Lecture Notes for

Engineering 5003 – Ship Structures I

Claude Daley, Professor, D.Sc., P.Eng.
Faculty of Engineering and Applied Science
Memorial University
St. John’s, CANADA
E5003 – Ship Structures I ii
© C.G.Daley

Table of Contents

Topic 1: Introduction to Ship Structures ............................................................................................ 2

Topic 2: Ship Structural Features ................................................................................................... 16
Topic 3: Material Behavior .............................................................................................................. 29
Topic 5: Longitudinal Strength: Murray’s Method............................................................................ 52
Topic 6: Longitudinal Strength: Wave Bending Moments ............................................................... 60
Topic 7: Longitudinal Strength: Inclined Bending / Section Modulus .............................................. 66
Topic 8: Beam Theory .................................................................................................................... 79
Topic 10: Indeterminate Beams – Force Method .......................................................................... 106
Topic 11: Indeterminate Beams – Displacement Method ............................................................. 117
Topic 12: Energy Methods in Structural Analysis.......................................................................... 127
Topic 13: The Moment Distribution Method .................................................................................. 135
Topic 14: The Moment Distribution Method with Sway ................................................................. 149
Topic 15: Matrix Structural Analysis.............................................................................................. 157
Topic 16 Overview of Finite Element Theory ................................................................................ 171
Topic 17: Hull Girder Shear Stresses ........................................................................................... 183
Topic 18: Shear Stresses in multi-cell sections ............................................................................ 194
Topic 19: Shear Flow in adjacent Closed Cells ............................................................................ 208
Topic 20: Torsion in ships ............................................................................................................. 210
Topic 21: Shear Center and Shear Lag in Ship Structures ........................................................... 218
Topic 22: Plate Bending................................................................................................................ 228
Appendix....................................................................................................................................... 242

Note: all images, sketches and photo's are © C. Daley unless otherwise noted

Cover image by CD using Sketchup

E5003 – Ship Structures I 1
© C.G.Daley

PART 1 : Introduction

St. John's Harbour

E5003 – Ship Structures I 2
© C.G.Daley

Topic 1: Introduction to Ship Structures

The course is intended to develop the student’s knowledge of ship structures. The
focus is on various types of intact structural behavior, building upon concepts from
mechanics of materials. The course project will involve the design, assessment,
drawing and reporting on the mid-ship scantlings (hull girder design) of a large
vessel. The follow-on course (6003) will move from the consideration of intact
behavior to the mechanics of structural failure.

Web Frame under Deck

hand drawn sketch

One of the aims of the course is for the students to develop the ability to make an
educated guess. Such guesses are not wild or random. Educated guesses are based
on sound reasoning, careful approximation and simplification of the problem. In
most cases the 'guess' starts by forming an idea of the problem in its essential form,
or in 'bounding' forms. Basic laws of mechanics are considered to determine what
E5003 – Ship Structures I 3
© C.G.Daley

fundamental principle might govern the outcome. Most problems are governed by
simple conservation laws, such as of forces, moments, momentum and/or energy.

A related aim of the project is for the students to develop the ability to sketch the
problem at hand, by hand and clearly. Sketching is a form of symbolic
communication, no less valuable than the alphabet or algebra.

Background
Humans have been constructing structures for a long time. A structure is a tool for
carrying (carrying what is in or on the structure). Ship structures have evolved like
all other types of structures (buildings, aircraft, bridges ...). Design was once purely
a craft. Design is evolving as we understand more about the structure itself and the
environment that we subject it to.

Traditional Design
 built by tradition (prior example)
 changes based primarily on experience (some analysis)
 essentially a builders “Craft”
 QA by proof test and use

Gondolas in Venice

Engineering Design
 incorporates analysis based on math/physics
 common designs are codified (building code, class rules..)
 new designs should follow the “Engineering Method”
E5003 – Ship Structures I 4
© C.G.Daley

early Finnish icebreaker (public domain - Wikipedia)

design, analysis, construction and regulation are separate specialties

design practice is evolving: In the 1950 tabulated requirements were found in Class
Rules. By the 70s all codes had changed to include prescriptive algebra. New trend
are towards "LRFD - load and resistance factored design", "risk based design" and
"goal based design". Current practice in large (novel) projects make extensive use of
"scenario based" design, with HAZIDs (hazard identification and mitigation).

The future of design will be "design by simulation" in which the many interacting
process and systems will be simulated numerically. In some ways this will
represent a return to the idea of proving a design by a "proof test", except it will be
a numerical proof test and will simulate the life of the design.

Purpose of Ship Structures

The structure of a ship or ocean platform has 3 principal functions:
 Strength (resist weight, environmental forces – waves + )
 Stiffness (resist deflections – allow ship/equipment to function)
 Water tight integrity (stay floating)
E5003 – Ship Structures I 5
© C.G.Daley

Warship (public domain - Wikipedia)

Bulk Carrier FLARE (from TSB report)

There are two other important functions

 provide subdivision (tolerance to damage of 1,3 above)
 support payloads
E5003 – Ship Structures I 6
© C.G.Daley

Shipbreaking on the beach at Chittagong

(Naquib Hossain - Wikipedia)

These functions are all interrelated, but should be considered somewhat separately.

Structural Arrangement
The particular arrangement of the structure is done to suit a variety of demands;
 Hull is shaped (reduce resistance, reduce motions, reduce ice forces, increase
ice forces, reduce noise)
 holds are arranged for holding/loading cargo
 holds are arranged for holding/installing engines
 superstructure is arranged for accommodation/navigation
 all structure is arranged for build-ability/maintainability
 all structure is arranged for safety
 all structure is arranged for low cost

Cruise ship Lifeboat

E5003 – Ship Structures I 7
© C.G.Daley

Types of Structural Work

Ship structural specialists are involved in a variety of work;
 Design
 Analysis
 Construction
 Maintenance
 Repair
 Regulation

While almost all Naval Architects get involved in structural issues, as with most
professions, a few focus on the area and tend to be involved in any advanced work.
This course aims to have you develop your ‘feel’ as well as your knowledge of
structures. In other words, you should work at developing you “Engineering
Judgment” in the area of ship structures.

Structural Behavior
Ship structural behavior, as with all structural behavior is essentially very simple.
Structures are an assemblage of parts. This distinguishes them from objects. A
beam or plate is a structural element, but only a collection of structural elements is
called a structure. The theory of structures builds upon the field of ‘mechanics of
materials’ (also called mechanics of solids, or strength of materials), by considering
the interactions and combined behaviors of collections of structural components. So,
much of this course will focus on techniques for understanding collections of
structural elements. We will also review and expand, somewhat, on the mechanics
of individual elements.
E5003 – Ship Structures I 8
© C.G.Daley
E5003 – Ship Structures I 9
© C.G.Daley

Levels of Structure

As a structure, a ship is an assemblage of components. At the largest scale a ship is

a simple beam, carrying weight and supported by buoyancy. The behavior or the
whole ship as a single beam is referred to as the behavior of the primary structure.
E5003 – Ship Structures I 10
© C.G.Daley

The hull girder is referred to as the primary structure. The strength and stiffness of
the hull girder depend on the properties of the cross sections of the ship. The key
section is the midship section.

Within the hull, as integral components of the hull, are large structural components
that are themselves made of individual structural members, and yet act as
individual systems. These are called secondary structures. For example, the whole
double bottom, between bulkheads, is a unit that acts as a sandwich panel,
behaving somewhat like a plate.

Locally a ship is comprised of frames and plate. These are called tertiary structures.
The tertiary structure are individual structural members.

Ships are a class of structures called "semi-monocoque". In a pure monocoque, all

the strength comes from the outer shell ("coque" in french). To contrast, in "skin-on-
frame" construction, the loads are all borne by a structure of framing under the
skin. In ships, the skin is structurally integral with the framing which supports it,
with the skin providing a substantial portion of the overall strength.

Newton's 3rd Law:

action = reaction

All the various parts and levels of a ship structure interact. Ships are "all-welded"
structures, meaning that it is all one single, complex, solid elastic body. The main
thing that structures (and all parts of structures) do is “push back”. i.e. across any
interface (across every patch of every plane, everywhere in the universe, always!)
the force acts in both ways. This powerful idea is the key to understanding what
happens in a structure.
E5003 – Ship Structures I 11
© C.G.Daley

Structural Design

The process of ship structural design varies depending on the specific issues.
Structural design occurs after the mission is set and a general arrangement is
determined. The general arrangement allows us to determine both the
environmental loads and the distribution of hull/outfit/cargo weights. The
establishment of scantlings (structural dimensions) is iterative. We assume that a
preliminary set of dimensions is settled upon from experience or by other choice.
The loads will cause a set of responses (stresses, deflections). The response criteria
are then compared to the responses. For any inadequacies we modify the structural
dimensions and repeat the response analysis. When all responses are satisfactory,
we are finished.

In cases where we wish to satisfy additional constraints (cost, performance..) we add

checks for these items after we have checked the structural response. Again we loop
until we have met the constraints, and reached optimal values for some measure.
E5003 – Ship Structures I 12
© C.G.Daley

As stated above, the structural design can only occur after the overall vessel concept
and arrangement is set, which is done during the preliminary design stage. The
structural design itself is a process that is comparable to the overall design. Just as
the vessels has a mission and a concept to satisfy that mission, so too does the
structure have a mission and concept to satisfy the mission. Prior to deciding on the
structural sizes (scantlings) , the designer must decide on the overall structural
concept and arrangement. In rule based design (Classification Society rules), the
loads and response criteria have been combined into standard scantling
requirements formulae. The user can use these formulae to determine minimum
dimensions for members and components. There can then be the need to check
additional criteria (e.g buckling, alternate loads). When this is complete the user
has a complete structural design, but not yet a final detailed design. The final
structural drawings also include detailed design features (e.g. bracket and weld
specifications). The image at left is taken from a structural drawing of a web frame
in an offshore supply vessel.
E5003 – Ship Structures I 13
© C.G.Daley

adapted for illustration from a design by Rolls Royce Marine

Load Types
We will define four general types of structural loads.
Static Loads (e.g. fixed weights)
Low Frequency Dynamic Loads (e.g. quasi static load, wave loads)
High Frequency Dynamic Loads (e.g. vibrations)
Impact Loads (e.g., blast, collisions)

With both static and quasi-static loads, we do not need to take inertial or rate
effects into account in the structural response. With high frequency loads we need
to consider structural vibrations which includes inertial effects and damping. For
impact loads, we have both transient inertial effects and rate effects in material
behavior. It is important to distinguish between loads affecting vessel rigid body
motions and elastic structural response. Wave forces may cause the vessel as a
whole to respond with inertial effects (heaving motions), but will seldom cause
anything but quasi-static response of the structure. The important determinant is
the relative frequency of the load and response. Local structure will respond
elastically at frequencies in the 100hz to 3000hz range. The hull girder will flex at
around the 1 hz rate. The vessel will heave and roll at around the 0.1 hz range.
(large vessels/structures will respond more slowly).
E5003 – Ship Structures I 14
© C.G.Daley

launch of MEXOIL, by John N. Teunisson, 14 February1918 (wikipedia)

In this course we will examine the structural response to quasi-static loads. The
hull girder is sized to resist the combination of self weights and wave forces.
E5003 – Ship Structures I 15
© C.G.Daley

Topic 1: Problems
1.1 Longitudinal strength is a primary concern during the design of a ship. Describe
the steps in the ship design process (in general terms) that must occur prior to
consideration of the longitudinal strength.

1.2 What is the difference between “low frequency dynamic” and “high frequency
dynamic” loads? Give examples.

1.3 Describe the types of loads that you would be concerned with during the launch
of a vessel on a slipway.

1.4 Loads on ships

The following is a table of load types. Identify each load as static, quasi-static,
dynamic or transient. Place a check mark  to indicate which categories apply to
each load type. If more than one type applies, explain why in the comments column.

static quasi- dynamic transient comments

LOAD static
Dry cargo
Liquid cargo
Engine
Propeller
Ice
Waves
Other: ______
Other:______

1.5 In preliminary design, when can the preliminary structural calculations be

made?

1.6 List 5 purposes of structure in a ship.

1.7 When is a load considered to be quasi-static?

E5003 – Ship Structures I 16
© C.G.Daley

Topic 2: Ship Structural Features

lifeboat on the Battleship Texas

Introduction
In this Chapter we will
Name and describe ships structural components.
Discuss some structural features and challenges for various vessels,
~~~~~~
Boats are made from a variety of materials, including wood, fiberglass,
composites, aluminum, steel and cement. Ships are built mainly from steel. In this
Chapter we will name and discuss the main structural features of steel ships. Ships
are much longer than they are wide or deep. They are built this way in order to
minimize resistance (fuel consumption), and yet maintain adequate stability and
seaworthiness. This geometry results in the ship being a girder (a beam built from
compound parts). The figures below show sketches of the structural details of the
midship section of a bulk carrier.
E5003 – Ship Structures I 17
© C.G.Daley

Figure 1.

This type of vessel is very common, and has many problems. Single side shell
vessels are being replaced with double hull vessels. The FLARE had this type of
construction.
E5003 – Ship Structures I 18
© C.G.Daley

Figure 2. Bow of Bulk Carrier FLARE

(from TSB report)

Figure 3 shows a 3D representation of the same x-section as show in Figure 1.

E5003 – Ship Structures I 19
© C.G.Daley

Figure 3.

Figure 4 Rhino Sketch of section of longitudinally framed double hull Container

vessel.

Ship Structural Photos

Terra Nova FPSO – Floating Production, Storage and Offloading vessel

(from wikipedia)
E5003 – Ship Structures I 20
© C.G.Daley

Terra Nova Hull FramingTerra Nova Structural Connection Details

E5003 – Ship Structures I 21
© C.G.Daley

Terra Nova Stringer with web stiffener bracket

Terra Nova Stringer with web stiffener bracket

E5003 – Ship Structures I 22
© C.G.Daley

Terra Nova Longitudinal angle frames

Transverse flat bar frames through stringer

E5003 – Ship Structures I 23
© C.G.Daley

Terra Nova flat bar frames

E5003 – Ship Structures I 24
© C.G.Daley

Terra Nova Flare tower

E5003 – Ship Structures I 25
© C.G.Daley

Terry Fox – Icebreaker

Bow framing in Terry Fox (photo by R. Frederking)

The Terry Fox is ~7000 tons displacement and capable of ramming thick old ice. It
has never been damaged.
E5003 – Ship Structures I 26
© C.G.Daley

Bow of Supply Boat

Reduta Ordona (Photo credit: Andrew Kendrick).

E5003 – Ship Structures I 27
© C.G.Daley

Local ice damage CPF superstructure plating

E5003 – Ship Structures I 28
© C.G.Daley

Topic 2: Problems

2.1 Read the SSC Case Study V and name all the parts of the Rhino sketch shown
below.

2.2 What was the basic cause of the “Recurring Failure of Side Longitudinal” in the
SSC report?

2.3 Sketch a X-section of a ship at mid-ships and label all features/elements.

2.4 Sketch, free hand, the structure in the double bottom of a ship. Keep it neat and
label the elements
E5003 – Ship Structures I 29
© C.G.Daley

Topic 3: Material Behavior

plastic frame response to ice load test

Introduction
In this Chapter we will
outline the material behavior models that are necessary to the analysis of
structures.

Hooke's Law
Hooke's law is a very simple idea. It just states that there is a linear relationship
between force and deflection in an elastic body;

where k is the 'spring constant' or the 'stiffness'

For a uni-axial state of stress we can also write Hooke's law in terms of stress
( normalized force) and strain ( non-dimensional deflection);

where E is Young's Modulus.

E5003 – Ship Structures I 30
© C.G.Daley

This law may seem to be too simple to deserve the term 'law'. However, this idea
was not easily found. The world, especially in the time of Hooke and before, was so
full of variability, inaccuracy and non-linearity that this idea was not obvious. Many
things were made from natural materials (stone and wood) and the idea of linear
behavior was radical. Hooke was a contemporary, and rival, of Newton. He
developed a coil spring for use in a pocket watch. In 1678 he published a discussion
of the behavior of his spring, saying: "ut tensio, sic vis" meaning "as the extension,
so the force". Hooke worked in many fields (architecture, astronomy, human
memory, microscopy, palaeontology), but it is only in mechanics that his name is
associated with a fundamental law.

How important is Hooke's contribution? For structural analysis it is the

fundamental idea, as important to structural analysis as is Newton's 2nd law
to the field of dynamics.
E5003 – Ship Structures I 31
© C.G.Daley

Hooke's law is important because linearity of behavior permits the use of

superposition. And only with the idea of superposition can we divide problems up
into parts, solve the parts and add them back together to get a total solution. The
whole field of structural analysis depends on Hooke's law.

Hooke's law can be expanded to describe 2D and 3D behavior. Consider a 2D sample

of elastic material. When a force is applied in one direction (x) the material
stretches in that direction and contracts in the lateral direction(y). So for a stress in
the x direction we get strains in x and y. This is Hooke's law in 2D for the case of
uni-axial stress;
E5003 – Ship Structures I 32
© C.G.Daley

When we consider a general state of stress, comprised of a combination of x and y

direction stresses , as well as shear stress we can write the relationship
amoung the stresses and strains Hooke's law in 2D for the general case;

or equivalently;

The above equations are used to describe isotropic materials (materials that are
similar in all directions, such as steel), which have the same value of E and n in all
directions.
Note: Anisotropic materials, such as wood and fiberglass have different values of E
for each axis. Hooke's laws for anisotropic materials have many more terms.

Hooke's law can be expressed in 3D as well, but 2D is sufficient for the problems
that we will examine.
E5003 – Ship Structures I 33
© C.G.Daley

Consider a small element of material with normal and shear stresses on vertical
and horizontal planes. We refer to these stresses as engineering stresses,
Now consider what the stresses would be on any other plane, so one that is rotated
by the angle from the vertical (from the plane for stress). Mohr showed that
the stresses on all planes, when plotted, will form a circle in vs. coordinates.

The stresses on the vertical plane, and , are plotted on the Mohr's circle (point
A). The stresses on the horizontal plane, and , are plotted at point B. These
two planes are physically 90 degrees from each other, but are 180 degrees apart on
the Mohr’s circle.

state of stress in 2D

The line joining A, B is a baseline. To find the stresses on a cut plane at angle 
from the vertical plane (the plane of A), we must move 2 from the 'A' direction
around the Mohr’s circle. This lands us at point C, where the stresses are , and
.
E5003 – Ship Structures I 34
© C.G.Daley

The general equations to find the stresses on a plane at angle from the plane of
are;

principal stresses

You can see from the drawing of Mohr's circle, that the largest value of occurs
where is zero. The largest and smallest values of are called and . They are
sufficient to define the circle, and are called the principal stresses.
E5003 – Ship Structures I 35
© C.G.Daley

We do not need to solve for 1 and 2 graphically. We can use the following
equations:

or

large strain behaviors

E5003 – Ship Structures I 36
© C.G.Daley

At low strains steel is a linear elastic material. However, when steel is strained to
large levels, the linear behaviour ends. Typical ship steels will follow a stress-strain
curve as shown at the left. After yielding the stress plateaus while the strains
increase significantly. At larger strains the stress begins to rise again, in a
phenomenon called 'strain hardening'. At even larger strains the material starts to
'neck' and eventually ruptures. Typical yield stresses are in the range 225 to 400
MPa. Typical ultimate stresses are in the 350 to 550 MPa range.
The initial slope is the Young's modulus which is about 200,000 MPa (200 GPa). So
the strain at yield is about 1200 to 2000 x10-6 strain (-strain). Rupture occurs at
around 25% strain (300,000 -strain).

yield criteria and equivalent stresses

E5003 – Ship Structures I 37
© C.G.Daley

In ships structures, made almost entirely of plate steel, most stress states are
essentially biaxial. In this case we need to have a criteria for any 2D state of stress.

The 2D von Mises criteria is plotted at left. The curve is normally represented in
terms of principal stresses and forms an oval. The oval crosses the axes ay the
uniaxial yield stress . The equation for the yield condition is;

The criteria can also be expresses in terms of engineering stresses;

To show whether a general 2D stress is at yield, the concept of an equivalent stress

is used (the von-mises equivalent stress). The equivalent stress is a uniaxial stress
that represents the same % of yield as the biaxial stress. In this way any 2 states of
stress can be compared. The equivalent stress is;

or

We will make use of equivalent stresses in the ANSYS labs.

E5003 – Ship Structures I 38
© C.G.Daley

Topic 3: Problems
3.1 A column is made of steel pipe with OD of 8", and ID of 7". It is 8 feet tall. The
column supports a weight of 300kips (300,000 lb). How much does the column
shorten under load?
(E for steel is 29,000,000 psi) (Ans: 0.843”)

3.2 A 2D state of stress is (200, -20, 45) MPa. What are the strains
? (Ans: 1030,-400,585 e)

3.3 For a 2D state of stress of (180, -25, 40) MPa, plot the Mohr's circle.
What are the principal stresses ? (Ans: 187.5,-32.5 MPa)

3.4 For a 2D state of stress of (100, -100, 60) MPa, what is the von-mises
equivalent stresses ? (Ans: 202 MPa)

3.5 For a 2D state of stress of (150, 100, 30) MPa, what is the von-mises
equivalent stresses ? (Ans: 142 MPa)
E5003 – Ship Structures I 39
© C.G.Daley

3.6 For a small cube of material with what is the maximum

shear on any plane? (Ans: 50 MPa)
E5003 – Ship Structures I 40
© C.G.Daley

PART 2 : Longitudinal Strength

St. John's Harbour

E5003 – Ship Structures I 41
© C.G.Daley

Topic 4: Longitudinal Strength: Buoyancy & Weight

Pompei
Introduction
In this Chapter we will
Discuss Still water bending moments, bonjean curves, Prohaska’s method and a
similar method for non-parallel midbodys
~~~~~~~~~~~~~~~~~~~~
Overview

