relatively small angle, thereby limiting the magnitude of the astern torque, as illustrated by Fig. 5.
An optimally balanced rudder typically requires a tradeoff
analysis between the maximum negative ahead torque, the
maximum positive ahead torque, and the maximum astern
torque. It may be noted that in Fig. 5 the ahead rudder torque
at 35 deg (which is usually the maximum design rudder angle)
is somewhat larger than the maximum negative ahead torque
and is approximately the same as the maximum astern torque;
of course, this ideal circumstance seldom occurs in practice.
There are several reasons for more concern regarding the
maximum negative ahead torque than the maximum positive
ahead torque at 35 deg. One is that the maximum negative
torque (with the effects of friction, etc. included) is encountered

Table 1 Steering gear pressure a data as recorded by trial observers for a


series of C4 sister ships [Ship E as described by Figs. 26 and 28(b) and Table 41

Ship of Class
Ahead trials:
highest peak pressure observed, psi
Second highest peak pressure observed, psi
Astern trials:
highest peak pressure observed, psi
second highest peak pressure observed, psi

First

Second Third b Fourth

1400
1400

1550
900

1900
1200

1190
1190

1000
850

1200
1000

1200
1000

1150
1150

Maximum pressure in the driving cylinder(s).


h Continuous recordings were also taken; see Figs. 75-77.
24

Rudder Torque Prediction

during zigzag maneuvers as the rudder leaves one hardover


position and moves toward the centerline position; as can be
imagined by looking at Fig. 1, the flow into the rudder as that
occurs causes abnormally high rudder forces. Another is that
during zigzag maneuvers, the maximum negative torque occurs
during the mid-region of the timing cycle; and if the relief
valves should lift or if a power limiter is installed and reduces
the hydraulic pump stroke, it may be impossible to compensate
for the lost flow of hydraulic oil and move the rudder at the
required average rate. Also, with a Rapson-slide steering gear
arrangement, more hydraulic pressure is required to develop
the same torque at smaller rudder angles (for the same pressure,
the torque developed varies inversely as the cosine squared of
the rudder angle). Additionally, specific rates of rudder
movement are not required to be demonstrated when developing the torque computed at the hardover rudder position.
And finally, as the 35-deg position is approached the stroke of
the hydraulic pump is reduced, thereby slowing the rate of
rudder movement and reducing the ram pressure which would
otherwise be developed.
The acquisition of trial data, such as given in Table 1, entails
considerable judgment and interpretation, and the use of the
data may introduce yet other anomalies. For example, due in
part to the misleading analytical procedures formerly used, until
relatively recent years there was a conviction that the maximum
ahead rudder torque would occur at the maximum rudder
angle. Such being the case, the sea trial data observer would
be inclined to discount all pressure data except those occurring
in the vicinity of the maximum rudder angle. To compound
the error, when the trial data were later correlated with the
rudder torque predictions, it was again assumed that the
maximum pressure readings did, in fact, occur at the maximum
rudder angle reached. As a consequence, in some cases, the
data that were accepted as being unchallengable were in substantial error and led to invalid conclusions.
Another problem faced by the data observer is that in order
for the data acquired to be useful, as a mimimum, the rightrudder ram pressure, the left-rudder ram pressure, the rudder
angle indicator, and the elapsed time must be monitored simultaneously with consistent data recorded for all four readings
at precisely the instances of greatest significance (which are
often recognized after they have passed). The situation is
clarified considerably by making continuous simultaneous recordings of the data required to provide meaningful results, and
Figs. 43 through 79 contain such data for some particular ships.
Of course, these data leave many unanswered questions regarding the rudder angle of attack, ship heading, the frictional
and hydrodynamic components of the rudder torque, etc., but
due to the expense associated with the acquisition of more informative trial data, it appears that these questions may never
be clarified.
A review of trial data leads to the recognition that due to the
inherent scatter of the data, even with sister ships, there is a high
probability that any steering gear rating selected Will be either
inadequate or excessive. Of course, neither extreme is desired;
but due to the expense associated with back-fitting additional
steering gear capacity, a modest amount of overcapacity is a
prudent objective.

Specification requirements
The manner in which the major design criteria (that is, the
torque rating and the rate of rudder movement) are defined
for steering gears by specification requirements is rather unusual. The general specifications for naval ships [2] give no
requirements regarding the rating of steering gears except for
the following provision:

-++'+L fT0.0UE
,.e,0

OL

:~ Lu ~,,~

I
1o

I
3o

,o

/RUDDERANGLEOFA"I'rACK

~ ~.~~ e~
>-

I/
~/

.ASTERN

Fig. 5 Simplified torque characteristics of a balance6 rudder

"'The steering gear shall be capable of moving stopping,


and holding the rudder at any angle, with the ship going
ahead at full power.
"If there are steering gear limitations for astern operations, suitable warning plates shall be installed in the Pilot
House and at the steering gear."
The military specification for steering gears [3] likewise does
not delineate either the steering gear torque rating or rudder
rate. However, the detailed specifications for specific ships
do place limits on the acceptable rudder rates. The exact
wording varies from ship to ship; but the rudder rate is specified
in the following general manner [4]:
"The steering gear shall have the capacity to move the
rudder from 35 degrees (hardover) on one side to 30 degrees on the opposite side at an average rate of 3 degrees
per second at any ship's speed up to and including the
maximum obtainable ship's ahead speed during sea
trials."
The specification requirements for merchant ships follow
a similar pattern. The standard Maritime Administration
(MarAd) specifications [51 limit the amount of overload permitted on the electric motor, but give no other restrictions or
guidance regarding the rating of a steering gear except to refer
to the Society of Naval Architects and Marine Engineers
(SNAME) test and trial codes. Of course, the SNAME codes
are not intended to establish design criteria, but instead to set
forth test procedures; the SNAME Code For Sea Trials [6] does
precisely that. However, be that as it may, in order to accomplish the steering tests specified in reference [6], the ship
must be capable of unrestricted rudder movement; that is, no
limits are placed on the rate of rudder movement, but the
rudder must be tested to the hardover positions while proceeding at maximum shaft rpm in both the ahead and astern
directions.
The classification societies, for example the American Bureau
of Shipping (ABS) Rules [7], stipulate that effective means for
steering is to be provided which "is to be capable of putting the
rudder from 35 degrees over to 35 degrees over with the vessel
running ahead at the maximum continuous rated shaft rpm."
Reference [7] further states that the steering gear "is to be capable of putting the rudder over from 35 degrees on either side
to 30 degrees on the other side in 28 seconds with the vessel
running ahead at the maximum continuous rated shaft rpm;"
however, astern steering requirements are not mentioned. The
Coast Guard regulations [8] agree with ABS for ahead operations, but for astern steering, also stipulate that the "main
steering gear and rudder stock shall be so designed that they are
not damaged at maximum astern speed."

Rudder Torque Prediction

25

-"='--RUDDER STOCK
CENTERLINE
_

I
I
I

1
I
h2
11- . . . .

W2

-----~

I
I
I

--~-b--~/

Bottomley have many shortcomings. For reasons not clearly


understood, the method proposed by Joessel gained rather wide
acceptance as a means of estimating rudder forces and torques.
The Joessel method is based on the much-publicized experiments conducted by the French naval constructor Joessel in
1873. Based on experiments conducted in the Loire River
[having a maximum current of 1.3 m/s (meters per second)]
with a rectangular plate of span 30 cm and chord 40 cm, Joessel
derived empirical relationships for the variation of torque and
the variation of the center of pressure with the angle of attack
[11 ]; these relationships, when corrected for the larger density
of seawater, are as follows:

x = (0.195 + 0.305 sin a)

(2)

where

1
I
I
I
1
I
r
I

W1

Fig. 6

(1)

and

hI
f'

Q = 0.811 Avew sin ce

. . . .

Q
A
v
w
a
x

Model of a horn-type rudder used with the Joessel method

=
=
=
=
=
=

rudder torque about leading edge of plate, ft-lb


area of plate, ft 2
water velocity, fps
plate width, ft
angle of attack, deg
distance from leading edge to center of pressure, ft

By combining equations (1) and (2), the resultant force on a


plate is determined to be
F-

In view of the foregoing, it may be concluded that ahead


steering requirements are adequately covered by existing
specifications; however, astern steering capability is not required to be a serious consideration in the design of either
merchant or naval ships. While the necessity of an unrestricted
astern steering capability can be debated, there can be no
question that an assessment of astern steering capabilities is
desired during the design stage.

Fundamental considerations
The rudder torque which must be controlled by the steering
gear is composed of two major elements. One is the hydrodynamic torque which is caused by the action of the water over
the rudder and, to be technically accurate, would require a
rigorous assessment of factors such as hull wake, propeller race,
hull drift angle, and change in rudder angle of attack as the hull
turns in addition to the effects of the arrangement of the particular rudder under analysis. The other torque component
is due to frictional losses in the rudderstock bearings and this
component tends to shift the hydrodynamic torque curves
upward an amount which is largely influenced by the type of
rudderstock bearings used and the rudder support arrangement.
Two approaches have been customarily taken when predicting rudder torque. One entails the use of the laws of similitude in conjunction with an overall coefficient, and the other
attempts to break the analysis down to basic principles in which
each factor is evaluated as a separate entity.
T h e doessel m e t h o d
A variety of empirical formulas has been offered for use in
estimating rudder forces and moments, and a description of the
more widely known ones is given in references [9, 10]. The
empirical formulas and coefficients developed by Baker and
Bottomley perhaps have a more rigorous technical basis than
others which have been proposed, but even those of Baker and
26

Q0.811Av 2sino~
x 0.195 + 0.305 sin a

(3)

where F is the resultant force on the plate in pounds.


These are the basic relationships as derived by Joessel, and
assuming that they were presented as being the results of the
particular experiments conducted, they are not vulnerable to
technical criticism. Of course, the manner in which these
equations have been interpreted and applied when making
rudder torque predictions is a different issue and may involve
technical inaccuracies of varying magnitudes. For example,
the plate used in Joessel's experiments was flat and rectangular;
therefore, for the method to be useful, consideration must be
given to other shapes. One approach has been to approximate
the geometry of a rudder with one or more rectangles having
the same total area as the rudder, then compute the hydrodynamic loads on each of the rectangles as though they acted independently, and then sum the loads on the individual rectangles to obtain the total effect.
Regarding the remaining considerations which must be dealt
with when using Joessel's equations, it has been argued that use
of the method is, in fact, an application of the laws of similitude.
Therefore, it the torque is a function of the rudder area and the
square of the water velocity, and the shape of the curve is predicted accurately, then if a correlation coefficient were available for a similar application, the same correlation coefficient
could be used for a prediction. On this basis, the Joessel
equations were used extensively for many years. Libraries of
correlation coefficients (or Joessel coefficients) were developed
for various ship types and the effect of many influences, such
as rudderstock bearing friction, were considered to be included
in the coefficients.
In order to illustrate the manner in which the Joessel method
has been used, consider the horn-type rudder illustrated by Fig.
6; this rudder dan be transformed into two rectangles as shown
by the dashed lines; and by applying equations (1), (2), and (3)
with the inclusion of a Joessel coefficient, expressions for the
ahead and astern rudder torques become

Rudder Torque Prediction

The DTMB Report 933 method

Qahd = Kahd 0.811 V2 sin c~


X [(w' (0'195 + 0.305 sin ) -- b) w,h '
0.195 + 0.805 sin c~

+ wzZh2]

0.811 v 2 sin v
Kant 0.195 + 0.305 sin o~
X l(a - wl(0.l.95 + 0.805 sin oz))Wlhl
+ w22h2(0.805 - 0.305 sin c~)]
As can be seen, the Joessel method is remarkably simple and
easy to use. If the shape of the rudder torque curve is predicted
accurately and there is previous experience which provides a
confident basis for the selection of ahead and astern Joessel
coefficients, then the predictions may be reasonably accurate.
However, the Joessel method may entail significant problem
areas, namely:
No experience with a similar application may severely
limit the confidence in a prediction.
The shape of the torque curve predicted may not be accurate.
The failure to accurately predict the shape of the torque
curve could be a more serious problem than a lack of experience
with a similar application. It is usually possible to obti~in some
data points which can be used to provide guidance in the selection of a Joessel coefficient in any particular case; but without
a reasonably accurate estimate of the shape of the curve, serious
problems can result. In general for ahead predictions the
Joessel equations understate the magnitude of the resultant
force and indicate a chordwise center of resultant pressure
which is much farther aft than do other model test data (see
Figs. 7 and 8) or trial data. Referring to Fig. 8, it is noted that
only the shape of the torque curve is important. That is, the
Joessel coefficient, or experience factor, K, is used to adjust the
magnitude of the curve at some point (usually the maximum
angle) based upon prior experience; and the Joessel equations
. are used to establish other points on the curve. However, due
to errors in locating the chordwise center of pressure, at small
angles of attack the resultant center of pressure is significantly
farther forward than predicted; and for balanced rudders, the
magnitude and range of the negative torque region may be
substantially larger than expected. For rudders with a small
amount of balance (which was the prevalent type of design for
many years), the negative region of the torque curve was unimportant and could be completely neglected. However, for
larger ships of higher power, the rating of the steering gear
becomes prohibitively large unless the rudder is reasonably
balanced, and this requires an accurate assessment of the negative-torque region.
The Joessel equations are widely recognized to be deficient
in many respects when predicting rudder torques during ahead
operations; however, these deficiencies are of less importance
when predicting astern torques. For example, during asternflow conditions the hydrodynamic clmracteristics of an airfoil
section would be, expected to be quite similar to those of the flat
plate used by Joessel during his experiments. Also, rudder
balance is not an important consideration when assessing astern
torque requirements; that is, as shown by Fig. 5, the entire astern torque curve has a negative sense and reaches a maximum
value near the maximum (or stall) angle of attack. In addition,
as discussed previously, the accuracy of the prediction is not as
critical for astern estimates. Consequently, the Joessel method
can provide useful guidance when developing astern torque
predictions. It is also acknowledged that, with proper regard
for its limitations, the Joessel method may be useful when
predicting ahead torque requirements.

Due to the inadequacies of the Joessel and similar methods


of predicting rudder torques, attention was directed towards
the development of more comprehensive prediction techniques
during the 1950's. The most important contribution in this
effort was made by Whicker and Fehlner as reported in David
Taylor Model Basin (DTMB) Report 983 [12], which provides
test data for a series of all-movable low-aspect-ratio control
surfaces of the spade type. Rudder torque predictions using
the Joessel equation generally take the form of single lines, such
as illustrated by Fig. 5, and are considered to include the
composite effect of a vareity of influences. However, an entirely different approach is used when making rudder torque
estimates based on the test data as reported in DTMB Report
988 or a n y o f the other test data for airfoil sections. Such test
data can be used as a basis for predicting the hydrodynamic
torque imposed upon a rudder; however, in addition to the
hydrodynamic torque, there are also frictional losses in the
rudder support bearings which must be considered.
Due to the frictional losses in the rudder bearings, the rudder
torque requirements when displacing the rudder (that is, when
going to larger rudder angles) are not equal and opposite to the
rudder torque requirements when restoring the rudder (that
is, when going to decreasing rudder angles). The frictional
component of the torque must be overcome by the steering
engine (which means that the frictional torque is considered
to be positive) regardless of whether the rudder is being displaced or restored; however, the sign of the hydrodynamic
torque must be reversed when the rudder is being restored.
Consequently, when displacing the rudder, the net torque to
be overcome is equal to the frictional torque plus the hydrodynamic torque; and when restoring the rudder, the net torque
to be overcome is equal to the frictional torque minus the hydrodynamic torque.
Figure 9 illustrates the theoretical rudder torque characteristics during a simple maneuver. Assuming that the rudder
is ordered from the centerline position, the Qo curve (which
is equal to the sum of the frictional and hydrodynamic components) would be followed until the ordered angle is reached
at Point a. After reaching the ordered angle, the rudder is held
in position by the hydraulic raln pressure and small movements
tend to dissipate the effects of friction; consequently, a transition
is made from Point a on the Qo curve to Point b on the QH
curve. With the rudder held in the ordered position, the ship
develops a drift angle which tends to reduce the rudder angle
of attack, thereby causing a movement from Point b toward
Point c on the Qtt curve. Assuming that the rudder is ordered
back to the centerline position when Point c is reached on the
QH curve, an immediate shift is made to Point c on the - Q H
curve because the direction of desired rudder movement is
reversed; then a transition is made to Point d on the QR curve
because friction must be overcome when moving the rudder;
and the QR curve is followed as the rudder is restored to the
centerline position. The two basic components of the curves
shown by Fig. 9, that is, the frictional torque and the hydrodynamic torque, are obtained from related calculations. The
hydrodynamic torque is equal to the moment of the hydrodynamic normal force about the rudderstock centerline, and the
frictional torque is derived from an analysis of the reactions in
the rudder support bearings caused by the rudder hydrodynamic normal force.
In addition to providing test data itself, Whicker and Fehlner
also developed semi-empirical relationships which can be used
in a generalized way for prediction purposes. The generalized
equations presented in DTMB Report 983 provide normal force
predictions which are in good agreement with the DTMB 988
test data as shown by Fig. 7; however, due to inaccuracies in the

Rudder Torque Prediction

27

1.2

/
1.0

DTMB93"3DATA~

a:

=o
.=,
o

J
~DTMB

933 EQUATIONS

.4

-JOESSEL
K=.S03

0
0

. 10

15

20

25

30

3S

RUDDERANGLE OF ATTACK (DEG)

Fig. 7

Comparison of rudder normal force predictions for Ship A using alternative


procedures

10

/,

/ ,'
JOESSEL---~
K=.503
/
~:j

/
,I

/ /

/
0

DTMB933DATA-

,I

/
oaz:

I
I

i//
--//
/

-4'

DTMB933 EQUATIONS
-6

-8
0

10

15

20

25

30

35

RUDDER ANGLE OF ATTACK (DEG)

Fig. 8

28

Comparison of hydrodynamic torque predictions for Ship A using alternative


procedures

Rudder Torque Prediction

TOTALTORQUEWHENRESTORING,
F

/
~
=

,,i

. - - - ~i

f..oRooYNA.,C

- TORQUEWHEN
RESTORliG,-QH

/'/

R=QF-QH

i
i

,,'
~, I

t\

!I

IX i

I \i
-- ~

T
OTALTORQUEWHEN q
olSPLACING'QO=QF*QH

II

I II

'

~b

- HYDRODYNAMICTORQUEWHEN
DISPLACING.QH
i

HI

15

20

:~5

30

35

RUDDERANGLEOFATTACK(OEG)
Fig. 9

Rudder torque elements during a simple maneuver

computed chordwise center of pressure, the hydrodynamic


torque predicted from the DTMB 933 equations is in substantial
error at the larger angles of attack, as can be seen from Fig. 8.
DTMB Report 933 also contains comparisons between the test
data and the results obtained from the semi-empirical equations
which, upon study, show that at small angles of attack, the
correlation between the torque data recorded during tests and
that computed from the semi-empirical equations is excellent;
but at the larger angles of attack, the equations overstate the
torque predictions, particularly for effective aspect ratios in the
vicinity of two (which is the area of most practical importance).
The magnitude of this shortcoming of the DTMB 933 semiempirical equations has not been widely recognized; however,
it is considered to be so serious as to limit the usefulness of the
equations when used in this manner. Another shortcoming of
the semi-empirical equations is that they provide no means for
determining when a breakdown of lift is to be expected; the
result of this limitation could be that provisions are made to
accommodate large rudder torques which are, in fact, impossible to develop. Nevertheless, as will be further discussed, the
DTMB 938 semi-empirical equations are useful in that they
provide relationships which can be used to make corrections
for taper ratios which differ from those corresponding to the
test data.

Spade rudder: ahead hydrodynamic torque prediction


Figure 10 illustrates the basic characteristics of a typical spade
rudder, which is the type of all-movable, low-aspect-ratio
control surface covered by DTMB Report 938. The basic test
data contained in DTMB 933 are self-evidently valid, but dif-

ficult to use in the form presented. That is, interpolations must


be made to account for variations in both the effective aspect
ratio and the sweep angle of the particular control surface under
study. In orderto facilitate the use of the DTMB 933 test data,
cross-plots such as Figs. 11 through 16 can be prepared. The
lift coefficient, drag coefficient, and chordwise center of
pressure can be read from these curves for zero and q-ll-deg
sweep angles and then interpolated to determine the values for
a specific sweep angle. If needed, data are available in DTMB
Report 983 to prepare similar curves for negative sweep angles.
Another problem associated with the use of the DTMB 933
test data is encountered when the test data indicate the onset
of stall. It is informative to know when a breakdown of lift is
to be expected, but it is also desirable to know the rudder performance which would occur in the event that the rudder does
not stall in the expected manner. In order to permit projections
beyond the stall angle indicated by the model test data, the basic
DTMB 988 test data were extrapolated, and then these extrapolations were plotted as broken lines in Figs. 11 through 16.
As shown by Fig. 9, the calculation results are also shown as
broken lines for rudder angles of attack exceeding the predicted
stall angle. Consequently, this procedure provides information
regarding the stall angle observed during snodel tests and also
provides insight regarding the risk involved if stall does not
occur as predicted.
Rudders which generally conform to the geometry shown
by Fig. 10 arereadily analyzed after several simplifying assumptions are made, and Table 2 illustrates the procedure used
to calculate the torque developed by a spade rudder using the
DTMB 933 test data and semi-empirical relationships.

Rudder Torque Prediction

29

~~o

- ~

In the calculations given in Table 2, tile velocity of the water


over the rudder is assumed to be uniform, and a correction is
made for the average wake effects and the augmented water
velocity due to the effects of the propeller race. The speed'of-advance calculation (wake effect) is familiar, but the dynamic pressure calculation (propeller race effect) may be unfamiliar. Here it is assumed that tile propeller actuator-disk
theory, as described in reference [131, applies and the propeller
thrust is expressed as

=i
I

HULL----/

iI
/

CENTER OF

il

PRESSURE
PRE'
IURE i/',1

MEAN k--_

T=pA
"

~+

where
r

~ LOCUS OF
QUARTER CHORD

ME~ * !i

T
A
p
v
z

CHORDWISE
'=='~ CENTE R OF
PRESSURE

=
=
=
=
=

propeller thrust
propeller disk area
water density
propeller speed of advance
velocity imparted to the water by the propeller

This equation can be manipulated so as to obtain the following


expression for the dynamic pressure, p:

Fig. 10

P=~P ( z

TIP
CHORD--,D,

- -

Basic characteristics of a Spade rudder

1.6,

1.4-

+ v) 2 = -pv 2 + T

This is the basic form of the expression used in Table 2.

NOTES:
.
BSWEEPANGLE OF ZERO
DEGREES
SQUARE TIPS
DEVELOPED-FROM.FIGURES
44, 55, AND 66 OF DTMR
REPORT 933

ill/r

SS]

S S" S ~

- -

;'/,,',4>-

_1

,, Y/,5"/l',,''.

W|JmR!

~.u
m
u.
uuJ

#
/
z~i

i i

,e

7"

24*

225

"//
l

i0

"

15

.......,,,.1,.,~~ ~

~
~

5"
1.0

1.5

2.0

EFFECTIVE ASPECT RATIO

Fig. 11

30

~-

Lift coefficient, sweep angle of 0 deg

Rudder Torque Prediction

2.5

3.0

an accurate assessment of the effective aspect ratio is Unimportant in the determination of the drag coefficient; however,
accuracy may be required when determining the lift coefficient
and the chordwise center of pressure. Accuracy is of particular
importance when evaluating the chordwise center of pressure
at the higher rudder angles for balanced rudders since a small
percentage error is magnified in the subsequent torque calculations.
For spade rudders, such as Ship A (see Fig. 22), at vary small
rudder angles a complete mirror effect is provided by the
fairing above the rudder. As the rudder angle increases, the
effective aspect ratio decreases; but the effect of the fairing
continues to larger rudder angles than may be expected. As
discussed in reference [14], test data [151 show that when an
all-movable control surface is mounted against a submarine hull
shape, a substantially complete mirror effect is achieved, at low
rudder angles, and although the effect diminishes with larger
rudder angles, there remains a substantial mirror effect
throughout a range o f rudder angles. For the specific tests
conducted, the ratio of the effective aspect ratio to the geometric aspect ratio varied from 2 at zero rudder angle to 1.5 at
the stall angle of 31 deg. Based on the test data reported by

The effective aspect ratio (slenderness ratio or span-to-width


ratio) of most rudders does not lend itself to an exact evaluation,
and this introduces unavoidable inaccuracies into the calculations. As can be imagined, a very high aspect ratio would
suggest a long, continuous lifting surface where the breakdown
of lift due to end effects is minimal and the flow conditions are
ideally suited for the development of lift with low angles of
attack.
One means of effectively increasing the aspect ratio of a
rudder is to develop an arrangement in which the rudder is
bounded on one end by a surface that is in the plane of the flow.
If the bounding surface completely seals the end of the rudder,
thereby eliminating cross-flows and end losses, a mirror-image
effect is achieved and the "effective" aspect ratio is equal to
twice the geometric aspect ratio. This condition may exist in
the case of the fairwater diving planes on some submarines
where the fairwater diving planes abut the sail, which is nearly
wall-sided, thereby providing mirror-image conditions.
However, the opportunity to provide a mirror effect at all angles of attack seldom arises in the design of rudders due to the
fact that the hull surface in way of the rudder is rarely flat.
From a study of Figs. 11 through 16 it may be concluded that

1-8

NOTES:
SWEEPANGLE OF +11 DEGREES
SQUARE TIPS
DEVELO~O FROM FIGURES
45. 60, AND 67 OF DTMB
933

1.6-

f
/
f
/

. , ,, , .

REPORT

.~
i~#
/

Pf

"",/', .'"".

.S J"
~,J'

,.P

**

,.**

1.4

"~

"-

i., J

,",,/.

J'

f . ,
j

~,~e

-P

,"

""
~ :"S 1

""

1.2

A
1.0

/Z y///, " / ' / S.


