High Performance Yacht Design Conference

Auckland, 4-6 December, 2002

IMPLEMENTATION, APPLICATION AND VALIDATION OF THE ZARNICK STRIP

THEORY ANALYSIS TECHNIQUE FOR PLANING BOATS

Christopher D. Barry 1, cbarry@elcbalt.uscg.mil

Debabrata Ghosh 2, dghosh@elcbalt.uscg.mil
Richard Akers3, dakers@shipmotion.com
Andrew Ulak 4, afu.mla@att.net

Abstract. Zarnick developed an approach to calculate the resistance, motions in waves, and resultant pressures of planing craft by
stripwise integration of the forces on transverse sections modelled as wedges entering the free surface vertically. A practical
software implementation of this technique was developed and compared to model test results for a variety of military, recreational
and working craft. Numerous modifications and improvements to the technique were made in the course of this study, most notably
to a series of coefficients to represent buoyancy and three-dimensional lift and drag factors, to improve the correlation of the method.
The final results of this study show that this method produces better evaluations of resistance for relatively high-speed vessels than
previous methods and compares reasonably well with model test results for motions. This method explicitly models variable beam
and deadrise, which previous methods cannot address. The basic background of the technique, its implementation and the
development and modifications of coefficients is discussed with a history of validation efforts to improve the correlation.

1. INTRODUCTION some cases the panelization near the free surface must be
modified to avoid numerical difficulties. Panel methods
Resistance and motions prediction is an ongoing problem
also cannot be used to directly model viscous effects
in the design of high speed planing craft. Standard
such as boundary layers or spray. Finally, panel
analytic methods such as the Savitsky-Brown technique
methods are extremely computationally intensive and
for resistance, and the Hoggard-Jones or Savitsky-Brown
require special skill to panelize the surface appropriately.
equations for motions only account for a very few
parameters of the design, generally maximum chine 2.1 Zarnick Methods
beam, a single characteristic deadrise, weight and
Zarnick [2,3], following the work of Martin [4],
longitudinal centre of gravity. Designers know that
developed a mathematical formulation for the
warping the bottom and varying beam down the length of
instantaneous forces on a planing craft. In Zarnick's
the vessel can have beneficial or adverse effects, but
method a planing craft is modelled as a series of strips or
these effects can only be evaluated during initial design
impacting wedges. At high speeds, the surge
based on comparisons with model tests and model data if
perturbation velocity of the water is small relative to the
comparable data is available.
speed of the wedge and can be neglected. The passage of
a planing hull with deadrise can therefore be modelled
Motions in waves are often as important as smooth water
as, from the water fixed system, a wedge entering the
resistance and even less reliable guidance is available for
water at some vertical velocity (Figure 1). The
either prediction or optimisation of craft design for
instantaneous trim and transom draft of the boat predicts
motions. Since motions govern sea loads, inability to
the immersed depth, and hence the shape of a stripewise
reliably predict motions is a structural issue as well.
entering wedge at any lengthwise position on the hull.
r
2. NUMERICAL APPROACHES (Figure 2) A characteristic outward normal vector n is
defined for each strip and submergence (Figure 3). The
A number of investigators have applied three- normal vector is arbitrarily defined at the longitudinal
dimensional panel methods to the problem of planing midpoint of the strip and at a height of 2/3 of the
craft with some success, for example, Jiminez [1] applied submergence above the keel line. The instantaneous
panel methods to sailboard resistance prediction and deadrise angle is defined as the angle of the normal with
optimisation. However, the free surface location, which respect to the baseplane as seen in the body plan view.
is required as an input to the panel equations is a function The hydrodynamic force on the strip per unit length can
of the solution of the method, so that the final solution then be determined as a sum of the added mass on the
requires iteration on the free surface position. (This is a impacting wedge and the hydrostatic force.
more significant problem for a relatively heavily loaded Since the instantaneous elevation and kinematics of the
powerboat than it might be for a lightly loaded water surface are readily calculated at any point along the
sailboard.) This not only adds to solution time, but in length of the vessel based on trim, heave and wave