Structural design starts from:

Principal Dimensions - L,B,T

Hull Form - CB, CWP, CM
General Arrangement – decks and bulkheads

Which is called preliminary design:

E5003 – Ship Structures I 42
© C.G.Daley

The first strength consideration is the longitudinal strength of the hull girder. The
hull girder feels vertical forces due to weight and buoyancy. For any floating body
the total weight must equal the total buoyancy, and both forces must act along the
same line of action. However, at each location along the ship, the weight will not
normally equal the buoyancy.

The weights are set by the combination of lightship and cargo weights. The locations
of the weights are fixed (more or less). The buoyancy forces are determined by the
shape of the hull and the location of the vessel in the water (draft and trim). The net
buoyancy will adjust itself until is exactly counteracts the net weight force. However,
this does not mean that each part of the vessel has a balance of weight and
buoyancy. Local segments of the vessel may have more or less weight than the local
buoyancy. The difference will be made up by a transfer of shear forces along the
vessel.
E5003 – Ship Structures I 43
© C.G.Daley

Bending Moment Calculations

The ‘design’ bending moment is the combination of Stillwater bending and wave
bending. To calculate these values we will make the following assumptions;
Ship is a beam
Small deflection theory
Response is quasi-static
Lateral loading can be superimposed

~~~~~~~~

Still Water Bending Moment (SWBM)

The still water bending moment is calculated from the effect of the weights and
buoyancy in calm water. The buoyancy force is a line load (e.g. kN/m). The local
buoyancy per meter is found from the x-sectional area of the hull at each location.
The x-sectional area depends on the local draft and are found from the ‘bonjean’
curves.
E5003 – Ship Structures I 44
© C.G.Daley

Bonjean Curves – Calculating the Buoyancy Distribution

Bonjean curves show the relationship between local draft and submerged cross-
sectional area. There is one bonjean curve for each station. There are typically 21
stations from the FP to the AP, with 0 being the FP. This divides the Lbp into 20
segments.

At each station we can draw a bonjean curve of the x-section area;

Bonjeans are drawn on the profile of the vessel. With these curves, we can find the
distribution of buoyancy for any waterline (any draft, any trim).

For hydrostatic calculations we need to know the distribution of buoyancy along the
ship. We need to be able to find this for every possible draft/trim. If we had a wall
sided vessel, it would be relatively easy to solve for the draft/trim (as in Assignment
#1). With shaped hulls, there is a non-linear relationship between buoyancy and
position. We use bonjean curves to find the buoyancies as follows.
E5003 – Ship Structures I 45
© C.G.Daley

For the typical 21 station ship, we divide the ship into 21 slices, each extending fore
and aft of its station. Using the bonjean curve for each station we calculate the total
displacement at our draft/trim;

For example, the displacement for station 3 is;

L BP
3  A 3  [m3]
20
The buoyant line load for station 3 is;

 3  3    g [N/m]

(assuming that area is in m2, g=9.81 m/s2 and  = 1025 kg/m3)

The above will provide a way of calculating the buoyant forces at each station. We
will now discuss the weights.

Calculating the Weight Distribution

E5003 – Ship Structures I 46
© C.G.Daley

We will discuss three methods for determining weighs.

If the weight distribution is known (even preliminarily), we use them directly. The
steps to follow are;
Calculate the weight at each station (+- half station)
(optionally) find the c.g. of weights for each segment
(optionally) place the weights at the c.g.

~~~~~~~
If the weight distribution is unknown and we need to estimate the distribution, we
can use the Prohaska method. Prohaska proposed a method for a ship with parallel
middle body (i.e. most cargo vessels). The weight distribution is a trapezoid on top of
a uniform distribution, as follows;

The weights are distributed according to the pattern above. With the average
Whull
weight/meter of the hull : W  the values of a and b are ;
L

a b
W W
Tankers .75 1.125
Full Cargo Ships .55 1.225
Fine Cargo Ships .45 1.275
Large Passenger Ships .30 1.35
Note that the values of a and b are related, so that the average is W . This gives
b a
 1.5  .
W 2W
To move the position of the center of weight (the lcg) the fore and aft ends of the load
diagram are adjusted by equal (and opposite) amounts.
E5003 – Ship Structures I 47
© C.G.Daley

7 lcg 54
lcg  x  L2 or, x
54 L2 7
~~~~~~
If the weight distribution is unknown and we have a vessel without a parallel middle
body (i.e. most sail yachts), we need a smoother distribution. The method below uses
a parabolic distribution on top of a uniform distribution. The two parts each have
half the weight.

The equation for the weight is;

W 3 2x
W  W (1  (  1) 2 )
2 4 L

To shift the total center of weight by ‘x’ we shift the c.g. of the parabola by 2x. This is
done by ‘shearing’ the curve, so that the top center, ‘D’, shifts by 5x. All other points
shift proportionally.
E5003 – Ship Structures I 48
© C.G.Daley

Topic 4: Problems

4.1. For the three station profiles shown below, draw the bonjean curves in the space
provided.

4.2. For a vessel with 4 stations, the bonjean curves are given at the 3 half stations. Lbp is
60m.
for the vessel to float level (no trim), at a 4.5 m draft, where is the C.G.? (Ans: )
What would the Prohaska distribution of weight be to achieve this? (plot)
If the C.G is at midships, and the draft (at midships) is 4.5 m, what is the trim?
(Ans: )
E5003 – Ship Structures I 49
© C.G.Daley

4.3. For the vessel body plan shown below (left), sketch the corresponding bonjean curves (on
the right).

4.4. For the bonjean shown below (right), sketch the corresponding vessel body plan curve
(on the left).
E5003 – Ship Structures I 50
© C.G.Daley

4.5. Bonjean Curves The following figure shows 5 potential Bonjean curves. Some of them
are impossible. Identify the curves that can not be Bonjean curves and explain why. For the
feasible Bonjeans, sketch the x-section that the Bonjean describes.

4.6. For the two ship stations shown below, sketch the corresponding bonjean curves on the
grid below.

20 m 20 m

12 m

145 m2
(a) (b)

12
10
z [m] 8
6
4
2
0
0 50 100 150 200
Area [m2]
E5003 – Ship Structures I 51
© C.G.Daley

4.7. You are supervising a preliminary ship design project. You have asked one of your team
to produce a net load (weight-buoyancy) diagram, so that bending moments can be
calculated. The diagram you are given is ;

Net Load Curve

20

-20
AP ¼ ¾ FP

why is this diagram impossible? Justify your answer. (hint: use SFD and/or BMD)

4.8. For the three station profiles shown below, sketch the corresponding bonjean curves
E5003 – Ship Structures I 52
© C.G.Daley

Topic 5: Longitudinal Strength: Murray’s Method

Battleship TEXAS
Introduction
In this chapter we will
Discuss Murray’s Method to estimate still water bending moments
~~~~~~~~~~~~~~~~~~~~
Murray’s Method

Murray’s method is based on the idea that forces and moments in a ship are self-
balancing (no net force or moment is transferred to the world). Any set of weight and
buoyancy forces are in balance.

Also, for any cut at x, the moment at the cut can be determined in two ways;
BM ( x)  y1 L1  y 2 L2

 y5 L5  y3 L3  y 4 L4

Murray applied this idea to a ship:

E5003 – Ship Structures I 53
© C.G.Daley

where
ff,fa are the distances from the to the centers of weight (fore and aft)
gf,ga are the distances from the to the centers of buoyancy (fore and aft)

The bending moment at midships is;

 Wa f a   a g a
or
Wf f f   f g f

These are two ‘estimates’ of the maximum bending moment. We can combine the
two, and increase our accuracy, by taking the average of the two;


1
Wa f a  W f f f   1  a g a   f g f 
2 2
 BM W  BM B
weight - buoyancy
To find the buoyancy part, Murray suggested

BM B 
1
 a g a   f g f   1   x
2 2
where x = average moment arm

Murray suggested a set of values for x , as a function of the ship length, block
coefficient and the ratio of draft to length;

x  L(a  CB  b)
where
E5003 – Ship Structures I 54
© C.G.Daley

T/L a b
.03 .209 .03
.04 .199 .041
.05 .189 .052
.06 .179 .063

This table for a and b can be represented adequately by the equation;

a  .239  T / L
b  .1.1T / L  .003

Example using Murray’s Method

Ship: Tanker L=278m, B=37m, CB=0.8

Assume wave bending moment is;

WBMsag = 583800 t-m
WBMhog = 520440 t-m

The vessel weights, and weight bending moments are as follows;

I TEM Weigh t lcg Mom en t
( t) (m) ( t -m )
Fwd
ca r go 62000 40 2480000
fu el & wa t er 590 116 68440
s t eel 12000 5 5 .6 667200
3 ,2 1 5 ,6 4 0
Aft
ca r go 49800 37 1842600
m a ch in er y 3400 125 425000
ou t fit 900 120 108000
s t eel 12000 5 5 .6 667200
 140690 t 3 ,0 4 2 ,8 0 0

BM w = 3 ,1 2 9 ,2 2 0

To find the buoyancy moment we need the draft;

W    CB  L  B  T  
E5003 – Ship Structures I 55
© C.G.Daley

 140690
T 
C B  L  B   0.8  278  37 1.025
 16.68 m

T 16.68
  0.06
L 278

Murray’s table gives;

a=0.179, b=0.063

x  278(.179  0.8  .063)  57.32 m

1
BM B   x
2
1
 140690  57.32  4,032,428 t-m
2

SWBM = BMW-BMB
hog sag
 3,129,220  4,032,428
 903,145 t-m (- is sag)

we need to add the wave bending moment in sag

Total BM = 903,145 + 583,800

= 1,486,945 t-m (sag)

Note that in this case the ship will never get in the hogging condition, because the
SWBM is so large.
E5003 – Ship Structures I 56
© C.G.Daley

Topic 5: Problems

5.1. Longitudinal strength is a primary concern during the design of a ship. Briefly explain the
idea behind Murray’s Method.

5.2. There is a ‘rectangular’ shaped block of wood, as shown in the image below. The block
weighs 200 N and has uniform density. It is 1 m long and 0.20 m wide. It is 20 cm thick and
is floating in fresh water.

draw the shear force and bending moment diagrams for the block.

Now consider the addition of a small 50 N weight on the top of the block, at a
distance 2/3m from one end. (hint - a right triangle has its centroid at 2/3 of its
length)

After the block settles to an equilibrium position -

Draw the bending moment and shear force diagrams
What is the max. bending stress on the transverse plane at the middle of the block (ie at 0.5 m from the
end)?

5.3. There is a ‘diamond’ shaped block of wood, as shown in the image below. The block
weighs 5.4 kg. and has uniform density. It is 60 cm long and 30 cm wide. It is 12 cm thick
and is floating in fresh water. Resting on the block are 2 weights, each small blocks of steel
weighing 1 kg. They are symmetrically placed and are 55cm apart.
What is the midship bending moment in units of N-cm ?
What is the maximum bending stress in the wooden block?
E5003 – Ship Structures I 57
© C.G.Daley

Draw the bonjean curve for a cross section of the wooden block at a point 15cm from the end. (show actual
units).
What is the block coefficient for the block?

ANS: a) 171.5 N-cm (hog) b) 23.8 MPa c) Straight and then vertical d) 0.5

5.4. Consider a 100m vessel resting in sheltered fresh waters (see below). The CG of all
weights fwd of midships is 23m fwd of midships (ff=23m). The CG of all weights aft of
midships is 25m aft of midships (fa=25m). The weights fwd and aft are 4200 and 4600 t
respectively. Two bonjean curves are given. Assume each refers to the average x-section area
for 50m of ship (fore and aft). The (fore and aft) buoyancy forces act at the bonjean
locations, which are 18m fwd and 20 aft (of midships). The buoyancy force aft is 4650 t.

Using the bonjeans, find

The vessel drafts at the two bonjeans.
The buoyancy force fwd.
The still-water bending moment at midships
E5003 – Ship Structures I 58
© C.G.Daley

5.5. Murray's Method Consider a 100m long vessel resting in sheltered waters. The CG of
all weights fwd of midships is 20m fwd of midships (ff=20m). The CG of all weights aft of
midships is 25m aft of midships (fa=25m).
- Describe how you would use Murray’s Method to determine the still water bending
moment for this vessel.
- What other info, if any do you need?
Note: you don’t need to remember the specific values for terms suggested by
Murray.

5.6. Hull girder strength The hull girder can be viewed as a beam. When floating in still
water, is the beam statically determinate or statically indeterminate? Provide reasons for your
answer.

5.7. You see below a sketch of a ship that is 200 m long. The displacement is made up of the
lightship plus the weight of cargo in two holds. The ship has stranded itself on a submerged
rock. Draw the various curves of load and response for the vessel (weight, buoyancy, net
load, shear, moment, slope and deflection) that are compatible with the information given.
The numerical values don’t matter. The intention is to draw a set of curves that are logical for
the ship as shown.
5.8.

5.9. You see below a sketch of a ship that is 200 m long. The displacement is made up of the
lightship plus the weight of cargo in two holds. The forward cargo hold is empty. Draw the
various curves of load and response for the vessel (weight, buoyancy, net load, shear,
moment, slope and deflection) that are compatible with the information given. The numerical
values don’t matter. The intention is to draw a set of curves that are logical for the ship as
shown.
5.10.