. /ll, Y/'// " / / "

kLI
U.

f ~

,. # Z / / , / / /

--I

./////
/"
"'///" /

.6

2t1'

24'
.4

"2 /
- /

,,f

.2

lo' s

0
0

.... I

.5

//
~,..r

1.5

EFFECnVE ASPECT RATIO

Fig, 12

2,5
.,

Lift coefficient, sweep angle of + 11 deg


Rudder Torque Prediction

31

38

NOTES:
SWEEP ANGLE OF ZERO DEGREES
oSQUARE TIPS
DEVELOPED FROM FIGURES
44, 55, AND 66 OF DTMB
REPORT 933

~.ANGLE OFA~, ACK_

36

.5-

34

o
v

32
J

.4,

uJ

30

Fig. 13 Drag coefficient,


sweep angle of 0 deg

8
,3"

iiIII
II
m!m_L.
I
IIIi
~ I I I

,2.

--7"--..-

.1.
15
10

O"

1.0

1.5

2.0

2.5

3.0

E FFECTIVE ASPECT RATIO

.7-

MOTES:

38
.6-

-...._

ANGLI

SWEEP ANGLE OF +11 DEGREES


SQUARE TiPS
DEVELOPED FROM FIGURES
45, 60, AND 67 OF DTMB
REPORT 933

)F ATT.4cI(

36

--

mI.

34

32*

z~

.4

w
t..)

30"

u..
uJ
C:)

Fig. 14 Drag coefficient,


sweep angle of + 11 deg

28"
26 j

24"
J

v
~

15
10

5_i
0

1.5

EFFECTIVE ASPECT RATIO

32

Rudder Torque Prediction

2.5

NOTES
.SWEEP ANGLE OF ZERO DEGREES
SQUARE TIPS
.DEVELOPED FROM FIGURES
44 55 AND 66 OF DTMB
REPORT 933

38=,~
36% ~
340 % ~
~

.32 -

. '44'L ~
.30-

28%

.28"
r~

.26 "

22.,
20*

.24.

.22 '

Fig. 15

Chordwise center of

pressure, s w e e p angle of 0 deg

lb'

N
=

.18

c~

/ J

.IS '

.14

10~

.12

.10

5"
,08
0

15

25

EFFECTIVE ASPECTRATIO

NOTES
SWEEPANGLE OF +11 DEGREES

38
34

3384:~ ~

32

.SQUARE TIPS
OEVELOPED FROM FIGURES
45 60 AND 67 OF OTMB
REPORT 933

32%

30
28
A

26

a:
o

t
Fig. 16 C h o r d w i s e c e n t e r of
pressure, s w e e p angle o f + 11 deg

24

lii

22

1S~
20
J

18

_......=.=,,m~...=...=..~.~

16

14

,o/

12
10
08
5

15

25

EFFECTIVE ASPECTRATIO

Rudder Torque Prediction

33

Table 2

NOTE:

Ahead rudder torque calculations for a typical spade rudder using cross-faired DTMB Report 933 test data

These calculations are for the 20-deg rudder angle-of-attack position for Ship A, as defined by Figs. 22 and 28(a) and Table 4.

Propeller thrust = 326 P e / V ( 1 - t)


Speed of advance = 1.69 V(1 - w)

lb

fps

Dynamic pressure = 0.994 v 2 +

psf

Taper ratio = X 5 / ( X 1 + X2)

251 900
33.61
1924.7
0.875
16.875

ft

M e a n c h o r d = 0 . 5 ( X 1 -~- X 2 Jc X5)

Sweepangle=tan-l[-O'25(X5-Xl-X2)+X2-Xqx3

ft

deg

Rudder angle

o~

deg

Effective aspect ratio = - ~ (2 - ~5 )

2.054

Data uncorrected for taper ratio:


lift coefficient (see Figs. 11 and 12)
drag coefficient (see Figs. 13 and 14)
center of pressure (see Figs. 15 and 16)

eLl
CDI

CP~,

0.942
0.163
0.220

Liftcoeff. corr.

ACL

0.041

Corrected lift coeff. = CL1 + ACL

CL2

0.983

Drag coeff, correction - C22 - C~,


2.83a
Corrected drag coeff. = Co~ + ACD
eLl COSOt q- CD1 sina
CL2 cosa + CD2 sina

ACD

0.014

CD2
CNI
CN2

0.177
0.941
0.984
0.00773

1"63~-0"73(
2
a
a )

1 A CL
(0.25-CP~,)CN1 - -~

pcX3CN2

20

CM~/42

Corr. center of pressure = 0.25 - CM~/42


N o r m a l force =

4.82

CPe2
F

CN2 +---X41

Hydrodynamic torque = 12F[?CP~ 2 X2


/
2
Rudderstock bearing friction"
dl F/0.42X3 + X g / +
d2 =/0.42X3 + Xs + Xg.I
=#'Y ~- ~
) ~'~T~ "
Xs
/
Rudder torque displacing = QF + QH
Rudder torque restoring = QF - QH

Q.

lb
in.-lb

QF

in.-lb

QD
QR

in.-lb
in.-lb

0.242
639 200
-3.84 X 10 6
2.32 X

-1.52 x 10 6
6.16 X 106

,oo]
==g

PUMPSTROKE~

35

~e

30

500]
~

iooo
1250 ~

131
Fig. 17

34

133

135
TIME(SECONrIS)

137

Ram pressure characteristics when ram takes charge of rudder (see also Fig. 47)
Rudder Torque Prediction

10 6

139

reference [15] and a correlation of rudder torque estimates and


trial data, the approach taken in Table 2 is to assume that the
ratio of the effective aspect ratio to the geometric aspect ratio
varies linearly from 2 at a zero rudder angle to 1 at a 75-deg
rudder angle. This assumption is considered to provide reasonable accuracy for the customary spade rudder arrangements
over the rudder angles of practical significance. Of course,
other rudder arrangements would require further study.
With the effective aspect ratio established, values for the lift
coefficient, drag coefficient, and chordwise center of pressure
can be read from Figs. 11 through 16 for zero and + l l - d e g
sweep angles. Then, to determine the values corresponding
to a particular quarter-chord sweep angle, ~2, an interpolation
is made.
The data in Figs. 11 through 16 correspond to a family of
control surfaces which have a taper ratio (ratio of tip chord to
root chord) of 0.45. These data must be corrected for other taper
ratios, and the semi-empirical equations in DTMB 933 are
useful for that purpose. Table 2 illustrates the process of correcting for taper ratio, and thereafter, the rudder normal force
and hydrodynamic torque are calculated in a straightforward
manner.

Spade rudder: frictional torque prediction


By assuming that the position of the spanwise center of
pressure is located 42 percent of the span distance from the root
chord and assuming that the rudder normal force is equal to the
rudder resultant force (which are sufficiently accurate assumptions for the purpose), the loads carried by the upper and
lower rudderstock bearings are readily determined; therefore,
with the bearing loads and stock diameter known, the bearing
frictional torque can be computed if the friction coefficient is
known. There is no difficulty in selecting reasonable values
for coefficients of friction which correspond to the materials
and lubricants normally used. Based on the information provided in references [16, 17] and the standard handbooks, a
friction coefficient of 0.2 for reinforced laminated phenolic
(such as Micarta) bearings would appear reasonable; this is the
type of bearing often used outboard where the lubricant is
seawater. For lubricated steel-on-bronze bearings--the type
sometimes used for the inboard upper rudderstock bearing--a
value of 0.1 should be a reasonable approximation, and for roller
bearings, which may be used inboard, a friction coefficient of
0.01 should be a close estimate.
It is noted, however, that it may not be necessary to rely solely
upon estimated values for the bearing friction coefficients since
sea trial data are available. From a study of Figs. 48 through
51, which are the ahead-trial data for Ship A, it may be observed
that in several instances, such as the 1S4-second mark of Fig.
47, the rudder is being held at an ordered-angle position and
an equilibrium condition is being approached when the rudder
is ordered to an opposite position. Figure 17 is an enlarged
view of the rudder angle and ram pressure shown by Fig. 47
and also shows the stroke of the hydraulic pump. As can be
seen from Fig. 47, when the 1S4-second mark is being approached, no hydraulic fluid is being pumped, the rudder angle
is constant at 35 deg, and the ram pressure is nearly constant.
As the hydraulic pump goes on stroke, it may be noted from Fig.
17 that there is no evidence of static, or breakaway, friction
since the ram pressure does not tend to peak or rise significantly
as the rudder begins to move. In fact, it can be argued that
when the rudder is being held at a constant angle, the alternating forces and slight movements between the sliding surfaces
result in the frictional forces within the steering gear and the
rudder support bearings being intermittent and hence incapable
of sustaining a continuous frictional torque. This argument
would suggest that as the 1S4-second mark is being approached

in Fig. 17, the rudder is being held in the hardover position by


the ram pressure, pl, only. As the hydraulic pump is ordered
to full stroke, the rudder begins to move and the hydraulic
pressure rises to a higher level. An observation to be made here
is that the torque required to accelerate the rudder is not a
significant consideration. (Calculations indicate that the hydraulic pressure required to accelerate the rudder is about 20
psi; and, incidentally, the frictional torque due to the static
weight of the rudder is of the same magnitude.) After the
hydraulic pump has gone on full stroke, but before the rudder
angle has changed significantly, it can be argued that the hydrodynamic torque being overcome is substantially the same
as before, but in this instance, the ram pressure, p2, must also
overcome any frictional losses in the rudder support bearings
and any frictional losses in the steering gear itself.
The data plotted in Fig. 17, and those shown by Figs. 43
through 79, are a reproduction of the trial data recorded with
a minimal amount of interpretation. Such being the case,
rudder torque is not plotted inasmuch as it is a derived value;
instead the ram pressure differential, that is, the difference
between the right-rudder and left-rudder cylinder pressures,
is shown. There is a relationship between the ram pressure
differential and the rudder torque; however, the friction between the various elements of the steering gear introduces assumptions into the relationship.
The relationship between the ram pressure differential and
rudder torque requires a detailed analysis of the forces and
frictional losses within the steering gear. A double Rapson-slide
steering gear arrangement is shown by Fig. 18. Figure 19 is
a more detailed view in way of the crosshead, and also shows
a diagram of the forces acting within the steering gear assembly.
Referring to Fig. 19, the force P is the force developed by each
ram and is equal to
p = r--dZ p
4
where

P = force developed per ram, lb


d = ram diameter, in.
p = ram pressure differential, psi
The force F is the resultant force exerted on the tiller. Neglecting frictional losses for the present time, as a simplification,
the horizontal component, Fh, of the resultant tiller force becomes equal to the force developed by the ram; therefore
F-

P
COSOL

where
F = resultant tiller force (neglecting friction), lb
ce = rudder angle, deg
T h e advantage of a Rapson-slide type of steering gear is due
in part to the fact that at the larger rudder angles the resultant
force, F, exerted on the ram is much greater than the ram force
developed and also the lever arm at which the resultant force
acts is greater than the crosshead radius by a similar factor.
Since the rudder torque is equal to the resultant tiller force times
the effective lever arm, the torque developed by the steering
gear is
ZPRE
(~ -- COS2(Z~
where

Q = torque exerted on rudderstock, in.-lb


Z = number of rams

Rudder Torque Prediction

35

cYLINDER,

xRAMh

CROSSHEADBLOCK

. . . .

CYLINDERBUSHING

Fig. 18 DoubleRapson-slidesteeringgeararrangement
R = crosshead radius, in.
E = ram-to-rudder efficiency
Not only does the Rapson-slide type of steering gear permit
smaller rams and cylinders to be used due to the magnified
torques developed at larger rudder angles, but also the double
Rapson-slide, such as shown by Fig. 18, is advantageous in that
there is no resultant unbalanced force exerted on the rudderstock by the steering gear. That is, the diagonally opposed
hydraulic cylinders are piped together such that they act in
unison, thereby imposing only a net couple, and no resultant
force, on the rudderstock. A single Rapson-slide (only one ram
with opposing cylinders) would develop a magnified torque but

would also exert an unbalanced resultant force on the rudderstock, thereby increasing the rudderstock frictional losses
somewhat. However, the disadvantages of an unbalanced
resultant force and increased rudderstock frictional losses, associated with a single Rapson-slide, are often more than offset
by the advantage of simplified equipment; consequently, single
Rapson-slide arrangements are commonly used. There are also
other types of steering gears, such as the link type or the vane
type, which have advantages under particular circumstances.
Reference [18] contains a more detailed description of the
characteristics of the alternative types of steering gear arrangements.
The ram-to-rudder efficiency, E, in the preceding equation

Fv- fay

fb

fc
fc

Fh

Fig, 19 Rapson-slideforce diagram


36

Rudder Torque Prediction

is an overall efficiency factor which includes the combined


effects of several loss components. One of the major losses is
due to the friction force, fa, between the crosshead and the tiller
as shown by Fig. 19. The magnitude of fa is

/'d

coso~

/2d

--

R
COS~

fa =/2aF

or

where

rd = 1 -- h/2d cosol
2R

fa = crosshead-tiller friction force, lb

,LLa

crosshead-tiller friction coefficient

The horizontal component, f , h , of the crosshead-tiller friction


force opposes the ram force, P, and the loss can be expressed
as

where h is the diameter of crosshead pin in inches.


The total ram-to-rudder efficiency, E, then becomes
E = rarbrcrd

E = (1 - / 2 a tan(~)(1 - / 2 b l t a n t x - ga])

P - f a h _ P - /2aF sino~
raP
P

X (X -- 4gC d) ( 1 - - )h#d
- ~ - cosc~

or

ra = 1

/2a tanc~

Another component of the ram-to-rudder efficiency is the


loss due to the friction in the cylinder bushings. The sum of
the two reactions in each pair of cylinder bushings is equal to
the difference between the vertical component, F~, of the tiller
resultant, F, and the vertical component, fay, of the crosshead-tiller friction force fa. The bushing losses, fb, which also
oppose the ram force, then become
/'b --

e -fb

_ e -

leo - f

vl/2b

or

rb = 1 -/2/,I tan~ - 120]


where/2/, is the ram-cylinder bushing friction coefficient.
There is also a friction force between the ram and the cylinder packing which opposes the ram force. Assuming that the
contact pressure between the ram and the packing is equal to
the cylinder internal pressure and assuming that this pressure
is maintained for an effective length, L, the packing loss can
be expressed as:

By multiplying the terms in the foregoing expression and


dropping secondary effects (terms containing products of
friction coefficients), there results
E = 1 --/2a tanc~ --/2b tano~ -- 4/2c Ld

h#d cosa

2R

In the customary Rapson-slide steering gear arrangement,


the crossheadradius is approximately three times the ram diameter and the diameter of the crosshead pin is approximately
two-thirds the ram diameter. If it is assumed that the effective
length, L, of the bushing contact pressure is 0.05 times the ram
diameter, d, and that all of the friction coefficients are equal,
the ram-to-rudder efficiency becomes as shown by Fig. 20. It
may be seen from Fig. 20 that the customarily assumed ramto-rudder efficiency of 0.8 corresponds to a friction coefficient
of about 0.12 at a 35-deg rudder angle, which is not an unreasonable expectation.
Now that an expression has been derived for the ram-torudder efficiency, E, it is possible to return to Fig. 17 and use
trial data to establish values for the various friction coefficients.
As may be expected, the 134-second mark of Fig. 47 was not

7r d2 p _/2TrdLp
p-fc_4
/'C

--

7r dZ p
4
90-

or

I.-

rc

t..i

4/2c j

ua

where

~ s0z

u = packing-ram friction coefficient

L = effective length of bushing contact pressure p, in.


A different approach must be taken when assessing the effeci
of friction between the crosshead pin and crosshead block. The
friction force fd between the crosshead block and crosshead pin
produces a moment about the pin which is balanced by a shift,
e, in the position of the tiller force F. The effect of this shift
is a reduction in the effective radiusof the tiller, thereby reducing the torque produced. Therefore, the crosshead pin loss
becomes
R

- - - - e
COSO~

"

~-r~
E
t:J

,-,,,

~ 701...

<~
= s0-

50.
10

20

30

40

50

l'd-

RUDDER ANGLE,or(DEGREES)

R
COSO~

Fig. 20

Rudder T o r q u e Prediction

Efficiency of a Rapson-slide steering engine


37

selected by chance for illustration purposes. Based on a careful


analysis of the potentially valid data points (those not obviously
affected by extraneous influences) in Figs. 43 through 51,
fourteen data points were selected and averaged with the
conclusion being that Fig. 17 is typical of all of the data points
observed.
As discussed previously, the pressure pl is the ram pressure
required to resist the hydrodynamic rudder torque being developed at the rudder angle c~ with no frictional influences;
therefore
QH COS2OL,
Pl-

-~-cl2Z R
4
and the pressure P2 is the ram pressure required to overcome
the hydrodynamic torque plus frictional losses in the rudder
bearings and frictional losses in the steering gear itself; therefore
p2 =

(O. + QF)cosZ-,
7r dzZR E
4

where
Q n = hydrodynamic torque, in.-lb
Ql~ = frictional torque, in.-lb
Since the rudder angle is substantially the same when the two
pressure observations are made, the ratio of p~ to pz can be
expressed as:
Pl _
Qtt
E
Pz Q n + Qe
For the case illustrated by Fig. 17, Pl = 600 psi and P2 = 1050
psi for an observed value of Pl/Pe equal to 0.571. With all
other values known (see Table 4), Qn can be determined to be
3.93 X 106 in.-lb from the foregoing expression for pt; this
corresponds to a rudder angle of attack of about 19.8 deg and
a rudder normal force of 680 000 lb. With these values determined, the equation for Qi~ shown in Table 2, the curves for
the ram-to-rudder efficiency shown by Fig. 20, and the assumed
coefficients of friction of 0.01 for roller bearings, 0.2 for phenolic bearings, and 0.1 for steel-on-bronze, a check can be made
to determine if the calculated value of p l / p z equals the observed value; that is
E
Qn + QF
8.98
(0.831)
8.93 + 4.50
= 0.387
This calculated value does not compare well with the observed
value of 0.571. The obvious conclusion is that the actual frictional losses are much less than those corresponding to the assumed coefficients of friction.
There are several possible explanations for the unexpectedly
low frictional losses. One is that the sliding surfaces may be
extremely smooth; reference [17] (and other sources) notes that
smaller friction coefficients are associated with extremely
smooth surfaces. However, this explanation is regarded as
being unlikely due to the observation that all five sister ships
of the Ship A class exhibited similar behavior and no special
effort was made to ensure that any of the sliding surfaces was
exceptionally smooth. A more likely explanation is that due
to the alternating nature of the rudder forces, these pulsating
forces are not capable of maintaining as much continuous

P21calculated

38

frictional contact as can more constant forces. There can be


no doubt regarding the existence of a large resultant mean
rudder force and the existence of mean forces between the tiller
and crosshead block, but the data clearly show that these forces
do not maintain the uninterrupted contact required to sustain
a continuous frictional torque. There are, however, some
frictional losses in both the rudder bearings and steering gear
and a means of allowing for these losses must be developed.
Returning to the originally assumed coefficients of friction,
that for the roller bearing is so small as to not influence the
calculation result; therefore, it can be accepted as valid without
a loss in accuracy. There is no reason to expect that the frictional effects in the steering gear are any less influenced by the
pulsating nature of the forces than those in the rudder bearing;
consequently, it is reasonable to assume that the ratio of the
coefficient of friction between the moving parts of the steering
gear to the coefficient of friction in the lower rudderstock
bearing maintains the same relationship of 1:2 as the values
originally assumed. With this simplification, it is seen that in
order for the calculated value of P~/P2 to equate to the measured value of 0.571, the friction coefficient for steel-on-bronze
must be 0.05, and that for phenolic bearings must be 0.10.
Referring back to Table 2, with the friction coefficients established, calculations for the frictional losses in the rudderstock
bearings can proceed, and the calculations may be concluded.
As discussed previously, the frictional torque must always be
opposed by the ram when moving the rudder; but the sense of
the hydrodynamic torque depends upon the direction of rudder
travel. When displacing the rudder (going to larger rudder
angles), the hydrodynamic torque is added to the frictional
torque; but when restoring the rudder, the hydrodynamic
torque is subtracted from the frictional torque. Calculations,
similar to those in Table 2, are made for rudder angles of attack
throughout the range of interest and the results are plotted as
shown by Fig. 29. As indicated by the dashed lines on Fig. 29,
according to the DTMB 933 model test data, the rudder for Ship
A is predicted to stall at rudder angles of attack exceeding S0
deg.
Horn rudder: ahead torque prediction

While the test data and analytical procedure outlined in


DTMB Report 988 are based on a series of control surfaces
which are generally similar to most rudders of the spade type,
such is not the case with rudders of the horn type. Aside from
reference [19], which is an empirical method for predicting the
torque required for semibalanced rudders not directly in the
propeller race, there is no reasaonbly comprehensive test data
or analytical procedure which deal with the horn type of rudder
as an entity; therefore, improvisations must be made.
One approach in estimating the torque characteristics of
horn-type rudders is to independently analyze that section of
the rudder below the horn and the remaining section of the
rudder abaft the horn, with the results of the two analyses added
to obtain the total effect. This is similar to the techniques used
when applying the Joessel method, where questions regarding
the aspect ratio never arise as explicit considerations. While
arguments can be made that the Joessel coefficient incorporates
aspect-ratio considerations, a total neglect of the interactions
between the two surfacesections when using one of the more
rigorous methods would introduce obvious errors. Another
consideration which requires careful study is the effect of the
rudder horn upon the hydrodynamic characteristics of the two
rudder sections.
It may be noted that the DTMB 933 semi-empirical equations have been used to compute the torque required for horn
rudders with some success; however, this practice is not technically rigorous and entails unnecessary risk. By considering

Rudder Torque Prediction

'

/
/

1.40

-1.2o

- 1.00

.40

- .80

CNu/'~

.30

.20

/
.10

~'

-.~ ~
u..

ix
_~

- .20

-0

10

15

20

25

30

35

RUDOERANGLE(OEG)

Fig. 21

Hinge moment and normal force coefficients of rudder area abaft horn

that part of the rudder below the horn to be a spade rudder and
using the DTMB 9;33 serni-empirical equations to compute the
required torque, the torques calculated at the higher rudder
angles of attack will be understated as shown by Fig. 8. But
interestingly, by considering that portion of the rudder abaft
the horn to be similar to a spade rudder (hence, neglecting the
effects of the horn upon the hydrodynamic characteristics of
the trailing rudder), the torque required by that part of the
rudder is overstated. The two inaccuracies therefore tend to
cancel, and the net result may be credible if the relative sizes
of the upper and lower rudder sections are favorable. However, a more rigorous analytical procedure is preferred.
That part of the rudder below the horn is similar to a spade
rudder; consequently if the effects of the horn and the upper
rudder section are reflected in the evaluation of the effective
aspect ratio of the lower section, the hydrodynamic characteristics of the lower section should be approximated with
reasonable accuracy by using DTMB 933 data cross-plots as
illustrated by Figs. 11 through 16. However, correlations with
trial data show that an assessment of the upper rudder section
cannot be handled in this manner.
An evaluation was made of a variety of known methods
which could be used to calculate the torque requirements of the
upper rudder section, but none was found to provide satisfactory results; consequently, alternative solutions were sought.
Since the rudder horn, together with the section of rudder abaft
the horn are, in effect, a flapped control surface, an investigation was conducted to evaluate the applicability of techniques
used for flapped control surfaces in this case. The University
of Maryland Wind Tunnel Report No. 259 [20] contains test
data for a series of flapped control surfaces, but the geometry
of the models used during the tests reported in [20] deviates
excessively from the configuration of typical horn rudders.
Therefore, the techniques described in [21, 22] were used to
develop corrections for the data presented by [20] such that the
data were applicable to horn rudders. The corrected data were
then plotted as a family of curves from which coefficients
corresponding to any ratio of horn width to trailing rudder

width could be selected. However, this degree of detail was


later found to be unnecessary because the range of the ratio of
horn width to trailing rudder width of practical importance is
so narrow that a single line is adequate for the customary rudder
arrangements. Unfortunately, the reference [20] data modified
in this manner did not provide satisfactory results.
Data for flapped control surfaces tend to be more erratic than
that for all-movable control surfaces, but the data in reference
[20] appeared to be particularly erratic. Suspecting that more
consistent results may be obtained if the data in [20] were less
erratic, a faired-line representation of those data, modifed to
agree with the common arrangements of horn rudders, was
developed as illustrated by Fig. 21. The data provided by Fig.
21 permit a simple and straightforward prediction of the hydrodynamic characteristics of that part of the rudder trailing
the horn for the customary rudder configurations, and the
predictions correlate favorably with trial results.
Table 3 is an illustration of the procedure used to calculate
the torque requirements for a horn rudder, and, as with the
spade rudder, the velocity of the water onto the rudder is assumed to be uniform and equal to the velocity in the propeller
race. It is noted that in some calculation procedures [23], only
that portion of the rudder which is within the projected outline
of the propeller disk is considered to be subject to the uniform
velocity in the propeller race; and parts of the rudder not within
projection of the propeller disk area are considered subject to
a water velocity equal to the speed of the ship. This approach
may be logical and consistent with other approaches which have
been taken; however, a correlation of results obtained with this
procedure and trial data will not support that assumption. As
can be seen from Figs. 24 through 27, parts of most horn rudders
are outside a projection of the propeller; such is particularly the
case with Ship B, where the rudder horn is well clear of the
propeller disk. In the case of Ship B, the velocity of the ship
is small compared with the water velocity in the propeller race;
therefore, to consider parts of the rudder subject to the ship
velocity only would, in effect, delete those parts from the calculation. However, trial data show that even that section of

Rudder Torque Prediction

39

Table 3

Ahead rudder torque calculations for a typical horn-type rudder

NOTE: These calculations are for the 20-deg rudder angle-of-attack position for Ship C, as defined by Figs. 24 and 28(b) and Table 4.
T
v

Propeller thrust = 326 P e / V ( 1 - t)


Speed of advance = 1.69V(1 - w)
2 4T
Dynamic pressure = 0.994v -IRudder angle

p
a

Lower Rudder Section:


Taper ratio = X 5 / ( X 1 + X2)
Mean chord = 0.5(X1 + X2 + X5)
+ X2 - X4]
Sweep angle = tan_ t [).25(X5 - XI - X2)

al

Aspect ratio =
2 X, + X2 751
Data uncorrected for taper ratio:
lift coefficient (see Figs. 11 and 12)
drag coefficient (see Figs. 13 and 14)
center of pressure (see Figs. 15 and 16)
1.63X- 0.73/[ a, ~2
/
Lift coeff, correction a7
~57.31

482 500
28.67

psf
deg

1800
20

t't "

0.752
17.38

deg

1.73

Xa

lb
fps

1.88

eL 1
CDI

...
...