1 Senior Naval Architect, Boat Engineering Branch, US Coast Guard ELC, Baltimore MD
2 Chief, Boat Engineering Branch, US Coast Guard ELC, Baltimore MD
3 Principal, Ship Motion Associates, Portland ME,
4 Principal, Mount Lebanon Associates, Pasadena MD
 Time
2-D Impacting
Wedge
Jet

Calm Water Wetted

Beam

Dynamic Wetted
Beam

Figure 1. Planing Hull Passage From Sea Fixed Origin Figure 2. The Entering Wedge Strip
separation at the stern decreases hydrostatic lift, and both
cause an increase in pitching moment. These effects are
speed dependent, and there is no single factor that can be
used to correct the hydrostatics calculations for flow
separation. In his work on rectangular planing surfaces,
Shuford [6] suggested that hydrostatic buoyancy should
be halved in a dynamic simulation in order to achieve the
correct total lift force. Zarnick found that use of an
additional factor of one-half for the hydrostatic moment
resulted in an accurate trim angle. The algorithm
includes a buoyancy force and a buoyancy moment
coefficient to correct the vertical force and pitching
moment. These coefficients can be set to 0.5 based upon
the recommendation of Shuford and Zarnick, or they can
be set empirically so that simulation results match tank
test results. Determining these coefficients has proven to
be an interesting problem, and the recent work with this
method has largely involved processes to derive
appropriate values of the coefficients.