5.11. You see below a sketch of a ship that is 200 m long. The displacement is made up of the
lightship plus the weight of ballast in 4 tanks. The cargo holds are empty. Draw the various
curves of load and response for the vessel (weight, buoyancy, net load, shear, moment, slope
and deflection) that are compatible with the information given. The numerical values don’t
matter. The intention is to draw a set of curves that are logical for the ship as shown.
E5003 – Ship Structures I 59
© C.G.Daley

5.12. Calculate the still water bending moment (in N-cm) for the solid block of plastic sketched
below. Assume the block has density as given and is floating in fresh water (density also
given). Is the moment hogging or sagging?

5.13. For the example of Murray’s method in the Chapter, remove the cargo weight and add
4000 t of ballast, with a cg of 116m fwd of midship. Re-calculate the maximum sag and hog
moments (both still water and wave).

5.14. For the example of Murray’s method in the Chapter, instead of using the weight locations
as given, assume that the weights are distributed according to Prohaska. Re-calculate the
SWBM.
5.15. Consider a 100m long tanker resting on an even keel (same draft fore and aft) in sheltered
waters. The CG of all weights is at midships and is 8000 tonnes.
Use Murray’s Method and Prohaska’s values to determine the still water bending
moment for this vessel (i.e. get both the weight and buoyancy BMs about midships).
E5003 – Ship Structures I 60
© C.G.Daley

Topic 6: Longitudinal Strength: Wave Bending Moments

Cape Spear
Introduction
In this Chapter we will
Discuss the shape of ocean design waves
The moments caused by waves
~~~~~~~~~~~~~~~~~~~~
Design Waves

Design wave forces are considered to be quasi-static. As a wave passes by a vessel,

the worst hogging moment will occur when the midbody is on the crest of a wave and
the bow and stern are in the troughs. The worst sagging moment will happen when
the bow and stern are on two crests, with the midbody in the trough between.

Whether for sagging or hogging, the worst condition will occur when the wavelength
is close to the vessel length. If the waves are much shorter,
E5003 – Ship Structures I 61
© C.G.Daley

or much longer than the vessel, the bending moments will be less than if the
wavelength equals the ship length.

Consequently, the design wave for any vessel will have a wavelength equal to the
vessel length. The wave height is also constrained. Waves will have a limited height
to length ratio, or they will break. This results in a standard design wave of L/20. In
other words the wave height (peak to trough) is 1/20th of the wave length.

Trochoidal Wave Profile

Note that the waves sketched above did not look like sinusoids. Waves at sea tend to
be trochoidal shaped, rather than simple sine waves. This has the feature that the
crests are steeper and the troughs are more rounded.

A trochoidal wave is constructed using a rolling wheel.

In the case of the design wave;

LW = LBP } for now we assume that this
HW = LBP/20 length and height or wave is
possible
We can see that;

LW = 2  R
HW = 2 r

Which gives;
E5003 – Ship Structures I 62
© C.G.Daley

L BP L
R  , r  BP
2 40

r 

R 20

To construct a plot of the wave, we start with a coordinate system at the crest of the
wave.

x  R   r sin  }  = rolling angle

z  r (1  cos  )

This is a parametric equation (  is a parameter). We can write;

L L
x   sin 
2 40
L
z  ( 1  cos  )
40
To plot the wave, it is a simple matter to calculate x and z as a function of  and then
plot z vs x. This is done in the spreadsheet below.

L 100
H 5

 x z
0 0 0
2
10 2.343657 -0.03798
20 4.700505 -0.15077
0
z

30 7.083333 -0.33494
-2
40 9.504142 -0.58489 -4
50 11.97378 -0.89303 -6
60 14.5016 -1.25 0 50 100 150 200
x
70 17.09521 -1.64495
80 19.7602 -2.06588
90 22.5 -2.5
100 25.31576 -2.93412
110 28.20632 -3.35505
120 31.16827 -3.75

1.1 L Wave
E5003 – Ship Structures I 63
© C.G.Daley

L/20 waves have been found to be too conservative for large vessels, esp. for vessels
>500 ft. A more modern version of the 1 . 1 L wave. In this case;

as before, LW = LBP

H w  1 . 1 L BP (in feet)
or
H w  0 . 607 L BP (in meters)

For trochoidal waves this gives;

L BP
R  , r  . 55 L BP (feet) or r  . 303 L BP (meters)
2

Calculating Wave Bending Moments

We can now calculate the wave bending moments by placing the ship on the design
wave. We can use the bonjean curves to determine the buoyancy forces due to the
quasi-static effects of the wave;

The steps to determine the wave bending moment are;

Obtain bonjeans
at each station determine the still water buoyancy forces, using the design draft.
Fisw = Aisw li g
at each station determine the total buoyancy forces, using the local draft in that
portion of the wave. Fiwt = Aiwt li g
The net wave buoyancy forces are the difference between wave and still water.
Fiwave=Fiwt-Fisw
E5003 – Ship Structures I 64
© C.G.Daley

This gives us a set of station buoyancy forces due to the wave (net of still water).
These forces should be in equilibrium (no net vertical force). We can calculate the
moment at midships from either the net effect of all forces forward, or all forces aft
(the two moments will balance).

There are other ways to do this kind of calculation. 3D cad programs such as Rhino
can be used to find the still water and wave bending moments. Assuming that we
have a hull modeled in Rhino, we can find the still water buoyancy forces for the fore
and aft halves of the vessel by finding the volume and location of the centroids of the
two submerged volumes.
The procedure would be as follows;

Produce solid model of hull

Cut the model at both the centerline and waterlines.
Find the volumes and centroids of the two halves.
Calculate the buoyant moments about midships.

A similar procedure would determine the wave values. The only difference would be
the need to draw the trochoidal wave as a surface.

The example below shows use of Rhino to calculate the Bouyant BM for a large
vessel. The centroids of the two half volumes are shown.

BMB = 109,000 x 1.025 x 53.97 (m3 x t/m3 x m = t-m)

= 6,029,798 t-m
or
BMB = 123,000 x 1.025 x 58.58 (m3 x t/m3 x m = t-m)
= 7,385,473 t-m

average: BMB = 6,707,376 t-m (sag)

E5003 – Ship Structures I 65
© C.G.Daley

The difference between this and the weight moment (hog) will give the SWBM.

Rhino model, showing slices and centroids

Topic 6: Problems

6.1. Using a spreadsheet, plot the design trochoidal wave for a 250m vessel, for the L/20
wave.

6.2. Using a spreadsheet, plot the design trochoidal wave for a 250m vessel, for the 1.1 L.5
wave.
E5003 – Ship Structures I 66
© C.G.Daley

Topic 7: Longitudinal Strength: Inclined Bending / Section

Modulus

a breaking wave in Lisbon

Introduction
In this Chapter we will
Discuss the calculation of bending of an inclined vessel
General calculation of hull section modulus/inertia
~~~~~~~~~~~~~~~~~~~~
Inclined and Lateral Bending

When a ship rolls the weight and buoyancy forces cause lateral as well as vertical
bending. Normally the bending moment vector is aligned with the ship’s y axis. My is
the bending moment that results from buoyancy and weight forces.

When the vessels rolls by an angle q, the moment vector remains horizontal. This is
because the buoyancy and gravity forces are always vertical. This means that the
bending moment is no longer aligned with the y,z axis of the vessel;
E5003 – Ship Structures I 67
© C.G.Daley

Moments are vectors, adding in the same way that force vectors do.

M z  M w sin  lateral bending

M y  M w cos  vertical bending

Stresses in the Vessel

Both My and Mz cause bending stresses in the x (along ship) direction.

M yz Mzy
V  H  
I NA I CL

Note: Sign convention: R.H.R., moment acting on +x cut face, compression is

positive.

In this case a +My causes tension (-) on the +z part of the vessel. A +Mz causes
compression (+) on the +y side of the vessel.

The total axial stress at any point on the vessel is the sum of the stresses caused by
the two directions of bending.

M yz Mzy
 X  V   H  
I NA I CL
 M w z cos  M w y sin 
 
I NA I CL
E5003 – Ship Structures I 68
© C.G.Daley

When we have bending moments in both y and z, there will be a line of zero axial
stress that we call the heeled neutral axis. This is not necessarily aligned with the
total moment. To find the heeled neutral axis we solve for the location of zero stress;

 M w z cos  M w y sin 
X  0   ,
I NA I CL

solving for z in terms of y , we get;

I
z  NA tan   y ,
I CL
I
where we define: tan   NA tan 
I CL
z  tan   y

 is the angle of the heeled neutral axis from the y axis;

E5003 – Ship Structures I 69
© C.G.Daley

Peak Stresses

The highest stresses will occur @ y=B/2 , Z=Zdeck

There are 2 section modulus values;

I NA I
Z NA  , Z CL  CL
z deck B 2

So that we can write;

 cos  sin  
 max  M w   
 Z NA Z CL 

This leads to the question: What is the worst angle of heel (cr)?

To find it we use;

d  max   sin cr cos cr 

 0  Mw    ,
d  Z NA Z CL 

which gives;
Z
tan cr  NA
Z CL

Typically Z NA Z CL  0 . 5 so cr  26 . 6 

Mw   sin 26 . 6 cos 26 . 6 
For example, if    0  then   26 . 6  M w   
Z NA  Z NA 2  Z NA 
M
 w 1 . 12
Z NA

i.e. for this vessel, there is a 12% increase in stress during the worst roll.
E5003 – Ship Structures I 70
© C.G.Daley

Section Modulus Calculations

Ships are largely built of plates. This means that the moment or inertia and section
modulus calculations normally involve a collection of rectangular parts. For any
individual plate:

Ina = 1/12 b t3

= 1/12 a t2

Ina = 1/12 t b3
= 1/12 a b2

Ina = 1/12 a d2
= 1/12 t b3 cos2

~~~~~~
For compound sections we need to be able to find the inertia about other axes. We
use the transfer of axis theorem:
E5003 – Ship Structures I 71
© C.G.Daley

Izz = Ina + a c2

The overall neutral axis (NA) is found by equating 2 expressions for the 1st moment
of area;
A hNA =  ai hi

The total area A is just the sum of areas.

A =  ai

This gives;
hNA =  ai hi /  ai = (a1 h1 + a2 h2)/(a1+a2)
The overall NA goes through the centroid of the compound area.

Moment of Inertia Calculation

Izz =  ai hi2 +  Inai

INA = Izz - A hNA2

or
INA = Inai + ai (hi - hNA)2)
A simple spreadsheet, as shown below, can be used to find the moment of inertia of a
ship;
E5003 – Ship Structures I 72
© C.G.Daley

See Assignment #2 for an application.

Section Modulus for Material Combinations

(e.g. Steel Hull, Al Superstructure)

Consider a section with 2 materials

When the section bends the sections remain plane, meaning that the strain field is
linear.
E5003 – Ship Structures I 73
© C.G.Daley

To determine the stress/strain/deflection relationships, we convert the x-section to

an equivalent section. The idea is to modify the section so that it is all made of one
material, but retains the distribution of axial force (and bending stiffness). We do
this by adjusting the width of one of the materials, in accordance with the ratio of
Young’s Modulus. For example, Aluminum is converted to steel, but made thinner by
Eal/Est.

For the modified section, ITR is calculated in the usual way. The strains and
deflections for any vertical bending moment will be correct.

M
i.e. v  
EI TR

The only error will be the stresses in the transformed region. The stresses in the
unmodified region will be correct, but the modified region will be wrong by the ratio
of modulii. We can correct this as follows;

My My E My
   1  and  2  2
I TR I TR E 1 I TR
E5003 – Ship Structures I 74
© C.G.Daley

Topic 7: Problems

7.1. Find the moment of inertia of this compound section:

1.1.

dimensions in mm
7.2. A box steel hull is 4m x 1m with a shell thickness of 10mm. It is inclined at 15 degrees,
and subject to a vertical bending moment of 2 MN-m. Find the bending stress at the emerged
deck edge.
1.2.

7.3. For a composite beam (Steel plate with Aluminum web/Flange) loaded as show below
a) find the central deflection.
b) find the maximum stress in the Aluminum

7.4. Consider a compound steel-aluminum beam, shown below. Calculate the deflection d
(show steps)
E5003 – Ship Structures I 75
© C.G.Daley

Ans: 0.112m
E5003 – Ship Structures I 76
© C.G.Daley
E5003 – Ship Structures I 77
© C.G.Daley

PART 3: Beams and Indeterminate structures

Sintra Tile Mosaic

E5003 – Ship Structures I 78
© C.G.Daley
E5003 – Ship Structures I 79
© C.G.Daley

Topic 8: Beam Theory

Test Grillage at Memorial University

Introduction
In this Chapter we will
Develop the elastic behavior of beams
Show the relationship among load, shear, bending, slope and deflection
~~~~~~~~~~~~~~~~~~~~
Coordinate System and Sign Convention

The standard coordinate system has the x axis

along the neural axis of the beam. The positive y
axis is pointed up. The sign convention for force
and moment vectors follows the right hand rule;

+ Forces and deflections follow the axes.

+ Moments and rotations follow the curl of the
fingers (on the right hand) when the thumb is
pointing along the axis.

Shear strain:

Bending moment:
E5003 – Ship Structures I 80
© C.G.Daley

To determine the equations for beam bending we

take a small section of the beam (which represents
any part) as a free body. We look at all the forces
and moments on the section and assuming that
the net force and net moment are zero (Newton!)
we derive the equations.

At this point we haven’t specified P,Q or M. They

can have any values. We will examine equilibrium
in SI units: conditions and see how these result in
P : N/m relationships among P,Q,M.
Q:N
M : Nm We start by summing vertical forces, which must
dx: m sum to zero for equilibrium;

[N]
which is simplified to;

and rearranged to give;

This is a differential equation that states that the

line load on a beam is equal to the rate of change
(slope) of the shear force. Next we sum moments
about the right hand end, which must also sum to
zero to show equilibrium of the free body.

which is simplified to;

note that dx is not just small, it is vanishingly

small, so that dx2 is vanishingly small by
comparison (i.o.w. we can remove the second order
terms, in this case with no loss of accuracy).
Therefore;

or;
E5003 – Ship Structures I 81
© C.G.Daley

This is our second (related) differential equation,

which states that the shear in a beam is the rate of
change (slope) of the bending moment.

We now have two differential equations;

and

We can re-express these relationships as integral

equations. The shear is;

In the form of a definite integral with a constant of

integration the shear is;

In words, this equation means: shear is the sum of

all loads from the start to x. Similarly, the
moment is;

which becomes;

Aside: The shear difference between any two

points on a beam will be exactly equal to the load
applied to the beam between these two points, for
any pattern of load. This leads to a very easy and
accurate way to measure force;

This principle has been used to design load cells,

and to instrument ship frames to measure contact
loads from ice or slamming.
E5003 – Ship Structures I 82
© C.G.Daley

Adding Deformations

So far we have differential equations for

load/shear/bending relationships. Now we add
deformations.

The shear force and bending moments are causing

stresses and strains in the beam. We make the
assumption that we can ignore the shear
deformations (this is part of what we call simple
beam theory), so that only the bending moments
cause distortions. This means that only consider
the shortening of the compression side of the beam
and the lengthening of the tension side. When this
happens, the beam deforms from being straight to
being a curve. The curve shape for any short
length is an arc of a circle, with a radius R. The
local radius, as we can show, turns out to depend
only on the local bending moment. The figure
below show a short length of a bending beam. The
curved shape is also presented in differential form,
meaning essential or limit shape for a very small
value of dx.

The neutral axis (NA) does not stretch or contract.

The upper and lower parts of the beam compress
E5003 – Ship Structures I 83
© C.G.Daley

and/or stretch. We can use the two ‘known’

relationships, the stress-moment equation;

and 1D Hooke's law;

For the top fiber (in the figure above) we see that
the strain is;

from the above we have;

which can be re-arranged to give;

or

We also have

Where is the 'radius of curvature' and is called

the 'curvature' (note the odd naming).

Note also that d is both the change in relative

angle of two cross sections separated by dx and
also the change in slope between two points
separated by dx along the beam. (x) is the slope of
the beam.
E5003 – Ship Structures I 84
© C.G.Daley

This gives us;

For prismatic sections, EI is constant, so;

Similarly, to find deflections v, we use the

relationship, assuming small deflections;

and

which lets us write;

and;

This completes the development of the differential

and integral equations for beams.
E5003 – Ship Structures I 85
© C.G.Daley

Beam Example 1: Cantilever beam with left end

free
The cantilever beam is sketched at the left. The
left end is free and the right end is fixed. The
shear force is found by integrating the load. In this
case the initial shear is zero, because there is no
reaction at the left had end (it's a free end) ;

The bending moment is similarly found by

integrating the shear. And again there is no initial
value of moment because the boundary condition
has no moment;

The shear is a straight line. We did not solve for

the right hand vertical reaction , but it is
and it opposes the shear in the end of the beam
(which we can see is ). The moment is a
quadratic function with a maximum value of
as is easily found from summing moments
about the right hand end.

Next we solve the equation for the slope.

by inserting the expression for bending moment

we get ;

which becomes;
E5003 – Ship Structures I 86
© C.G.Daley

At this point we can either carry forward the

unknown initial slope or solve for it. We know the
slope at L is zero, so we can write;

which can be solved to get;

therefore the complete equation for slope is;

Now we can find the deflection. The integral

equation is;

which becomes;

which becomes;

The deflection at L is zero, letting us write;

which gives;

so the total equation for the deflection is;

which completes the solution.