CP~

...

0.901
0.16
0.219

ACL

...

0.032

Corrected lift coefficient = CL, + ACL

CL2

...

0.933

Drag coefficient correction - C2= - C2n


2.83at
Corrected drag coefficient = Col + ACD
CNI -- CL1 COSC+ CD, sina
CN2 CL 2 cosa + CD2 sina

ACD

...

0.011

CD2
CN,
CN2

...
...
...

0.171
0.902
0.936

CMJ42

...

0.012

CP~2

...

0.237

Ft

lb

QHt

in.-lb

-15.1 X 106

QF~

in.-lb

1.4 x 106

Cu

ft

CNu
CHM

...

F.
QH.

il~
in.-lb

257 000
9.7 106

QF:

in.-lb

0.4 X 106

QH

QF

in.-lb
in.-lb

-5.4 106
1.8 X 106

...

QD

in.-lb

-3.5 X 106

QR'

in.-lb

7.3 X 106

CM~/42 =

(0.25 -- C ~I)CNI -- ~

ACL

Corrected center of pressure = 0.25


Normal force

CM~/42
CN2

p'cX3CN2

Hydrodynamic torque = 12FI(KCP~2


Bearing friction

X2 + X4
2

503 000

=~3~-d3 Fl {1 + 0 . 4 2 X 3 ~ X121
.]-{'- F, (0.42/3:X12)
2Xs ~ Xl0 + N i l

] [~2d2(Xs + Xg)

pldlX9]

Upper Rudder Section:


Mean chord = 0.5(X1 + X6)
Normal force coeff. (see Fig. 21)
Hinge moment coeff. (see Fig. 21)
Normal force = pcuX~eNu
Hydro. torque = 12pcuX7CHM
Bearing friction
= t.ta ~ F u [ 1
0.42X7
+~XI2].]
Fu

+2X8

13.605

0.75
0.174

0.42X7 + XI2 [p2d2(X8 + X9 ) + /ZldlXg]


Xlo+Xu

Total Rudder:
Hydrodynamic torque = QH~ + QH.
Bearing friction = Qpt + QF.
Single-ram correction - # t d l cosec
2R
Rudder torque displacing
= QF + QH + rlQF + QHI
Rudder torque restoring

0.018

= Q p - QH + r I QF - QHI

the Ship B rudder trailing the horn is effective and cannot be


neglected.
Following logic similar to that used in connection with the
calculations for a spade rudder, the expression given in Table
3 for the effective aspect ratio of the lower rudder section is an
approximation which makes allowances for the influence of the
upper rudder section and the horn. T h a t is, to the extent that
the upper and lower rudder sections are contiguous, the seal
provided to the lower section by the upper section is considered
40

to be complete; and the seal provided by the rudder horn to the


rudder i m m e d i a t e l y below it is considered to be c o m p l e t e at
a zero r u d d e r angle but to d e g r a d e in a linear m a n n e r until
there is no effect at a rudder angle of 75 deg. Calculations for
the h y d r o d y n a m i c torque of the lower r u d d e r section are otherwise similar to those given for a spade rudder in Table 2.
Test data indicate that the friction coefficients for spade and
horn rudders generally have similar characteristics; therefore,
having previously established friction coefficient values in

Rudder Torque Prediction

connection with the spade rudder, this issue is resolved.


However, the rudder bearing loads are not as simply determined as with the spade rudder. In order to avoid an indeterminate problem involving a beam with a variable cross
section on three supports, the expressions used in Table 8 to
determine the rudderstock and pintle bearing loads are based
on the assumption that the rudder pressure forces are transmitted to the three support bearings in two steps. That is, for
the upper and lower rudder sections, respectively, the resultant
rudder force is first considered to be transmitted to the heavy
rudder gudgeons into which the pintle and rudderstock are
embedded; then the load on the pintle gudgeon is considered
to be transmitted to the pintle bearing by shear. The loads on
the two rudderstock bearings (transmitted from the rudderstock
gudgeon) are found by summing moments in the classical
manner.
As noted previously, calculations for the upper section are
simple and straightforward. The separate calculations for the
upper and lower sections of the rudder are summed, as shown
in the summary of Table 8, to obtain the total hydrodynamic
and frictional torques. Since Ship C had a single Rapson-slide
steering gear (instead of a double unit with balanced forces),
the steering gear exerts an unbalanced force on the rudderstock.
In the case of Ship C, as is usually the case, the upper rudderstock bearing is adjacent to the tiller; therefore, virtually all of
the unbalanced tiller force on the rudderstock is resisted by the
upper rudder bearing. A correction for the effects of a single
ram is readily determined by dividing the effective steering
gear lever arm (that is, R/coso0 into the net rudder torque to
obtain the upper bearing force; then multiplying that force by
pj(dl/2). Since the frictional torque cannot be negative, the
absolute value of the net rudder torque is used.
The calculation is repeated over the range of rudder angles
of attack of interest and plotted as illustrated by Figs. 82 and
81.

Astern torque predictions


The test data and analytical procedures available for use in
making astern torque predictions are very limited. As discussed
previously, accuracy is not as important when making astern
torque predictions, and, for this reason, use of the Joessel
method is often advocated; it being further argued that the
rudder sections have no aerodynamic shape in the astern direction and thus are, in effect, the same as the flat plate used
by Joessel when conducting his experiments.
No corrections are made for any perturbations in the water
flow into the rudder; that is, the velocity of the water flowing
onto the rudder is considered to be uniform and equal to the
astern speed of the ship. Considerable difficulty may be encountered in establishing an estimate of the astern speed. A
rule of thumb which may provide useful results is to assume that
the power required astern is twice the power required going
ahead at the same speed, and that the power-rpm relationship
is the same going ahead or astern. However, this approximation can be considerably in error for some types of ships. In
some cases the astern power may be more than three times the
ahead power required at the same ship speed.
Horn rudders are divided into two sections for analysis purposes: the upper section abaft the horn, and the lower section
below the horn. The results obtained for the two sections are
then added.
Aspect-ratio considerations must again be dealt with when
making astern predictions. In the case of spade rudders such
as that for Ship A (see Fig. 22), the effective aspect ratio is taken
to be twice the geometric aspect ratio at all rudder angles, for
two reasons. One is that at low angles of attack the close
proximity of the adjoining hull structure would support that

assumption, and the other is that stall conditions are reached


at a relatively low angle of attack when going astern, such that
allowances for a reducing aspect ratio with increasing rudder
angles would not be of practical importance. In the case of
horn rudders, both the upper and lower rudder sections are
considered to have effective aspect ratios equal to twice the
geometric aspect ratios at all rudder angles due to the sealing
effect that each section provides the other and the effect of the
horn on the lower section; no allowance is made for the increasing gap between the horn and lower rudder section at
larger rudder angles.
Predictions for the frictional components of the astern rudder
torque are handled as outlined in Tables 2 and 8, for spade and
horn rudders, respectively, for the ahead case.
Figures 87 through 42 are plots ofastern torque predictions
using both the Joessel method and also cross-faired DTMB
Report 988 test data, such as illustrated by Figs. 11 through 16
for the ahead case. As with the ahead predictions, estimates
are made for the rudder being beth displaced (frictional torque
plus hydrodynamic torque) and restored (frictional torque
minus hydrodynamic torque). The.hydrodynamic~torque is
shown for Ship B only since this is the only ship analyzed with
sufficient frictional torque to make a significant difference
between the hydrodynamic and restoring or displacing
torques.

Predictions for specific ships


The analytical procedures which have been outlined were
used to make ahead and astern rudder torque predictions for
the six ships described by Figs. 22 through 28 and table 4; the
rudder torque predictions are shown by Figs. 29 through 42.
The only ship which warrants further detailed discussion is Ship
B. The arrangement and size of the rudder for Ship B require
special consideration. For example, while the frictional torque
resulting from the static weight of the rudder is usually of a
negligible magnitude, as mentioned in the foregoing, such is
not the case with Ship B. The rudder assembly for Ship B
weighs about 500 tons, and after allowances are made for
buoyancy, the frictional loss in the support bearing is about 1
X 106 in.-lb (this loss for other ships is typically about 50 000
in.-lb).
Another exception for Ship B is due to the support arrangement. Due to the flexibility inherent with the palm type of
rudderstock and the shoe support, there is no simple way to
obtain a reasonable approximation for the rudder bearing reactions, and more sophisticated analytical techniques must be
used. To illustrate the point, despite the massive cross section
of the shoe on Ship B (50-in. width and 84-in. depth), due to the
shoe's large span, the shoe deflects approximately 0.88 in. at full
power, and it is this flexibility which is considered to create an
additional exception (or lack of exception) for Ship B. When
discussing rudder bearing and steering gear friction coefficients
for Ship A, it was observed that seemingly reasonable assumptions for these quantities did not apply, and the discrepancy was attributed to the pulsating nature of the loads on the
rudder support bearings. In the case of Ship B, correlations
indicate that the originally assumed coefficients of friction do
agree with trial data and this agreement is considered to be due
to the much greater rudder assembly flexibility and, hence,
more continuous loading on the support bearings.

Correlations with trial data


Figures 48 through 79 are plots of ram pressure differential
and rudder angle versus time as recorded during sea trials for
the six ships previously described. Special note is made of the
fact that rudder angle, and not rudder angle of attack, is plotted,

Rudder Torque Prediction

41

Fig. 22

Ship A stern arrangement


Fig. 25

Ship D stern arrangement

I
Fig. 23

Fig. 24

Ship B stern arrangement

Ship C stern arrangement

and that ram pressure differential, and not rudder torque, is


plotted. Due to the myriad of complicating factors involved,
no attempt was made to plot the rudder angle of attack. It is
possible, by making a series of assumptions, to provide an indication of rudder torque on Figs. 43 through 79; however, it
was considered preferable to present the data as recorded with
minimal interpretation, thus making it possible for other interpretations to be made on the basis of different experiences
or points of view.
Ship A trial results
The ahead-trial curves for Ship A, Figs. 43 through 51, are
an interesting set of curves which refute the long-accepted
axiom that steering gear pressure patterns fluctuate in an unpredictable manner and do not repeat. This series of nine runs
provides the results from the port and starboard pumping unit
tests on five sister ships (no data were obtained during one run).
Not only do these data show that the characteristics of the
pressure traces tend to repeat for the same ship, but also that
42

Fig. 26

Ship E stern arrangement

Fig. 27

Ship F stern arrangement

the traces tend to repeat for sister ships. This suggests that
variations due to construction tolerances are not as large as may
be expected; it also suggests that the pressure traces are following well-structured laws of physics.
Correlation of the ahead-torque estimates shown by Fig. 29
and the trial data in Figs. 43 through 51 requires considerable
study and interpretation. With the initial throw of the rudder
to the 25L position, right-rudder pressure is required to restrain the rudder, and right-rudder pressure continues to be
required as the ordered angle is reached. Before reaching the
25L ordered angle, the pressure values range from 400 to 700
psi, which corresponds to torque values of -2.1 to - 3 . 7 X 106
in.-lb on the Qo curve. After the ordered rudder angle is
reached and held, two influences occur. One is that the frictional forces in the steering gear and rudder bearings lose their
effect, thereby causing a transition from the Qo to the QH
curve in Fig. 29 and resulting in a rise in the pressure curve; and
another is that the ship assumes a larger drift angle, which reduces the rudder angle of attack, thereby also causing the

Rudder Torque Prediction

~d

dl--'"

II

X8
X8

x2--4

Xl--I~

-!0

X9

m m

Xl .

d3m
~

X5

(a) SPADE RUDDER

(b) HORN RUDDER


Fig. 28

Table 4

-1

(c) SHOE RUODER

Definition of rudder dimensions and support details given in Table 4

Descriptive data for six ships instrumented during steering trials (see also Figs. 22 through 28)

Ship
Hull: length
beam
draft
displacement (trial)
Rudder type
Steering gear: type
number of rams
ram diameter
crosshead radius
Ship ahead speed (trial)
shaft horsepower
performance coefficient
t h r u s t deduction factor
wake fraction
propeller diameter
Ship astern speed (trial)
Rudder dimensions:

Rudder support details:

Upper stock OD
Upper stock brg. type
Upper stock brg. frict, coeff.
Lower stock OD
Lower stock brg. type
Lower stock brg. frict, coeff.
Rudder pintle OD
Rudder pintle brg. type
Rudder pintle frict, coeff.

ft
ft
ft
tons
..
.

Z
d
R
V
P
e
t
w
D
U
X1
X2
X3
X4
X5
X6
X7
Xs
X9
Xlo
Xzz
X12

m.
in.
knots
hp
...
...
ft"
knots
ft
ft
ft
ft
ft
ft
ft
ft
ft
ft
ft
ft

550
82
19.9
13 800
spade
Rapson
2
l0
28
22.6
22 000
0.697
0.122
0.12
20
10
12.29
5.71
20
3.46
15.75
...

1143
228
74
450 600
shoe
Rapson
2
18.625
55.5
15.5
45 000
0.676
0.267
0.668
31.5
4.6
25.25
10.75
41.75
10.75
36
25.25
6.25
34.9

916
135
36
97 098
horn
Rapson
1
16.25
37
19.5
40 000
0.635
0.12
0.13
25
9.5
12.34
7.5
17.17
5.75
14.92
14.87
14
18.42
6.3
10.43
2.2
1.1

570
83
24
13 000
horn
Rapson
1
9.5
26
20.9
13 400
0.69
0.12
0.08
18
10.5
8.49
5.42
15.25
2.5
9.5
9.32
8.52
9.0
5.0
6.9
2.1
1.0

528
76
29.8
11 700
horn
Rapson
2
9
27
23.1
19 250
0.725
0.165
0.180
22
12.25
8.35
5.44
14.33
2.5
9.5
9.4
11.17
9.85
3.89
7.83
2.3
1.0

529
75
28.7
12 800
horn
Rapson
2
10
25.25
23.6
18 150
0.71
0.16
0.205
20.5
12.25
8.73
5.9
13.41
4.0
11.0
10.09
11.09
9.9
4.5
8.0
2.2
1.1

43
bronze
0.10
32
phenolic
0.20
32
phenolic
0.20

25.75
bronze
0.05
25.5
phenolic
0.10
24
phenolic
0.10

12
bronze
0.05
17.5
phenolic
0.10
16.25
phenolic
0.10

20
bronze
0.05
20.25
phenolic
0.10
14.25
phenolic
0.10

20
babbit
0.05
23
phenolic
0.10
18.75
K-Monel
0.15

12:1
3.8
...
44164
. . . . . .
. . . . . .

dl
in.
......
#1
:..
d2
in.
......
~t2
:..
d3
m. . . .
. . . . . . . . .
#3
. . . . .

22
roller
0.01
35
pheonlic
0.10

Rudder Torque Prediction

43

14

)2

10

I
I
I
I
l
I

.I
I

I
I
I

o
o

//
2

(_

o~"~

o
o=

o,j /

I
I

A
\

'
\

,I
I

I
I

I
I

-4

-6

I
I

-8

-10

10

15

20

2S

30

35

RUDDER ANGLE OF ATTACK (DEG)

Fig. 29

Ahead torque curves for Ship A; ship speed of 22.6 knots

pressure curve to rise due to the negative characteristics of the


QH curve in this region. After the ship begins to settle into the
turn, the pressure curve tends to reach a value ranging from 700
to 1100 psi, which corresponds to torques from - 3 . 8 to - 5 . 9
106 in.-lb on the QH curve. As discussed previously, when the
rudder is ordered to the 30R position, there is an increase in
pressure due to the transition from the - Q H curve to the Qn
curve and it is seen that the pressure increases to the range Of
1100 to 1400 psi, corresponding to 5.6 to 7.1 106 in.-lb on the
QR curve. At the time the transition is completed, it appears
that the maximum point on the QR curve has been passed, and,
as the rudder progresses in the right direction, the QR curve is
followed toward the origin with decreasing pressure. However,
instead of the ram pressure continuing to decrease and reverse
as the zero rudder angle is passed and the Qo curve is followed
to the opposite side, the ram pressure declines initially, but then
increases in a smooth continuous manner until it increases
rapidly with large undulations as the ordered angle is approached. This consistent occurrence is impossible in the absence of an influence as yet unaccounted for. That is, when
44

the rudder departed the 25L position, the resultant pressure


was forward of the rudderstock on the port side, hence requiring right-rudder pressure to control the rudder. When the
rudder reached the 30R position, the resultant pressure was
aft of the rudderstock on the starboard side (which is where it
is predicted to be for angles of attack greater than 24 deg),
hence, requiring right-rudder pressure to control the rudder.
However, there is an apparent anomaly: the resultant pressure
position made this transition without passing through either the
rudder forward region on the starboard side or the rudder after
region on the port side, either of which would have required
a left-rudder pressure. Obviously, there is another influence.
It is possible that the influence could be due to the oblique
flow of water onto the rudder. Figure 80 is a plan view (the
elevation view is shown by Fig. 22) of Ship A and indicates the
positions of the rudder when hardover. When the rudder is
in one position and a drift angle has been established, the flow
of the water into the propeller enters at an oblique angle and
can reduce the rudder angle of attack to a value much less than

Rudder Torque Prediction

60-

50-

I
I

40-

" - "~

!
I

30"

o
o
o

I
I
20-

10.

Fig, 30 Aheadtorque curves for


Ship B; ship speedof 15.5knots

//

O'

-10

QD

~.

-20

I
,

-30

"

I
i

-50
5

10

15

20

25

30

35

RUDDER ANGLE OF ATTACK (DEG)

30

" --'~'

~QR

20

~\
10

Fig. 31 Aheadtorque curves for


Ship B; ship speedof 10.2knots

=.

/'
%.

"~

/I

/,

QF

oc
-10

%
%
-20

-30
5

10

15

20

25

30

35

RUDDER ANGLE OF ATTACK (DEG)

Rudder Torque Prediction

45

20

16
I
I

I
12

la

~.....~QR

I
I

I
I

.,i

f/

,1
i

. ...~t ,.~"

\,\,
-12

-16
0

10

15

20

25

30

35

RUDDER ANGLE OF ATTACK(DEG)

Fig. 32

Ahead torque curves for Ship C; ship speed of 19.5 knots

the rudder angle itself. However, when the rudder is ordered


to the opposite position, the drift angle which previously reduced the angle of attack then augments the angle of attack and
may cause other effects. It may be seen from Fig. 80 that the
hydrodynamic characteristics of the rudder could be influenced
by the cross-flow of water into the rudder even before the
centerline position is r e a c h e d . . I t is suspected that the impingement of the obliquely flowing water onto the aft region
of the rudder may cause a continuously increasing pressure
force in that region which, in effect, is superimposed upon the
rudder hydrodynamic forces developed by the flow in the
propeller race. The data suggest that the expected decline in
ram pressure as the rudder leaves the 25L position becomes
masked by the buildup in hydrodynamic pressure forces on the
after-starboard region of the rudder as the rudder passes the 0
position, and thereafter the cross-flow causes a further pressure
buildup.
As discussed in reference [24], the lift coefficient can fluctuate widely and irregularly at angles of attack just above that
corresponding to maximum lift (that is, the onset of stall);
therefore, the strong pressure pulsations recorded as the ordered
position is approached indicate that stall conditions have been
reached. Such would indeed be expected in view of the very
large angle of attack the rudder has at that time. The rightrudder ram pressure is seen to reach values ranging from 1800
to 2400 psi, corresponding to torques of 9.1 to 12.1 X 106 in.-lb
on the Qo curve shown by Fig. 29. The 12.1 X 106 in.-lb
torque value represents a 35-deg apparent angle of attack,
which is 5 deg greater than the predicted stall angle. These
data support the observation made in reference [25] that rud46

ders can continue to develop lift at angles of attack considerably


greater than the stall angle normally expected.
As the 30R ordered position is gradually approached, the
ship continues to respond to the ordered rudder position,
thereby reducing the rudder angle of attack and restoring more
orderly flow conditions over the rudder. The rudder angle of
attack continues to reduce and the QD curve is followed until
the 30R rudder angle is finally reached, at which time a
transition is made to the Qn curve with a consequent further
reduction and eventual reversal of the ram pressure. The ram
pressure then trends toward a left-rudder value of about 600
psi (corresponding to - 3 . 5 X 106 in.-lb on the QH curve) when
the rudder is ordered to the 35L position. A transition is then
made to the QR curve, resulting in a pressure rise to about 1100
psi (6.0 x 106 in.-lb) and then a reduction in pressure as the Qn
curve is followed to the origin, and the Qo curve is followed to
increasing rudder angles. On this occasion the cross-flow due
to the ship's drift angle is not of sufficient magnitude to completely mask the negative-torque region of the Qo curve;
however, there is an influence. In addition to substantially
reducing the negative region of the Qo curve, the rudder angle
of attack is not as large as during the previous maneuver such
that a left-rudder ram pressure of about 1000 psi (6,0 X 106
in.-lb on the Qo curve) is required to reach the 35L position
with no evidence of the rudder stalling; this is in considerable
contrast to the violent stall conditions experienced during the
25L to 30R maneuver. However, as may be seen from Fig.
29, a breakdown of lift would have been predicted at higher
rudder torque values.
After the 35L rudder position is reached (in the vicinity of

Rudder Torque Prediction

I
I

I
u
I

/ /

_.....

-~-"

I
f

Fig. 33 Ahead torque curves for


Ship C; ship speed of 15.5 knots

'x'
%
%

-6

-8

-10
0

10

15

20

25

30

35

RUODER ANGLE OF ATTACK (DEG)

4.

I
I
3'

2,

n
~

l
I

QF ,.,,,.~

Fig. 34 Ahead torque curves for


Ship D; ship speed of 20.9 knots

""

/"~

,,

\
\

-3
0

10

15

20

25

30

35

RUDDER ANGLE OF ATTACK (DEG)

Rudder Torque Prediction

47

I
I
e
I

'

-'

I
I

I
I

1
I

-2

,~

,~
\ ,

I
-4

-5
0

10

15

20

25

30

35

RUODERANGLE OF ATTACK (OEG)

Fig. 35

Ahead torque curves for Ship E; ship speed of 23.1 knots

the 80-second mark), the changing ships heading again reduces


the rudder angle of attack, a transition is made from the Qo to
the Qn curve, and the QH curve is followed until the rudder
is ordered to the 35R position, whereupon a transition is made
to the QR curve and the cycle is repeated.
As a general observation it is noted that the predicted magnitude of the negative torque region of the Qo curve is not
consistently reflected in the pressure recordings. The crossflow of water onto the rudder may be partially responsible for
the failure of the negative torque to appear during Z-maneuvers; but this explanation does not apply in other cases. During
the initial rudder movement, for example, the predicted
magnitude of the negative region of the Qo curve does not
consistently appear. There has been no entirely adequate explanation for this occurrence, but it is evident that the predicted
negative torques cannot be ignored or discounted because on
occasions they do appear as predicted. During restoring rudder
movements (when the negative hydrodynamic torque during
displacing movements becomes positive), there is a high
probability that the torque observed will be of the predicted
magnitude. Such being the case, negative torque predictions
must be given due respect.
The astern trial data presented by Figs. 52 through 60 for
48

Ship A indicate that after an initial positive (left-rudder) torque


is used to displace the rudder from the center-line position in
the left-rudder direction, the rudder takes charge, with restraining (negative) torques being required when subsequently
displacing the rudder. The heading of the ship changes very
little during astern maneuvering, and as the rudder maneuvers
are made, the unsteady characteristics of the pressure traces
indicate that the rudder promptly reaches stall conditions in
the alternate positions. As may be seen from Fig. 37, stall
conditions are expected to occur at angles of attack exceeding
approximately 21 deg and the astern trial data plotted in Figs.
52 through 60 generally support that prediction.
It may be seen that the maximum astern pressures recorded
for Ship A range between 800 and 1400 psi, which corresponds
to a torque range of 4.8 to 8.4 106 in.-lb. There is, however,
one pressure peak of 2100 psi (12.6 106 in.-lb) that occurs near
the 127-second mark in Fig. 60; in no other instance does the
pressure reach that magnitude. Following earlier practices,
the only pressure data recorded for that maneuver would have
been a reading of 2100 psi, and the design of subsequent similar
ships would have been based upon that data point. No explanation can be given for that pressure peak unless it resulted from
a rise in pressure, which often occurs when the rudder is ordered

Rudder Torque Prediction

#
#

4'

/f/,-

j/

I
I

I
I
QF
~,

Fig. 36 Ahead torque curves for


Ship F; ship speed of 23.6 knots

~-,"

0
(:3

,x,.

...