3. IMPLEMENTATION
Akers [5] has implemented this method in the form of a
Figure 3. The Stripwise Panel and Its Normal
computer program, POWERSEA, which runs in the
profile, the Zarnick method can be used for the Windows 95, 98 or NT environment. The program
calculation of resistance and dynamic stability, (i.e. provides a graphic interface for inputting hull
freedom from porpoising) in still water and waves, and parameters as well as methods for data input from other
motions, velocities, accelerations and forces in waves. file types. The hull geometry is ultimately converted of
The method also gives the longitudinal distribution of splines representing the keel, chine and deck edge. 201
forces (for calculation of global loads) directly and can stripwise stations are derived from this model for
be used with suitable assumptions of the transverse subsequent analysis. The program includes both regular
distribution of pressure to estimate pressures. Akers [5] waves and irregular seas including Pierson-Moskowitz,
provides details of the calculation of forces and JONSWAP, ISSC and Ochi spectral density
moments, the stripwise integration, the solution of the formulations. The actual wave histories developed from
equations and proposes a process to estimate sealoads. the spectral formulations are synthesized by combining
1024 frequency components. The software is also
2.2 Hydrostatic Forces capable of modelling ship wakes by specifying leading
Hydrostatic forces and moments must be included in the and following packets. Powering options include
analysis, but are difficult to predict. Water rise at the constant velocity and constant thrust applied at a location
bow of a planing vessel increases hydrostatic lift, flow and vector as appropriate for the propulsor simulated.
Modelling of appendages such as skegs, struts, rudders
and trim tabs are by standardized formulations. An air
drag appendage is also included, and can be very
important, especially for matching model test data.
(Note though that the air drag appendage does not
currently include lift, which can be significant for
extremely high-speed craft.) The input also includes
instructions to report velocities and accelerations at any
desired locations. The user also generally has to input an
initial condition, and this can be a source of problems, as
it is possible to select an unrealistic condition that does
not converge to a stable or realistic condition. Finally, the
program includes standard Savistky planing and pre-
planing analyses operating from the same database. Figure 4. Typical Fridsma Hull
was appropriate for accurate pitch and trim above
The following parameters can be evaluated: Volume Froude Numbers of about 2.0, whereas
modelling the full chine beam produced better powering
Calm Water Seakeeping predictions through the speed range. Double chine or
round bilge hull forms also allow a projected chine
Keel and Chine Offsets Calm Water Data technique (the intersection of a vertical at the assumed
CG and Displacement Calm Force File waterline and the continuation of the deadrise near the
Thrust Location and Angle Wave Spectrum centreline), and an effective chine technique (the
Radius of Gyration Definition intersection of the hull and the assumed waterline). The
Thrust Mode (Constant Or chine point can also vary with length. The combined
Thrust Or Constant Wave Amplitudes chine technique uses a transition from an outer chine
Velocity) and Wave toward the bow. This technique recognizes that the aft
Speed for setting default Lengths sections are generally wetted, especially at low speeds,
Buoyancy Coefficients * Or whereas the forward sections are generally dry.
Appendage Characteristics, Ship Wake
Location, Centre of Effort Characteristics Appendages, especially wedges and flaps that
substantially change trim have substantial effects on the
The result of the analysis is a time history of position results of the analysis. Running the model with and
(trim and heave), velocity, acceleration at desired without the appendages is wise to develop a sensitivity
locations, and effective horsepower. This is the case for analysis of the analysis for these appendages.
both calm water and wave runs. In the case of calm
water, the run is usually started at a speed below those of 4. VALIDATION
interest and accelerated to the top speed.
4.1 Initial Work
3.1 Hull Form Modelling
The initial validation of the algorithm and the program
The basic theory of Zarnick assumes a simple wedge was performed on the data of Gerard Fridsma [8, 9].
entering the water. Most modern planing craft have Fridsma constructed and tested a range of idealized
either extended flat chines forming a spray rail for some prismatic models (Figure 4) at different speeds and in a
percentage of the length. Other modern forms, especially range of regular seas as well as calm water at a range of
those for larger craft such as patrol craft, have double weight and LCG conditions. This study found a
chines with a narrow segment between the two chines at significant dependence on the buoyancy coefficients as
an intermediate angle between the bottom and the side. noted above. Each configuration and wave period/height
The analyst has to select a single point to use at each combination in Fridma’s tests was run using the initial
station to model the effective characteristics of the hull coefficients of 0.5 (the “low buoyancy” option”) and
form, and this requires some significant thought and coefficients intentionally set to accurately reproduce the
perhaps an iterative process of modelling, running and calm water trim and resistance (the “Fridsma option”).
comparison with other methods and data. The latter coefficients were typically on the order of 0.7
to 1.0. However the results using the higher buoyancy
For a typical flat chine there are three basic choices, the options generally over predicted the motions. A typical
inner chine, the outer chine, and an average of the two. plot of this effect is shown in Figure 5 from Akers [5].
Since the immersed section changes with draft and trim, The algorithm was also compared to tests in irregular
two models might be appropriate, one for lower speeds waves for a limited set of cases, and it was found that
and one for higher speeds to represent appropriate wetted pitch and heave response correlated reasonably well.
shape of the hull at the intended speed. As a general Centre of gravity and bow accelerations were somewhat
rule, Ulak [7] found that using the inner chine, thereby (30%-34%) under predicted . A method was developed
modelling the lower portion of the section accurately, to predict structural loadings on the hull panels, and
 running trim angles are still significantly higher than the
3.5 observed trim angles, but the Fridsma studies
demonstrated that in many cases underestimating the
3.0 hydrostatic forces gave better dynamic results than by
using the exact hydrostatic forces. (Note that
2.5 overestimating trim in the analysis results is equivalent to
qp/(2pH/l)

2.0 requiring a higher angle of attack and a deeper

immersion on the hull to maintain equilibrium. Hence
1.5 the analysis is underestimating vertical forces actually
generated for a given kinematic condition of the hull in
1.0 steady conditions. This may occur because most
0.5 transient behaviour of a high-speed vessel is the result of
hydrodynamic forces, not hydrostatic forces, so more
0.0 accurate results are obtained if the hydrostatic forces are
0.00 0.05 0.10 0.15 0.20 0.25 0.30 understated.
CLambda The wake set up by the other boat also had to be
modelled and since the test boat slowed as it encountered
Pitch RAO
the wake, the exact condition of speed with respect to
Pitch RAO (low buoyancy) time as the boat passed through the wake had to be
Fridsma approximated by an analytic function. The speed
change was determined from actual measurements and
was ramped linearly down from 37 feet per second to 33
Figure 5. Typical Fridsma Comparison feet per second in about ten seconds. The results of a
reasonable values, comparing well to those used by typical experiment at 22 knots are shown in Figure 6.
Spencer were determined. 4.3 Coast Guard Studies
4.2 Wake Impact Studies In response to a perceived need to support numerous
Akers, Hoeckley, Peterson and Troesch [10] compared upcoming procurement initiatives, (which have increased
the results of this algorithm to performance of a 25-foot in tempo since 9/11), the U.S. Coast Guard Engineering
utility boat in calm water and during transition of a wake Logistics Centre, Boat Engineering Branch has been
set up by another boat. This effort was intended to get engaged in a series of Survey and Design initiatives to
reliable physical data in a controlled full-scale improve the reliability of performance predictions of
environment with a reliably reproducible and repeatable high speed craft. The branch had previously co-
wave environment. The time history of the wave sponsored an initiative relating to Zarnick methods with
generated was measured by videotaping a buoy and the the U.S. Navy, and based on that and the papers