E5003 – Ship Structures I 87
© C.G.Daley

Example 2: Pinned-pinned beam

In this case the initial value of shear is the

reaction at the left end. We can solve for this from
static equilibrium at the start. So the shear is;

The bending moment is;

The plot at the left shows the shear and bending

solutions. In this case, we were able to use statics
to solve for one unknown at the start, which
simplified the problem.
Next we solve the equation for the slope, as before,
by inserting the expression for bending moment
we get ;

which becomes;

At this point we can either carry forward the

unknown initial slope or solve for it. We know,
from symmetry, that the slope at is zero,
so we can write;

which can be solved to get;

E5003 – Ship Structures I 88
© C.G.Daley

therefore the complete equation for slope is;

Now we can find the deflection. The integral

equation is;

which becomes;

which becomes;

The deflection at L is zero, letting us write;

which completes the solution.

E5003 – Ship Structures I 89
© C.G.Daley

Topic 8: Problems

8.1. Consider a beam made of steel joined to aluminum. The steel is 10 x 10 mm, with 5 x 10
mm of Aluminum attached. Esteel = 200,000 MPa, EAl = 80,000 MPa. The beam is fixed as a
simple cantilever, with a length of 100mm and a vertical force at the free end of 2 kN.

convert the section to an equivalent section in steel and calculate the equivalent
moment of inertia.
What is the deflection of the end of the beam (derive from 1st principles).
What is the maximum bending stress in the Aluminum at the support?

8.2. For elastic beam bending, derive the equation:

d M

dx EI
where  is the slope of the deflected shape, M is the moment, E is Young's Modulus,
I is the moment of inertia. You can assume the =E and =My/I. Use at least one
sketch.

8.3. Find and draw the shear force and bending moment diagrams for the following beam.
Find the values at supports and other max/min values.

8.4. There is a 3m beam. The shear force diagram is sketched below.

Sketch the load, moment, slope and deflection diagrams (9)
What are the boundary conditions and discuss whether there can be more than one option for the boundary
conditions.(6)
E5003 – Ship Structures I 90
© C.G.Daley

8.5. For elastic beam bending, complete Figure 1. The shear force diagram is sketched. You
need to infer from the shear what the load (including support reactions) may be, as well as an
estimate of the bending moment diagram, the slope diagram and the deflected shape. Draw
the support conditions and the applied load on the beam, and sketch the moment, slope and
deflection is the areas given.

8.6. Beam Mechanics. For the beam sketch below:

a) sketch by hand the shear, moment, slope and deflection diagrams

E5003 – Ship Structures I 91
© C.G.Daley

b) Assuming the beam is a 10cm x 10cm square steel bar, solve the problem to find
the bending stress at the fixed support. Use any method you like.

8.7. There is a length of steel that is 3.1416 m long, 50mm wide. It has a yield strength of
500 MPa (N/mm2), and a Young’s Modulus of 200 GPa. If the steel is thin enough it can be
bent into a perfect circle without yielding.
What is the maximum thickness 't' for the steel to be bent elastically (and not yield)?
If the steel thickness is 1mm, what is the stress when it is bent into a 1m Dia circle.
What would the shear force diagram look like?
(Hint :this relates directly to the derivation of the differential equations for beam
bending)

8.8. Sketch the shear, bending, slope and deflection patterns for the four cases shown below.
No numerical values are required.
E5003 – Ship Structures I 92
© C.G.Daley

Topic 9: Solving Beam Equations

A Train Station in Lisbon

Introduction
In this Chapter we will
Review the differential equation set derived in the last Chapter and discuss
solutions using Macaulay functions and Maple.
~~~~~~~~~~~~~~~~~~~~
Family of Differential Equations

Simple beam behavior considers only the

deflections due to bending, and only in 2
dimensions. Torsion, shear and other elastic
distortions are neglected (for now).

Consider a beam between two supports. We

describe the deflections with the variable v(x).
E5003 – Ship Structures I 93
© C.G.Daley

The analysis of bending in Chapter 8, developed

the following differential equations;

These can be re-expressed into a set of related (not

coupled) differential equations, of increasingly
higher order;

Seen in this way, the key behavior is deflection,

with all other quantities being derived from it.
There is a similar set of equations, expressed in
integral form, starting from load;
E5003 – Ship Structures I 94
© C.G.Daley

The set of derivative equations show that if the

deflected shape is known, all other quantities can
be determined. In such a case there is no need for
any boundary conditions. (to do: think of a
situation where the deflected shape is fully known,
while other quantities are not.)

Normally we would not know the deflected shape.

Instead we would know the load and would want
to determine the deflected shape. In that case we
would employ the integral equations. One
significant issue with the integral equations is
that the 'constants of integration' must be found.
These are found from the boundary conditions. All
types of end conditions can be represented as some
derivative of deflection being zero. More
specifically, two of the derivatives will be zero at
each end of the beam. This gives four known
boundary conditions for any beam (2 ends!), and so
the four integral equations can be solved.

At this level of consideration, there is no difference

between a determinate and an indeterminate
beam. All beams have 4 integral equations and 4
boundary equations (or it could be said that all
beams are represented by a fourth order ordinary
differential equation with four boundary condition
equations, regardless of the type or loading or
supports).

In the previous chapter we solved two beams by

progressively solving the integral equations. Those
cases were relatively simple, both because they
were determinant systems, and they had simple
load patterns, and in one case was symmetric.
Solving non-symmetric cases of indeterminate
beams with discontinuous loads (patch loads) can
E5003 – Ship Structures I 95
© C.G.Daley

involve a lot of algebra. We will solve one such

system in three different ways; 1) directly with the
integral equations, 2) with Macaulay functions
and 3) with the help of the Maple program.

Example 3: Solving Piecewise Beam Equations

The beam sketched at left is fixed on the left end,

guided on the right and with the loading and
properties shown. A qualitative sketch of the
solution is plotted, indicating that the solution is
in three parts. The load is a patch load, so the
solution must be in parts. The points labeled 'B'
and 'C' represent break points in the solution. The
various quantities at these points represent
ending values for the partial solution to the left of
the point and starting values for the solution to
the right of that point.

The boundary conditions create a set of unknown

loads on the ends of the beam, which are sketched
in the 'Loads' diagram. For a fixed end we know
that the deflection and rotation are zero. For a
guided end we know that the shear (reaction) and
rotation are zero. These conditions give two
unknown loads at the left end of the beam. There
are two known movements (deflection and slope
are zero) at the left end of the beam. At the right
end the moment and deflection are unknown while
the shear and slope are both zero (recall that there
are always 2 known and 2 unknown values at each
end, in some combination of loads and
displacements). In this particular beam we know
that RA is the only vertical support and must
balance all the applied load (which is 4x5=20). We
also know that there is no shear in the right end of
the beam (the vertical force must be zero because
the roller has released it). So the shear solution is
as follows;

part 1:

part 2:
E5003 – Ship Structures I 96
© C.G.Daley

part 3:

The moment solution is;

part 1:

part 2:

part 3:

The slope solution is;

part 1:

part 2:

part 3:

Therefore
E5003 – Ship Structures I 97
© C.G.Daley

The deflection solution is;

part 1:

part 2:

part 3:

Summary of solution:
E5003 – Ship Structures I 98
© C.G.Daley

This completes the manual integration method for

solving example 3. To check this we will be solving
the same problem in 2 other ways.

Macaulay Functions

Macaulay functions (also called singularity

functions) are simply a generalization of the idea
of a step function. These functions provide a
convenient way of describing point forces,
moments and piece-wise continuous functions. And
when a few special rules of integration are
employed, it becomes very easy to use Macaulay
functions to solve beam problems.

The fundamental Macaulay functions are shown

on the left. Each function in the sequence
represents the integral of the previous function
(with the small exception noted later). Any of the
functions can be multiplied to a constant to change
the magnitude.

For example, a unit moment at is described

as;

and a moment of magnitude M at is;

Similarly, a point for of magnitude at is;

The triangular brackets are just a way of saying

that the function is meant to be seen as "one
sided". In simple terms :

Two examples of how Macaulay functions can be

combined to describe various piecewise curves are
shown below;
E5003 – Ship Structures I 99
© C.G.Daley

Integrating Macaulay Functions

The integration of Macaulay functions is very

similar to normal functions with an exception. If
the exponent is positive then the normal rules of
integration apply. If the exponent is negative, then
we just add one to the exponent. The rules are
shown at the left.

So for example;

but

It likely makes sense to the reader that the

integral of a point force is a step and the integral
of a step is a ramp. Does it make sense that the
integral of a point moment is a force? To explore
this idea, consider the functions sketched at the
left. In the first case we have function with a small
patch of load in one direction followed by a small
patch of load in the opposite direction we have no
net force but we do create a small point moment
E5003 – Ship Structures I 100
© C.G.Daley

(in the limit). When we integrate this we get a

small triangle, which when integrated again gives
a step.
To Illustrate Macaulay functions, we start with an
example of a pinned-pinned beam with a central
force:

Example 3: Solved with Macaulay Equations

E5003 – Ship Structures I 101
© C.G.Daley
E5003 – Ship Structures I 102
© C.G.Daley

Solving Example 3 using Maple

Maple is a computer program that is capable of

solving a wide variety of mathematical problems,
including differential equations.

Here is a very simple example of Maple’s ability to

solve and plot differential equations. This is the
solution of a cantilever beam (EI=1, L=10) under
uniform load (p=-1).

The basic differential equation;

The boundary conditions are;

Below is the full Maple input and result, showing

the shape of a deflected cantilever;
E5003 – Ship Structures I 103
© C.G.Daley

Example 3 using MAPLE 14 to solve differential equations for beam

by: Claude Daley

> restart;
The following aliases simplify the definition of loads.
> dist_load := (w,a,b) -> w*Heaviside(x-a)- w*Heaviside(x-b): # distributed force
Length, Stiffness, Load at "a", Load at end, Location of "a"
> L := 10: EI := 10^6:
> wa:=5:we:=5:a:=2: b:=6:
> loads := -dist_load(wa,a,b)-(x-a)/(b-a)*dist_load(we-wa,a,b);

> plot(loads,x=0..L,title=`LOADS`, color=blue);

> supports := {y(0)=0, D(y)(0) = 0, D(y)(L) =0, D(D(D(y)))(L)=0}:

> de := EI*diff(y(x),x$4) = loads; # Form differential equation

> dsolve({de}union supports ,y(x)): # Solve boundary value problem

> yy := rhs(%): # Extract deflection
> th := diff(yy,x): # Extract slope
> m := EI*diff(yy,x$2): # Extract moment
> v := EI*diff(yy,x$3): # Extract shear
> plot(v,x=0..L,title=`Shear`, color=blue);

> plot(m,x=0..L,title=`Bending Moment`, color=blue);

> plot(th,x=0..L,title=`Beam Slope`, color=blue);

E5003 – Ship Structures I 104
© C.G.Daley

> plot(yy,x=0..L,title=`Beam Deflection`, color=blue);

> evalf(subs(x=0,m));evalf(subs(x=L,m));evalf(subs(x=L,yy));

The manual, Macaulay and Maple solutions are all

the same, as expected.
E5003 – Ship Structures I 105
© C.G.Daley

Topic 9: Problems

9.1. Solve the following beam by direct integration. What is the maximum deflection (mm)?
What is the maximum stress (MPa) ?

ANS: .000136mm, 140 Pa

9.2. Solve the following beam using Macaulay functions. What is the maximum deflection
(mm)? What is the maximum stress (MPa) ?

ANS: .000484mm, 253 Pa

E5003 – Ship Structures I 106
© C.G.Daley

Topic 10: Indeterminate Beams – Force Method

part of the superstructure on an FPSO

Introduction
In this chapter we will
Review the idea of indeterminate beams and one way to solve them
~~~~~~~~~~~~~~~~~~~~
Transverse and Local Strength

Most of the local structure in a ship exists to resist lateral loads.

Example: The sketch below shows a bulkhead between the deck and inner bottom,
supported by one intermediate deck. The bulk cargo (liquid or granular) will exert a
lateral pressure on the bulkhead.

We can model the bulkhead frame as a pinned frame over 3 supports, subject to a
lateral load;
E5003 – Ship Structures I 107
© C.G.Daley

To solve this type of structure we need a method to solve indeterminate structures.

What does indeterminate mean? Determinate structures have a simple set of

supports, such that the support reactions can be found from considerations of rigid
body equilibrium alone. This means that there are just enough supports for
equilibrium to exist. This is normally 3 for 2D structures and normally 6 for 3D
structures. The number of supports is also the number of equilibrium conditions that
need to be satisfied.

The sketch below illustrates the difference between determinate and indeterminate
for a 2D beam.

Determinate Indeterminate

 Find the Reactions  Find the Reactions

 
 Then find the deflections  Then find the deflections

Reactions don’t depend on deflections The reactions depend on the deflections

 Equations for Reactions Equations for Reactions


 coupled
Equations for Deflections Equations for Deflections
There are two approaches for solving indeterminate systems. Both approaches use
the principle of superposition, by dividing the problem into two simpler problems,
solving the simpler problems and adding the two solutions.
E5003 – Ship Structures I 108
© C.G.Daley

The first method is called the Force Method (also called the Flexibility Method).
The idea for the force method is;

step  release internal forces* or external reactions until we have one or more
determinate systems
step  solve each determinate system, to find all reactions and deflections.
Note all incompatible deflections
step  re-solve the determinate structures with only a set of self-balancing
internal unit forces* (at internal releases) or unit reaction forces at removed
reactions. This solves the system for the internal or external forces removed in .
Observe the magnitude of incompatible deflections that occur per unit force.
step a scale the unit forces to cause the opposite of the incompatible
deflections noted in 
step  Add solutions (everything: loads, reactions, deflections…) from  and
a. Note that this will result in no incompatible deflections.
*note: forces include both forces and moments

Overview of Force Method

The structure: a beam over multiple supports:

step  cut the structure to have one or more determinate systems

step  solve each system. Note  – an incompatible deflection.

E5003 – Ship Structures I 109
© C.G.Daley

step  re-solve the cut structures with self-balancing internal unit forces*
step a scale these forces (moments) to cause the opposite of the incompatible
deflections noted in 

step  Add solutions (everything: loads, reactions, deflections…) from  and

a. Note that this will result in no incompatible deflections.
E5003 – Ship Structures I 110
© C.G.Daley

Example of the FORCE Method:

Manual Solution

Problem:
1 - Find MA, RA, RB in terms of w, EI, L
2 – Find maximum displacement

Solution:
Part 1 – solve with MA released (denoted ’ ). The reason we do this is because the
structure is statically determinate.

The line load function is:

Reactions are found from static equilibrium:

E5003 – Ship Structures I 111
© C.G.Daley

The shear is found by integrating the line load:

The moment is found by integrating the shear:

The slope is found by integrating the moment:

And finally the deflection is found by integrating the slope:

This leaves us with one left unknown to find, which is the slope at A . We use the
boundary condition:

which is solved to give;

Substituting back gives;

Slope:

Deflection:
E5003 – Ship Structures I 112
© C.G.Daley

The gives us the first half of the solution. Now we need to ‘correct’ the solution, by
removing the rotation at A (in Part 2). This is done by applying a moment at A, of
just sufficient magnitude to cause . This moment will be the true reaction
moment at A. All other responses in Part 2 are added to the Part 1 responses
(deflections, shear, moments, etc). Responses can be added because the systems are
linear (superposition holds).

Part 2 – solve with just (the * denotes the corrective solution ). will cause a
rotation opposite to , which when added to the results of Part 1 will create a
‘fixed’ condition (no rotation) at A. Initially is unknown.

Reactions are found from static equilibrium:

is negative, so is negative. is positive.

The shear is found by:

The moment is found by integrating the shear:

The slope is found by integrating the moment:

E5003 – Ship Structures I 113
© C.G.Daley

And finally the deflection y’(x) is found by integrating the slope:

To find *A and M*A , we use:

Substituting back gives;

Reactions:

Shear:

Moment:

Slope:

Deflection:

This gives us the second half of the solution.

E5003 – Ship Structures I 114
© C.G.Daley

Now we sum the two parts together for the complete solution:

This is the answer to the first question. The maximum deflection is found where the
slope is zero. The full expression for the slope is:

We can create a new normalized variable , which ranges between 0 and 1. This
gives us slope in a simpler form:

where

To find the location of zero slope we set the term inside the brackets above to
zero, which can be simplified to:

The solution of this equation will be the location of maximum deflection. One way to
solve this (which can be done without derivatives or computers) is to solve the
E5003 – Ship Structures I 115
© C.G.Daley

equation for z iteratively. This can be done on any hand-held calculator. We pick one
of the z terms (the first term here), and express z as a function of z:
0.62

0.6

0.58

This iterative equation might be expressed as:

0.56

0.54

0.52

0.5

0.48
0.48 0.5 0.52 0.54 0.56 0.58 0.6 0.62

Recall, z ranges from 0 to 1. So any value between 0 and 1 is a possible starting

value. We can guess that the maximum deflection will be at z >.5, so we could start
with a guess of 0.6. It doesn't really matter, except that the better the initial guess,
the quicker the solution will converge. Starting with z =0.6, we iterate to 0.5975 in 7
iterations.

Note: there is another possible iterative version of the z equation;

Unfortunately, it won’t converge to an answer in the 0-1 range.