\,

/ // '

-1

-2

'

'

-4
15

20

25

30

35

30

35

RUDDER ANGLE OF ATTACK (DEG)

10
JOESSEL
K=I

j
/;
.._1
z
..~

QR, QF, AND QD ARE BASED


ON DTMR 933 TEST DATA

//

%
Fig. 37 Astern torque curves for
Ship A; ship speed of 10 knots

rhr J

QF
0

-2

-4

-6

-8
0

10

15

20

25

RUDDER ANGLE OF ATTACK (DEG)

Rudder Torque Prediction

49

20

,
K=I
16

S'

12

fw

r
QR, QH' OF'AND QDARE BASED
ON DXMB933 TEST DATA

//

x
z

QF
0

-4

-8

-12

-16

F'---C:

-18
5

10

15

20

25

30

35

RUDDER ANGLE OF ATTACK (DEG)

Fig. 38

Astern torque curves for Ship B; ship speed of 4.6 knots

from a fixed position, being superimposed upon an undulation


peak (perhaps caused by a wave slap) which happened to occur
at that instant
Attempts to use the astern data recorded for Ship A in the
evaluation of friction coefficients, and other investigations, were
unsuccessful due to the erratic nature of the pressure recordings,
which is believed to have been caused by a breakdown of rudder lift.
Ship B trial results
Due to the very large size of Ship B, the response to even
hardover rudder angles is very slow. For example, when the
rudder is thrown hardover to 35 at 15.5 knots, the ship's
heading changes 2 deg at the end of 20 seconds and only 8 deg
at the end of 40 seconds It would appear, therefore, that
during the initial rudder movement, the rudder angle of attack
should nearly equal the rudder angle
Figures 61 and 62 are the full-power trial runs for Ship B with
the rudder evolutions reversed. Questions often arise with
regard to the symmetry of the rudder torque requirements, and
these figures are interesting to study from that standpoint.
During the 0 to 85L rudder movement in Fig. 61, the recording began after the rudder motion had started, but it appears that a left-rudder pressure of about 500 psi may have been
required to initiate rudder movement, after which the pressure
reversed and about 250 psi was required at the 25L position;
this point corresponds to a torque of - 1 0 . 5 106 in.-lb on the
50

Qo curve shown by Fig. 80. As the ordered rudder angle is


held and the ship slowly turns, a transition is made to the Q n
curve and the pressure rises to about 600 psi ( - 2 7 X 106 in.-lb
of torque) when the rudder is ordered to the opposite side.
It is believed that Ship B was not fully settled into a straight
course with zero rudder angle when the evolution shown by Fig.
62 began. Instead, it appears that about 250 psi of right-rudder
pressure was required to hold the rudder in the centerline position, and this effect gradually diminished until it finally dissipated around the 40-second mark, after which the pressure
required to hold the initial hardover rudder angle tends to
converge to the same value in Figs. 61 and 62.
It is apparent that there is a breakdown of rudder lift during
the first rudder movement in both Figs. 61 and 62. It is difficult to identify the rudder angle at which the breakdown occurs
due to the pressure required to hold the centerline position
when the movements began, but the predicted angle of 27 deg
appears to be a good estimate.
When the opposite hardover angle is ordered at the 45-second
mark of Fig. 62, a transition is made to the QR curve and the
pressure goes to 1000 psi, corresponding to a torque of 37.4 x
106 in.-lb; and as the QR curve is followed to the origin, the peak
pressure reached is 1350 psi (43.6 106 in.-lb). The presence
of the negative-torque region of the Qo curve is clearly evident
in the vicinity of the 60-second mark of Fig. 62 with the
right-rudder pressure reaching 600 psi (18.7 X 106 in.-lb). The
pressure then reverses, with 500 psi of left-rudder pressure re-

Rudder Torque Prediction

12

'OES E'--

/
6

~t

. . . .

OR, OF,AND QD ARE BASED


ON DTMB933 TEST DATA

A
%
, = ~ . =QF~

.==v,,,== = ~ = = ~ .'= n

-2

\
\

QD

-8

-10

10

15

20

25

30

35

RUDDERANGLE OF ATTACK (OEG)

Fig. 39

Astern torque,curves for Ship C; ship speed of 9.5 knots

quired to reac h the 35L position; this corresponds to a rudder


torque of 18.7 106 in.-lb and an apparent rudder angle of
attack of 29 deg which is 2 deg greater than the predicted stall
angle. In Fig. 61, the pressure trace was expected to follow the
negative region of the Qo curve in the vicinity of the 50-second
mark, but the pressure trace did not reverse in that area, as is
often the case. The rudder is seen to stall near the 60-second
mark of Fig. 61 at a pressure of about 200 psi, which is equivalent to a torque of 6.9 106 in.-lb on the QD curve; Fig. 30
shows that this is very close to the predicted stall angle.
It is interesting to compare the data from the first Z-maneuvers of Ship A and Ship B. The pressure patterns at first
appear to be quite similar, but upon closer observation it is seen
that the pressure peaks occur at different stages in the maneuvers and for different reasons.
At the end of the first Z-maneuver in Figs. 61 and 62 (near
the 75-second mark), it is seen that the hardover angle is not
held long enough for an equilibrium trend to be established
before a reversal is ordered. When the reverse hardover angle
is ordered, the characteristics of the pressure trace for the second
Z-maneuver in Fig. 62 are observed to be nearly the same as
those for the first Z-maneuver in Fig. 61, and conversely. The
final evolution of returning the rudder to the centerline position
also repeats tl~e earlier patterns.

Trial data were recorded also for Ship B at a reduced power


of 15 000 hp and these data are shown by Figs. 63 and 64. The
effects of friction are seen to be relatively large in comparison
with the pressure levels otherwise required, and this fact should
be considered when evaluating the importance of specific data
points. The pressure rises to about 500 psi as required to establish right-rudder motion, then reverses with about 100 psi
of left-rudder pressure required at the 30R position; this corresponds to a torque value of -4.7 106 in.-lb on the QD curve
shown by Fig. 31. After the ordered angle is reached, a transition is made to the Qn curve, and as a drift angle is developed
with a consequent decrease in rudder angle of attack, a pressure
value of about 300 psi (-13.5 X 106 in.-lb on the Q n curve) is
reached when the rudder is ordered to the 35L position. At
that time the pressure rises to about 600 psi, which corresponds
to a torque of 22.5 106 in.-lb on the Qn curve. As the QR
curve is followed to the origin, the pressure reaches a maximum
value of about 700 psi at a rudder angle of 25R; the equivalent
torque at that time is 22.6 X 106 in.-lb. The negative region
of the Qo curve is completely masked as the rudder progresses
in the left-rudder direction to the 35L position. Although not
clearly evident from the data, the rudder lift has broken down
well before the 35L position is reached; but due to the lowpressure values involved, the rudder angle at which the

Rudder Torque Prediction

51

,/
#

y1

K=I

7----

OR' OF' ANO OD ARE 6~ED


. ON OTMB 933 TEST DATA

QF
z
c3
i=:

-1

-3

\
10

15

20

25

30

35

RUDDER ANGLE OF ATTACK (BEG)

Fig. 40

Astern torque curves for Ship D; shipspeedof 10.5 knots

breakdown occurs cannot be determined. While the 35L


angle is held, the pressure slowly rises to about 100 psi rightrudder (corresponding to about - 4 . 5 X 106 in.-lb on the Qn
curve); and the cycle is repeated when the :35R angle is ordered.
The torque required to accelerate the rudder up to the steady
turning rate is sustained for only about one second and is generally so small as to be negligible. However, for Ship B, due
to the size of the rudder and the otherwise low pressure levels
during the astern trials shown by Fig. 65, the pressure peaks
required to accelerate the rudder are clearly evident. With
an allowance made for the entrained water, the torque required
to accelerate the rudder on Ship B is 4 106 in.- lb, which
corresponds to 107 psi at a hardover position; therefore, this
consideration should be borne in mind when evaluating the
pressure peaks.
Figure 65 shows that about :300 psi of left-rudder pressure
is required as the rudder initially passes the 20R position
during astern trials, and this corresponds to a torque of -11.5
X 106 in.-lb on the Qo curve shown by Fig. 88. After the 35R
ordered angle is reached and held, the pressure slowly declines.
to 200 psi, corresponding to - 9 106 in.-lb on the Q n curve;
then the rudder is ordered to the :35L position with a pressure
rise to 600 psi. Recognizing that about 100 psi is required to
accelerate the rudder, the corresponding torque is 18.7 106
in.-lb on the QR curve in Fig. :38.
A negligible amount o[ pressure is seen to be required to hold
52

the rudder in the 35L position, and the only pressure rise of
significance occurred when the rudder was ordered to the 35 R
position. The pressure traces for the subsequent maneuvers
are nearly identical to those recorded earlier.
Ship C trial results
The data presented for Ship C provide an opportunity to
make several observations which are noteworthy. As a means
of improving maneuverability, Ship C was designed with a
45-deg hardover capability, and such extreme rudder angles
should give an indication of the stall characteristics of the
rudder. Also, comparisons may be made with hydrodynamic
torque extrapolations as made by two model basins on the basis
of model test results. Both model tests were conducted by restraining the model such that it could not turn, and then measuring the observed hydrodynamic torque as the rudder was
held throughout a range of angles. A plot of the model basin
predictions and the comparable hydrodynamic torque calculated (see Table 8) is shown in Fig. 81. Both model basins are
seen to predict some degree of asymmetry in the basins are seen
to predict some degree of asymmetry in the data, but the calculated prediction is not in sufficient detail to consider asymmetry. As indicated by the dashed lines, calculations based on
cross-plotted DTMB 988 test data indicate an onset of stall
conditions at a 81-deg angle of attack. The curvature in the
model basin data, which commences in the vicinity of :31 deg,
(text continued on page 66)

Rudder Torque Prediction

K=I

f
~

QR' OF' AND Oo ARE BASED


ON DTMB933 TEST DATA

QF

z
-.~ 0

cJ
-2

-4

-6

-8
0

10

15

20

25

30

35

RUDDERANGLE OF ATTACK (DEG)

Astern torque curves for Ship E; ship speed of 12.25 knots

Fig. 41

8"

o. ~........---~

JOESSEL
K=I

6.

4'

QR'QF, AND QD ARE BASED


ON DTMB933 TESTDATA

QF

i,-

-2

-4

-6'

-8

10

15

20

25

30

35

RUDDERANGLE OF ATTACK (DEG)


Fig. 42

Astern torque curves for Ship F; ship speed of 12.25 knots


Rudder Torque Prediction

53

C,,n

3000 .

@ 2000-

<

- - - -

IO00

"~

~E--

"

~-,'~--~'-~

--k

1000 -

\ f "

/
"~,,_~1/

_m

I = 2o0o-~,
3000 -

- 0

w
0

lJ

<

xJ

- -

I0

20

40

30

50

60

70

80

90

100

110

130

140

150

160

170

leO

TIME ($ECGNOS)

Fig. 43

3000- i

/ r

2000z

~ . ~

APJ~

--RUDDER ANGLE

lOOO-

_8

(--

Ahead trial data for Ship A1; ship speed of 22.6 knots

J
t

0-

:l

I000-

2ooo-

\...,

t.%.--RAM

j/

%___

50

-25

-o

-25

-50

<

PRESSURE

~"

3000-

~-

10

20

30

40

60

50

70

80

90

I00

no

120

130

140

150

160

170

leo

TIME (SECONDS)

Z3

Fig, 44

Ahead trial data for Ship A1; ship speed of 22.6 knots

3000-

2000--

fv'

RUDDER A N GLE
~

lO00--

o-

I000-

f-

%.

t.F,

,~,.\

. iJ f' ~ , ~

~'~

(
R A M PRESSURE ~ ? "

<
e~

~-"

2000-

3000

I0

30

30

40

50

60

70

8to

90

100

It0

120

130

T~ME (5ECONOS)
Fig. 45

Ahead trial data for Ship A2; ship speed of 22.5 knots

140

15o

160

170

160

3000

v ~ 2000
/"IF,

~ ~

RUDDER ANGLE

lOOO

/'

//

.__~-J
_

j.

I000 -

/I "~\
R A M PRESSURE

\/-\

\ .11

~J
f

v ~ .

i/

""-'-

<
=.

2000-

3000
10

20

30

40

SO

60

70

80

90

100

110

120

130

140

150

160

170

IlO

190

TIME (SECONDS)

Fig. 46

Ahead trial data for Ship A2; ship speed of 22.5 knots

3000
~

2000-

N I000-

C~

- -RUDDER ANGLE

0
IOOO -

~ . J

<
'~-~RAM

PRESSURE~

2000C~

3OOC
10

20

30

40

50

70

6O

80

90

100

110

120

130

140

150

160

170

180

190

TIME (SECONDS)

Fig. 47

Ahead trial data for Ship A3; ship speed of 22.2 knots

3O00
--

2000-

~ ~

looo-

o-

f~

__.~/~RU

DDER ANGLE

....

i"

. / ~
v

/" ~.'~

"

x_.,_

z~
<

I000~ R A M

PRESSURE

2000300~

10

20

30

40

SO

60

70

80

90

lO0

|10

120

130

.ME (SECONOS)
C,n
C~

Fig. 48

Ahead trial data for Ship A3; ship speed of 22.2 knots

14o

15o

160

170

18o

19o

On

3000 - - -

.l~.

~ ~ 2000

50
-

1000-

RUDDER A N G L E

~--~-"

o ~

~
w

"\~"~ /

\_/

1000

g~

, 25

/.p--

~...,jl/

' ~ .

. ~

<

j//

2~

- R A M PRESSURE

5o

I ~ 2OOD
3000"

10

30

20

40

50

60

80

70

90

110

100

120

140

130

I$0

160

170

180

TIME (SECONDS)

Fig. 49 Aheadtrial data for Ship A4; ship speed of 22.6 knots

3000-

213
E
el.
Q_

I
r,

2000
Z

1ODD

fOOD

"0

~4.

/,\~

"P\

RUDD'ER ANGLE

, ~._~,

_J

I~

" ~'~--~.~/

rI

5o

,'X

\1\

~ k l ' \

',~
\

-*~

<

i=

K--iAMPmESSU
I

2000

Q.
3000
20

40

30

50

60

70

80

90

I00

110

130

120

140

150

160

170

UME ~SECONOS)

Fig. 50 Aheadtrial data for Ship A4; ship speed of 22.6 knots

30002000-

i
o
w

fj

1000

::III
I1~

0- /
\

1000

"./

RU DD ER A N G L E

,,,...,,

-'<<2

.i

~,,~,~.~

<

RAM P ~ E S S U R E ~

2000NOTE: N O D A T A FOR THE INITIAL ZERO TO 25 L M A N E U V E R

3000
O

10

20

30

40

5O

60

70

80

90

IO0

110

TIME (SECONDS)

Fig. 5 1

Ahead trial data for Ship A5; ship speed of 21.9 knots

120

130

140

150

160

3000

~'~ ~ 2000

50 ~
~RUDDER

IOO0
f

i" "~J~ ~~.~ " ~

ANGLE

\.-~,

,'~\_

.~J

25 ~ ~

~...~
^\

'

F\v/.J

,..,

1000

~..>-

/,,,

",, .,.,, J

25

<
~

$0

R A M P R ES $ U

~ ~I. 2000
300

I0

20

30

40

60

so

Fig. 52

70

80
90
TIME (SECONDS)

100

I10

120

130

140

150

170

160

Astern trial data for Ship A1; ship speed of 10 knots

3000

< m

C
Q.
Q.

1000

;,

"~

i-

PRESSURE

j~/"\l~" ~..I ~%. \ _ /

,~

~RUDDER

--25

= ~

/5~'xl - - O

1000

=, 2000

--5o

.~

<

ANGLE
--SO

Q.
3000

10

20

40

30

SO

60

70

00

90

100

110

120

130

140

150

160

17o

TIME(SECONDS)

Fig. 53

3000-

2000-

m 1000-

Astern trial data for Ship A1; ship speed of 10 knots

RUDDER ANGLE

,,

0<

1000-

20003000-

I0

20

30

40

50

"/
~-RAM

PRESSURE

60

~ ' - -

70

BO

90

100

110

120

T~ME(SECONDS)
c.n
--4

Fig. 54

~,--.

Astern trial data for. Ship A2; ship speed of 10 knots

130

140

150

160

170

O"1
CO

7 "........
;;

_,. ooo
fi

so ~
~ / ~ R U D D E R ANRLE

I ~ /'~I"

-2S

<

10DO-

2000

~-~,~

---

2s
PRESSURE
SO

<

3000

IO

20

30

40

so

7o

6o

80
90
TIME (SECONDS)

IOO

11o

12o

130

140

ISO

160

170

180

Astern trial clara for Ship A2; ship speed of 10 knots

Fig. 55

3000- I "
2000
ANGLE

~, ,DO0

(1)

_RUDDER
if

,J"-

"-'4
o
_O
C
D
-.13

"-"

N
<

I000

/~RAM

2000-3000-0

10

PRESSURE t

20

30

40

50

60

70

90
80
TIME (SECONDS)

100

I10

120

130

140

150

160

170

Astern trial data for Ship A3; ship speed of 10 knots

Fig. 5 6

3000

2000-

_~./--RU0t0ER ANGLE
=~

1000-

'

'

~\

FJ/',~,-'x~.,-~
o
~

~ " -

o- ~

~,/"v/

lOOOI

jf/'~"~--~_/~-

j ~ ,
<

j'

2OO03ODD- O

10

20

30

40

50

60

Fig. 57

70

80
90
TIME (SECONDS)

100

II0

120

Astern trial data for Ship A3; ship speed of 10 knots

130

140

ISO

160

170

3000~

-50 ~

2000-

.~s~R
Z

/f

U DDER ANGLE

1000-

o.

-2S

\J

\J

-0

7o
<

lOGO

2000

3000-

10

20

30

40

SO

60

70

80

90

100

120

110

130

140

150

160

-25

-5o

)RO

170

TIME (SECONDS)

Fig. 58

Astern trial data for Ship A4; ship speed of 10 knots

3000~3
E
0.
o_

2000-

1000-

--SO ~
. ~ ' - - R U O D E R ANGLE

IwC3

o-

"-'0

lOOOI

J
\,/" V~V..j

s
<

.j'"~jf~ j

--2s

20003000-

c)

20

Io

30

40

50

60

70

RO

90

100

110

120

130

140

150

,7o-

160

UME (SECONDS)
Fig. 59

Astern trial data for Ship A4; ship speed of 10 knots

3000
~
<

O
N

2000

1000

~_ "~

~./f~.

/',,r~

i ~ ,

I"

--RAM PRESSURE f . / ' ~


,~%

% /

f f ~ .

s ~ j,.l

~-jP\

o-

<

.=

1000

\t

~..,r
IE _,~

<

3000
0

I0

20

30

40

50

60

70

RO

90

100

I10

120

TIME (SECONDS)

(D

:i

L/

2OOO

Fig. 60

Astern trial data for Ship A5; ship speed of 10 knots

130

140

1SO

160

170

180

tSO0-

lOOO-

'"

_g
Z
w

-/~,',
"','k

-so

~ ~,

SO0-

i %,

/<--.ODD. .O!,

SOO-

~,,~

-is

..=.

'~,/'~,"

IOOO-

I
1500

20

10

30

50

4O

60

70

80

90

120

II0

100

130

140

T,ME(SECONDS)
Ahead trial data for Ship B; ship speed of 15.5 knots

Fig. 61

1SO0-

,jf ~ , ~ , ~j'~

OOO-

C~

- ,Do-.J \.. /--'f-

-t
o
t-CD

o- /

--

~,r\/.j,~v./-\d\_

rf \

S00-

"I3

IOOO-

<

"Ut

ISO0-

IO

20

30

40

SO

60

70

80

" 90

lOO

110

,120

130

140

1S0

TIME (SECONDS)

Fig. 62

Ahead trial data for Ship B; ship speed of 15.5 knots

1500-

w
N.

1000-

5DO-

O-

'~'JV

SO0
<

~N

RUDDER A N G L E

,~f

20

g
w

<

-2s

-SO

t
10

-25

"~ - - - - ~ .

10001500-

-50

50

60

70

80

90

.
I00

I10

TIME (SECONDS)

Fig. 63

Ahead trial data for Ship B; ship speed of 10.2 knots

120

130

140

150

160

,io

1500~

IO00-

500-

---....
~
w

RAM PRESSURE

k~

O-

sOo .
~

<

~J

RUDDER ANGLE

~ 1000-

150010

20

40

30

$0

60

70

80

90

I00

130

120

110

140

150

TIME (SECONDS)

Fig. 64

Ahead trial data for Ship B; ship Speed of 10.2 knots

1500-

30

CE
Q.

~g

--

lO00-

ff

a8

--I
o
..0
CD

<

soo-

~, ~

~
~---.M

PRESSURE

"~

1000C),.

I50O-

10

20

30

40

SO

60

70

90

80

100

Fig, 65

3000-

--

120

130

"

140

Astern trial data for Ship B; ship speed of 4.6 knots

I
--50

2000-

.M PR.SU,E--Y"

I000-

<

" ~

25 ~
--SO

200030000

//~"

i "'-o-~
10

20

30

40

50

60

70

80

90

100

TIME (SECONDS)
Ob

110

TIME (SECONDS)

Fig. 66

Ahead trial data for Ship C; ship speed of 19.5 knots

I10

120

130

140

03
I",,.1

3000-

2000-

z = ~ooo- -

/_.j

j\

....

7-,

o-

~ j ~
~

iooo-

"~'~"--~

z~<

\~..~

~ -

\\j q

=.. ~
2000-

~-'~- RUDDER ANGLE

<
3000- 0

10

20

30

40

SO

60

70

80

90

100

110

120

TIME (SECONDS)

Ahead trial data for Ship C; ship speed of 19.5 knots

Fig. 6 7

3000C
OC~

~ 200Oz

--t

Iooo-

o-

c-

/'~RAM

1000-

(I)

f
f

PRESSURE
J

<

2000-

~E
3000-

10

20

30

40

60

5O

70

80

90

100

110

120

130

140

IS0

TIME (SECONDS)

Ahead trial data for Ship C; ship speed of 19.1 knots

Fig. 6 8

3000--! -

2000

-50

-25

~ ~

, ~ R U D O E R ANGLE

/AJ ~ , . ~ .
--

<

3000

.
0

.
10

20

60

50

70

80

90

100

110

TIME (SECONDS)

Fig. 6 9

Ahead trial data for Ship C; ship speed of 15.5 knots

120

130

140

150

160

170

-o

o
z
<

-2s

~=

-SO

30002000z
=

IOOO

....

PRE.URE

I ~ f~'~.<

1 / ~ ~ .

--5O

~
z

--25

0-

,f/

1000-

~'~

z
<

I"

--2s

."

--50

RUODER ANGLE
~E
<
=

2ooo3000-

10

20

30

40

50

60

70

80

90

100

120

110

130

140

TIME (SECONDS)

Fig. 70

Ahead trial data for Ship C; ship speed

of 15.5

knots

3000~ 2000-

CQz

CD

--t
o

~-

IOOO-

cCD

>

~"~'~/<~

~RAM

PRESSURE

--SO

--25

~-"

~ J ' ~ j
o

1000-

R DOER ANGLI

----50

2000-

-~
3000-0

10

20

30 '

4C

SO

60

70

80

90

100

"

120

110

130

140

150--

TIME (SECONDS)

Fig. 71

f-

Ahead trial data for Ship C; ship speed of 15.5 knots

"RUDDER ANGLE

-....

c~

J m m m

mmjmll

m m m m

~>,.,

.=
~RAM

PRESSURE

=.
<
m
I0

20

30

40

SO

60

70

80

90

100

110

120

TIME (SECONDS)

O~
CO

Fig. 72

Astern trial data for Ship C; ship speed of 9.5 knots

130

140

150

160

170

180

(:33

3000~_. ~

2000-

"_~U"_"f

I000-

.~

~,l~

o-

-- ~"~'~\

10

20

IO

-0

,~

,/~/

~" 2OOO-

-so

t\ ~ / ~

lOOOI

RAM IPRESS

20

30

IO

20

30

-2S

I0

20

-SO

TIME (SECONDS)

Fig. 73

Ahead trial data for Ship D; ship speed of 20.9 knots

3000~

2OOO-

<d ~

C3O-

~- O
z ~" 1OOO-

_a

-=

cco

~=
.=

1 200o-

|\ / /

....

O-

I'~

iooo-

~ooo

~ j / r\

-RUDDER ANGLE

\,,,j,

- I~( .j~'

i-/-

P|ESSU~IE

/IIAM
\-

~'--~

L,'nL/"\j .'~"V""

20

30

40

,.

/^\ / "t . i

10

20

30

Fig. 74

40

50
0
TIME (SECONDS)

IO

20

30

40

10

SO

:=i
20

30

Astern trial data for Ship D; ship speed of 10.5 knots

15OOI000-- -

/f"i

\\_

RAMPRESSURE~ f~
.N

~.,r.

o-

-0

5OOI000-

1SO0-

,
0

IO

20

<

"V ',, ""

,.J

I0

.~

30

40

50

Fig. 75

-7s

-SO

\j

60

70
80
TIME (SECONDS)

90

1OO

Ahead trial data for Ship E; ship speed of 23.1 knots

110

120

130

140

is;

ISOO
~

I
/'\

IOOO-

i
Z

\I~..

-SO

/
"--

~,

$0~

~..~RUODER

ANGLE

RAM PRESSURE ~

-25

/.- ~></~"/~

\
\

\/-

"C~

-O

500-

-25

1000-

-50

:g

1500-

10

20

30

SO

4O

Fig. 7 6

60

70

80

90

IOO

liD

120

130

i
i

150

140

Ahead trial data for Ship E; ship speed of 23.1 knots

15OO-

c-

CD

--t
o

1000-

SOD-

/'

o-,

RUDDER ANG L

"RAM PRESSURE

P\

//

\ f"

I~

~-.~-

"~'/~

I=

~' i \
.,, -,,_,. \ d

"

,,--

",~/~

$O0-

c~E

1000-

"0

Q.

~E

1500-

10

20

30

40

SO

70

60

6"

Flg. 77

IO

90
TIME (SECOND/)

rJ

SOD-

\ ,,..

/...,

110

130

liD

l-RAMP.ESSURE/'~.
/

-o
I

\ / _ . , t/

rJ

-SO

500-

20

IO

20

30

I0

20

30

T~ME(SECONDS)

Fig. 78

160

-SO

t u l Yi'

)2

150

/r ~

o-

I000-

140

Astern trial data for Ship E; ship speed of 12.25 knots

I000-

"

1OO

Ahead trial data for Ship F; ship speed of 23.6 knots

20-

z~

170

IRO

(text continued from page 52)

[ S ] ] I I O ] Q ) ] ' I ~ N Y ~I]C]QI'I~I
/H~I~I

L:I:I1

J,\
-3

J
C~

L~i~

c~

(
\
(/
o

\\

i.~.

tOg

-!