20
Sim Pitch Rate
10 (degr/sec)
Wave Elevation
0
at FP (in)
-10 Meas Pitch
Rate
-20
2 3 4 5 6 7 8 9 10
Time (seconds)

Figure 6. Wake Crossing Comparison

boat was instrumented with accelerometers and a gyro discussed here, tasked Band, Lavis, Associates (BLA) to
inertial measurement unit. evaluate a contemporary Zarnick based program. BLA
determined that POWERSEA was the only commercially
At 22 knots the observed trim of the boat ranged from available software for this type of analysis.
2.2 to 4.0 degrees for different runs, and 2.0 to 2.6
4.4 Initial Studies
degrees at 30 knots. The calm water simulation with the
default values of 0.5 resulted in about 6 degrees at 22 The report, by Ulak [7], compared a range of boats for
knots and 4 degrees at 30. Increasing the buoyancy force which the U.S. Coast Guard had model test data with the
and moment coefficients to 0.8 and 1.0 respectively predictions of a Zarnick method. The specific vessels
resulted in 4.0 degrees at 22 and 3.5 at 30 knots. These were chosen to cover a range of speeds and hull shapes
 BLASport Fishing Hull, Light Load, LCG 23ft FWD
(Includes Air Drag)

4000 5

Tank Test Trim

4.5
3500
Savitsky Trim
4
3000
3.5

Running Trim (Degree s)

2500
3
EHP (HP)

Tank Test EHP

2000 2.5
Powersea Trim

2
1500

Savitsky EHP 1.5

1000
Powersea EHP 1

500
0.5

0 0
Fb=2
10 15 20 25 30 35 40 45 50 55 60 65

Speed (KTS)

Savitsky EHP Tank Test EHP PS EHP Savitsky Pitch Tank Test Pitch PS Trim

Figure 7. Sport fishing Hull Calm Water Comparison

including types that would not be well represented by the 60 knot single chine boat with considerable rocker
algorithm. and some warp to the bottom running all the way aft.

The vessels included in the analysis were: The Island class studies were dropped after some initial
· 47 MLB, a 47 foot, 28 knot hard chine surf rescue work that confirmed that there was no reliable way to
boat with a full length warp to the buttocks, a model a round bilge boat of this type. Initial studies of
somewhat rounded stern and a full width wedge at the the other four designs brought out a number of issues
stern. relating to the use of the initial version of the software.
· The Chernesky, a 46 foot, 26 knot survey boat. This The modelling issues discussed previously affected the
boat has a wide flat chine and a prismatic hull form results significantly. The Heritage analysis resulted in
aft of midships. It initially failed to make speed and good correlation at speeds between 27 knots and 35 knots
was model tested with a prismatic extension and a based on a combined chine technique, but a projected
wedged extension and a variety of loading chine worked better at lower and higher speeds.
conditions, so it represented a range of systematic
cases, however only calm water data was available for The sport fishing hull, since it was closest to the ideal
this boat. assumed by the algorithm, was probably modelled best
· The Heritage is a 120 foot, 30 knot patrol boat. It has overall, however, even these predictions were sensitive to
a double chine. the selection of buoyancy coefficients. Typical
· The Island class patrol boats are 110 feet at 30 knots, comparisons are shown in Figure 7 and Figure 8.
and have a “classic semi-displacement” hull form Though reasonably good, the calm water prediction is not
with a deep vee rounded shape forward worked into a much better than a standard Savistsky analysis.
nearly flat bottom with hard chines aft. These Extensive simulation of the 47 MLB (which was
methods were not expected to model the 110 well, but probably had the most complete set of test data) was also
it was included to confirm the boundaries of the somewhat disappointing.
method. 4.5 Improvements
· BLA was also required to select an additional craft for
the comparison matrix and chose a high-speed sport The basic problem is that the Zarnick technique is strictly
fishing boat of their own design. This was a 61 foot, two-dimensional. This means that each section has
 BLASport Fishing Hull, Full Load, LCG 22ft FWD
Significant and RMS CGAccel. @ H1/3=4.7ft