The equation for deflection is:

The final step in the solution, is to find , which is at :

This answer can be checked in Roark, which gives the same answer. This completes
the problem.
E5003 – Ship Structures I 116
© C.G.Daley

Topic 10: Problems

10.1. Solve the below by removing the reaction RB (as shown). This creates ‘cut’
problem that is a cantilever beam.

10.2. Force Method.

a) Sketch 3 alternative approaches to solving this indeterminate problem using the

force method. For each approach, you will need two sketches of the auxiliary
systems.

b) Using one of the approaches sketched in a) , solve the system to find the reaction
at B (in kN)
E5003 – Ship Structures I 117
© C.G.Daley

Topic 11: Indeterminate Beams – Displacement Method

Cruise Ship in Adriatic

Introduction
In this chapter we will
introduce the displacement method used to solve structural problems
introduce the standard stiffness components for a beam in 2D and 3D
~~~~~~~~~~~~~~~~~~~~
Indeterminate Problem

We start by considering the indeterminate beam as shown below. This could be

described as a fixed-pinned beam or a cantilever with a pinned end.

Find MA RA and RB

To solve this problem with the displacement (stiffness) method we create two sub-
problems, each simpler than the whole problem. Rather than removing a support
(removing a force or moment), we remove a movement (i.e we completely fix the
structure). This becomes the problem marked * below. To the * problem, we add a
second problem, the ** problem, that fixes any errors that we created with the *
problem. In this case we have a moment MB* that should not exist, while we have a
B* that should not be zero. So, in the ** problem, we impose B**, (and only a B**)
sufficiently large to cause a moment MB** that is equal and opposite to MB*.
E5003 – Ship Structures I 118
© C.G.Daley

fixed-fixed beam applied moment at pin

known solution the moments and forces can be found
MA*=- MB*=pL2/12 from the “stiffness” terms, as shown
RA*=RB*=pL/2 below:
MB**= B** 4EI/L
MA**= B** 2EI/L
RB**= - B** 6EI/L2
RA**= B** 6EI/L2

To solve the problem we use;

MB**+ MB*=0
which gives;
B** 4EI/L - pL2/12 = 0

from this we can solve for B**;

B** = pL3/(48 EI) = 0

from this we can find all other ** terms;

MA**= pL3/(48 EI) 2EI/L = 1/24 pL2
RB**= - pL3/(48 EI) 6EI/L2 = - 1/8 pL
RA**= pL3/(48 EI) 6EI/L2 = 1/8 pL

from this we can find the reactions;

MA =MA* + MA** = pL2/12 + pL2/24 = 1/8 pL2
RB = RB* + RB** = - pL/8 + pL/2 = 3/8 pL
RA = RA* + RA** = pL/8 + pL/2 = 5/8 pL

The terms used to find MB**, MA**, RB** and RA** are called stiffness terms because
the are an ‘action per unit movement’, such as a force per unit displacement or
moment per unit rotation. They can also be a kind of ‘cross stiffness’ such as a force
per unit rotation or a moment per unit displacement. In the case of the example
above, with the equations;

MB**= B** 4EI/L

MA**= B** 2EI/L
RB**= - B** 6EI/L2
E5003 – Ship Structures I 119
© C.G.Daley

RA**= B** 6EI/L2

The stiffness terms 4EI/L, 2EI/L, -6EI/L2 and 6EI/L2 are forces and moment ‘per unit
rotation’. We will define these stiffness terms in the next section.

Stiffness Terms

When using the stiffness method, we always need to find a set of forces and moments
that occur when we impose a movement at a support. The movement will correct a
situation that involved the suppression of a movement at a support. In our case here,
the structure is a beam, and the supports are at the ends of the beam. The supports
prevent the ends of the beam from moving. There are 3 possible movements at a
support for a 2D problem, and 6 for a 3D problem. Because of this we will define a
standard set of ‘degrees of freedom’ for a beam. A ‘degree of freedom’ can have either
a force or displacement, or a rotation or moment. The standard 2D degrees of
freedom for a beam are shown below;

2D beam = 6 degrees of freedom

The degrees of freedom follow the Cartesian system, with the right-hand rule. These
are essentially x, y, rotation (called rz). In general, to impose a unit movement in one
(and only one) of these degrees of freedom, we need to also impose a set of
forces/moments, The forces/moments must be in equilibrium. These forces/moments
will be ‘stiffnesses’.

The mechanics are linear. This means that the set of forces/moments corresponding
to each movement can be added to those of any other movement. A general solution
for any set of movements of the degrees of freedom can be found by superposition.

For now we will just consider the 2D case and derive the stiffness terms. There are 6
degrees of freedom. For each degree of freedom, there are potentially 6 forces or
moments that develop. This means that there are a total of 36 stiffness terms. Any
single term would be labeled kij, meaning the force/moment at i due to a
displacement/rotation at j. For example;

k11 = force at 1 due to unit displacement at 1

k41 = moment at 4 due to unit displacement at 1
k26 = force at 2 due to unit rotation at 6

All the terms can be written in matrix form as;

E5003 – Ship Structures I 120
© C.G.Daley

 k 11 k 12 k 13 k 14 k 15 k 16 
k k 22 k 23 k 24 k 25 k 26 
 21
k k 32 k 33 k 34 k 35 k 36 
K   31 
k 41 k 42 k 43 k 44 k 45 k 46 
k 51 k 52 k 53 k 54 k 55 k 56 
 
k 61 k 62 k 63 k 64 k 65 k 66 
We will now derive these 36 terms. Luckily they are not all unique.

Axial Terms

The axial terms are found by asking what set of forces is required to create a unit
displacement at d.o.f. #1 (and only #1);

For axial compression, the deflection under load is;

F1L F AE
1   1  1  k11 
AE 1 L

the force at d.o.f. #4 is equal and opposite to the force at #1;

F4  AE
F 4  F1   k 41 
1 L

There are no other forces (at #2, 3, 5, 6), so we have;

F2
 k 21  0 and k 31  k 51  k 61  0
1
A displacement at 4 would require a similar set of forces, so that we can also write;
E5003 – Ship Structures I 121
© C.G.Daley

AE  AE
k 44  , k 14  , k 24  k 34  k 54  k 64  0
L L

This has given us 12 terms, 1/3 of all the terms we need. Next we will find the terms
for the #2 and #5 direction.

Shear Terms

The shear terms are found from the set of forces is required to create a unit
displacement at d.o.f. #2 (and only #2);

For shear of this type, the deflection is;

F2 L3 F 12 EI
2   1  2  k 22 
12 EI 2 L3

Note: to derive this easily, think of the beam as two cantilevers, each L/2 long, with a
point load at the end, equal to F2.

The force at d.o.f. #5 is equal and opposite to the force at #2;

F5  12 EI
F 5  F 2   k 52 
2 L3

Following from the double cantilever notion, the end moments (M3, M6) are ;
L 6 EI
M 3  M 6  F2  k 32  k 62  2
2 L
There are no axial forces, so;
E5003 – Ship Structures I 122
© C.G.Daley

k 12  k 42  0

A displacement at #5 require a similar set of forces, so that we can also write;

12 EI 12 EI  6 EI
k 55  3 , k 25  3 , k 35  k 65  , k 15  k 45  0
L L L2
This has given us 12 more terms, for 2/3 of all the terms we need. Next we will find
the terms for the #3 and #6 direction.

Rotary Terms

The rotary terms are found from the set of forces/moments required to create a unit
rotation at d.o.f. #3 (and only #3);

For illustration and to find these stiffness terms we will solve the system. We can
draw the shear force, moment, slope and deflection diagrams as below;
E5003 – Ship Structures I 123
© C.G.Daley

Q( x)  F2

M ( x)  M 3  F2  x

1  2 
 ( x)   3    M 3 x  F2  x 
EI  2 


1  2 3
 ( x)   2   3 x    M 3 x  F2  x 
EI  2 6 


We can use the boundary conditions (3=1, 2=0, (L)=0,(L)=0) to find M3 and F2.
1  2 
 ( L)  0  1    M 3 L  F2  L 
EI  2 

1  2 3
 ( L)  0  0  L    M 3 L  F2  L 
EI  2 6 


These two equations can be solved to get;

4 EI 6 EI
M3  , F2 
L L2
from these we can find;

2 EI  6 EI
M6  , F5 
L L2

This allows to find the stiffness terms;

4 EI 2 EI 6 EI  6 EI
k 33  , k 63  k 23  2 , k 53  , k 13  k 43  0
L L L L2

A rotation at #6 require a similar set of forces, so that we can also write;

E5003 – Ship Structures I 124
© C.G.Daley

4 EI 2 EI 6 EI 6 EI
k66  , k36  k 26  2 , k56  2 , k 16  k 46  0
L L L L

We can collect all these terms in the matrix;

 AE  AE 
 L 0 0 0 0 
L
 12 EI 6 EI  12 EI 6 EI 
 0 0 
 L3 L2 L3 L2 
 6 EI 4 EI  6 EI 2 EI 
 0 L2 L
0
L2 L 
K  
  AE 0 0
AE
0 0 
 L L 
  12 EI  6 EI 12 EI  6 EI 
 0 L3 L2
0
L3 L2 
 6 EI 2 EI  6 EI 4 EI 
 0 0
 L2 L L2 L 

Note that the matrix is symmetrical. This means that terms such as k35 (moment at
#3 due to displacement at #5) is equal to k35 (force at #5 due to rotation at #3). This
may seem quite odd that these two items would be equal. We will examine this in
the next Chapter.

The standard 3D degrees of freedom for a beam are shown below;

3D beam = 12 degrees of freedom

The K matrix for a 3D beam is a 12x12 (144 terms).

E5003 – Ship Structures I 125
© C.G.Daley

Topic 11: Problems

11.1. Solve the pinned-pinned beam by using the displacement method as sketched below. The
solution for the fixed-fixed beam is the same as above. Then it is necessary to show that
MB*+MB**+MB***=0 and MA*+MA**+MA***=0. Note: MA** = ½ MB**, and MB*** =
½ MA***.

11.2. Describe how you would solve the beam shown below by using the displacement method.

11.3. For the simple beam shown below, derive the shear stiffness terms (i.e k15 to k65)

11.4. Solve the beam shown below using the stiffness method. Find the reactions at A and B,
and the deflection at B.
E5003 – Ship Structures I 126
© C.G.Daley

ANS: MA= 166667 N-m, MB = 83333N-m B = -.2082m

11.5. Stiffness method .

sketch a 2D beam and show the degrees of freedom.
Describe the meaning of the terms (any, all) in the 6x6 stiffness matrix for a 2D
beam, and give 2 examples.

11.6. Explain the difference between the “Force” method, and the “Displacement” method.

11.7. In the stiffness method for a 2D beam, the standard value for the k22 stiffness term is;

Derive this equation (Table 1 in appendix may be useful).

E5003 – Ship Structures I 127
© C.G.Daley

Topic 12: Energy Methods in Structural Analysis

Coliseum
Introduction
In this chapter we will
Discuss application of energy methods in structural analysis
Show how conservation of energy conservation to the symmetry of structural
stiffness terms
~~~~~~~~~~~~~~~~~~~~
Energy Methods
Structural analysis is concerned with forces,
deflections, stresses and strains. All these involve
energy. An analysis of energy can be a way to
simplify structural analysis. Energy is a scalar,
and must be conserved, somehow. In some cases
the mechanical work done by a force is converted
to heat by friction:
E5003 – Ship Structures I 128
© C.G.Daley

In some cases the mechanical work done by a force is converted to elastic potential
energy in a spring. Potential energy (in a spring or in a gravitational field) can later
be recovered:

Consider a body subject to a simple axial load:

The above is correct for situations where axial stresses dominate, as in column
compression or simple beam bending. This does not take shear strain energy into
account.
E5003 – Ship Structures I 129
© C.G.Daley

Example: derive formula for Cantilever beam deflection using energy methods.

Consider a simple cantilever with rectangular cross section.

Start with Energy Balance equation:

External Work (EW) done by the applied load P is balanced by the elastic potential
energy (EPE) stored in the beam;

In this case we assume that the stress is the result of bending and we find the
stress from;

and

which lets us write;

We can re-write dvol as w dx dy and use :

E5003 – Ship Structures I 130
© C.G.Daley

The last part of the above equation is the moment of inertia:

This simplifies the problem to:

Which gives the final and correct answer:

Betti-Maxwell Reciprocal Theorem

The Betti-Maxwell theorem states that for any linear elastic body (also called a
Hookean body), that the movement at a d.o.f. A, caused by the application of a
force/moment F at a d.o.f. B, is exactly the same as the movement at a d.o.f. B,
caused by the application of a force/moment F at a d.o.f. A. In the sketch below,
refers to the movement at due to the application of a force at . So we can write the
Betti-Maxwell theorm as;
E5003 – Ship Structures I 131
© C.G.Daley

Proof:
As a linear system, superposition will hold. The structure will assume the same final
position regardless of the order of application of the forces. This means that the same
stored elastic energy will exist in either case. These are ‘conservative’ systems,
meaning that all work done by the loads is converted to elastic potential energy (and
is ‘conserved’ to be recovered later). We will apply F to the structure in two places,
and compare the work done when we change the order in which we apply the forces.

When F is applied at both 1 and 2, the total deflection at 1 and 2 will be;

If we imagine applying F at 1 first, and then at 2, the work done will be;

If we imagine applying F at 2 first, and then at 1, the work done will be;

The work done will be the same, so;

Hence Betti-Maxwell is proven.

Example 1 of Betti-Maxwell

For a simple cantilever, the deflection at x2 caused by a force F at x1 should be the

same as the deflection at x1 when F is applied at x2 :

Solution:
The beam deflection tables (see Appendix) can be used to find 12 and 21 .
To find 21 we first find the deflection at x1 . The beam to the right of x1 has no shear
or bending. Consequently it is perfectly straight. It slopes downward at the same
E5003 – Ship Structures I 132
© C.G.Daley

angle as the slop at x1 , which is 11 . The addition deflection past is just equal to
the slope angle times the distance. The total deflection at x2 found as follows:

To find 1 2 we use the general equation for the deflections in a cantilever of length x2
and solve for the deflection at x1 .

The two results are identical, as Betti-Maxwell predicted.

Example 2 of Betti-Maxwell

For a simply supported beam, the rotation at the right hand end caused by a unit
vertical force F in the center should be the same as the vertical deflection at the
center caused by a unit moment at the right hand end :

Solution:
The beam deflection tables (see Appendix) can be used to find 12 and 21 .
The rotation 21 is as follows:

To find 1 2 we use the general equation for the deflections in a simply supported
beam with an end moment and solve for the deflection at L/2 .
E5003 – Ship Structures I 133
© C.G.Daley

The two results are identical, as Betti-Maxwell predicted.

E5003 – Ship Structures I 134
© C.G.Daley

Topic 12: Problems

12.1.
of Betti-Maxwell.

12.2. Illustrate the Betti-Maxwell theorem using the beam load cases shown below. Use the
deflection table on pg 8 at the end of the paper.
E5003 – Ship Structures I 135
© C.G.Daley

Topic 13: The Moment Distribution Method

Venice

Introduction

In this chapter we will describe the moment distribution method for solving
indeterminate beams
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

Overview

The moment distribution method is a type of

displacement (stiffness) method because it makes
use of the stiffness terms we derived earlier. It is
particularly useful for solving problems involving
beams over multiple supports, and frames with
moment connections. It is what can be termed a
‘relaxation’ method. This refers to the iterative
way that errors are ‘relaxed’. The method can be
solved manually on paper with a simple calculator,
and was once the dominant method used in
professional practice. These days it can easily be
solved with a spreadsheet, but is seldom used
professionally. Its current value is in helping
students develop an understanding of structural
behavior. The essence of structures is the
interconnected behavior of structural elements.
The moment distribution method is all about the
way neighboring elements interact.

Hardy Cross The method was developed by Prof. Hardy Cross in the
(Wikipedia) 1920s and 30s. Cross studied at MIT and Harvard, taught
at Brown, Illinois and Yale and consulted
extensively.
E5003 – Ship Structures I 136
© C.G.Daley

Prof. Hardy Cross described his procedure as follows:

" The method of moment distribution is this:

1. Imagine all joints in the structure held so that they cannot rotate and
compute the moments at the ends of the members for this condition;
2. At each joint distribute the unbalanced fixed-end moment among the
connecting members in proportion to the constant for each member defined as
"stiffness";
3. Multiply the moment distributed to each member at a joint by the carry-over
factor at the end of the member and set this product at the other end of the member;
4. Distribute these moments just "carried over";
5. Repeat the process until the moments to be carried over are small enough to
be neglected; and
6. Add all moments - fixed-end moments, distributed moments, moments carried
over - at each end of each member to obtain the true moment at the end."

Description of Method

The moment distribution method is a way to solve

indeterminate structures comprised of beams. The
method works for continuous beams over multiple
supports and for frames. In its basic form it does
not consider joint translation. All joints are only
assumed to rotate, as would occur at a pin or roller
support, or at a frame connection (beams to
column) where sway is prevented. Subsidence of a
support can easily be handled. An extended
version can treat sway of a frame system.

Fixed End Moments – FEM : To start the procedure,

all joint are considered fixed and all fixed-end
moments are calculated. One example of fixed end
moments is shown below for a beam with a central
point force. The moments are expressed as true
moments acting on the supports. This is an
important point. Note that both end moments in
the sketch cause concave downward bending, and
would this have the same sign in a bending
moment diagram. But here they have opposite
true senses (clockwise on left and counterclockwise
on right) and so have opposite signs. And we keep
E5003 – Ship Structures I 137
© C.G.Daley

tract of the moments acting from the beam, not the

reactions by the support.