E
ft

<

.-U.
0

C_

114OlU

L:I]I

(ISd) 1VILN3~I3JJI(] ]llNSS3~ld WVIJ

66

suggests a breakdown of lift; however, the lift breakdown is not


as abrupt as that indicated by the DTMB 933 model test
data.
Figures 66 and 67 are plots of the full-power-ahead trial data
for Ship C. Figure 66 shows that with the initial throw of the
rudder, the pressure reaches a maximum negative value of
about 300 psi (corresponding to a torque of about - 2 . 8 x 106
in.-lb on the Qo curve in Fig. 32); but the sense of the pressure
reverses, with 900 psi (9.4 X 106 in.-lb on the Qo curve) being
required to reach the 35L position. The pressure trace
suggests that the rudder has stalled; and referring to Fig. 32, it
is seen that stall may have occurred approximately as predicted.
As the 35L ordered angle is held, a transition is made to the
Q n curve and 200 psi right-rudder pressure (-2.3 106 in.-lb)
is required when the 35R order is given and the pressure immediately rises to 500 psi (5.2 X 106 in.-lb on the Qn curve).
The QR curve is followed to the origin and a restraining pressure
of 450 psi (-;3.8 x 106 in.-lb on the Qo curve) is required as the
rudder crosses the 10-deg right-rudder position. However, the
pressure reaches 1750 psi (16.6 X 106 in.-lb on the Qo curve)
as the rudder approaches the 95R position with indications that
there has been a breakdown of rudder lift. The maximum
apparent angle of attack observed was about 35 deg, which is
4 deg more than the predicted value. The rudder is reversed
to the &5L position before equilibrium is established, with 1000
psi (7.7 X 106 in.-lb) being reached at the peak of the QR curve.
The peak of the QR curve was reached at a torque value of 7.1
106 in.-lb when the rudder was subsequently restored to the
centerline position.
The run illustrated by Fig. 67 was intended to be the opposite
of Fig. 66; however, no data were obtained for the initial maneuver. During the 35R to 35L evolution, 700 psi (7.3 X 106
in.-lb) was recorded at the peak of the QR curve, and this is
consistent with other recordings; however, a maximum rightrudder pressure of 1200 psi (-9.4 X 106 in.-lb) was measured
as the Qo curve was followed out from the origin, and then the
pressure reversed, with a heavily pulsating pressure peaking
at 2150 psi (20.4 X 106 in.-lb on the QD curve) being required
before reaching the 35L ordered angle. Referring to Fig. 32,
it may be seen that the apparent angle of attack corresponding
to 20.4 X 106 in.-lb of torque is about 37 deg, or about 6 deg
beyond the stall angle predicted for more orderly flow conditions. During the subsequent 35L to 35R maneuver, the
pressures are generally low, with the largest being 600 psi (4.9
X 106 in.-lb) recorded at the peak of the QR curve. During the
final maneuver, 900 psi (9.4 X 10 in.-lb) is required to clear the
QR curve, and 1400 psi (10.9 X 106 in.-lb) is needed to hold the
rudder in the centerline position due to the ship's drift angle.
Figure 68 is a plot of data taken for Ship C with the rudder
exercised to alternate 45-deg rudder angles while at full power.
Although to opposite hand, the data for the initial rudder
movement in Figs. 66 and 68 are similar. After requiring 400
psi (-3.7 X 106 in.-lb) to clear the negative peak of the Qo
curve, there is a complete breakdown of rudder lift at a rudder
angle of about 30 deg (7.6 X 106 in.-lb of torque), which compares favorably with the prediction. Due to the breakdown
of rudder lift, the ram pressure decreases as the rudder angle
continues to the 45R position. As the ship develops an increasing drift angle, the rudder angle of attack is seen to decrease such that lift appears to have been restored near the
42-second mark when the rudder is ordered to the centerline
position., Before the order is received, 200 psi of lift-rudder
pressure (-3.1 X 106 in.-lb on the Qn curve) is required to hold
the 45R positin, and this corresponds to a 23-deg apparent
angle of attack. Immediately after the order is given, the
pressure rises to 500 psi (6.3 X 106 in.-lb), indicating a 22-deg
Rudder Torque Prediction

angle of attack on the Qn curve, and the QR curve is followed


to the origin, peaking at a torque value of 7.1 X 106 in.-lb. It
is seen that 1300 psi (10 106 in.-lb) is required to hold the
centerline position in the vicinity of the 70-second mark, and
this is approximately the same as that required at the end of the
run shown in Fig. 67.
The ship continues to have a substantial drift angle when the
rudder is ordered to the 45L position in Fig. 68, and the rudder
stalls at an angle of about 20 deg. The pressure peaks at a
pressure of about 1600 psi (16.8 X 106 in.-lb), which indicates
an apparent 35-deg angle of attack on the QD curve. The actual rudder angle of attack around the 100-second mark of Fig.
68 is the sum of the 45-deg ordered rudder angle and the remaining effects of the previous drift angle, and the sum of these
two is seen to be so far beyond the rudder stall angle that there
continues to be a breakdown of lift when the rudder is ordered
to the centerline position. About 700 psi of left-rudder pressure
(10.7 X 106 in.-lb on the Q n curve) is required to hold the
rudder in the 45L position when the centerline order is given;
the waves in the pressure curve after the centerline order is
given correspond to 430 psi (-6.8 106 in.-lb), 350 psi (-5.1
106 in.-lb), and 250 psi (-3.1 X 106 in.-lb, all on the QR curve)
before lift is reestablished. The QR curve is then followed to
the origin, peaking at 950 psi (7.8 X 106 in.-lb). No pressure
was required when the rudder was restored to the centerline
position, and this is a significant observation since it indicates
that there was, at most, a small ship drift angle when lift was
resumed. Therefore, the rudder angle at that time would be
approximately equal to the rudder angle of attack.
Figures 69 and 70 are pressure traces at 16 000 shp for Ship
C. The first maneuver in Fig. 69 was actually the first test for
that ship, and it is seen that the rudder was exercised at a cautious rate. However, the design rudder rate was used to reverse
the rudder near the 40-second mark, with 600 psi (5.7 106
in.-lb) being required to cross the peak of the QR curve, and 800

Fig. 80

Plan view of the rudder arrangement for Ship A

psi (-6.7 X 106 in.-lb) of right-rudder pressure being required


at the negative peak of the Qo curve. The rudder stalls as the
35L ordered angle is approached, with the ram pressure pulsating to a maximum value of 1300 psi (12.3 X 106 in.-lb). As
was the case during the full-power run (Fig. 67), the apparent
angle of attack reaches 37 deg, which is 6 deg greater than the
predicted stall angle corresponding to normal ahead operations.
The data following the 70-second mark of Fig. 69 are similar
to established patterns.
Approximately 1000 psi of right-rudder pressure was required to hold the rudder in the centerline position when the
run shown by Fig. 70 began, but this effect is seen to dissipate
with Fig. 70, becoming a substantial repeat of Fig. 69.
The Ship C rudder was also exercised to the 45-deg capability
at reduced power, and Fig. 71 shows that the required ram
pressure is lower than may have been expected. The data
clearly indicate that the rudder stalls, with a maximum of 600
psi (8.2 106 in.-lb) being required to attain the initial 45L
rudder angle; this corresponds to an apparent angle of attack
of about 34 deg. When the rudder is ordered to the 45R position, the rudder again stalls as the ordered angle is approached,
with 1000 psi (11.8 X 106 in.-lb) being required as the rudder
passes the 40R position; this corresponds to an apparent angle

2O

'/

16

PREDICTION

-MODEL
BASINY

50
.10E

IOE01

~,%~ , ~ /

-8

~'--MODEL
BASIN2

-12
Fig, 81

Comparison of hydrodynamic torque predictions from tests conducted at two model basins
with the calculated prediction for Ship C at a ship speed of 19.5 knots
Rudder Torque Prediction

67

of attack of about 37 deg, which agrees with the data taken


during the full-power trials. It is seen that Fig. 71 generally
follows established patterns for Ship C.
The astern trial data for Ship C (Fig. 72) indicate that the
rudder stalls at an angle of attack which approximates the
21-deg prediction shown by Fig. $9. However, the rudder
torque corresponding to the recorded data ranges up to about
8.5 x 106 in.-lb, which is lower than the maximum torques
predicted. It is believed that the astern trial speed was less than
the estimated 9.5 knots used in the calculations.

Ship D trial results


Figures 73 and 74 are the trial data for Ship D and reflect the
point of view held at the time the data were taken. That is, the
o n l y data considered meaningful were those taken while the
rudder was moving, and sufficient data were considered to be
recorded during one series of maneuvers. The amount of
confidence that can be associated with the ahead-trial data in
Fig. 73 is not known; however, they are the only data available
for Ship D. It appears that the ship was not on a straight course
when the ahead trials began in that approximately 800 psi of
right-rudder pressure was required to hold the rudder on centerline; however, once right-rudder movement was established,
the pressure reversed, with 350 psi of left-rudder pressure being
required at the 25R position (-0.8 X 106 in.-lb). Referring
to Fig. 34, it is seen that a breakdown of lift was predicted to
occur at rudder angles of attack exceeding 26 deg, and it appears that this prediction is substantiated. When the 35L
order is given, the pressure initially rises to about 1200 psi
(corresponding to 8 x 106 in.-lb on the QR curve), but the
pressure later reverses and peaks at 1500 psi (-;3 X 10# in.-lb)
as the QD curve is followed out from the origin. About 25
seconds into the second maneuver, the pressure reverses again
.as the Q o curve is followed out to angles of attack exceeding
the stall angle, and about 350 psi (0.8 X 106 in.-lb or torque) is
required at the S0L position. The pressure pattern for this
maneuver is seen to have the same characteristics as that of the
third ahead maneuver for Ship A.
The third maneuver for Ship D also has a pressure pattern
similar to thesecond maneuver for Ship A in that the trace
initially rises to about 2100 psi (4,8 X 106 in.-lb on the QR
curve), then turns in the negative direction; but instead of reflecting the negative torque region of the Qo curve, the trace
turns again to peak at 1600 psi (8,6 x 106 in.-lb on the OD
curve). Referring to Fig. 34 it may be seen that the apparent
angle of attack reached is about 95 deg, or 9 degrees greater
than the stall angle predicted for more orderly flow conditions.
The fourth maneuver for Ship D is seen to be largely a repeat
of the first half of the second maneuver.
No data were recorded for the initial 25L astern maneuver
in Fig. 74; however, it appears that Ship D follows the characteristics of other ships in going to alternate stall conditions
during astern maneuvers. The maximum pressures recorded
are seen to generally be in a range of from about 1700 to 2300
psi. These pressures correspond to torques ranging from 4 to
5 X 106 in.-lb. This range of maximum torques is in good
agreement with the astern-torque predictions plotted in Fig.
40.
Ship E trial results
A comparison of Table 1, which is a summary of steering gear
pressure data as recorded by trial observers, and the pressure
traces shown in Figs. 75, 76, and 77 leads to a situation typical
of those encountered when attempting to draw firm conclusions
from steering gear trial data. The highest arid second-highest
pressures recorded by the trial observer during the astern trials
68

for the third ship of the Ship E class are shown in Table 1 to be
1200 and 1000 psi, respectively. Figure 77 was recorded
during the same astern trial; therefore, it should be possible to
confirm these data points. It is seen that the highest differential
pressure reached on Fig. 77 is 1000 psi and occurs near the 9,
33, and 42-second marks. Since a back pressure of 200 psi is
a reasonable value, the 1200-psi peak pressure is confirmed.
Also, the back pressure can drop to very low values; consequently, the second-highest peak of 1000 psi cannot be refuted.
Similarly, the second-highest peak of 1200 psi which was reported for the ahead direction is confirmed by the recording
at the 60-second mark of Fig. 75. However, the 1900-psi
reading given in Table 1 as the highest peak pressure during
ahead steering corresponds to the 1100-psi differential pressure
peak at the 13-second mark of Fig. 76, and cannot be confirmed
because an 800-psi back pressure is not credible. A choice must
be made between the 1900-psi observed pressure data point in
Table 1 and the recorded data in Fig. 76. In view of the numerous opportunities for errors by data observers when reading
data from pressure gages, the 1900-psi data point must be discounted. Without the continuously recorded pressure traces,
however, the apparently extraneous 1900-psi data point could
easily have influenced the rating of future steering gear machinery.
In both Figs. 75 and 76 an unusually large positive torque is
seen to be required to reach the initially ordered hardover angle;
800 psi or 3.4 x 106 in.-lb is required in Fig. 75 and 1100 psi or
4.7 X 106 in.-lb is required in Fig. 76. Referring to Fig. 35,
these torque values are seen to correspond to apparent angles
of attack of 81 and 34 deg, respectively, which are well beyond
the predicted stall angle of 27 deg. Data for some of the earlier
ships showed that during Z-maneuvers the breakdown of rudder
lift may be delayed to angles of attack significantly greater than
that predicted on the basis of model test data, but the lift
breakdown occurred nearly as predicted during the initial
rudder movement. However, the data for Ship E, which is a
C4 cargo ship (basically a Mariner hull), show that the onset of
stall for that ship is also delayed during the initial rudder
movement to approximately the same extent as during subsequent Z-maneuvers. That is, the same pressure recording of
1100 psi was made at the 60-second mark of Fig. 75 and the
1S-second mark of Fig. 76 even though the angle of attack was
much larger in the former case. Displacing torques required
to reach the hardover position range between 8.3 and 5.2 X 106
in,-lb for the maneuvers shown in Figs. 75 and 76. Figures 75
and 76 also offer the opportunity to study the symmetrical aspects of the data, and it is seen that the pressure traces are very
similar.
Figure 77 indicates that during astern maneuvers the rudder
for Ship E experiences a breakdown of lift at rudder angles
above about 20 deg. The maximum operating pressure required for astern maneuvering ranges from about 600 to 1000
psi, corresponding to rudder torques of from 3.5 to 4.75 X 106
in.-|b. Referring to Fig. 41, it is seen that the maximum predicted torque is about 7 X 106 in.-lb; with such a difference
between the predicted and actual torques, the only plausible
explanation is that the actual astern trial speed was less than the
estimated speed of 12,25 knots.

Ship F trial results


Figure 78 provides the ahead trial data for Ship F, and the
first evolution has the same characteristics as those in Figs. 75
and 76 for Ship E. That is, a pressure of 1100 psi (5.4 106
in.-lb) was required to reach 35R during the first evolution,
and from Fig. $6 it is seen that the apparent angle'of attack
corresponding to this torque is about 36 deg, which is 7 deg
greater than the predicted stall angle. Torques required to

Rudder Torque Prediction

reach hardover positions during subsequent maneuvers were


2.9 and 3.8 106 in.-lb (32- and 33.5-deg apparent angles of
attack, respectively), which are approximately as expected.
Due to the small amount of data taken, no judgment can be
made regarding the repeatability of the first maneuver, but
considering the similarity of the stern arrangements of Ships
E and F (see Figs. 26 and 27) and the pattern of.the pressure
traces recorded for beth ships during the first evolution, the data
points must be accepted as valid and characteristic of b o t h
ships.
The astern trial data, Fig. 79, for Ship F supports the hypothesis that full astern speed was not established for Ship E
when the astern steering.trial for that ship was conducted. For
Ship F the maximum astern torques ranged from about 3.6 to
7.8 106 in.-lb, which compare favorably with the astern
torque predictions plotted in Fig. 42. Due to the close similarity of Ships E and F, the torque predictions are nearly the
same and, therefore, the astern trial data would be expected to
correspond to similar torques if the ship speeds had been as
anticipated.

a rudder which are outside of a sternward projection of the


propeller disk area.
10. During astern operation, ship rudders generally stall
at rudder angles of attack above approximately 21 to 27 deg.
11. Envelopes of maximum astern torque requirements can
be established with good confidence by using either cross-faired
DTMB Report 933 astern data or the Joessel equations.

References
1 Schoenherr, K. E., "A Program for an Investigation of the
Rudder-Torque Problem," Marine Technology, Vol. 2, No. 3, July
1965.
2 "Steering Gear," Section 561, General Specifications for Ships
of the United States Navy, Department of the Navy, Naval Ship Engineering Center, 1 Jan. 1978.
3 "'Steering Gears, Electro-Hydraulic, Marine," Military Specification MIL-S-17803D, Department of the Navy, Naval Ship Engineering Center, 29 April 1960.
4 "Steering Gear,'" Section 9220-2, Specifications for Building
Attack Cargo Ships LKA 113, 3 Nov. 1969.

5 Standard Specifications for Merchant Ship Construction,

Conclusions
The more significant conclusions that can be drawn from this
investigation are as follows:
1. The semi-empirical equations for rudder torque prediction, which are presented in DTMB Report 933, increasingly
understate the hydrodynamic torque requirements at rudder
angles of attack exceeding approximately 15 deg, resulting in
a substantial error at the higher angles of attack.
2. Cross-faired plots of the test data presented in DTMB
Report 933 can be used as the basis for rudder torque predictions with reasonable accuracy.
3. The breakdown of rudder lift, or stall, is an important
consideration in the establishment of rudder torque requirements. A correlation of calculated rudder torque predictions
based on cross-faired DTMB Report 933 model test data with
sea trial experience shows that there is good agreement between
the predicted angle of attack at which stall occurs and trial data
for very large crude carriers; however, for smaller and more
maneuverable ships, the onset of stall may be delayed to angles
significantly larger than predicted.
4. Due to the pulsating nature of the rudder forces, the
effective friction coefficients in rudder support bearings and
between the moving parts of steering gear ram units are approximately 50 percent of the values normally expected for
rigid rudder-support arrangements typical of the spade and
horn types. However, more flexible rudder support arrangements, such as those using a shoe support, may entail bearing
friction coefficient values of the expected magnitude.
5. Ram pressure recordings during steering trials indicate
that the pattern of the pressure traces is generally consistent for
repeated runs.
6. Ram pressure recordings for five sister ships show that
pressure variations due to construction tolerances are not significant.
7. Semi-empirical relationships, based on model test data
recorded for flapped control surfaces, can be used to predict
the force and hinge-moment on that part of a horn rudder
trailing the horn.
8. Predicted regions of negative rudder torque (which in. dicate that rudder restraint is required), are not consistently
reflected in trial data to the magnitude calculated, but such
negative torques do appear sufficiently to confirm their existence.
9. Correlations with trial data show that the augmented
water velocity in the propeller race can impinge upon parts of

United States Department of Commerce, Maritime Administration,


Office of Ship Construction, Dec. 1972.
6 "Code for Sea Trials 1973," Technical and Research Code C-2,
prepared by Panel M-19 (Ship Trials), SNAME, Jan. 1974.
7 Rulesfor Building and Classing Steel Vessels, American Bureau of Shipping, New York, 1980.
8 "Code of Federal Regulations," 46 Shipping, Part 58.25, U.S.
Coast Guard, Department of Transportation, 1 Oct. 1977.
9 Jaeger, H. E., "'Approximate Calculation of Rudder Torque and
Rudder Pressures," International Shipbuilding Progress, Vol. 2, No.
10, 1955.
10 Attwood, E. L., Pengelly, H. S., and Sims, A. J., Theoretical
Naval Architecture, Longmans, Green and Co., London, 1953.
11 Schoenherr, K. E., "Steering" in Principles of Naval Architecture, Vol. 2, H. E. Rossell and L. B. Chapman, Eds., SNAME,
1958.
12 Whicker,L. F. and Fehlner, L. F., "Free-Stream Characteristics
of a Family of Low-Aspect-Ratio, All-Movable Control Surfaces for
Application to Ship Design," David Taylor Model Basin Report 933,
Dec. 1958.
13 Schoenherr, K. E., "Propulsion and Propellers" in Principles
of Naval Architecture, Vol. 2, H. E. Rossell and L. B. Chapman, Eds.,
SNAME, 1958.
14 Mandel, P., "'Ship Maneuvering and Control" in Principles of
Naval Architecture, J. P. Comstock, Ed., SNAME, 1967.
15 Harper, J. J. and Simitses, G. J., "Effect of a Simulated Submarine Hull on the Characteristics of All-Movable Control Surfaces,'"
Georgia Institute of Technology Report 439, Aug. 1959.
16 Taplin, A., "Notes on Rudder Design Practice" in Proceedings,
First Symposium on Ship Maneuverability, David Taylor Model Basin,
DTMB Report 1461, Oct. 1960.
17 Smith, Watt V., "Properties of Lignum Vitae and Its Use as a
Bearing Material," T&R Bulletin No. 3-21, SNAME, 1967.
18 Smith,I. W. and Nickerson, A. M., "Hull Machinery" in Marine
Engineering, R. L. Harrington, Ed., SNAME, 1971.
19 Gover, S. C. and Olson, C. R., "A Method for Predicting the
Torque of Semibalaneed Centerline Rudders on Multiple-Screw Ships,"
David Taylor Model Basin Report 915, Nov. 1965.
20 Bowers,A. A., "Wind Tunnel Investigation of the Characteristics of a Flapped Control Surface Mounted on a Simulated Submarine
Hull," University of Maryland Wind Tunnel Report No. 259, June
1959.
21 Ames,M. B. and Scars, R. I., "Determination of Control-Surface
Characteristics from NACA Plain-Flap and Tab Data," NACA Report
721, National Advisory.Committee for Aeronautics, 30 Dec. 1940.
22 Cahill, J. F., ' Summary of Section Data on Trailing-Edge
High-Lift Devices," NACA Report 938, National Advisory Committee
for Aeronautics, 1949.
23 Sheffield, R., "'Development of a Technical Practice for Rudders and Diving Planes, Part II, Torque Predictions," NAVSEC Report
61:36-74-272, Naval Ship Engineering Center, 30 Aug. 1974.
24 Hoerner, S. F. and Borst, H. V., Fluid-Dynamic Lift, Heerner
Fluid Dynamics, Bricktown, N.J., 1975.
25 Eda, H. and Crane, C. L., Jr., "'Steering Characteristics of Ships
in Calm Water and Waves," TRANS.SNAME, Vol. 73, 1965.

Rudder Torque Prediction

69

Discussion
J. R. Kane, Member

Although it is only one of a number of factors affecting the


turning of a ship, rudder torque can be very important, parficularly if you don't have enough steering engine provided to
put the rudder hard down at full speed of the ship in the specified time. This has been known to happen on at least one
important naval capital ship, which may account for the interest
the U.S. Navy has taken in research on this matter.
Hydrodynamically, the problem is bound to be quite complex involving as it does highly unsteady and turbulent flow
regimes, low-aspect-ratio control surfaces which are not satisfactorily treated by two-dimensional flow theory, frequent large
angles of attack such that the rudder is operating in the stall
condition during many emergency maneuvers, and unknown
interactions between both the ship's wake and the rudder(s) and
the propeller(s) and the ruddei(s).
Steering test data taken on sea trials are notoriously unsteady
and difficult to repeat closely, making it difficult to interpret
into precise theory. This is not necessarily because of inadequate instrumentation or poor conduct of such tests, but rather
because of the inevitable factors which apply in practical operation at sea, such as the effects of wind and seas, the angle of
encounter with waves, the directional stability of the ship and
the possibility of initial swing or drift, the coordination of
rudder movements with multiple rudders, and the actions of
the helmsman.
Thus, the engineer charged with making the rudder torque
prediction must work as best he can with the theory, experimental tests and trial results available to him, with results substantially less elegant and reliable than he would like. I think
the engineering approach taken by the author is excellent, and
that he has very carefully indicated those areas where the
analysis and synthesis are least certain. His conclusions appear
to be well supported in the text, and quite useful.
I like the author's statement, " . . . but due to the expense associated with back-fitting additional steering gear capacity, a
modest amount of over-capacity is a prudent objective." No
doubt many ships carry about a steering engine a bit oversize
for what is normally required. But when the wind blows its
hardest and the rocks show up on the leeward bow, it probably
seems like a good idea to the man on the bridge.
Edwin T. Cangin, 3 Visitor

[The views expressed herein are the opinions of the discusser and not
necessarily those of the Department of Commerce or The Maritime
Administration. ]
The author has provided a comprehensive study on rudder
torque. Not only did he review and advance analytical predictions, but also wrestled with correlation of shipboard data
with the predicted values.
The detailed analysis of rudder torque predictions leading
to a method for calculating torques on horn-type rudders using
DTMB 93;3 is definitely an advance in the state of the art. In
fact, this method could be applied with the rudder divided into
three sections. The area behind the rudder horn could be one
section; the second section could be the portion aft of the pintle
bearing, with the third section being the lower portion of the
rudder. The reason for this suggestion is that the area behind
the pintle bearing has less movable area, and by including it in
the upper section the result could be an overstatement of the
required torque.
3 Mechanical engineer, Maritime Administration, Washington,
D.C.
70

In correlating the shipboard data with the predicted values,


the author might want to investigate why the hydraulic
equipment in the steering system is one of the most susceptible
susceptible items to damage even though the unit operates
below full capacity. Does the cyclic nature of the system in the
steering gear flat prevent the use of smaller steering gears that
can be occasionally overloaded to handle peak demands? The
data should be able to give some clues to the nature of this
problem. Determining whether or not steering gears can be
built based on the lower torque values seen most often by the
system or one which is to be designed for peak values in the
system is an important result that could come from additional
study of these accumulated data.
We have posed the preceding question for the author to encourage him to continue his work in the area of the rudder
torque and to look into the steering system required to overcome
this torque.
A. Taplin, Member

Mr. Harrington has written a very comprehensive paper.


My discussion is limited to two questions and four comments.
The author defines negative and positive torques, and the
definition involves knowing what direction of movement is
desired. My first question is: Has the author considered using
a more simple definition, namely, that negative torque exists
when the center of pressure is forward of the stock, and positive
torque exists when the center of pressure is aft?
Hydrodynamic torque predictions from tests at two model
basins are shown in Fig. 81. Presumably these were run at
Fr0ude scaling, to get propulsion simulation. This of course
means that the Reynolds number for the rudder is not simulated, so that there is low confidence in the stall and torque data.
My second question is: Could the author tell us something
about the size of the models and his opinion on the usefulness
of the torque prediction?
The first comment is in connection with the laws of similitude, discussed by Mr. Harrington. There are some problems.
Similitude in flow around rudders is needed to get good predictions of forces, centers of pressure, and stall effects. The laws
of similitude require getting close to the full-scale Reynolds
number for the ship rudder. Figure 7 of the author's reference
[12] gives some numerical values of the effects. (Reynolds
number is, of course, the flow velocity times the rudder chord
divided by the kinematic viscosity, which is about 1.2 10 -5
fte/sec for water.) Taking the data for Ship A from Table 4,
and ignoring propeller race effects, the Reynolds number for
sea trials was

(ft)

(22.6) 1 . 6 9 ~

IX1
+Xz+XSft)
[
2

RN -

- 5.4 X 107

1.2 X 10 -5 ft2
see

The wind tunnel tests of reference [12] were run in the Reynolds
number range of from 1 to 3 million, an order of magnitude
away from reality. And the Joessel test was run at a Reynolds
number of about 4.7 X l0 s, still another order of magnitude
lower. A low confidence level should therefore be expected
with the Joessel predictions.
The second comment concerns steering gear efficiency. The
author has made a most thorough presentation of the theoretical
factors involved. His Fig. 20 is a very useful contribution, and
comparing it with full-scale ship data may be of interest. This
discusser was involved in such a test, made in drydock on USS
(FF 1052). The rudderstock on
was linked to ex-

Knox

Rudder Torque Prediction

Knox

ternal measured weights, which were raised and lowered by


the steering gear action; a description of the test is presented
in reference [26] (additional references are listed after the author's closure). The measured efficiency, on a well-tuned
single-ram Rapson slide, was 72 percent at about 40-deg rudder.
This is quite consistent with Mr. Harrington's Fig. 20.
My third comment concerns Mr. Harrington's list of references. I should like to make it even more extensive by
suggesting two recent sources of information: the paper by
Kerwin and Mandel on flapped rudders [27] and a paper by this
discusser concerning rudder torque and force measurement
[28]. These references do not solve all of the problems, but they
do add to the body of knowledge.
My final comment is one of gratitude to Mr. Harrington for
reparing this paper. He has put together a great range of test
ata analysis and scattered pieces of information (such as ram
pressure to overcome inertia). In addition, he has carefully
shown the pertinent dimensions and stern arrangements of each
test ship, a feature skipped by many authors. Mr. Harrington
has made a notable contribution to the profession.