5
POWERSEASig. CG

4
Accelerati on (g's)

Tank Sig. CG

POWERSEACG Pitch

Tank RMS CG

0
Fb=2
10 15 20 25 30 35 40 45 50
Speed (Kts)

TTCGRMS PS CGRMS TTCG 1/3 PS CG1/3

Figure 8. Sport fishing Hull Rough Water Comparison

constant deadrise, i.e. no warp or longitudinal camber.
As a result the effective lift produced by the angle of Figure 9 shows the results of a typical calm water
attack of the section due to for and aft curvature is comparison with the automatically optimised
ignored. The method therefore underestimates lift and coefficients. Figure 10 shows a comparison of motions
thus generally requires more trim to achieve equilibrium. in irregular seas. This is clearly an improvement,
In addition, the forces on the sections, are strictly based especially since the point of the program is not so much
on added mass in the basic theory because the force is calm water performance as motions and loads.
due to the substantial derivative of added mass-like
4.6 Results
“imaginary” forces. Especially aft, the change of vertical
velocity of wedges is zero, hence there are no mass-like The usefulness of Zarnick's strip theory approach was
forces. The strict interpretation of the algorithm would demonstrated to a limited degree – it works best when
therefore result in no forces aft, which is clearly the simulation is tuned to match tank test data. In
incorrect. The standard algorithm thus adds the cross- general, with tank test data for verification of simulated
flow drag coefficient to produce force due to the results, POWERSEA should outperform the Savitsky
essentially constant vertical velocity of the wedge as well planing method for single, hard-chined hulls at speeds
as the added mass coefficient. However, this is an corresponding to a beam Froude number of 2.0 or
arbitrary value. As a result of all of these greater.
approximations, corrections are required. (In addition to
the buoyancy forces and moments.) Though strictly Each hull examined in the study was unique, so there was
speaking, they are not all buoyancy forces, there is no no measure of the repeatability and applicability of the
real point in separating them. hull modelling procedures developed for the multi-chine
and round bilge hulls. The recommended approach
Based on these early results, BLA and Ship Motions requires “tuning” the model with calm water resistance
Associates worked together to develop an improved data, then using it as a tool for performing critical
version of the software that addressed some minor use response and motions/accelerations analysis. "Tuning"
issues, but most importantly, had an internal correction involves making changes in the optimised buoyancy
coefficient matrix so that the program automatically coefficients and hull geometry until tank test results are
optimised the “quasi-buoyancy” coefficient based on reproduced within an acceptable error margin. This
speed and loading. This later version of the program was might suggest that the program is of little use, but it is
used to rerun the 47 MLB data. difficult to do motion testing, so the ability to determine
 2.0