Moment Distribution factors - : At each joint

where two or more beams connect, each beam
provides part of the rotary stiffness. When an
external moment is applied to the joint, it rotates
as a unit, with each of the connecting beams
resisting part of the total moment. The portion of
the total is called the moment distribution factor -
. For each beam the moment will be :

where is beam end rotation stiffness (see

Ch10);

for beam i

The moment distribution factor is;

Carry-Over factors - CO: As we saw earlier, when one

end of a bean is rotated, the other end of the beam
experiences a moment as well. This is the
moment. In other words, when a moment is
applied to one end of a beam, and the far end is
fixed, that other end experiences a moment.
Because is half of , the far end moment is
always half of the near end moment. Therefore the
carry over factor is always 0.5.
E5003 – Ship Structures I 138
© C.G.Daley

Steps in the Moment-Distribution Method

The steps in the MDM are shown on the left.

The steps are discussed in more detail below.

Step 1: sketch the structure:

Sketch the structure, show the loads and number

the joints. In the case of two or more members
connected at a joint, there is one 'end' for each
beam. Any correcting moment applied to the joint
is divided among the ends according to the
moment distribution factor.
E5003 – Ship Structures I 139
© C.G.Daley

Step 2: plan the solution strategy and determine if

the structure will sway

In the standard type of problem the joints do not

translate, they only rotate. Axial and shear
deformations are ignored. Only bending
deformations are considered. If the model supports
permit one or more joints to translate, and the
load is such that it will cause such a movement,
we need to consider sway. The example structures
at the left show both types (no-sway and sway).

Note: And 'imposed' joint movement, as would

occur when a support 'settles' a fixed amount, is
not a sway problem. Imposed movements are just
as easy to solve as are applied loads.

In cases where there are redundant parts of the

structure (a determinant sub-structure), such as
cantilever portions as shown at left, these should
be removed and replaced with the moments or
forces that they cause on the remaining structure.

Step 3: Find moment distribution factors  :

For each joint we find the set of moment

distribution factors. In general;

The moments will tend to be larger in the stiffer

members, where rotary stiffness is . Thus the
shorter members will tend to have the higher
factors.
E5003 – Ship Structures I 140
© C.G.Daley

Step 4: Find fixed-end moments:

In this step, we find the fixed end moments for

each beam end. In the example at left, we have 3
beams connected in a frame. The top two have
loads and so have fixed-end moments. The vertical
beam (the column) is unloaded so its FEM are
zero.

Steps 5, 6, 7, 8, 9: Perform iterative calculation to

correct end moments. The fixed-end moments
found in step 4 are the first estimate of the
solution. The moments are in equilibrium with the
external loads, with the only problem being that
some of the joints are incorrectly fixed, when they
should be free to rotate. We will set up a
calculation table that will allow us to add a
correcting moment to each joint. We will perform
the corrections iteratively and the solution will
converge to the correct answer.

The table with the solution is shown on the next

page. With two beam, there are 4 end and so there
are 4 columns in the table. The first row contains
the moment distribution factors. The second
contains a note describing the target moment (this
is an extra feature normally not included). The
third row contains the fixed end moments. The
fourth row shows the total correction (later
ignored), with the fifth row dividing the correction
among the beam ends. The sixth row adds the
carry-over moments from the neighboring ends.
And then the seventh row add the third, fifth and
sixth row terms to get a new estimate for the end
moments.

The whole process is repeated until the solution is

sufficiently converged.
E5003 – Ship Structures I 141
© C.G.Daley

Steps 10: Solve for the other reactions and beam

responses.

Once the end moments on a beam are known, the

vertical reactions can be found from static
equilibrium.

Remember that the end moments found in the

MDM are moments acting "on" the supports.
Moment reactions "from" the supports are opposite
to these.

Once the vertical reactions are found, all other

responses (distribution of shear, bending, slope
deflection, stress) can be found using normal beam
theory.
E5003 – Ship Structures I 142
© C.G.Daley

Example 2: Here is a simple case that solves fully

in 1 iteration. This will happen when there is only
one joint that needs to rotate to bring the problem
into equilibrium.

Also note that this example shows a case of

different EI values.
E5003 – Ship Structures I 143
© C.G.Daley

Example 3: Here is a case that shows a frame with

two columns. This is a relatively complex case,
though without sway.

w
i
t
h

t
h
e

e
n
d

m
o
m
e
n
t
s

solved, the full set of horizontal and vertical

reactions can be found using force and moment
equilibrium.
E5003 – Ship Structures I 144
© C.G.Daley
E5003 – Ship Structures I 145
© C.G.Daley

With the reactions found, the shear force and

bending moment diagrams can be sketched as
follows:

The bending moments above are drawn on the

compression side of the beam. Deflections can be
found by double integration of the moment
diagram.

Exercise: What is the slope at joint #3?

E5003 – Ship Structures I 146
© C.G.Daley

Topic 13: Problems

13.1. Moment distribution method

13.2.

13.3. Moment distribution method

13.4. Moment distribution method. For the case shown on the attached page (Figure 1), fill in
the first two cycles of the MD calculations.
13.5.

13.6. For the statically indeterminate beam shown below, with the loads, properties and end
conditions as given,
a) Solve using the moment distribution method.
b) What is the vertical reaction at the middle support
E5003 – Ship Structures I 147
© C.G.Daley

13.7. A 3 bar frame is shown below.

Solve for the moments using the moment distribution method.
Sketch the deformed shape.
Find the vertical reaction at the pin (the right hand end).

13.8. Solve the frame using the MDM method (suggest you use a spreadsheet).

13.9. Solve the frame using the MDM method (suggest you use a spreadsheet).
E5003 – Ship Structures I 148
© C.G.Daley

13.10. For the case shown below, set up and fill in the first two cycles of the Moment
Distribution calculations.

13.11. A 2 bar structure is shown below.

Solve for the moments using the moment distribution method.
Find the vertical reaction at the pin A (the left).
E5003 – Ship Structures I 149
© C.G.Daley

Topic 14: The Moment Distribution Method with Sway

Introduction

In this chapter we will

extend the application of the moment distribution
method for solving frames with sway
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

In the previous chapter we dealt with beams and frames in which joints could not
translate due to bending. In this chapter we all add the possibility of sway motion.
For simplicity we will only consider one sway motion.

a Quadrant

The solution of a sway

problem takes two parts. In
the first part a unit sway
sway is imposed on the
structure (call this the *
problem). The imposed motion
causes initial fixed end
moments, which relax as the
solution progresses, just as
happens with applied forces.
The force required to impose
the unit sway can be found
once the solution is found, just
like the other reactions. In the
example at left this is F*2x .

In the second problem (the 'f'

problem) the sway is
prevented, and the problem
solved.
E5003 – Ship Structures I 150
© C.G.Daley

To get the total

solution we need to
scale the * problem
by  (we call this the
** problem) and add
it to the 'f ' problem.

How large is  ?

 is chosen so that the

conditions at the
"false" sway support
are corrected.

If there is no direct
force at the false
support, (as in the
example at left), we
want:
F*2x = -F f2x

If there is a direct
force at the false
support, we would
want:
F*2x = F2x - F f2x

Example of MDM with Sway

E5003 – Ship Structures I 151
© C.G.Daley

To illustrate the moment distribution method with

sway, we will use a problem similar to Example 3
in Topic 13. In this case the problem has a roller
on the left, instead of a pin. As a result the frame
can sway.

To solve the problem we need to split the problem

into two component problems. The first problem
has sway prevented (by a pin on the left support).
The complementary problem has an imposed sway
which will create a reaction of opposite magnitude
to the first problem.

The first problem was solved in Example 3

above. The reaction at the left hand pin was (see
pg. 130);

Now we solve the second problem with a unit

displacement  applied to the roller. For the
imposed unit displacement, we have the initial
fixed end moments as shown at the left. For
example the moments in the right column are;

Once we have solved the second problem, and

found the reaction at the roller, we scale the whole
solution to match the reaction with the 17.1 kN we
need. The final answer is the sum of the scaled
solution of second problem and the solution of the
E5003 – Ship Structures I 152
© C.G.Daley

first problem. All the solutions needed are

presented below in the form of spreadsheets.

The solution of Problem #1:

This is the solution of Problem #2:

E5003 – Ship Structures I 153
© C.G.Daley

This is the solution of Problem #2, scaled to counteract the pin force from problem
#1 (call this #2a):

This is the sum of Problem #1 + #2a, which is the frame with roller solution. The
values are moments at the locations indicated.

1 2 3 4 5
A B C D E F G H
MDM 0.0 183.89 -294.13 110.24 127.84 - 82.27 5.59
127.84
BEAM3D 0.0 183.8 -294.4 110.6 127.3 -127.3 83.9 6.47

To confirm these values independently, the same problem was analyzed in the DnV
program BEAM3D. The values shown above correspond very well with the MDM
results. The plots from BEAM3D are shown below;
E5003 – Ship Structures I 154
© C.G.Daley

1 2 3

Z
4

bending moment diagram, and reactions.

1 2 3

X 4

shear force (red), bending moment and deflections (exaggerated)

E5003 – Ship Structures I 155
© C.G.Daley

Topic 14: Problems

14.1. Solve the frame using the MDM method (suggest you use a spreadsheet).

14.2. A 3 bar frame is shown below.

Solve for the moments using the moment distribution method.
Sketch the deformed shape.
Find the vertical reaction at the pin (the right hand end).

1.3.
E5003 – Ship Structures I 156
© C.G.Daley
E5003 – Ship Structures I 157
© C.G.Daley

Topic 15: Matrix Structural Analysis

Introduction
In this chapter we will
Discuss a very general method to analyze structures, to give bending moments and
axial forces in general frame structures.
~~~~~~~~~~~~~~~~~~~
The behavior of a structure can be expressed in
matrix form as;

This type of equation is 'discrete'. It represents a

set of relationships among a finite set of degrees of
freedom (dof).

For a general structure or arbitrary shape, the

behavior can be adequately described by
describing the behavior of a set of points. In such a
E5003 – Ship Structures I 158
© C.G.Daley

case, all forces would have to be applied at the

points and all responses would be determined at
the points. Actions and responses at other points
can be considered, as long as there is a way to
gather actions to points and to interpolate
response to locations between points.

We might define arbitrary degrees of freedom, for

which we could write;

But how would we find the kij terms? For an

arbitrary body (a violin, a rock, a teapot ...) the kij
terms would be hard to find. There would be no
table of standard values.

The kij terms could be found by experiment.

- apply a test force at dof "i", measure all
displacements at dofs "j":

But is it even possible to apply a force at "i" and

only "i" ? Remember that includes reactions as
well as applied forces (there is no difference as far
as the structure is concerned!)

Determining experimentally is not practical.

The best one can do is to attempt to validate the
matrix experimentally by measuring responses
and comparing to predictions.

To make the determination of a structural

stiffness matrix practical, we normally describe a
structure using regularly shaped parts, with
standard degrees of freedom.
E5003 – Ship Structures I 159
© C.G.Daley

For the 3 bar frame at left, we can define the

define local degrees of freedom for each member
using the same standard approach that was
described in Chapter 11. We will start from the
local element stiffness matrices and assemble the
full structural global stiffness matrix, just to
illustrate the process.

The local degrees of freedom follow the individual

members, while the global degrees of freedom are
all aligned to the Cartesian (x-y) system. The
other aspect is that global degrees of freedom refer
to nodes of a structure, rather than to ends of
members. This means that several member ends
can share a single set of degrees of freedom.

The matrices below show the local and global

versions of the stiffness matrix for beam 1. The
difference is the way the degrees of freedom are
defined. In this case the global degrees of freedom
are just versions of the local dofs.
E5003 – Ship Structures I 160
© C.G.Daley

Aside: There is a general way to find the global

stiffnesses for a rotated bar. The rotation matrix
can be used to find the stiffness terms for a rotated
beam. In a rotated beam dof 1 is partly axial and
partly shear, as is dof 2. But as
superposition holds, any movement
along dof 1 can be expresses as some
axial and some shear, and the resulting
axial and shear forces can be resolved
back into the 1 and 2 directions.

The matrix below and the matrix operation

expresses the mix of effects in a concise way.
E5003 – Ship Structures I 161
© C.G.Daley

(suggestion: derive the rotation matrix using

vector algebra).

In the case of a 90 degree rotation, the rotation

matrix has the effect of doing row-column swaps.
For other angles the effect is more complicated.

Beam 2 has a local [k] that is similar to beam 1

except that area is 2A and modulus is 4I. The
global [k] looks similar to the local [k], except that
the numbering is shifted.
E5003 – Ship Structures I 162
© C.G.Daley

Beam 3 has a local [k] that is the same as beam 1.

The global [k] also looks similar to the global [k],
of beam 1 because a rotation of +90 produces a
similar effect to -90. The only change is that the
numbering is shifted.
E5003 – Ship Structures I 163
© C.G.Daley

The structural stiffness matrix is just the sum of

the global versions of the member stiffness
matrices. Where two terms share a dof, the two
values are added. This is again reflecting the
simple idea of superposition in linear systems that

Hooke first saw.

Stiffness matrices are symmetrical. This is a

curious property, especially when you think about
the off-diagonal terms. Some of the terms refer to
forces per unit rotation and moments per unit
translation.

With the whole stiffness matrix assembled, we

have a single equation that relates all actions
(forces and moments) with all movements
(translation and rotations):
E5003 – Ship Structures I 164
© C.G.Daley

To solve the system of twelve by twelve equations

we need to identify the twelve unknowns. It is
(almost) never the case that we would know twelve
deflections and want to know twelve forces. Nor
would we know twelve forces and look for the
deflections. Typically we know some forces (mostly
zero) and some
deflections (zero at supports):

We should have some combination of unknown

loads and deflections that adds up to twelve. If we
don't, we can't solve the system.

Note that the structure does not know what is an

applied force and what is a reaction. All the
structure know is whether it is in equilibrium.

There are a variety of ways of solving matrix

equations like:

There are various numerical strategies used in

linear algebra that are used to solve such systems.
E5003 – Ship Structures I 165
© C.G.Daley

Gaussian elimination is one common method. We

can assume that if we have N equations in N
unknowns that we can solve it.

To solve these in Maple (see 3bar_frame.pdf or

3bar_frame.mw), we would just expand the matrix
expression into a set of 12 simultaneous equations;

Maple will solve these equations in either

numerical or algebraic form, giving expressions for
all results in terms of the variable. For example,
for this problem, Maple will give;

Q1: With the above solution for force and

deflections at the nodes (the dofs), how would we
find the stresses in each member?

A1: To find the stresses we have to return to the

individual beams. We use the global stiffness
matrix of a single member. For example, for the
cross beam in the previous example (beam 2), we
find the member forces as follows;
E5003 – Ship Structures I 166
© C.G.Daley

The forces are not the same as found above. They

are only the forces that act on the individual
member.
The beam forces are found as follows:

Note that there is no axial force (would be F4, F7) in

Beam 2. This is because the roller at bottom of
beam 3 releases all horizontal force. The applied
load of 10 must all be transmitted to ground
through Beam1. With these forces and moments
we can find the shear force and bending moment
diagrams, along with the axial, shear and bending
stresses:
E5003 – Ship Structures I 167
© C.G.Daley

Because there was no load along the member, the

maximum stresses in the above case occurred at
the ends of the beam.

Q2: How are loads along a beam dealt with?

A2: Loads that are act between dofs are dealt

with in three steps. In step 1, the fixed end forces
and stresses that the loads cause are found. In
step 2, the fixed end actions are placed on a full
structural model and solved. All responses,
including deflection, stresses, strains, for the full
structure (including the beam where the loads
acted) can be found for the whole structure.
The complete solution comes from adding the two
solutions (step1 + step2):
E5003 – Ship Structures I 168
© C.G.Daley
E5003 – Ship Structures I 169
© C.G.Daley

Topic 15: Problems

1.4. 15.1 Frame Structures can be analyzed by "Matrix Structural Analysis" or by solution of
sets of continuous differential equations. Compare and contrast these two approaches.
1.5.
1.6. 15.2 The stiffness matrix for a 2D beam is said to have axial, shear and rotary terms. Give
examples of each of the 3 types of stiffness (i.e. 3 examples of the individual kij terms), with
a sketch of the terms.
1.7.
1.8. 15.3 Describe what is meant by the “rotary stiffness terms” in the stiffness matrix of a
beam. Explain which terms in the matrix are rotary terms and how they are derived.
1.9.
1.10. 15.4 For the 4-bar frame shown below, the 2D solution is found by solving 12 equations
in matrix form shown beneath. For the case of the loads and boundary conditions as shown,
fill in the 14 columns (there is 1 column for forces, 1 for displacements and 12 in the
stiffness matrix), with any known values. In the force and displacement vectors, write in a
zero (0) for known zero values and the letter X or variable name for other unknown values. In
the stiffness matrix write a 0 for the zero terms and the letter K for a non-zero stiffness terms.
You only need to fill in the upper half of the stiffness matrix. You don’t need any equations
or numbers (other than 0).

15.5 A 2 part frame is shown below.

Construct the full structural stiffness matrix for the structure. Describe the steps you take to do so.
Write the force-deflection equation for the structure in matrix format, showing all terms (ie include all
terms in the matrices or vectors). Explain which, if any, terms are unknown.
E5003 – Ship Structures I 170
© C.G.Daley

15.6 Assuming that you are using a program that performs matrix structural
analysis, explain concisely how the global stiffness terms for the joint circled in the
sketch below are determined. You don’t have to solve this frame.
E5003 – Ship Structures I 171
© C.G.Daley

Topic 16 Overview of Finite Element Theory

Introduction
In this chapter we will
introduce the 2D finite element called the constant stress triangle (cst)
show how to derive the element stiffness and all output values from energy
considerations
~~~~~~~~~~~~~~~~~~~~
Finite element method

Recall that for a beam, we can relate the end loads

by a stiffness equation in matrix form;

F   K x
We can find the K terms for a beam by solving the
beam bending equation for various end
movements. To find the displacement of some
point along the beam (at x ) we could solve the
system for the displaced shape. We would find that
the displacements would be;

dx  d 1  x ( d 4  d 1 ) (why so simple?)
and
dy  d 2  f ( x , d 3 , d 5 , d 6 ) (why is this
more complex?)
E5003 – Ship Structures I 172
© C.G.Daley

For this beam element, we made use of what is

called ‘beam theory’, to solve for the loads and
deflections under certain loading conditions.