A. Rem, 4 Visitor

First of all I would like to compliment the author for his


paper, which must have been a laborious task to prepare.
Though a great part of it deals with information about the
mechanical part of the rudder design, I think that many people,
interested only in the hydrodynamic behavior of the rudder,
also can find useful information. Along such lines, I would
make the following comments:
It was noticed from the full-scale torque measurements that
stall sometimes appears later than predicted. Shiba [29] showed
that stall starts at a later stage on a rudder behind a ship-propeller system than on a rudder in the free stream, such as the
DTMB 988 profiles. Further, he found that the stall angle
increases with increasing slip ratio of the propeller.
The Netherlands Ship Model Basin (NSMB) has also found
differences in the onset of stall if rudder forces are measured
on a rudder behind a ship model with a propeller as compared
with a rudder model in the free stream. The aforementioned
phenomena are well explained in the literature by considering
the rudder-propeller-ship as a "wing system with flap."
Also, the influence of the Reynolds number on the onset of
stall may be of importance. Crossfairing of the stall angle
versus RN of the DTMB 988 profiles shows that some profiles
have an increasing stall angle with increasing RN. As most of
the full-scale rudders have a Reynolds number about ten times
higher than the highest value for the DTMB 933 profiles, it may
be assumed that for full-scale rudders, stall will appear in a later
stage than predicted. I think, therefore, that the observed
discrepancies in the onset of stall between the full-scale measurements and the predictions can partially be explained by the
foregoing.
Another comment concerns the angle of attack versus the
rudder angle. Taplin [16] used about the same method of
calculation and the same DTMB 988 profile data; however, he
introduced the factor 5/7 as the relationship between the angle
of attack and the rudder angle. That is, 5/7 times rudder angle
gives the angle of attack. This angle of attack is used for the
calculation of the hydrodynamic rudder torque. Of course
there will be a difference between the rudder angle and the
angle of attack, and I am given the impression that this factor
is used to correlate the predicted rudder torque with the fullscale torque. Has the author found a corresponding value
based on the predicted torques and the full-scale measurements?
4 The Netherlands Ship Model Basin, Wageningen, The Netherlands.

A. Fairlie-Clarke, Member

It must seem incongruous, in the face of all the advances in


methods of analysis of naval architectural problems over the
past two decades, that such a practical requirement as a method
of obtaining an accurate steering gear torque prediction was
not met long ago. As the paper describes, one need not look
for long at the many factors involved to realize why this is so,
and the author has therefore been courageous in putting before
the Society the results of his obviously extensive experience and
research on the subject. The more so for the practical and indepth reports that he has given which make the paper a valuable document both for the practicing shipyard naval architect
and for the researcher. The wealth of data presented and the
attention to detail in the analysis particularly commend the
paper, and the clever application of the DTMB 988 data provides a valid alternative to the Joessel method of torque estimation.
Some experiences gained within the companies of the Vickers
Marine Engineering Division in Scotland, and elsewhere, may
be of interest in relation to this method. Considerable cavitation tunnel testing has been carried out in association with fin
stabilizer developments. The tests show that, when tested in
water, the control surfaces develop about the same initial liftcurve slope as given in DTMB Report 988, but the maximum
lift is around 20 percent lower, due to a reduction in lift curve
slope at higher angles and an earlier stall at equivalent Reynolds
numbers. The cavitation tunnel results also give the chordwise
center of pressure about 5 percent farther aft than indicated
by DTMB Report 988 and the spanwise center of pressure at
49 percent of the span.
On the question of friction coefficients for steel~on-bronze
and phenolic bearings, our experience is in line with the author's
conclusions and the figures we use are similar to his recommended values.
Whether the author's method of torque estimation will replace Joessel is uncertain. Both are basically methods of similitude and the task of re-correlating large quantities of
often-not-well-documented data to a new basis is difficult,
time-consuming, and uncertain. The Joessel method has
normally served well, yet a few undersized steering gears have
been installed and, in the other extreme, a number of oversized
steering gears have been installed.
Any torque estimation based on the Joessel or DTMB 9:38
method will necessarily be limited in accuracy since it does not
address in detail all the variables involved. In particular, the
flow conditions at the stern, as influenced by the ship's motions,
propeller and rudder, must be known at all stages of a turn.
Some insight into this problem is given by Fujii and Omori [80],
who show in particular the importance of considering the
rudder characteristics when operating in a stream of water
disturbed by the ship's wake and the propeller slipstream. Tests
carried out in Japan to measure the hydrodynamic characteristics of rudders operating in a propeller slipstream show that
the effect is not simply one of a change to the flow rate, but that
the center of pressure moves appreciably and also the stall angle
increases, and these effects are dependent on the slip ratio of
the propeller at any instant. It may prove therefore that a full
analysis of the rudder problem cannot be based on freestream
rudder characteristics, but that some systematic series of experiments to establish the characteristics of rudders when operating in propeller slipstream will be required.
The trial results included in the paper show the benefit of
continuously recording data during steering trials, and we
would encourage such recordings on every possible occasion.
The addition of yaw rate, ship speed, and propeller rpm traces
would greatly enhance the value of the records. Possibly the

Rudder Torque Prediction

71

taking of such records could be coordinated by a central body


that would supply a suitable instrumentation package, and that
later would publish a comprehensive data base.
C. F. Butler, 5 Visitor

The supporting data for Figs. 43 through 79 of Mr. Harrington's paper were aceumulated during full-scale testing of
steering gear engines and the data span many years of ship
construction. During this period, a considerable amount of
data from several types of vessels were recorded and analyzed.
Since the accumulation of operational data plays such an important part in the verification of analytical techniques, it is
considered appropriate to briefly discuss the instrumentation
and procedures used to acquire full-scale trial data on steering
engines.
The instrumentation engineer is faced with two basic considerations when preparing a sea trial instrumentation
package:
5 Test supervisor,Newport News Shipbuilding, Newport News, Va.

selection of suitable equipment to ensure an accurate re-

cording of all pertinent data, and


presentation of the data in a form that.Jail interested parties
can interpret.
For those not familiar with the arrangement of steering gear
spaces, consideration must be given nolfonly to the high noise
levels caused by steering gear machinery and propeller cavitation, but also to the high levels of propeller blade-rate vibration and deck heel caused by hard rudder throws. In other
words, the environment is harsh for delicate instrumentation.
Hence, when selecting equipment to be used as steering gear
instrumentation, it must be not only accurate, but rugged as
well.
When selecting transducers and interface locations, the importance of the steering gear system to the operation and safety
of the ship cannot be overlooked. This consideration presents
a limitation when trying to measure any control signals internal
to the steering system itself. In no way can an instrumentation
failure be allowed to affect system operation.

HIGH POINT
VENT

HIGH POINT
VENT

PORT
RAM

}4

STBD
RAM

~4

6
PT

118 VAC,
SO Hz

60 Hz
SLDC

:C)

ill
SCE
PWR
SUP

T-

PT-

S"

ORDEREO POSITION

I1

TACH GENERATOR

VlSlCORDER

PRESSURE TRANSDUCER

SYNCHRO

I~ . ISOLATION

VALVE

Fig. 82

72

Typical steering gear instrumentation diagram


Rudder Torque Prediction

(3=

The recorded data are used not only to verify design predictions, but also to assure proper operation of the steering gear
machinery. The following parameters are normally instrumented:
(a) ram pressure,
(b) pump stroke,
(c) rudder angle,
(d) pump motor rpm,
(e) ordered angle, and
(.f) peak electrical load (not normally recorded continuously)
In order to demonstrate proper system operation and to establish torque loads, simultaneous and continuous recordings
of Parameters (a) through (e) are normally required. Examples
of equipment which may be used to record these parameters
are illustrated by Fig. 82 accompanying this discussion and
discussed as follows:
The ram pressure is measured by using bridge-type pressure gages which are normally calibrated in the range 0 to 5000
psi.
The pump stroke is recorded by using a synchro-transmitter which is coupled to the pump stroke indicator and feeds
a synchro-to-linear dc converter (SLDC).
The rudder angle is recorded by using a tap on the ship's
rudder angle indicator which is fed into an SLDC.
The pump motor rpm signal is obtained from a tachometer
generator which is coupled to the motor shaft.
The ordered angle can be recorded using the same
equipment used for the pump stroke, or an angular potentiometer can be coupled to the ordered angle indicator.

immediately, thereby facilitating a comparison of design versus


actual rudder torques.
It is gratifying to the test engineer to see his many hours of
steering gear testing and data accumulation utilized for design
studies and system improvements as has been done by Mr.
Harrington. Not only is the steering gear design enhanced by
such a paper, but the test engineer benefits by having a very
excellent reference document for planning future full-scale tests
and interpreting the resultant data.
W. S. Vorus, Member
This is an excellent paper. I can find only one aspect that
requires further discussion. This pertains to the derivation of
the steering engine efficiency, E(c~), used in the author's torque
formula. The Rapson-slide force diagram, Fig. 19, applies only
when the engine drives the rudder; the directions of the frictional forces imply a counter-clockwise rotation of the crosshead
produced by the applied force P at the crosshead pin. When
the rudder drives the engine, the force direction P shown in Fig.
19 opposes a clockwise rotation of the crosshead, and all of the
frictional forces are of reversed direction. Following through
the analysis, the ram-to-rudder efficiency for the driving rudder, denoted E'(c~) is

E'(oe) = 1 + #a tantx + #b tancx + 4pc d + ~hp d cosc~


This is the same as the form of E(e) given for the engine driving, except for sign reversals; the efficiency for the rudder
driving is greater than 100 percent, reflecting the reversal of
losses in the engine..
The plot of E(o0, Fig. 20, shows that the frictional losses are
relatively small for the recommended coefficient of friction of
0.10. Denoting the losses as e, E(c~) and E'(c) can be written

The electrical output from these various transducers is fed


to a bank of signal conditioning equipment and, in turn, to a a s
high-speed recorder which uses a light-sensitive paper and
provides internally generated timing marks.
The accuracy of recorded data is ensured through the use of
precision instruments and a program of calibration checks. and
That is, after the entire package is installed on the ship, each
E'(a) = 1 +
function is calibrated throughout its entire range by a controlled
operation of the steering gear components and the introduction Then to first order
of known test pressures. Accuracy is limited only by recording
1
E'(~) resolution and transducer repeatability and is typically 1 percent of the full-scale reading for any given function.
Although not elaborate, the instrumentation system described by small argument expansion. The torque formula therefore
herein had remained basically unchanged for many years and
has two different forms depending on whether the rudder or
has proven to be reliable and accurate during full-scale testing.
engine is driving:
However, the addition of a differential pressure transducer
across the ram is planned in the near future to facilitate both
= Z__....RR.J E(e)
engine driving
on:the-spot torque calculations and the final data reduction
Q
after trials are completed.
cs2e / l
rudder driving
Once a set of steering gear data has been recorded, the data
.E(.)
must be labeled and interpreted. In order to accomplish this,
the instrumentation engineer must provide a complete set of where E(oz) is the ram-to-rudder efficiency given by Fig. 20.
Now it appears that the author has made the proper allowcalibration curves and scale factors. During the course of one
sea trial, it is not unusual to accumulate 100 ft or more of re- ances for engine friction in the several displacing torques which
he converts from the ship pressure data, although his calculacorder records. Upon completion of the sea trial, the test entions are not shown. It is important that others attempting
gineer normally transcribes enough data from the recorder
records to substantiate the satisfactory accomplishment of all similar correlations also recognize the distinction cited in the
preceding.
test requirements, but he would not document and analyze all
I have one other minor criticism of the paper. It left me
of the useful information until sometime after the completion
hanging a bit with regard to the comparisons of calculated
of sea trials. An apparent solution to the problem of "neglected
data" would be real-time data reduction. With the advent of . rudder torques with the recorded full-scale data. The correlations are discussed in some depth for each of the six ship cases.
microprocessor controlled test equipment and desktop comHowever, a few additional figures, or the superposition of a few
puters, it is anticipated that in the not-too-distant future we will
be plotting real-time rudder torque curves during full-scale data points on some of the existing figures, would have allowed
more ready judgment of the validity of the conclusions retesting. Not only will this technique facilitate total data regarding the utility of the proposed calculation methods. Figure
duction, but test results will be available to the trial director
Rudder Torque Prediction

73

,/

/J
!
I

~.,,~. CALCULATED
PREDICTION
"~ 12
- MODEL
BASINY

=0

z
so

+.

10

2o

[I

10

~/l

E,T

I0//

40

'."'~"2E.~OEO,

RUODER(DEG)
+

~ /

-8

--MODEL
BASIN

-12

Fig. 83 Comparison of hydrodynamic torque predictions from tests conducted at two


model basins with the calculated prediction for Ship C at a ship speed of 19.5 knots annotated to include trial data points taken from Figs. 66 and 68.

8;3 is the type of display suggested.

This figure is basically Fig.


81 of the paper; it compares the rudder hydrodynamic torque
characteristic for Ship C as calculated by the proposed method
against model test predictions conducted independently by two
model basins. The data points superimposed on the figure were
converted from the ram pressure data for Ship C from Figs. 66
and 68. The formula used is that stated in the foregoing, with
Q = Qu + QF. In applying the formula, the rudder angle in
Figs. 66 and 68 was assumed to be equivalent to the angle of
attack. This allows QF to be estimated from the friction torque
curve of Fig. ;3;3. E(~) was taken from Fig. 20 for an assumed
coefficient of friction of 0.10.
Due to an increasing ship drift and rotation with time, the
rudder angle of attack decreases relative to the rudder angle
as the rudder angle increases. This has the effect of stretching
the angle scale for the data points spotted on the figure; that is,
the data points actually lie closer to the origin of zero angle of
attack than the figure implies. But even disregarding the
rudder angle difficulties, the figure suggests that the differences
among the model test, calculated, and measured full-scale
rudder torque values are all of the same order. Judging by this
single correlation, model tests should not be expected to provide
superior accuracy over a calculation performed in accordance
with the method proposed.
The data also suggest, however, that there is still some room
for improvement in rudder torque prediction procedures (and
perhaps in the interpretation of full-scale pressure data, as well).
The failure of the rudder to consistently develop negative displacing torques of the expected order is particularly intriguing;
the reason is not at all obvious, and probably not simple. My
question regards incentives for further study and development
in this area. We research types are forever looking for interesting problems to which someone needs the solution. The data
presented in this paper and the questions raised provide the
basis for numerous good research projects and technical papers.
From a practical point of view, then, c~in it be concluded, in
light of this paper, that a better understanding or rudder torque
74

intricacies and more refined prediction methods are really


needed, or not?
H. T. McVey, Member

The following is, essentially, a commentary rather than a


discussion of what, in my opinion, is an excellent paper.
Some twelve years ago, while this discusser was working at
a well-known shipyard, we experie~ed some problems, in the
first ship of a series, in the bolting of the steering gear carrier
bearing pedestal. To verify the loads on the pedestal, we installed strain gages in the upper portion of the rudderstock, in
addition to ram pressure recorders, to enable us to work back
to try and determine the rudder forces and thus the loads on the
rudderstock's radial bearings. Much to our surprise, in trying
to plot torque, we could not come close to the predicted values,
using either Joessel's derivation or the DTMB Report 988
method. This in spite of our making the necessary allowances
for shape, friction, etc., for static rudder torque at discrete angles or for hydrodynamic torque, all at ship's design speed. The
rudder is of the spade type, of the configuration shown by Fig.
10 of the paper, with a cutaway stern, but without the hull seal
shown in Fig. 22. Since the vessel was already in service, we
never pursued the matter to fully establish what was happening;
we had to get on with tile fix. The pedestal bolting was
strengthened by accepting the calculated loads, taking the strain
gage/ram pressure readings at face value, plus some margin!
A few years later, for experimental purposes, we installed
copper-nickel cladding on tile rudder of one of the sister ships.
The cladding was applied after the rudder had been built,
conventionally, of steel. The ship's rudder was inspected
regularly and, to make a long story short, we found the cladding
was being stripped off even though we had allowed for high
(negative) airfoil suction pressures. The point I would like to
mention in connection with Mr. Harrington's paper is that there
was a discernible stripping pattern and intensity between port
and starboard rudder sides, as well as vertically.
Our in-service testing of the experiences related herein al-

Rudder T o r q u e Prediction

though not intending to establish torque requirements, did show


that
(a) There was repeatability in the torque measurements
but no predictabilityT
(b) There was no symmetry between right and left rudder
angles.
We felt we were facing certain aspects that we did not know
how to allow for in the calculations, mainly the effect of the
cutaway stern on cross-flows, coupled with the variability of
flow velocities in the propeller race, all of which only confirm
what Mr. Harrington has covered in a very thorough manner
in his paper. I would hazard to say, though, that no test result
torque characteristics are given in the paper because they would
be disconcerting.
Incidentally, the typographical error appearing on page 40,
regarding the rudder angle at which the seal effect disappears,
should be corrected.
One other point which I feel is currently important to highlight is the matter of astern torque requirements, which are well
covered in the paper. In the past, astern power for merchant
ships has been limited by the well-known requirements stipulated in MarAd specifications, and the need to keep astern
turbine losses down when going ahead. Recently, because of
the introduction of diesel and gas turbine propulsion, where
astern power can be equal to ahead power, the prediction of
astern speed becomes critical so that astern rudder torques do
not become controlling, and if they do, then everything must
be sized accordingly. Unfortunately, steam turbine manufacturers have also increased the astern power availability,
perhaps in an effort to be competitive and, also, because of
third-party liability fears, in the simplistic belief that they thus
improve the stopping ability of the ship.
I appreciate the opportunity to discuss this very fine paper,
which, in my opiiiion, pri~sents the most comprehensive analysis
of the state of the art.
C. D. Davis, Member

The author has attempted to find a more sound basis for


determining rudder torque in the ship design process. Ship
designers have known for years that the available methods
should not be universally applied. Lacking an alternative,
there was no choice available except to start with the existing
methods, apply some judgment based on past performance and
finally apply the engineers old standby of adding some more
power to cover the unknowns. Learning how to do the design
correctly and accurately will obviously reduce the urge to add
that extra punch and consequently reduce weight, size, and cost
of the steering system.
The use of DTMB 9;33 has yielded conflicting results on
drastically different ship forms in our experience. A brief
comparison of results is offered here.
A fast, light-displacement ship with fine lines for which the
rudder torque was calculated using DTMB 938 experienced
130 percent of calculated torque on trials. Fortunately, margins existed in the mechanical and structural elements, but the
pumps did reach stroke compensation levels and caused a
slowing of the rate of rudder swing. Required time to swing
was met after some stroke adjustment, but considerable effort
went into proving the system to the satisfaction of our customer.
On another ship of heavy displacement and moderate speed,
having propeller and rudder arrangements similar to the ship
just described, the same design method was used and the results
were surprisingly close to the calculated values.
So, where is the error? The experience described in the
paper seems similar to ours for the first ship, but not for the
second. The rate of ship turn versus the rate of rudder turn

must play a significant role as suggested by the author. It


would seem prudent to this discusser that model testing could
determine that there is a proper relationship of rudder angle
to ships turning rate at any instance throughout a given maneuver. If this relationship could be determined, then the
position of the rudder could be matched for best performance.
At present the;only control on rudder swing is to reach
hardover within the specified time limit. This criterion for
acceptance is obviously unrelated to the real problem. The
problem is to maneuver the ship, not to prove that the rudder
can go hardover in some time frame regardless of what the ship
is doing. Experienced racing sailors using tillers to steer their
boats know that maneuvers can be executed more rapidly and
with less loss of boat speed by "feeling" the rudder through the
turn. If that sense could be reduced to a design method resulting from model testing, we would have made a great advancement. It is suggested that SNAME take on a task to find
the answer.
In the mean time, we can benefit from the comparative data
presented in this paper. Congratulations and thanks to the
author for his contribution.
H. G. Acker, Member

Figure 84 shows the results of one of the rudder tests conducted about 20 years ago on a single-screw tanker built by
Bethlehem Quincy. The torsional strains were measured on
the rudderstock above the bearings and, consequently, would
have some bearing friction included. Although the shape of
the torque curve is essentially the same as those shown in the
paper for similar type rudders E and F, it is noted that the
highest torque was observed during the second hardover
swing.
How do these measured torques compare with calculated
values? One very simple method for calculating the hydrodynamic torque for this type of rudder is to use the Joessel
equations with a rudder balance of about 22 percent. The
calculated center of pressure from the centerline of the rudderstock will be about 10 percent of the chord length just below
the horn. I would hesitate to use a center-of-pressure value
much lower than 10 percent of the chord length regardless of
the calculated value. In this particular case, using the 10 percent figure for the center of pressure and using the Joessel relationship for a 35-deg rudder angle, the calculated hydrodynamic torque is about 9.0 X 106 in.-lb, which appears to be
slightly on the high side for this run. In other cases, however,
the 10 percent figure gave torques a little on the low side.
As long as the astern speed is not more than about one-half
the maximum ahead speed, the astern torque is not likely to be
greater than the ahead torque.
I think it is reasonable to say that there is no theoretical
method available for such calculations. There is, however,
some hope for calculations based on reliable model work and,
particularly, on actual ship performance.
It should be realized that the so-called Joessel coefficient (0.7
in this case) is really a mask for unknowns in both the pressure
force on the rudder and the center of pressure and not for just
one of these factors. As regards the pressure force, we also
measured the bending stresses in the horn and found that the
maximum transverse bending moment in the horn came close
to that calculated based on the pressure portion of the Joessel
formulation, and occurred shortly before the maximum torque
occurred, as might be expected.
In conclusion, I must say that this is the first paper of which
I am aware that gets to the nuts and bolts of this nebulous but
most important subject. My congratulations to Mr. Harrington.

Rudder Torque Prediction

75

STRAIN GAGES

%, 00ER

JOESSELHYOROOYNAMICTORQUE= 3.57 (.7) AV2 x (C.P.=10%)


= 9.0 x '106 IN-LBS

Q=7.8
, ~

//

RUDDERANGLE

HORN

\
TIME, SEC.

TORQUE x 1061N-LBS

'..,,,.

35o LEFT

Q=9.8

Fig. 84

Rudder torque measurements for a tanker

Deborah W. Berman, Member and Everett C. Hunt, Member

[The views expressed herein are the opinions of the discussers and not
necessarily those of the Department of Transportation or the U.S.
Merchant Marine Academy.l
Teachers of marine engineering and naval architecture are
indebted to Roy Harrington for this rational exposition of
methods for rudder torque prediction. Available textbooks do
not include these methods and the design data and test results
presented are always difficult for teachers to obtain. We will
make good classroom use of this material which is worthy of
publication in The Society's TRANSACTIONS.
The limitations of the Whicker and Fehlner report are not
always mentioned because their study provides the best and
sometimes the only semi-empirical predictive equations. The
author notes shortcomings of DTMB 933 particularly for effective aspect ratios near 2. It might be useful in this regard
if future theoretical work on predicting the hydrodynamic
forces was directed to closer investigation of the influence of
the span and the area aft of the rudder post rather than the aspect ratio.
The author states that an interpolation of Figs. 11 through
16 indicates that an accurate assessment of effective aspect ratio
is unimportant in calculating the drag coefficient. However,
if it is induced drag, due to side force or lift, then the assessment
of effective aspect ratio could become important, suggesting
that a more accurate prediction method is desirable.
Is the author's assumption of spanwise center-of-pressure
location related to a specific spade rudder shape or is this independent of shape?
The author is encouraged to further develop his argument
in which the upper and lower sections of a horn rudder are
considered to have effective aspect ratios equal to twice the
geometric aspect ratios at all rudder angles. Why are we
confident of the sealing effect that each section provides the
other?
The discussion of the reversal of rudder pressure at large
76

angles noted during trials is perplexing. This appears to be a


fruitful area for further investigation both experimental and
analytical.
The author's conclusions are a useful summary of the reliability of the equations and curves of the Whicker and Fehlner
report. We now have good evidence that the breakdown of
rudder lift, or stall, is of prime importance.
Perry W. Nelson, Member

As pointed out by Mr. Harrington, the accurate determination of precise rudder-torque predictions is not yet within the
reach of the practicing naval architect. However, the author
has provided an excellent summary of the problems that are
encountered and provides additional insight into the way in
which the data and analysis techniques currently available may
be used in the design of rudders.
The data provided on full-scale trials is most useful in attempting to substantiate model tests and theoretical analyses.
In this regard, it is possible that additional information on the
conditions that existed during the trials would aid in explaining
the variations between predicted performance and the trial
data.
For example, the following full-scale conditions could influence the flow around the ship's rudder:
1. Draft of the ship during trials.
2. Sea conditions during trials.
3. Trim of the ship if other than the design condition.
4. Heel during turns.
Again, I would like to congratulate the author for an excellent
paper on a most difficult subject.
Stanley G. Stiansen, Member

The author should be commended for an informative and


useful paper. As indicated by the author, the achievement of
precise rudder-torque predictions will be out of reach for many
years to come. The shortcomings of some of the prediction
methods are discussed in the paper. In addition, it is worth-

Rudder Torque Prediction

while to note that model testing for measuring forces and moments is generally conducted in a guided straight running
condition. Due to its difference with the ship maneuvering
condition and the lack of establishment of correlating model
and ship responses, the actual rudder torque cannot be confidently extrapolated from the model data. Consequently, the
only reliable rudder torque data are the full-scale measurements
during sea trials.
,:, .
Although the problem of rudder-torque prediction has been
confronted by naval architects for many years, little improvement has been achieved in recent years. This paper, which
summarizes the current status of the prediction methodology
and presents a refined prediction method based on both model
experimental and full-scale measured data, is undoubtedly a
significant contribution to the industry and provides useful data
and guidelines to the design engineer.
With regard to the proposed method. I would offer the
following comments for the consideration of the author:
1. The prediction method as recommended by the author
takes all of the important hydrodynamic and operating parameters into consideration and should be regarded as a refined
procedure. The accuracy of the prediction apparently depends
on the selection of each individual variable or coefficient. The
accurate values of some coefficients may have to be deduced
from model test data which are generally not available at the
early design stages. Therefore, it would be of great help to the
designers if the author can offer some guidelines as how to
determine the required coefficients in the calculation and what
are the limitations of the recommended formulas.
2. From the viewpoint of a classification society, the emphasis of the rudder-torque prediction focuses on the strength
of the rudderstock and the associated apparatus and the capability of the steering gear. Consequently, the method pursued
by classification societies may be generally limited to the
maximum rudder torque required for the rudderstock and
steering gear. " The American Bureau of Shipping (ABS) is
currently examining a number of the available empirical formulas for predicting rudder torques, and is also working together with other members of the International Association of
Classification Societies (IACS) in an attempt to unify the rudder
torque requirements. A final decision in this regard has not
been reached at the present time. In the course of this development, clue consideration will be given to the method proposed
by the author.
3. In the ABS study on rudder torques, comparison of the
actual steering gear torques based on empirical formulas has
been made for a number of existing ships of different types and
built by various shipyards. Although an appreciable variation
was found in the ratio of steering gear capacity to estimated
rudder torque, it is, however, feasible to establish a minimum
requirement for the rudder torque by modifying a selected
empirical formula. A tentative in-house guideline developed
at ABS gives torques generally comparable with those predicted
by the author for the six example w;ssels, Ships A to F.
In order to account .for the variation of ram pressures as
shown in the sea trial records for different runs and different
sister ships and also the limitation of a semi-empirical formula,
it would seem advisable to consider a proper design margin in
determining the required rudder torque based on the predicted
values. The author's comments in this regard will be appreciated.
L. Vassilopoulos, Member

This is an excellent paper on a subject that has not received


the attention it deserves, and the author is to be thanked for
sharing with us the fruits of his research on this important
problem.