47 MLB Calm W ater Heave

1.5

1.0

0.5

Speed, Knots
0.0
10 15 20 25 30 35

-0.5

PowerSea Model Tests Savitsky

-1.0

600

47 MLB Calm W ater EHP

500

400

300

200
PowerSea
Model Tests
100
Savitsky

Speed, Knots
0
10 15 20 25 30 35

2.0

47 MLB Calm W ater Heave

1.5

1.0

0.5

Speed, Knots
0.0
10 15 20 25 30 35

-0.5

PowerSea Model Tests Savitsky

-1.0

Figure 9. 47 MLB Calm Water Comparison with Optimized Coefficients

good motion responses from calm water information is It is also important, though, to remark that though the
actually a powerful tool. Even if motion testing is current algorithm is limited in its usefulness, it is the only
accomplished, the speed of planning boats is such that it methodology currently available for motions of planing
is difficult to get a statistically valid record of waves – hulls in waves.
the run times are too short. Thus running an abbreviated 5. FUTURE EFFORTS
series of tests in waves can allow very good predictions
5.1 More Correlation
for realistic exposures.
The most obvious improvement is continued work to
POWERSEA can be used with caution as a design and develop better and more proven correlation constants.
modification tool, especially for conceptual designs, This work is ongoing by the various authors. There are
particularly when calm water trim and EHP data can be also differences between the forces in calm water and in
developed. Since this data may be available from other waves. This is probably because of the differences in
sources such as standard series data as well as model three dimensional pressure distribution on the panels
tests of the particular design, a wise user could develop themselves and on adjacent panels in waves and calm
useful data at an early stage, provided it is used with water. Understanding the detailed physics of this process
caution. will improve the correlation.
 47-FT MLB Percent Diffe rence Betw ee n POWERSEA and Tank Tes t Res ults
for Se ak ee ping Simulations
POWERSEA
Under Prediction
30.000

25.000
20.000

15.000
Percent Difference

10.000

5.000
0.000
-5.000
- 10.000
- 15.000
- 20.000
20 Kts 30 Kts 15 Kts 20 kts
- 25.000 SS3 SS3 SS5 SS5
0 1 2 3 4 5
POWERSEA
Cas e Num ber
Over Prediction

Sig Pitch Sig. Heave Sig. CG Accel Sig. Bow Accel

Figure 10. 47 MLB Rough Water Comparison with Optimized Coefficients

Hulls in the Vertical Plane”, New England Section,
5.2 Structural Data
SNAME
Sectional load calculation tools are not fully
6. Shuford, C., Jr. (1958) “A Theoretical and
implemented in POWERSEA in the current release.
Experimental Study of Planing Surfaces Including
While one can save a sectional load data file, this feature
Effects of Cross Section and Plan Form”, NACA
is not yet fully implemented for general use. This is a
straightforward improvement that can come with time. Report 1355
7. Ulak, A., (2001), “Evaluation of POWERSEA, A
5.3 Hydrofoils And Catamarans
Software Simulator for Planing Hull Motions,
Hydrofoils have been used as motion control devices, Powering, and Accelerations”, BLA, RPT 681-2
and to augment lift, thereby decreasing resistance, in a 8. Fridsma, G., (1969), “A Systematic study of the
number of different ways. The effect of hydrofoils on
Rough-Water Performance of Planing Boats” Report
lift, drag and motion is available from hydrofoil design
1275, Davidson Laboratory
sources. The resistance and motion of a planing
catamaran in head seas can be simulated by modelling 9. Fridsma, G., (1971), “A Systematic Study of the
half the craft. However, especially high speed Rough Water Performance of Planing Boats
catamarans have relatively close hull spacing, and the (Irregular Waves -- Part II)”, Report 11495,
hulls may interfere with each other and change their Davidson Laboratory
added mass. 10. Akers, R., Hoeckley, S., Peterson, R., and Troesch,
A., (1999), “Predicted vs. Measured Vertical Plane
References
Dynamics of a Planing Boat”, Fast ’99, SNAME
1. Jiminez, J., (1988), Northern California Section,
SNAME
2. Zarnick, E., (1978), “A Nonlinear Mathematical
Model of Motions of a Planing Boat in Regular
Waves”, David W. Taylor Naval Ship Research and
Development Center, DTNSRDC 78/032
3. Zarnick, E. (1979), “A Nonlinear Mathematical
Model of Motions of a Planing Boat in Irregular
Waves”, David W. Taylor Naval Ship Research and
Development Center, DTNSRDC/SPD 0867-01
4. Martin, M., (1976) “Theoretical Predictions of
Motions of High-Speed Planing Boats in Waves”,
David W. Taylor Naval Ship Research and
Development Center, DTNSRDC 76/0069
5. Akers, R., (1999), “Dynamic Analysis of Planing