However, in the case of most finite elements, such

as 2D planar elements, plate elements, and solid
elements, we will not start from some general
analytical solution of a loaded membrane, plate or
solid. These solutions are too complex and will not
give practical results. Instead, we assume some
very simple behaviors, highly idealized, but which
satisfy the basic requirements for equilibrium (i.e.
forces balance, energy is conserved). With this
approach, the single element does not really model
the behavior or a comparable real solid object of
the same shape. This is ok, because the aggregate
behavior of a set of these simple elements will
model the behavior quite well. This is something
like modeling a smooth curve as a series of
straight lines (even horizontal steps). This is
locally wrong, but overall quite accurate.

Constant Stress Triangle

To illustrate the way that finite elements are

formulated, we will derive the full description of
an element called the constant stress triangle (cst).
This is a standard 2D element that is available in
most finite element models.
Consider a 2D element which is only able to take
in-plane stress. The three corners of the triangle
can only move in the plane.

For this element the force balance is;

F  K e
6x1  6 x66x1

We want to determine the element stiffness matrix

Ke , and we want it to be valid for any triangle;

So, while we have six degrees of freedom, as we

did in the beam case, we don't have any hand
analytical solutions. To create a general solution
E5003 – Ship Structures I 173
© C.G.Daley

that will apply to all triangles we will make some

very simple assumptions which will allow us to
model 2D stress problems (such as a web in shear,
or stresses in plane around a cutout in a web. .

We will follow the outline in Hughes (p. 245-253).

Step 1 - select a suitable displacement function.

Consider the movement of a general triangle. Each

corner moves differently, and every point inside
moves.

The movement in x is defined as u and the

movement is y is defined a v. Both u and v
are functions of x and y ;

Assuming that the material in the triangle is

isotropic (no preferred direction), then we
would expect the two displacement functions
and to look similar.

The functions for u and v can only depend on the 6

nodal displacements (that all the info that we have
to define movement), so we can have no more than
6 unknown coefficients for both functions.

A trial function;
a) lets try:

is this ok? No! Why? Because it means that at

(0,0) (the origin) there is no movement. It would be
as if all elements are pinned to the origin.

b) lets try:

is this ok? No! Why? same problem.

E5003 – Ship Structures I 174
© C.G.Daley

The simplest viable functions for u and v that has

6 coefficients is;

Occam's razor, in latin: "lex parsimoniae " (the

law of simple), is a principle that says: from among
alternative explanations, the one that works, but
makes the fewest new assumptions is usually
correct. The concept is central to rational thought.
William Occam was a 14th century English Friar
and writer.

This provides a very simple but viable general

description of the displacement field. We can re-
write the displacement function in matrix form;

Now we have the displacement function.

Step 2 - Find the constants in C

at the corners we can write;

The total displacement of the corners can be

written;
E5003 – Ship Structures I 175
© C.G.Daley

or

A is called the connectivity matrix. It contains the

geometric information, the coordinates of the
nodes of the triangle. The terms in the C vector
can be found;

is a 6x6 matrix;

where is the determinant of the 3x3

coordinate matrix;

where:

which happens to be 2x the area of the triangle (ie

is the area of the triangle).

We can now go back to;

which we can re-write as;

where is the displacement of any point in

the triangle, contains information on the
geometry of the triangle and contains the
E5003 – Ship Structures I 176
© C.G.Daley

displacements of the corner nodes of the triangle.

This lets of find the displacement anywhere by
just tracking the displacements of the nodes.
Remember that the finite element method lets us
model a continuum by modeling a discrete system
of connected nodes.

Step 3 - Find the strain in the element

We need to find the stress and strain in the

element so that we can determine the stiffness of
the element.
The (2D) strains at any point in the element have
3 components;

where the strains are found from the partial

derivatives of the displacement field:

recall that;

so that we have;
E5003 – Ship Structures I 177
© C.G.Daley

which allows us to write;

Note that the strains in the triangle are just

constants, and do not vary with x and y. This is
the reason that this element is called the CST or
constant stress triangle.
We can write the strains in matrix form;

and simplified to be;

We can substitute for C to get;

This is the strain fully described in terms of nodal

coordinates and nodal displacements. We can
collect terms;

where B is called the strain coefficient matrix, and

so write;

G is a 3x6 matrix. A-1 is a 6x6, so B is a 3x6 matrix

that relates the 3 strains to the 6 nodal
displacements.

Step 4 - Find the element stresses (and forces)

Start by defining the stresses;

E5003 – Ship Structures I 178
© C.G.Daley

We can write Hooke's law in matrix form as;

or in terms of stress;

In simpler form we write the stresses as;

where D is called the elasticity matrix. Now we

can use to let us write;

or

where and is called the stress matrix.

Step 5 - Obtain the Element Stiffness Matrix

Idea: To obtain the element stiffness we will use

the principle of virtual work. The principle of
virtual work states that for a body in equilibrium,
the virtual work done by real forces fi acting
through any viable pattern of virtual
displacements * will be zero. In our case we wish
to equate the work done by the real nodal forces
with the work done to distort the element.
E5003 – Ship Structures I 179
© C.G.Daley

The external work done for a set of 6 virtual

displacements will be;

or

Note that, for example, f3 only does work when 3*

moves. And the work is the full amount of, f3 3*,
as f3 is fully active during the whole of 3* .
Remember that f3 does not cause 3*. We just
imagine that 3* occurs even as the nodal forces
stay acting.

The internal work done is equal to the integral of

the stress time the strain over the volume;

which in the case of the virtual work done one

element becomes;

which when making use of the strain coefficient

matrix and the elasticity matrix can be written as;

In this equation refers to virtual displacements ,

while refers to real (existing) displacements.

So if we say;
E5003 – Ship Structures I 180
© C.G.Daley

we can obtain;

which simplifies to;

and;

where t is the element thickness and is the

element area. The term in the brackets is the
element stiffness;

is a 6x6 matrix ( is 6x3 x 3x3 x 3x6 =

6x6)

Numerical Example: Consider this triangular

element with properties shown.

For this case the matrices are;

E5003 – Ship Structures I 181
© C.G.Daley

This is the stiffness matrix for a specific CST

element.
E5003 – Ship Structures I 182
© C.G.Daley

Topic 16: Problems

16.1 The displacement functions of the constant stress triangular element are:
u(x,y) = C1 + C2 x + C3 y
v(x,y) = C4 + C5 x + C6 y

where u represents the x-translation of any point (x,y) and v represents the y-
translation of the point.

16.2 A beam has only one coordinate (x). However, most beam models would allow
a point on the beam to rotate as well as translate. So, construct 3 simple
displacement functions;
u(x),
v(x),
(x),
of a ‘beam element’, using the same logic as was used to create the displacement
functions of the constant stress triangular element.
E5003 – Ship Structures I 183
© C.G.Daley

Topic 17: Hull Girder Shear Stresses

Italian Stone
Introduction
In this Chapter we will
Examine vertical shear in a ship
Describe the idea of shear flow.
~~~~~~~~~~~~~~

Ships are made of steel plate. This means that

ships are thin walled shells. Even for the local
components such as individual frames the width of
a plate is much greater than its thickness;

Overall, the cross section of a ship contains long

sections of connected plate. Such sections transfer
shear very effectively. Ships are generally very
stiff in shear, and need to be.

We wish to be able to determine the shear forces

and stresses everywhere in the cross section of a
ship. We will start by examining the shear that is
associated with the vertical bending stress. In a
later chapter we will examine torsion.
E5003 – Ship Structures I 184
© C.G.Daley

Recall from beam theory that shear is the slope of

the bending moment:

There is significant shear is a ship. How is it

distributed in the cross section? Shear is not just
in the vertical plates. There is shear in all parts of
the vessel. The average shear stress can be found
by dividing the shear force by the cross-section
area;

How is Q distributed around the x-section of the

ship? Is the shear stress uniform? Is it only in
vertically oriented members? To find the pattern
of stress, we construct a free body diagram of a
part of a slice of the ship's cross section.

To find the shear on the cross section, we cut the

section longitudinally and note that the shear
stress on the cut must be the same as the shear
stress on the cross section at that point. We can
assume;

there is no shear on the centerline

the shear force on the cut is

We find the force on the cut by integrating all

horizontal forces on out slice atarting from the
centerline (keel). We integrate along the shell
plating, using the path variable 's'.
E5003 – Ship Structures I 185
© C.G.Daley

Define:

m : 1st moment of area, about the neutral axis, of

all the material from the start to the cut at S
(where  is determined)

Define:

The units of shear flow is N/m.

There is an analogy between shear flow and fluid
flow. At an abrupt change in section, the shear
flow remains constant, while the stress abruptly
changes. This is analogous to water flow where at
a change in pipe size the mass flow rate (kg/s)
would stay constant while the velocity would
abruptly change.

We can combine the above concepts into one

equation;
E5003 – Ship Structures I 186
© C.G.Daley
E5003 – Ship Structures I 187
© C.G.Daley

Shear Flow Example 1: a rectangular steel bar

subject to a shear force Q.

Summary:

Shear flow acts along the cross section of a plate.

There can be no significant shear across a thin
plate, because there is no shear on the inner and
outer surfaces. The shear flow is found by
determining the value of 'm' (a path integral)
along with Q (the total shear force) and I (the
moment of inertia);
E5003 – Ship Structures I 188
© C.G.Daley

Example 2: Shear Flow in a simple box-girder.

Consider the simple box girder with the

dimensions as show below. This is like a simple
barge without the frames. The overall vertical
shear Q is 20 MN. To find the pattern of shear
flow and then the shear stresses we first calculate
the location of the neutral axis, and I.

g and h are the distances from the deck and

bottom to the neutral axis;

The moment of inertia about the base can be

approximated by;

The moment of inertia about the neutral axis is;

(whole ship)

Now we can determine m

Next we find m. We will start at the centerline on

the bottom, where s1 starts;
E5003 – Ship Structures I 189
© C.G.Daley

Next we find m on the side shell, The initial value

for the side is the same as the final value for the
bottom. The shear flow continues around the
corner. We integrate along s2 (note: y = h - s2 ) ;

This is a quadratic equation in s2. To find the

location of the maximum value, we set its
derivative to zero;

This shows that the maximum shear flow is

occurring at the neutral axis;

Continuing the integral to the deck gives;

Next we continue the integral along the deck,

along s3, to the centerline;

With the shear force of 20 MN (about 2000 tonnes)

The maximum shear stress is;
E5003 – Ship Structures I 190
© C.G.Daley

Branching Shear:

At a T junction, the shear flow branches. As long

as there are no closed loops between the points of
zero shear (ie. pts A, B and C in the sketch at left)
the shear flow can be found easily. Such situations
are statically determinate.
E5003 – Ship Structures I 191
© C.G.Daley

Topic 17: Problems

17.1 An open section is shown below. This is the cross section of a long folded steel
plate. The cross section is subject to a shear force of 2 MN

Solve the shear flow, plot it and then also show the shear stress values.
If this is a section of a long cantilever (fixed at one end and free at the other)
explain what types of deformations would you expect to see.

17.2 An open section is shown below. This is the cross section of transverse frame
in a ship. The shear force of 200kN.
Solve the shear flow, plot it and then also show the shear stress values.
The web is welded to the shell plate. What shear force must be resisted at this joint?
E5003 – Ship Structures I 192
© C.G.Daley
E5003 – Ship Structures I 193
© C.G.Daley
E5003 – Ship Structures I 194
© C.G.Daley

Topic 18: Shear Stresses in multi-cell sections

Croatian Coast
Introduction
In this Chapter we will
Discuss indeterminate shear flow
Calculate shear slip in a cut section.
Do an example of shear flow in a ship
~~~~~~~~~~~~~~~~~~~~
E5003 – Ship Structures I 195
© C.G.Daley

Shear in Multi-cell Sections

Consider a tanker with two longitudinal

bulkheads;

There will likely be two spots in the cell where

m=0. The shear flow will look something like the
sketch to the left.

To solve the statically indeterminate problem, we

apply the same kind of technique that we used in
the Force Method to solve indeterminate beams.

We will cut the structure, releasing the shear force

and allowing shear deflection (called 'slip'). We will
then determine how much shear we have to apply
to the cell to remove the slip.

This is qualitatively similar to the correction of

movements in the force method.

For any case where the loops are not adjacent, the
steps in the solution process are;

1) Make n cuts to make the problem into a

statically determinate problem.
2) Solve the statically determinate problem.
3) Find the N incompatible deflections (slips).
E5003 – Ship Structures I 196
© C.G.Daley

4) Apply N internal forces (actually torsions) to

reverse the incompatible deflections
5) Add #2 and #4(s) to get the solution
The above steps are sketched below;

The cuts and the slip at the cuts are in the

longitudinal direction;

The shear flow occurs on the cross section, which

is a transverse vertical plane. The shear stresses
on this plane will also occur on a longitudinal
plane at right angles to the transverse plane. The
longitudinal plane may be horizontal or vertical or
inclined. The stressed plate will respond to the
shear by distorting into a 'diamond' with relative
E5003 – Ship Structures I 197
© C.G.Daley

movement in the longitudinal direction, which

creates a differential slip over a small part of the
cross section.

The total slip is found by integrating the slip over

the whole loop from one side of a cut to the other.
If the loop is symmetrical , the fore and aft slip
will cancel out and result in no slip. In an
unsymmetrical section there is a net slip.

s = the path variable (length) around any loop

shear strain
= a cyclic or loop integral

The slip can be found from the shear flow;

To correct the slip in a cut loop, we impose a

correcting shear flow , such that;

is a constant so we can find it as;

is constant around the loop and zero elsewhere.

is a determiate solution, found in the usual way.
The total solution is;
E5003 – Ship Structures I 198
© C.G.Daley
E5003 – Ship Structures I 199
© C.G.Daley

Shear Flow Example #2

Find the shear stresses in the section below. The

total shear is 10MN (5 MN on the half section.

First we find the section properties:

I D w h I a
t e o
e s
m c
.
1 d 1 . * .
e 0 0 3
c 3
k
2 w 4 . .
. 2 1
t 3 7

p
l
.
3 s 1 2 .
i 0 . 3
d 5
e
4 b 1 . * .
o 0 0 3
t 3
.
Σ 2 1
. .
7 0
3 7

The centroid and moment of inertias are (for half

section) ;
E5003 – Ship Structures I 200
© C.G.Daley

The shear flow and stress in the half section can

be found from;

where

So to find , we just need to find m. To find m we

need to integrate along the 5 branches of the
problem.

Because we have a loop, the problem is

indeterminate and we need to cut the loop, find
the slip and add a correcting shear flow.

The solution to the cut problem is called q*. The

correcting flow is called qc.

For s1 (along deck);

y = 10 - 5.48 = 4.52, y t = 0.1357

m = .1357 s1
= 0.814 (@ s1 = 6)
= 1.357 (@ s1 = 10)

For s2 (side shell above wing tank);

y = 4.52 - s2 ,
m = 1.357 + .03 (4.52 s2 - s22 / 2 )
= 1.357 + .1357 s2 - .015 s22
= 1.658 (@ s2 = 4) ( at wing tank plate)
E5003 – Ship Structures I 201
© C.G.Daley

For s3 (inclined plate of wing tank);

y = 4.52 - s3/2 , s3 = 0 -> 42

m = .03 (4.52 s3 - s32 / 22 )
= .1357 s3 - .0106 s32
= .428 (@ s3 = 42) ( at side shell)

For s4 (side shell below wing tank);

y = 0.52 - s4 , s4 = 0 -> 6
m = 2.086+ .03 (0.52 s4 - s42 / 2 )
= 2.086 + .0156 s4 - .015 s42
= 1.64 (@ s4 = 6) ( at bottom)
= 2.09 (@ s4 = .52) (max value at n.a.)

For s5 (along bottom);

y = -5.48 , s5 = 0 -> 10
m = 1.64 - .164 s5
= 0 (@ s5 = 10) ( at centerline) ok

Now we can calculate the corrective shear needed

to close the slip that occurs at the wing tank cut;

is a constant so we can find it as;

In this case t is a constant so;

E5003 – Ship Structures I 202
© C.G.Daley

where S is the length around the loop. S = 8 + 42.

We can use the definition of shear flow to get;

m*deck = .814 +.1357 s

m*side = 1.357 + .03 (4.52 s2 - s22 / 2 )
m*wt = .03 (4.52 s3 - s32 / 22 )

= 4.34 + 6.188 - 1.53

= 9.00
Note that the m*wt part is subtracted beacuse we
are integrating in the reverse direction. With m*
we can calculate qC;

qC = -0.1736 [MN/m]

We have m* and qC.

q = q* ± qC = 0.2364 m* ± qC

We can plot q* (solid lines) and q (dashed lines);

and we can plot the shear stress ;

E5003 – Ship Structures I 203
© C.G.Daley

The values of shear stress have been checked

against an ANSYS model, and show good, though
not perfect, agreement. A sketch of the ANSYS
model is shown below.