With the increased attention on ship safety and regulations


regarding steering gear systems, the naval architect is faced with
the problem of predicting with greater confidence than ever
before the forces and moments acting on ship rudders under
various maneuvers and the resulting responses of the rudderstock and steering systems.
The subject of rudder loads has been under investigation by
our firm for some ~ime, using both analytical and experimental
technique. We have found the material in this paper to be very
absorbing but, unfortunately, due to time limitations, we have
not been able to digest all of its contents. Accordingly, this
contribution is by necessity one of a general nature. The author
should be assured, however, that we would like at some future
time to provide additional comments after we have investigated
more thoroughly his procedures and results.
The forces and moments on a rudder and the resulting torque
on the rudderstock can be estimated with analytical methods
or tests carried out with ship models, in wind tunnels and aboard
ships. As the author points out, a completely analytical approach for an arbitrary rudder shape still leaves much to be
desired. This is not because the control surface hydrodynamics
cannot be handled effectively (for example, by employing a
computer program such as that in reference [81]), but because
the input flow field is very difficult to define for arbitrary
rudder angles.
We have used with similar success the Joessel and DTMB 983
methods, and, in the latter case, the interpolation procedure for
the test data is being accomplished automatically within a
specially written computer program. Incidentally, the original
DTMB publication contained certain typographical errors in
some of the approximate formulas. Has the author allowed for
these mistakes in his assessment of the utility of those expressions?
The data in the DTMB 933 report pertain, strictly speaking,
to spade-type rudders. The application of this information to
other forms of rudders, such as horn rudders, is by no means
clear and the procedures suggested by the author are worthy
of further study. In this context, it may be worthwhile to make
use of the more recent data on flapped rudders such as those
presented in references [27] and [321.
During the past few years we have.participated in trials of
certain vessels in which the need arose to measure rudderstock
loads and responses. The instrumentation system that has been
developed and applied to spade- and shoe-type rudders measures axial loads in the stock, and torques and bending moments
at selected locations, from which the stock bearing loads can
be determined. The package includes a transducer for recording the rudder angle automatically and the basic transducers consist of strain gages suitably spot-welded on the
stock.
There are very few full-scale data available in the literature,
and those published by the U.S. Navy in references [28] and [88]
suggested that the ram pressures may not be as accurate as
torques recorded on rudderstocks via strain gages. What is the
author's view on this issue? It would appear that both ram
pressures and stock strains are desirable; but if a choice had to
be made, this discusser would favor the latter.
Recent full-scale trials carried out aboard the Polar Star and
Polar Sea icebreakers of the U.S. Coast Guard, both in open
water and in ice, have provided some rather illuminating data
on rudder loads and these are currently being processed. The
lessons learned to date include the fact that rudder loads on a
triple-screw ship vary dramatically depending on whether the
ship takes a port or starboard turn, that ice-induced random
loads need not exceed open-water loads during Z-maneuvers
at high speeds, and that the stock shaft alignment (for systems
with more than two laterial bearings) has to be selected with

Rudder Torque Prediction

77

care to avoid hearing failures. It is hoped to have this information available to the profession in the not-too-distant future.
Future students of this interesting subject may find some of
the additional material listed as references [841 through [51] to
be of interest.
This paper has aroused considerable interest and will be
studied in greater detail before all of its conclusions and recommendations can be appreciated.
Robert Taggart, Member

In some relatively recent designs of destroyer type ships, the


torque required to rotate the rudder under extreme maneuvering conditions has exceeded predicted levels. Since rudderstocks and steering machinery must be sized on the basis of
design torque predictions, this situation is cause for concern in
that future torque overloads may exceed safety factors.
A second cause for concern is the apparent inconsistency
between alternative methods of measuring rudder torque on
operating ships. Torque derived from strain-gauge measurements on the rudderstock is seldom identical to that derived
from pressure measurements in the cylinders of the rams which
drive the tiller. Thus, there is an uncertainty as to whether
either of these measurement techniques gives a true indication
of the torque which is actually imposed upon the system.
The study referred to in this discussion was undertaken in
1968 in an effort to understand the behavior of all'elements of
a ship steering system and to attempt to derive a rational means
of predicting the maximum loadings which may be imposed
on the system during the operating life of the ship. Also, it was
intended to search for a reasonable explanation of the inconsistency between different full-scale measurement techniques
in evaluating the loads imposed.
The torque in a ship steering system, when the rudder is at
any fixed angle, is generated by a hydrodynamic force acting
on the rudder. The magnitude of this force is a function of the
velocity and direction of the flow around the rudder and the
size and shape of the rudder itself. The moment arm of this
force with respect to the rudderstock axis is another function
of these same factors and of the location of the rudderstock with
respect to the rudder. The torque is the product of the hydrodynamic side force and the moment arm. This torque is
transmitted to the ship hull through the rudderstock, tiller, rams,
and the ram cylinders and their foundations.
When the rudder is being rotated, additional torque elements
are introduced. A small, and probably insignificant, hydrodynamic torque is developed by the rotation of the rudder in
the surrounding fluid. As the rudderstock turns in its bearings,
a frictional torque is added. The magnitude of this frictional
torque is a function of the resultant force being applied to the
bearings and the type of bearings employed. The magnitude
of this force is, in turn, a function of the resultant hydrodynamic
force developed by the rudder and its vertical center with respect to the upper and lower bearings which support the rudderstock.
In order to arrive at any rational prediction of what the
maximum values of stock torque and of the required ram
pressure will be, the logical starting point is with the hydrodynamic forces generated at the rudder. This requires a definition of the flow field in which the rudder will be acting
during ship maneuvers which invoke the most drastic conditions
likely to be encountered.
Once this flow field has been defined, the next step is to derive values of the maximum force generated by the rudder and
the center at which this force is acting. From this information
can be derived the side force and drag acting on the hull that
actuate the maneuver as well as the resultant forces and moments that the steering machinery must produce.
78

With a knowledge of the geometry of the rudderstock, the


supporting bearings, and the ram and tiller assembly, it is possible to calculate the bearing loads and to estimate the total static
forces and torques required to hold the rudder in a fixed position. This can then provide a basis for evolving the dynamic
forces and torques that are the crux of the problem.
In the dynamic situation, where the maximum loadings are
encountered, practically all of the elements are undergoing
change. The tiller is being driven by the rams, which in turn
are being moved by the pumping of hydraulic fluid. The load
on the ram pistons changes with variations in the hydrodynamic
torque, with variations in the frictional torque developed in the
stock bearings, and with variations in the rudder angle, which
affect the relationship between tiller torque and piston force
as well as the magnitude of the frictional force developed in the
Rapson slide.
Large variations also occur in the hydrodynamic torque due
to several factors. The rudder generally operates in the race
of the propeller. The axial and tangential velocity components
of the propeller race that act on the rudder vary with the rudder
angle and with the speed of advance and loading of the propeller. It should be noted that for a single-screw ship with the
rudder abaft the propeller, the angle of attack of the propeller
race on the lower part of the rudder is in the opposite direction
from the angle of attack on the upper part when the rudder is
amidships. The magnitude of this differential angle of attack
varies with propeller loading. The resultant flow angle to the
rudder also changes as the rudder angle is altered and as the ship
responds to the rudder action.
Thus, in order to analyze the problem and to predict the
torques and ram pressures to be developed by the steering
machinery, it is necessary to take into consideration the dynamics of the ship response to steering system actuation and the
propeller loading as well as the dynamics of the steering system
itself. This is an extremely complex procedure but one which
cannot be avoided if a rational design prediction method is to
evolve.
The solution of any problem is always more realistic if the
answers are known at the outset. This does not necessarily
mean that the solution is expedited, particularly where the
answers are conflicting, confusing, and susceptible to experimental error. Although Mr. Harrington has presented us with
a number of such answers, this discussion is based upon yet
another set of data. Our approach was to select a class of ships
where a problem was known to exist and one for which a wealth
of full-scale maneuvering data and numerous steering system
measurements were available. This was the DE-1040 class of
destroyer escorts.
Maneuvering data included acceleration runs, turning circles,
and Z-maneuvers. Thrust measurements were made on one
of these ships during straight runs over the entire speed range.
Also, measurements of steering system ram pressures and of
rudderstock torques together with rudder angles were made
during four Z-maneuvers. Unfortunately, the Z-maneuvers
for which heading and track data were available were run at
lower speeds than those during which steering machinery
performance measurements were made, so it was necessary to
calculate the ship maneuvers during the higher-speed rudder
torque tests.
There are two types of maneuvers which can be considered
to impose the greatest demand upon a ship steering system.
These are hardover rudder at full speed astern and a Z-maneuver at full speed ahead with maximum rudder angle.
Historically it has been considered that hardover rudder at
full speed astern would impose the greatest torque load on the
rudderstock and would require the greatest ram pressure to shift
the rudder. In fact this condition has frequently been used as
the basis for steering machinery specifications. The reason has

Rudder Torque Prediction

been the assumption that the center of pressure where the hydrodynamic forces are applied is well toward the trailing edge
of the rudder and the moment arm from the center of the
rudderstock is at a maximum.
However, experience has proved that maximum torques are
seldom developed during this maneuver. Although the moment arm is great, the actual forces developed are relatively
small. This is due primarily to the fact that the~velocity around
the rudder is low when the ship is going astern--first because
astern ship speeds are low and second because the rudder is not
benefiting from the augmented velocity of the propeller race.
In addition, the rudder foil shape is inefficient when operating
in reverse and the lift forces generated are correspondingly
smaller than in ahead operation. Since a ship with a rudder
is totally incapable of controlling itself, when going astern at
any speeds that are used in normal operations, I would strongly
recommend deletion,of astern steering trials. Furthermore,
the rudder should be locked on dead center whenever the ship
is going astern since it cannot steer.
Generally it is in ahead Z-maneuvers that the greatest steering
system loadings have been measured. It, therefore, appears,
logical that this is the maneuver which should be selected as the
basis for predicting the maximum conditions under which a
ship steering system might be expected to operate. In the development of a torque prediction procedure, which was the
subject of this study, four sets of DE-1040 Z-maneuver trial data
were selected for detailed analysis It was reasoned that if the
results of these trials could be duplicated by a logical calculation
procedure, this procedure would then be applicable to the
prediction of steering system loadings for future designs

These four sets of trial data are shown in Figs. 85 and 86. In
Fig. 85, the Z-maneuver approach speed was 93 percent of
maximum. The two pairs of curves represent torques developed with an initial right rudder swing of 38 deg and an initial
left rudder swing of 40 deg. One curve of each pair is the
torque derived from strain-gage measurements on the rudderstock as a function of time. The other curve of each pair
is a corresponding torque derived from the difference in ram
pressure between the forward and after ram cylinders. Figure
86 presents similar pairs of curves for right and left Z-maneuvers at maximum approach speed
The lack of symmetry in the right and left torque values is
due to the fact that the DE-1040 rudderstock is offset to starboard of thepropeller centerline to facilitate removal of the tail
shaft. With left rudder, the majority of the foil is exposed to
the propeller race, whereas with right rudder a portion of the
trailing edge swings outside the propeller race.
These four sets of full-scale trial measurements graphically
illustrate the problem. The predicted maximum torque which
was used for designing the steering system and for setting the
ram pressure relief valves was 6.8 x 106 in.-lb. In spite of the

/./
w

/
a:

/
/
i /

/
s/

i J

TORQUE F~OM STRAIN GAG

< / o N RO*OERSTO.

,5
&

/
o

,/
'\\

" t~\

"\

N
//

-2

i/

iI

\\/

'i
"~q \
;~\.4 \\
ROM STRAIN GA

\
\\

l TORQUE
ON R[ DDER STOCK
\,
\

//

/',,

TORQUE FF OM RAM PRESSU

:~Ei

~,

\/.],j

IN IT IAL L E F T

Fig. 85

o.

-~..-..TORQUE F R O , RAM PRESSURE

INITIAL
40

"

UDDER

20
30
TIME FROM EXECUTE IN SECONDS

'"

TORQUE FR~OMESTRsAIHc G ,GE

' iJ '-I

;,

-i \~\/

-6

/S/i -/

'

i\ \

-4

'1

.\

,/
z
o
J
J

i\

/ /

//

;AGE

/./

/ '/

TORQU

,,//'

/
?

;'-

j. 4

) /

//

. . . . . . . . o. . . . . . PRE
...

t,

11 /

TORQUE FROM RAM PRESSURE


/

//

UDDER

\
",

INITIAL RIGHT RUDDER


\

,//' ~ -/\\ ~
/

INITIAL RIGHT

~
\

/
/

50

Trial data: DE 1040 Z-maneuvers at 93 percent maximum


approach speed

Fig. 86

I0

L E F T RUDDER

20
30
TIME FROM EXECUTE IN SECONDS

40

50

Trial data: DE 1040 Z-maneuvers at maximum approach

Rudder Torque Prediction

speed
79

FORCESANDTORQUES
ONBEARINGSANDTILLER

UPPER
BEARING

F$

Fig. 87 Forces and torques


on rudder and stock
F
II
i

LOWER
BEARIt4G

,j

Q4
6

fact that the ship master restricted the rate at which hardover
rudder angles were attained, this predicted maximum torque
level was exceeded in all but one of the trials. Furthermore,
the torques derived from strain-gage measurements and those
derived from ram pressure measurements are only occasionally
in agreement. This leads to a natural question of the validity
of one or both means of arriving at an experimental conclusion
as to the torque actually experienced by the steering system.
The steering machinery system for this class was studied in
detail, including the control system, the hydraulic system, and
the mechanical elements. A mathematical model was devised
to simulate the dynamic behavior of this system in order to
analyze its response to various control inputs and output torque
loadings. The mechanical elements of this model are depicted
in Figs. 87-89.
In order to understand the flow field in which the DE-1040
rudder operates, an exhaustive study was made of the propeller
performance. This included analysis of model and full-scale
tests of the propeller and hull interactions. It also included the
expansion of existing information on the magnitude and distribution of tangential and axial velocity components in the race
of a screw propeller. In parallel with this, the hydrofoil characteristics of the DE-1040 rudder were evolved from available'
experimental data. A means was then explored for calculating
the forces and moments acting on the rudder as it operates
under varying flow conditions at the stern.
For the purpose of analyzing the four selected Z-maneuvers,
it was necessary to develop a realistic estimate of the flow
conditions around the stern of the DE-1040 as it executed those
maneuvers. Although rudder angle versus time data were
presented for these maneuvers, there were no available data
on corresponding ship headings, speed, and lateral motion.
This required developing the equations of motion of the ship,
calculating the forces applied to the ship by propeller and
rudder, and then combining the two to predict the ship motions
during the Z-maneuvers from the known approach speeds and
changes in rudder angle. Involved were analyses of thrust
measurements, turning circle trials, and Z-maneuvers conducted at lesser approach speeds and rudder angles.
Finally, all of these elements were combined to simulate all
of the hydrodynamic conditions existing during the four Zmaneuvers being studied. The hydrodynamic torque as a
function of time was calculated together with the forces and
moments applied to the hull by the rudderstock. Using a series
of estimates of the friction coefficients in the upper and lower
80

,t

J
L

I:l ~

f j

]
I
t

/I
r

stock bearings, in the Rapson slide, and in the ram cylinder


bushings, the stock torques and ram pressure torques were
calculated and compared with the measured values given
earlier in Figs. 85 and 86. It could be anticipated that the relationships of hydrodynamic torque, stock torque, and ram
torque would approximate the curves shown in Fig. 90.
It was found that the data as measured aboard ship could be
duplicated with acceptable accuracy by a logical process of
calculation using realistic estimates of the unknown elements
involved. These comparisons for the four Z-maneuvers are
shown in Figs. 91-94.
An interesting aspect of these calculations is that they revealed an unsuspected result of stall at maximum rudder angles.

Rudder Torque Prediction

I;!lll " * _ J " ~ I N

Fig. 88

' ..

"~,,I

-J~~H~--"n" ~

Forces and torques on rarn and slide

that the rudder torque actually decreased when the stall point
was first reached. This was due to a considerable reduction in
the side force applied by the rudder. Although at stall the
center of pressure moves rapidly aft to the center chord point,
the movement is not sufficient to compensate for the differences
in side force.
Stalling of the rudder is detectable by a sudden drop in the
ram differential pressure measurements and strain-gage
measurements which appears in the majority of the torque
curves, beth measured and calculated. It is evident that there
was no maneuvering advantage in the use of a 40-deg rudder
swing instead of the customary 85 deg in this ship design since
the side force tended to drop off at the higher rudder angles due
to stall.
Based upon this demonstrated ability to calculate what actually occurred on one class of destroyer escorts, it is logical to
assume that the calculation process is valid. The process
therefore can be applied with reasonable confidence to' the
prediction of steering system loadings in future designs.
Since 1968, when this study was made, the technique of
predicting ship maneuvers from data available in the design
stage has advanced rapidly. Basic hydrodynamic and inertial
coefficients are being obtained routinely from model tests for
use in ship motion simulators. These could be applied directly
to calculating Z-maneuver ship, rudder, and propeller performance; maximum rudder torques could then be derived for a
Variety of rudder configurations, positions, and control angles.
In this way it would be possible properly to size the machinery
in the design stage to arrive at an optimum combination of
rudder, steering engine, and operational controls to achieve tl~e
maneuvering performance dictated by the ship's mission.

Author's Closure
Mr. K a n e has succinctly reviewed the various facets of the
considerations which must be evaluated when predicting
rudder torques; his observations reflect a clear understanding
of the complexities and uncertainties associated with rudder
torque predictions.

~r~~

-,~1- - ' \

,~ ~ I

Mr. C a n g i n ' s suggestion to divide a horn rudder into three


sections for analysis purposes has merit. The assumption frequently used with horn rudders is to consider that the rudderstock centerline is the leading edge of that part of the rudder
abaft the horn. In the case of flapped rudders, the chord of the
flap is customarily considered to be equal to the distance from
the stock centerline to the trailing edge, but the control surface
area is taken to be equal to the total flap area. After considerable investigation during the course of this study, it became
evident that any assumption entailed compromises; but in order
to minimize ambiguities, it would be preferable to follow the
common practice of assuming that the rudderstock centerline
was the leading edge of that part of the rudder trailing the horn.
This assumption required appropriate corrections to be made
to data reported for flapped control surfaces. Nevertheless,
in the case of submarine control surfaces, for example, the
suggestion of Mr. Cangin may provide a more accurate approximation.
It is true that steering gears are seldom subjected to a significant fraction of their rated load when in service. For this
reason, features such as electric motor overload and horsepower-limiting mechanisms are often used as a means of minimizing the cost of the steering engine. However, as noted by
Mr. Cangin, it would appear that there should be additional
opportunities for economy in this respect. One possibility is
an increase in the allowable ram pr.essure required to develop
maximum torque. Due to the relatively infrequent requirement to develop maximum torque, hydraulic pressures higher
than those customarily permitted to develop maximum torque
should be satisfactory.

Rudder Torque Predictiofi

81

Fa

F8

Fp

RIGHT RUDDER -- RIGHT ROTATION

RIGHT RUDDER-- LEFT ROTATION


F9

/, !
J
LEFT RUDDER -- LEFT ROTATION/
Fig. 89

ROTATION/

Torques applied by pin and slide to tiller

The first question by Mr. Taplin concerns the definition


of positive and negative rudder torques. During the displacing
movement of the rudder, it would be more simple to consider
the torque to be negative when the center of pressure is forward
of the rudderstock and positive when aft of the stock; however,
the frictional torque, which is always positive, complicates the
matter, and the sign convention for the hydrodynamic torque
would have to be reversed during the rudder restoring movement. There appears to be no way to truly simplify the situation, and the convention discussed in the paper was considered
82

--

to be the least ambiguous of those which could be devised.


It is interesting to note that in his April 1974 ASNE paper [26],
Mr. Taplin reported the observed efficiency of the well-tuned
steering gear for the USS Knox to be about 72 percent at a
rudder angle of 40 deg "in the relative quiet of a drydock," and
stated that "No conclusive measurements of underway efficiency were obtained." Referring to Fig. 20 of the paper, it
is seen that the measured efficiency of 72 percent corresponds
to an average friction coefficient of about 0.15 between the
moving elements of the steering gear, and that value is ap-

Rudder T o r q u e Prediction

o
4
Lo

NAMIC TORQUE,Qa

RAM TORQUE,.Q~- - ~ ~ , , , ~

INITIAL RIGHTZ-MANEUVER

RAM TORQUE,Q: "

J J J

STOCK TORQUE,O ~ / ~ ~

"

Y/A"

HYDRODYNAMICTORQUE,QI

INITIAL LEFT Z-MANEUVER


Fig. 90

Idealized torque relationships for initial right and left Z-maneuvers

proximately as expected for the materials and conditions involved. However, if when underway, the conditions aboard
the Knox are similar to those aboard other ships having rigid
rudder support arrangements, then, as discussed in the paper,
trial data show that due to the alternating nature of the loads
on the rudder support bearings, the effective friction coefficient
is approximately one-half the value corresponding to more
steady loading conditions. Therefore, for the Knox the effective friction coefficient when underway would be 0.075,
which corresponds to a ram-to-rudder efficiency of about 85
percent, which is within the range of the values often assumed.
The lengths of the models used to obtain the test results shown
by Fig. 81 were 25 and 22 ft for Model Basins Y and Z, respectively. Mr. Taplin noted the unavoidable disparity between
the Reynolds numbers of the model and the ship and questioned
the validity and usefulness of the model test results. The usefulness of the model test results Would depend upon the credibility of the calculated predictions. In cases where little excess
margin is provided in the installed steering gear rating, model
test results may be useful in corroborating estimated torques.
With regard to the validity of the model test results, a study of
the model test and sea trial results for Ship C indicates that the
full effects of stall are not reflected in the model test results, and
this limitation is most unfortunate since stall considerations so
heavily influence the required steering gear rating.
Mr. Rein confirms that the onset of stall is expected to be
delayed for the higher Reynolds numbers associated with
full-scale ships, and both he and Dr. Fairlie-Clarke referred to

experimental data [29, :30] which show that rudder stall is expected to be delayed for a rudder behind an operating propeller
as compared with a rudder in a free stream. These data are
helpful in that they support the validity of the observed trial.
results. It is, however, noted that test data clearly show that
the breakdown of rudder lift occurs later for a model of a rudder-propeller-hull arrangement than it does for a full-scale ship
even though the full-scale ship has a much higher Reynolds
number. No explanation is given for this occurrence, but a
comparison of the trial data for Ship C and the corresponding
model test results, provided by Dr. Vorus as Fig. 88, leads to the
same conclusion.
Use of the factor 5/7 as the relationship between the effective
rudder angle of attack and the actual rudder angle was suggested in reference [16]; as noted by Mr. Rem; however, this
suggestion is applicable only in particular circumstances. The
basis of this suggestion may have been the results from the USS
Norfolk tests [33] where the lift on the ship's rudder declined
(attributed to flow separation) above a rudder angle of about
25 deg during single rudder movements to 35 deg. Use of the
5/7 factor therefore translates the 35-deg rudder angle to the
25-deg angle of attack effectively observed at that time. Obviously, the exact ratio of rudder angle of attack to rudder angle
varies continuously until the ship settles into a turn, and ships
which establish a drift angle at rates greatly differing from that
of the USS Norfolk would require the use of other factors. For
example, reference [23] suggests 3/4 for surface ships, 5/7 for
submarine rudders, and 1 for submarine stern and fairwater
planes. However, for large ships, such as Ship B, there would

Rudder Torque Prediction

83

CO

I
a:

O
O

o
6

CALCULATED RAM TORQUE


) P ' ~ 2 4 o STALL ANGLE
1~23

STALL ANGLE

t,~,
CALCULATEDSTOCK TORQUE
"~z~'24 STALLIANGLE

~.--~23 STALi ANGLE

,__ .\~.,,~-~MEASURED RAM TORQUEI


' ~ . ~ - MEASURED STOCK TORQUE

/," A

'

133
C
CL
0.
Cb
"--I
O

z
,4

.8

c
CD
-(3

"

!i

/I/I/ ~"~ I

_J

..i

CL

o
0-2

UA

O.
O
o(

o
o

8o

-S

-6
O

a
a

~-s

I
~-7

uJ

tu

CALCULATID RAM TOROUE

.J

.J

10

Fig. 91

-OcI~

20
TIME FROM E X E C U T E

30
IN SECONDS

40

Calculated versus measured torques for 26-knot initial right Z-maneuver for
23-deg and 24-6eg stall angles

-8

10

20

30

40

50

TIME FROM EXECUTE IN SECONDS

Fig. 92

Calculated versus measured torques for 26-knot initial left Z-maneuver for
24-deg stall angles

-CALCULATED RAM TORQUE

STOCK TORQUE
A A

-7
/

.j

G>

z
Q.
Q.
CD

z
_J

MEASURiD STOCK TORQU i

z
MEASURED RAM TORQUE

0
c-

~-2
I.-

"0

-I
-J

Q.

_z

~,~CALCULATED RAM TORQUE

M E A S U R E D RAM T O R Q U E

~.,~.--CALCULATED STOCK TORQUE

MEASURED STOCK TORI

5
-7

-SI-

i,t
-7o

~.8

-9
10

20

30

40

S0

Fig. 93
CO
C.~

Calculated versus measured torques for 28-knot initial right Z-maneuver for
24-deg stall angles

10

20

TIME

TIME FROM EXECUTE IN SECONDS

Fig. 94

FROM

30

40

S0

EXECUTE IN SECONDS

Calculated versus measured torques for 28-knot initial left Z-maneuver for
24-deg stall angles

14NOTE: I

INOICATES RANGE OF DATA POINTS

-~- INDICATES DATA POINT

/I

10,

I
I

I
I

'I

rudder torque and the steering gear characteristics is different


when the ram drives, when neither the ram nor rudder drives
(as when the rudder is held at a given rudder angle and follows
the hydrodynamic torque curve), or when the rudder drives.
That is:
ZPRE

Q - cos2o~

I i

ZPR

Q - cos2o~
ZPR

Q - cos2c~E

//

:
-2.

-4.

\
-6'

l
l

-8.

-~0

"

10

15

20

25

30

35

RUDDER ANGLE OF ATTACK (DEG)

Fig. 95

Comparison of predicted ahead torque curves for Ship A and


trial data points; ship speed of 22.6 knots

be little difference between the rudder angle and rudder angle


of attack during the initial throw of the rudder because the
ship's heading changes only 2 deg in the time required to reach
the hardover position; therefore, the ship's drift angle would
be even less. Also, torque estimates which include such allowances for drift angle corrections would be misleading when
considering Z-maneuvers and evaluating the conditions at stall.
Either the positive peak of the QR curve or the conditions at stall
during the first rudder angle reversal of a Z-maneuver is a major
determinant of the required steering gear rating. Since the use
of such factors in either of these instances would be inappropriate, their use is not recommended.
Dr. Fairlie-Clarke observers that for steering trial data to
be useful for analysis purposes, continuous recordings must be
taken. There is no disagreement on this point, and the discussion of Mr. Butler concerning the techniques used to record
continuous trial data is appreciated. Additional traces which
provide an indication of ship speed, for example, would certainly be beneficial when analyzing the data.
The derivation for the ram-to-rudder efficiency, which
is plotted in Fig. 20, applies only to the case when the ram is
driving the rudder as stated by Dr. Vorus. The role of the
ram-to-rudder efficiency, E, in the relationship between the
86

ram driving
neither ram nor rudder driving
rudder driving

The terms in the preceding equations are as defined in the


paper. Dr. Vorus was correct when he observed that the appropriate relationship between the rudder torque and ram
pressure was used during the analysis of the trial data presented,
and elsewhere in the paper; however, his belief that this consideration may not be obvious to others attempting similar
correlations is well founded as evidenced by the manner in
which the ram pressure data recorded for the USS Norfolk [33]
were handled. In that case, the authors did not recognize any
inefficiencies in the steering engine. When questioned on this
point by the discussers, the authors stated that the available
formula had the efficiency factor in the numerator, but that
equation did not appear to be applicable since the rudder was
always driving; therefore, inefficiencies in the steering engine
were neglected. Of course, the efficiency factor should have
been in the denominator, and the torque curves given in reference [33] that were derived from pressure data are correspondingly in error.
The suggestion of Dr. Vorus to include additional illustrations
which present a comparison of estimated torque data and trial
data was considered when the paper was written. As stated in
the paper, the decision was made to present the data as recorded
with minimal interpretation such that other interpretations can
be made on the basis of different experiences or points of view.
But in order to provide guidance in correlating the trial data
with predictions, an analysis of some of the more important
characteristics of the trial data was included, recognizing that
the torque data provided in the analysis could be referred to
the predicted torque curves presented. To illustrate the intended use of the data provided, the torque data given in the
analysis of the trial results for Ship A have been indicated in Fig.
95, which is otherwise a replot of Fig. 29. Intermediate points
could also be indicated, but it is seen that the salient characteristics of the torque curves have been well-defined by the data
plotted. There are not as many data for the other ships tested;
therefore, representative ranges for those data are not known
with as much confidence. Nevertheless, the predictions and
trial results are easily compared.
The data points plotted in Fig. 83 by Dr. Vorus are not as
readily determined as those shown on Fig. 95 since the frictional
torque component must be deducted in order to have a proper
comparison. Dr. Vorus's approach is correct; and if some data
points which correspond to restoring maneuvers had been included, it is believed that the negative part of the curves would
be more accurately defined (that is, more negative).
Since rudders normally do not reach angles as high as 75
deg, Mr. McVey assumed that there was a typographical error
on page 40. Actually, it is not a typographical error, but the
point being made is easily misunderstood. A similar point was
discussed more thoroughly later in connection with spade
rudders; that is, it is assumed that the ratio of the effective aspect
ratio to the geometric aspect ratio varies linearly from 2 at a zero
rudder angle to 1 at a 75-deg rudder angle. The rudder cannot
reach a 75-deg angle, but correlations show that the assumption

Rudder Torque Prediction

provides good results over the range of rudder angles having


practical significance.
Inconsistent results are easily obtained when using DTMB
933, and the experience described by Mr. Davis is very similar
to that of others who have used that report. DTMB 933 does
not prescribe a procedure to be used when making torque
predictions, and several assumptions must be made in order to
apply the data provided. Each of these assumpti.ons (such as
that required to quantify the effective aspect ratio) can individually have a substantial effect upon the results obtained;
consequently, the use of DTMB 983 entails many opportunities
for inaccuracies and inconsistencies to be introduced.
There are several aspects of tile rudder torque prediction
process which require more extensive research before the calculations involved can be reduced to a routine process, and Dr.
Vorus questions the need for additional research to develop
solutions in ambiguous areas. There is a genuine need to have
a better understanding of the circumstances which cause ships'
rudders to stall. The lack of a better understanding of this
phenomenon is considered to be the major impediment to
further significant advances in the rudder torque prediction
process. Also, when displacing the rudder, the failure of negative torques to consistently appear of the magnitude expected
remains unexplained. If the reason for this occurrence were
known, it is possible that other avenues for advancement would
be opened.
It would be reasonable to expect a rudder which is developing
maximum lift to be more effective in maneuvering a ship than
a rudder which has stalled, and Mr. Davis suggests that the
rudder turning-rate requirements be related to the effectiveness
of turning the ship as opposed to a set time limit for rudder
movement. Steering gear tests, conducted in this manner, are
known to have been attempted on one occasion, but the considerations involved had not been studied in sufficient depth
beforehand. As a result, the rudder turning rate was not adequately coordinated with the motion of the ship, and the tests
did not provide meaningful results. However, the lack of
success with this test is no reflection on the soundness of the basic
idea. This would appear to be a promising area for further
study of the type suggested by Dr. Vorus.
The discussion by Mr. Acker is welcomed in that is provides an indication of the experience, background, and innate
"feel" for the subject which, in conjunction with the Joessel
method, can serve as the basis for successful torque predictions.
By limiting the rudder balance ratio (the ratio of the rudder area
forward of the rudderstock to the total rudder area), negative
torques will not be a significant design consideration and the
inability of the Joessel procedure to accurately predict negative
torques will not be of practical significance. With the rudder
balance so limited, the maximum torque will occur either at the
maximum rudder angle (as assumed by Mr. Acker) or at
stall.
The measured rudder torque data for the Bethlehem Quincy
tanker is seen to have the characteristics of a moderately balanced rudder during tile initial maneuver to 35R, and the
maximum torque reading of 9.8 106 in.-lb is recorded during
the subsequent Z-maneuver. Mr. Acker notes that the sterns
of the Bethlehem Quincy tanker and Ships E and F are similar,
but on occasions the highest torque readings for Ships E and F
(see Figs. 76 and 78) occurred during the initial maneuver. The
ship's speed would be somewhat higher during the initial maneuver than during subsequent maneuvers, thus providing the
capability for higher torques; and it is believed that, on occasions, the rudders for Ships E and F stall during the initial maneuver. Figures 35 and 36 shows that the rudders for Ships E
and F were predicted to stall at rudder angles of 27 and 28 deg,
respectively. The rudder for the Bethlehem Quincy tanker
appears to have had a relatively low effective aspect ratio, re-

suiting in a delayed breakdown of rudder lift to higher rudder


angles of attackmthis is an advantage of rudders having low
aspect ratios.
One of the simplifying assumptions suggested in the paper
is to consider the spanwise center of pressure to be located 42
percent of the span distance from the root chord. In the
comments of Ms. Berman and Mr. Hunt, further details are
requested regarding this assumption. As shown in DTMB
Report 988, the theoretical location of the spanwise center of
pressure is approximately 42 percent of the span distance from
the root chord for the family of all-movable control surfaces
tested. Procedures are given to compute the spanwise center
of pressure for other lifting surfaces having an elliptical load
distribution; however, experience shows that the values computed for the usual rudder configurations do not vary significantly from the 42 percent value. Since the only use made of
the spanwise center of pressure is to compute the reactions in
the rudder support bearing (and subsequently the frictional
torque component), a small compromise in the accuracy of the
estimate of the spanwise center of pressure does not entail any
consequences of importance in that the inaccuracies are reflected only as a minor error in the frictional torque component.
However, by estimating the spanwise center of pressure in this
manner, the calculations are simplified considerably.
Another simplifying assumption suggested in the paper is to
consider both the upper and lower rudder sections to have effective aspect ratios equal to twice the geometric aspect ratios
during astern operations. Ms. Berman and Mr. Hunt also
requested more details regarding this assumption. Of course,
the upper rudder section is completely sealed by the lower
rudder section, but the converse is not the case. That part of
the lower section below the rudder horn is sealed at a zero
rudder angle by the horn, but the seal becomes progressively
degraded at larger rudder angles. Referring to Fig. 80 of the
paper, where the rudder for Ship A is shown in the center-line
position as a solid line and in alternate 85-deg positions as
dashed lines, if that part of the rudder shown in the centerline
position forward of the rudder stock is considered to be the
rudder horn, then it is seen that the rudder must be displaced
to a rather large angle before the rudder horn and the underlying movable part of the rudder are separated. Nevertheless,
a correction could be made for the reduced sealing effect provided by the rudder horn with increasing rudder angles. But
recognizing that during astern operations the rudder stalls at
angles of attack exceeding 20 to 25 deg, it was reasoned that the
correction would not be of practical importance.
There are several conditions associated with the sea trial
itself which could have influenced the results obtained during
the steering tests, and Mr. Nelson mentioned several which may
help to explain the differences between predicted performance
and trial data. Records were made of the ship's draft and the
sea state during the trials, and they are as follows:
Ship
Draft, fwd, ft
Draft aft, ft
Sea state

A
17.6
22.2

B
' 74
74
2

C
36
86
2

D
15.2
21.4
2

E
11.9
24.8
8

F
17.5
24.8
8

However, definitive data were not recorded for the other


conditions cited.
Mr. Stiansen requested some guidance concerning the
selection of the coefficients to be used in the suggested calculation procedure and requested more details with regard to the
procedure's limitations. In order to avoid introducing unnecessary complexities in tile calculation procedure, several
approximations are suggested; for example, one deals with the
augmented water velocity in the propeller race, others facilitate
the determination of rudder support bearing loads, and yet
others concern the effective aspect ratio. The principles upon

Rudder Torque Prediction

87

which the various approximations are based are described in


sufficient detail to permit an independent evaluation of their
validity and, in this regard, no serious shortcomings are known.
Several coefficients are required to compute the augmented
water velocity in the propeller race, and these are best approximated on the basis of data for similar ships; fortunately,
highly accurate data are not required for this purpose. Also,
values must be assumed for the bearing friction coefficients.
As discussed in the paper, there are several sources of guidance
in the selection of expected values for the various combinations
of materials; however, the data and analysis presented show that
for ships having rigid rudder support arrangements, the effective in-service coefficients are approximately one half of the
values corresponding to steady loading conditions. The friction
coefficient data given in Table 4 reflect the experience derived
from the data analyzed and can be regarded as suggested values
for ships similar to the six described. It is noteworthy that the
experience of Dr. Fairlie-Clarke is in agreement with the~
bearing friction data in Table 4.
With regard to the limitations of the procedure, the major.
shortcoming is the inability to accurately predict the angle of
attack at which stall occurs for full-scale ships. Another limitation is the failure of full-scale negative torques to appear of
the predicted magnitude; however, since it is most unlikely that
the negative torque region would ever be a design constraint,
this limitation is not considered to be of great concern.
Two thoughts are offered for the consideration of Mr.
Stiansen when establishing classification society rules concerning rudder torques. One is that unless the rate of rudder
movement is controlled in some manner (perhaps as suggested
by Mr. Davis), the rudder will stall when following the displacing curve during Z-maneuvers, thereby limiting the maximum torque which can be developed. Another is that for
heavily balanced rudders, the positive peak of the restoring
curve can be the maximum torque required. In view of these
considerations, it may be feasiable to develop one expression,
which could be along the lines described by Mr. Acker, to
evaluate the maximum torque required to displace the rudder,
and then perform a check to ensure that the maximum restoring
torque required by rudders having a large amount of balance
is not overlooked.
The proper amount of margin to be added to predicted
torques when establishing design ratings is difficult to quantify
in a general sense, for several reasons. The underlying necessity
for margin is a lack of confidence in a torque prediction for a
specific ship. The level of confidence in a torque prediction
would depend upon the extent and nature of related experience
with similar ships. The large amount of consistent data
avail~ible for Ship A, for example, would permit a relatively
small design margin to be used when making predictions for
similar ships; similarly, however, the lack of consistent data for
Ship D would obviously require more margin. Another important consideration is that, with other factors the same, more
margin is appropriate for the more heavily balanced rudders
where the design rating of the steering engine is related to the
restoring torque curve. The reason for this concern can be seen
from Fig. 62, which is a pressure trace for Ship B. The peak
of the restoring-torque curve corresponds to the 50- and 103second marks, and if the steering engine were unable to develop
the torque required at these times, the rudder could not be reversed during the Z-maneuvers. (Note that for the less-balanced rudders, where the maximum torque occurs at stall, the
maximum ram pressure may occur near the end of rudder
travel where the inability to develop the required torque is not
as serious; this case is illustrated by the 39-second mark of Fig.
44 for Ship A and by Fig. 84, which was provided by Mr.
Acker.) Also, for the more balanced rudders, the distance
between the center of hydrodynamic pressure and the rud88

derstock centerline is less and the effect of small errors is correspondingly magnified, thus necessitating more margin.
There are typographical errors in DTMB 933, as noted by
Mr. Vassilopoulos, but they were recognized and are not the
serious shortcoming of the semi-empirical equations cited in
the paper. Figure 7 shows that the DTMB 9:3~3semi-empirical
equations provide an entirely satisfactory approximation of the
DTMB 933 test data, but Fig. 8 shows that the semi-empirical
equations do not adequately represent the chordwise center of
pressure test data. The comparison in Fig. 8 is specifically for
Ship A, but as can be seen from Fig. 96, similar results will be
obtained with other ships. Figure 96 is a reproduction of Fig.
35 in reference [12], and it is seen that particularly for aspect
ratios in the vicinity of two (the middle set of curves), which are
of the most practical importance, there is an increasing disparity
between the semi-empirical equations and the test data above
angles ofattack of about 16 deg. At first glance the disparity
may not appear to be large, but upon closer examination it is
seen that the scale which is used tends to obscure the true
magnitude of the disparity; and it is in fact, very large. This
may explain why the inaccuracy associated with torque predictions obtained with the semi-empirical equations at the
higher rudder angles has not been widely recognized.
Both Mr. Taplin and Mr. Vassilopoulos refer to the work done
with flapped rudders by Kerwin et al. These references were
reviewed when the problems associated with the prediction of
torques for horn rudders were being studied. The only reason
for using the University of Maryland report [20] as the basis for
the curves developed was the greater familiarity with that data;
if instead, the data provided by Kerwin had been used as the
basis, the final results shown by Fig. 21 would not be significantly different. This is not to suggest that Fig. 21 is not subject
to modifications--it certainly is. Figure 21 is the best representation of all the test data presently available, but obviously,
as more data are acquired, Fig. 21 should be reviewed and
modified as appropriate.
Mr. Taggart and Mr. Vassilopoulos make note of the different results often obtained when deriving rudder torque
values from strain-gage measurements and from ram pressure
measurements. There was a great deal of discussion on this
subject in connection with the Norfolk tests [33], but those tests
must be regarded as a pioneer effort into an area of unknowns.
As we now know, steering gear inefficiencies were neglected
when deducing the rudder torque from the ram pressure data;
but this is not the only correction to be considered when evaluating that data. The Norfolk was twin-screw and it is believed
that the ship had only a centerline rudder, which would make
a great deal of difference in the prediction procedures used. Be
that as it may, the tests were an attempt to measure what we
now know to be the highly erratic negative torque region of the
displacing curve and were destined to provide confusing and
misleading results.
It goes without saying, however, that the effect of the same
torque is reflected in either the strain-gage data or the ram
pressure data. In many cases where different results were
obtained from ram pressure data and strain-gage data, the basic
problem may have been that the engineering principles had
not been studied in sufficient depth to permit a rigorous
translation of strain-gage data and ram pressure data back into
rudder torque values. Mr. Vassilopoulos staes that he would
prefer to work with strain-gage data as opposed to ram pressure
data and inasmuch as he apparently has good confidence in his
instrumentation techniques, that is understandable. But good
arguments can also be made in favor of working with ram
pressure data by those who feel more comfortable with hardware. Handled exactingly, the results obtained with either set
of data should be the same.
Mr. Taggart reviewed the considerations which were eval-

Rudder Torque Prediction

pressure data, was about 8.8 x 106 in.-lb; but the design rating
was given by Mr. Taggart as 6.8 106 in.-lb. The normal force
on the DE-1040 rudder at stall is estimated to be about 700 000
lb; therefore, if the rudderstoek had been moved aft in the
rudder about 3 in., the design rating would not have been exceeded. Also note that if the rudderstock had been moved aft
6 in., the steering engine may have been considered to be overrated. Bearing in mind that the meaia chord of that rudder
is about 11 ft, it is seen that considerable accuracy is required
of the naval architect when making rudder torque predictions.
The discussers have provided to the Society the benefit of
their varied experiences which cover all aspects of the rudder
torque prediction process. Their contribution should be of
considerable benefit to those making sueh predictions in the
future.

uated in connection with the problems experienced with the


DE-1040, and there appears to be a good correlation between
the calculation results presented and the trial data for this ship.
A more detailed description of the procedures and assumptions
used in the analysis would have been welcomed. Of particular
interest would be an understanding of the analytical techniques
used to assess rudder stall. It is noted that the DE-1040 is a
lightly balanced rudder with no significant negative torque
characteristics, it would also be interesting to use the same
procedure to analyze a highly balanced rudder which has an
extensive region of negative torque.
Even though the DE-1040 rudder may stall at a 23-deg angle
of attack, a 40-deg-angle rudder capability may still be advantageous since the higher rudder angle will reduce the ship's
turning circle diameter, once the ship has fully settled into a
turn. Of course, if a reduced turning circle diameter is not
important for the DE-1040, then the 40-deg rudder angle capability may provide no benefit.
An appreciation of the accuracy required for successful
rudder torque predictions can be obtained from Fig. 86, which
was provided by Mr. Taggart. That figure shows that the
maximum torque observed during sea trials, based upon ram

Additional references
26 Taplin, A., "Efficiency of Electro-Hydraulic Rapson Slide
Steering Gear," Naval Engineers Journal, April 1974.
27 Kerwin, J. E. and Mandel, P., "An Experimental Study of a

NACA

0015 SECTION

SHAPE

C D =O.DO
c
SQUARE
R-2.7

TIP

x 106

EXPERIMENTAL
SEMI-EMPIRICAL
EQUATIONS
JOESSE L

0.1

,11.

..--U

----" ~-"

" "

"~'-- -----

-"--

l--l---

Z
m

G
~.
uJ
0
U

-3.0

-O.1

0.1

Z
w
m

-2.0

Z
U
-0.1
0.1

,=-Z_

|||E|,

~ ~ . ,

,.-lo

-O.1
0

12

16

20

24

28

32

36

40

A N G L E OF A T T A C K , Q '
Fig. 96

Comparison of experimental and calculated pitching moment coefficient for square-tip control surfaces of various aspect
ratios

Rudder Torque Prediction

89

Metric Conversion Table

Original experimental data herein were measured in U.S. customary


units. Following is a list of standard factors for conversion to metric
values:
1 in. =
1 ft =
1 ft2 =
I psi =
1 psf =
1 lb =
1 long ton =
1 hp =

25.4 mm
0.3048 m
0.092 m z
6.894 kPa
47.880 Pa
0.45 kg
1.016 047 metric tons (t)
0.7457 kW

Series of Flapped Rudders," Journal of Ship Research, Vol. 16, No.


4, Dec. 1972.
28 Taplin, A., "'Sea Trials for Measuring Rudder Torque and
Force," in Proceedings Fourth Ship Control Systems Symposium, The
Hague, Vol. 5, Oct. 1975.
29 Shiba, H., "Model Experiments About the Maneuverability of
Ships," First Symposium on Ship Maneuverability, David Taylor Model
Basin, DTMB Report 1461, Oct. 1960.
30 Fujii, F. and Omori, H., "Experimental Research on Rudder
Performance; Parts 1 and 2," Shin Mitsubishi Heavy Industries Engineering Technology Reports, Vol. 4, No. 2 and Vol. 4, No. 4, available
in English as British Ship Research Association Translation No.
3;372.
31 Kerwin, J. E., "'A Lifting Surface .Program for Trapezoidal
Control Surfaces with Flaps," Department Of Ocean Engineering,
M.I.T., Cambridge, Mass., Aug. 1974.
32 Kerwin, J. E. et al, "'Experiments on Riadders with a Smaller
Flap in Free-Stream and Behind a Propeller," Department of Ocean
Engineering, M.I.T., Cambridge, Mass., Report No. 74-16, Oct.
1974.
33 Becker, L. A. and Brock, J. S., "'The Experimental Determination of Rudder Forces During Trials of USS Norfolk," TRANS.
SNAME, Vol. 66, 1958.
34 Lotveit, M., "A Study of Rudder Action with Special Reference
to Single Screw Ships," Trans. NECIES, 1959.

90

35 Hoerner, S. F., "'The Hydrodynamic Torque of Ship Rudders,"


Second Shp Control Systems Symposium, 1969.
36 Norman, D. W., "Jet Flaps and Jet Assisted Rudders for Ship
Control," Second Ship Control Systems Symposium, 1969.
37 English, J. W. and Steele, B. N., "Ship-Borne Maneuvering
Devices" in Proceedings, First International Tug Conference, 1970.
38 Kennard, E. H. and Leibowitz, R. C., "Theory of Forces and
Moments oil Rudders as Affected by Ship Turning, Steady Change of
Rudder Angle and Propeller Race," NSRDC Report No. 3246, Jan.
1970.
39 Kwik, K. H., "Systematic Wind Tunnel Tests with Ship Rudders," Schiffstecknik, May 1971 (in German).
40 English, J. W. et al, "'Some Maneuvering Devices for Use at Zero
and Low Ship Speed," Trans. NECIES, 1971-1972.
41 Hagen, G. R., "A Contribution to the Hydrodynamic Design
of Rudders," Third Ship Control Systems Symposium, Sept. 1972.
42 Gregory, D. L. and Dobay, G. F., "The Performance of HighSpeed Rudders in a Cavitating Environment," SNAME Spring Meeting, 1973.
4;3 Hendrickson, W. A., "A Systematic Investigation into the
Concepts and Applications of a Foil with a Spinning Cylinder at its
Leading Edge," SNAME, New England Section, 20 Sept. 1974.
44. Mathis, P. B. and Gregory, D. L., "Propeller Slipstream Performance of Four High-Speed Rudders under Cavitating Conditions,"
NSRDC Report 4361, May 1974.
45 van Berlekom, W. B., "'Effects of Propeller Loading on Rudder
Efficiency," Fourth Ship Control Systems Symposium, 1975.
46 Kerwin, J. E. and Zolotas, A. B., "Experimental Optimization
of Propeller Rudder Orientation for Minimum Vibratory Loading,"
Department of Ocean Engineering, Report 75-6, M.I.T., Cambridge,
Mass., Feb. 1975.
47 "Rudder Analysis by Lloyd's Register," The Naval Architect,
March 1976.
48 Osouf, J., "Comportment Dynamique des Gouvernails," (Dynamic Behavior of Rudders), Bulletin, ATMA, 1976.
49 "High Performance Rudders for Improved Ship Handling,"
The Naval Architect, March 1979.
50 Laskey, N. V., "Design of Steering Gears, Rudders, Rudderstocks, and Propeller Protection for Canadian Arctic Class Vessels,'"
Marine Technology, Vol. 17, No. 3, July 1980.
51 Goodrich, G. J. and Molland, A. F., "Wind Tunnel Investigation
of Semi-Balanced Ship Skeg-Rudders,'" The Naval Architect, Nov.
1979; also, Trans. RINA, 1979.

Rudder Torque Prediction

Vous aimerez peut-être aussi