See next page for ANSYS results.

E5003 – Ship Structures I 204
© C.G.Daley

image from ANSYS

E5003 – Ship Structures I 205
© C.G.Daley

image from ANSYS Workbench

E5003 – Ship Structures I 206
© C.G.Daley

Topic 18: Problems

18.1 Solve the shear flow in the following section of a tanker. Ignore the radius of
the bilge.

18.2 Solve the shear flow in the following section of a tanker.

18.3 Solve the shear flow in the following section of a tanker.

E5003 – Ship Structures I 207
© C.G.Daley

18.1 Solve the shear flow in the following frame section. What are the shear forces
transferred through the welds in details A and B (in kN/m)?
E5003 – Ship Structures I 208
© C.G.Daley

Topic 19: Shear Flow in adjacent Closed Cells

In a double sided or double bottom vessel there are

often many adjacent closed cells. Adjacent close
cells present an added complexity when solving
shear flow. The complexity is that the corrective
shear flow in once cell causes a corrective slip in
the adjacent cell, because of the common side.

When we add a corrective shear flow in one loop

we can't help but get some flow and slip in
adjacent loops.

Consequently, in order to ensure that we have no

net slip at each and all cuts we need to satisfy a
set of coupled equations. For example, in the case
of two adjacent loops we have;
E5003 – Ship Structures I 209
© C.G.Daley

and are unknown constants. is the

determinate shear flow in the cut section. For N
adjacent closed cells, we have to solve N
simultaneous equations.

Topic 19: Problems

19.1 Solve the shear flow in the following section of a tanker. Ignore the radius of
the bilge.

19.2 Solve the shear flow in the following section of a tanker.

19.3 Solve the shear flow in the following section of a tanker.

E5003 – Ship Structures I 210
© C.G.Daley

Topic 20: Torsion in ships

Ships as a whole and many individual members

within ships experience torsion.

The overall design torsional moment is given in

various classification society rules;
E5003 – Ship Structures I 211
© C.G.Daley
E5003 – Ship Structures I 212
© C.G.Daley

Local structural torsion can be found throughout

ships. Bending of a frame can result in a torsion in
a supporting frame.

Torsion Review

Consider a solid circular shaft subject to a

torsional moment. The longitudinal axis of the
cylinder x axis. A torsion is a moment about the x
axis. In such a case we get an ideal torsional
response. Every circular cross section remains
plane and remains centered on the x axis. Each
plane rotates slightly in comparison to its
neighboring cross sections. Assume that two
planes (1 and 2) are separated by a distance dx. In
comparison to their original orientations, the
planes are rotated

or

note similarity to the deq. for bending:

For solid sections like the circular shaft shown at

left, the shear stress is;

Thin Walled Torsion

E5003 – Ship Structures I 213
© C.G.Daley

Torsion in thin walled sections differs greatly

between ‘open’ and ‘closed’ sections.

To examine the difference between open and

closed sections we first make some simplifying
assumptions;

sections are prismatic

no in-plane deformation (cross sections only rotate)
small out of plane deformations (warping)

Thin Walled Torsion – Open Sections

Consider an open section, built-in at its base and

subject to a torsion at the free end.

The section rotates about a point called the shear

center. Point ‘p’ moves in the y and z direction due
to rotation and in the x direction due to ‘warpage’.

The displacements of point ‘p’

For ideal open sections with no warping restraint;

E5003 – Ship Structures I 214
© C.G.Daley

J = St. Venant torsional constant

For an open section;

For example, for a pipe of thickness t, radius r, cut

longitudinally;

Thin Walled Torsion – Closed Sections

Closed sections carry torsion in an entirely

different way from open sections. Because the loop
is closed, shear can flow around the loop. The
shear stress is uniform over the full thickness of
the wall. The shear flow is also constant over the
full loop. Once again;

We can also write;

As q is constant we can write;

E5003 – Ship Structures I 215
© C.G.Daley

where;
A = enclosed area of the loop

For a pipe (a circle);

Using the general formula for torsion;

We can use this to find

Compare this to ;

For example, consider a pipe of 1m dia., with a

10mm wall thickness;

The difference is so dramatic that it is easily

illustrated by seeing what happens when a
cardboard tube (eg paper coffee cup) is cut open
longitudinally.

Thin Walled Torsion – warpage restraint

E5003 – Ship Structures I 216
© C.G.Daley

Warpage is the term to describe axial

displacements due to torsion. In a closed circular
section the axial symmetry prevents all warpage.
In an open circular section, the warpage is
unrestrained (ie. The section is free to warp), so no
warpage stresses arise.

In sections with corners such as a box section, the

twist of one face is, to a degree, incompatible with
the twist of the connecting face. Each face wants to
warp differently, but is constrained at the corner.
This results in stresses on both faces. The
treatment of these effects requires the use of
warpage functions. This topic will not be
considered any further here. We will limit our
attention to simple torsion theory.
E5003 – Ship Structures I 217
© C.G.Daley

Topic 20: Problems

20.1 A hollow closed section is made of plate of uniform thickness ‘t’ . A torsional
moment of 80 MN-m is applied. To have the maximum shear stress equal to 135
MPa, what value should t be?
E5003 – Ship Structures I 218
© C.G.Daley

Topic 21: Shear Center and Shear Lag in Ship Structures

Topsides supports on an FPSO

Introduction
In this Chapter we will
Discuss the idea of the shear center of a frame
Describe the idea of shear lag and the notion of effective width.
E5003 – Ship Structures I 219
© C.G.Daley

Consider a channel section. A channel is a common

structural profile, but one that is asymmetric on
one axis. The center of area (centroid) and the
‘shear center’ are not at the same location.

What is the centroid? For one thing, it is a

property of the cross sectional area. But what does
it mean for the channel section? If we were to
want to use the section as a column and apply an
axial force that would only compress (and not
bend) the column, we would apply the force at the
centroid ‘g’. This is because a uniform stress in
the cross section would have a ‘center of force’ at
‘g’.

To find ‘g’ we use the standard formulations;

If the end of the column had an end cap, the load

would naturally find its way to the centroid.

However if the end were connected with a bold

through the web, the load would be applied off the
centroid and the axial load would cause bending.
In this case the end load would not only cause
bending, but the bending deflection would increase
the moment arm to further increase the bending.
This is a kind of self-excited response called the
p-delta effect, and is the subject of a special
analysis.

The above discussion is about axial loads. What is

the connection to shear? The connection is the idea
of the shear center. When a load is applied at the
shear center of a beam, the load will only cause
shear and bending, and no torsion. If the load is
applied anywhere else, a torsion will result.
E5003 – Ship Structures I 220
© C.G.Daley

Consider a shear force Q =10000 N applied to the

channel section on the previous page.

We will need Q/I;

Now we find the values of m. On the top flange;

So at B;

The force on the top flange is;

In the web;

The force on the web is;

E5003 – Ship Structures I 221
© C.G.Daley

The lower flange is symmetrical with the upper

and will have a shear flow of the same magnitude
but opposite in direction.

The shear flow as drawn shows the directions of

shear in the direction of the applied force. If we
think instead of the reaction to the applied force,
we have the sketch at left.

In this case the applied force is shown pushing

directly down on the web. In this case the vertical
forces oppose each other and produce no moment.
However, the horizontal forces, while equal in
magnitude, are separated by 190mm and produce
a couple of 1879 x 190 = 355300 N-mm. This couple
is a torsion acting on the section.

In order to eliminate the torsion, we would need to

apply the load Q at the shear center ‘e’ to the left
of the web. We can find the location of ‘e’ as
follows;

General formula for shear centers of channels

The following derivation is only valid for

symmetrical channels with constant wall
thickness.

The force in the top flange;

E5003 – Ship Structures I 222
© C.G.Daley

Now we can find ‘e’ by setting the sum of the

torsional moments to zero. The flange forces create
one couple and the applied load, opposed by the
reaction in the web, creates another couple. The
two couples will sum to zero when the load is
applied at the shear center.

For the previous example

(Q? – why would there be a slight difference

between the above result and the previous
example? )

Shear Lab / Effective Width

We normally assume that bending in a frame of a

ship or the hull girder can be modeled with what
we call ‘simple beam theory’. This means that we
assume that as the beam bends, plane sections
remain plane. When we make this assumption, we
E5003 – Ship Structures I 223
© C.G.Daley

are implicitly assuming that the flange is

uniformly compressed (or stretched), and that the
compressive or tensile stresses are uniform in the
flanges. Recall that ‘standard’ formula;

This formula says that all stresses at the same

value of y will be the same (i.e. all stresses in the
flange are the same!).

While the simple beam assumption is ok for beams

with relatively narrow flanges, the assumption is
not valid for sections with wide flanges such as are
sometimes found in ships.

In the case of very wide flanges, the compressive

stresses drop off away from the web.

To find the true pattern of flexural stress in a wide

flange beam, and the consequent effective width, is
a complex analysis, easily done in a finite element
model, but difficult to obtain analytically. The idea
of the behavior is presented below.

When we a lateral load (a bending load) to a beam

or ship frame, the web carries the load and tends
E5003 – Ship Structures I 224
© C.G.Daley

to bend. The flange is attached at the edges of the

web and as the web bends, its edge shortens (or
lengthens) and tends to pull the flange with it. To
pull on the flange, a shear stress of applied to the
edge of the flange. As shown in the sketch, the
flange is acted upon along its edge. Its as if the
flange is pinched along its edge, causing the flange
to compress more near the web and less away from
the web.

Unfortunately there are no general analytical

solutions for shear lag and effective width. Certain
approximate solutions have been postulated (see
PNA, VI, pp 247-250)

Shear lag and diminished effective width are most

important in cases of ;
wide flanges (large b)
short frames (small L/b)
proximity to free ends
proximity to concentrated loads

Finite element programs, when shell or brick

elements are used to model the frames, will
naturally show the shear lag effects.
There have been experiments on hull girder
models that have shown not only a variation in
deck stresses, but actual stress reversals. This
means that even when the average deck stress is
compressive, there may be a part of the deck (at
center) where the stresses are tensile, with the
deck edges in exaggerated compression. (PNA p
250)
E5003 – Ship Structures I 225
© C.G.Daley

Classification society rules have various

approaches to account for effective width. We will
consider deck plate buckling in the next ship
structures course (6003). In that case we will
consider another type of effective width of plating,
but one that describes a buckled plate’s reserve
capacity.

ANSYS analysis results

E5003 – Ship Structures I 226
© C.G.Daley

Topic 21: Problems

21.1 The following figure shows 4 x-sections. Identify the location of the shear
center in each case (i.e. which letter?). You should sketch the shear flow to help
identify the location.

21.2 When the vertical force F is applied to this section, how will the cantilever
beam deform? Explain

21.3 Where is the shear center of a 300 x 150 x 15fl x 10w mm ?

E5003 – Ship Structures I 227
© C.G.Daley
E5003 – Ship Structures I 228
© C.G.Daley

Topic 22: Plate Bending

Wexford Ireland

Introduction
In this chapter we will
Discuss the mechanics of plate bending
~~~~~~~~~~~~~~~~~~~~
E5003 – Ship Structures I 229
© C.G.Daley

Plates are the essential structural components in

ships. Almost all the structural weight in ships is
from the shell plating, the bulkheads, decks and
webs of large frames.

This section will examine the lateral deformation

of a single plate panel subject to a uniform
pressure. We will limit our problem as follows;
rectangular plate
constant thickness (t<<a, b)
simple edge conditions (fixed, pinned, free)
linear elastic material behavior
steel material (isotropic, homogeneous)
pressure normal to surface
no membrane stresses (no in-plane stress)

Recall that with beams we describe the

deformation and strains as follows;
E5003 – Ship Structures I 230
© C.G.Daley

Plates can bend in 3 ways;

x-bending
y-bending
twist

X and y bending are similar to beam bending.

Recall that there are no membrane stresses,

therefor no x and y stresses at the mid-plane.
Stresses only arise from bending, and are equal,
opposite and maximum on the bottom and top of
the plate.

Twist is a behavior that does not occur in beams,

although it is something like torsion.

Twist causes a shear strain in the top (and bottom)

of the plate, and results in curvature on
diagonals. When we twist a dx x dy portion of a
plate we get;

therefore

The above equation can be stated as;

the change in x-slope with change in y
= the change in y-slope with change in x
E5003 – Ship Structures I 231
© C.G.Daley

What about the curvature on diagonals?

Consider a view of the plate normal to the

diagonal.

We now have a variety of relationships for

deflection, curvature and strain.

The x direction movement 'u' is the result of

bending deflection w in the y direction.

We can find the strain from derivatives of the

movement;

In the y direction the movement is called 'v';

E5003 – Ship Structures I 232
© C.G.Daley

When u and v vary in x and y we can get shear

strains.

CONCEPT: displacement field

In solid mechanics it is useful to describe how all

points move relative to their original positions as a
'displacement field'. In the example below we just
consider how points along an x axis move. We call
the movement in the x direction u. A point at some
original position moves to a displaced position
. The displacement . we describe as
a function of , or . We could also write this as
because we think of the displacement as
dependent on the original position.
If all points move the same amount, then
. In such case the derivative of the
displacement field is zero and there is no strain
anywhere. We call this 'rigid body movement'. If
the movement is a linear function of the
coordinate, (such as ) then the
derivative of the displacement field is and the
strain is everywhere. The sketch below
illustrates the concept. The concept can be
extended to 2D and 3D problems.
E5003 – Ship Structures I 233
© C.G.Daley

ANSYS analysis results

For

we can use our definitions of u and v to get;

E5003 – Ship Structures I 234
© C.G.Daley

or

We can use the 2D version of Hooke's Law to get

the stresses.

Clearly when z = 0 (middle of plate), all stresses

vanish. Also, there are no average in-plane
stresses, only bending moments and torsion.

By using the expressions for , and we can

write;
E5003 – Ship Structures I 235
© C.G.Daley

where

The derivation of these equations is as follows;

So far we have expressions for stress and strain (2

axial and shear) and for moments (2 bending and
torsion) expressed as the derivatives of the
deflection w.

We now want to derive the differential equation

relating the deflection to load. The load is a
pressure acting normal to the plate. Consider a
small section of the plate subject to a uniform
pressure p.

Summing the vertical forces ;

E5003 – Ship Structures I 236
© C.G.Daley

Summing moments about x axis (about center of

plate) ;

now divide by dy;

and by dx;

which gives;

Using the previous expressions for and we

can write;

Similarly;

Now, using

we can write;
E5003 – Ship Structures I 237
© C.G.Daley

which simplifies to;

and can be written in the short hand got a general

4th derivative in 2 dimensions;

Note the similarity to the differential equation for

a beam of . Now we need to solve
for the appropriate boundary conditions
to get w(x,y) and the other results (stress, moments
etc.)

Example #1:
A long plate, simply supported with a pressure in
the shape of a half sine wave.

Check the pressure equation;

E5003 – Ship Structures I 238
© C.G.Daley

Note that nothing varies along the x axis, so all

derivatives of x are zero. Therefore, the differential
equation becomes;

assume the solution has the form;

so

which becomes;

and lets us solve for C;

which gives the deflection as;

with

The stress can be found using;

which simplifies to;

E5003 – Ship Structures I 239
© C.G.Daley

The stress at the top of the plate , @ z = t/2 ;

The stress as the edge is;

The stress in the center is;

Similarly, we can find;

General Plate Problems

The solution for a general plate problem requires

the solution of the 4th order partial differential
equation;

Such solutions can be complex, even for simple

load patterns. Even in the case;

The solution is found by expressing the load as a

Fourier equation;
E5003 – Ship Structures I 240
© C.G.Daley

where

For this load pattern and simply supported edges,

the deflected shape can be derived as;

Note that a sine pattern of load has been shown to

produce a sine pattern of response. So a group of
sine shaped loads will produce a group of sine
shaped responses. Hence the Fourier approach
should work. It all depends on the elegance of
super-position (hurray for Hooke!)

We will leave the general solution of more complex

problems to a specialized course in palates and
shells. See Hughes for solutions to some typical
problems.
E5003 – Ship Structures I 241
© C.G.Daley

Topic 22: Problems

E5003 – Ship Structures I 242
© C.G.Daley

Appendix
E5003 – Ship Structures I 243
© C.G.Daley
E5003 – Ship Structures I 244
© C.G.Daley

Typical spreadsheet to solve Moment Distribution problems.

M o m e n t Distribu tio n M e th o d
table o f v alu e s Example #3
param . v alu e u n its 1 problem all E I, no sway
L1 10 m 10m 20m
L2 20 m
L3 20 m w=8 kN/m
L4 12.5 m 32
12 21 23
3 35
EI1 1 rel
1 2 24
EI2 1 rel
EI3 1 rel
12.5 m
EI4 1 rel 20m 53
w 8 kN / m 5
all E I
42
4
e12 e 21 e 23 e 24 e 42 e 32 e 35 e 53
 1 .0 0 .5 0 .250 0 .250 0 .0 0 .38 46 0 .6 1 54 0 .0
FE M -6 6 .7 6 6 .7 -26 6 .7 0 .0 0 .0 26 6 .7 0 .0 0 .0
net 6 6 .7 20 0 .0 0 .0 -26 6 .7 0 .0
dist 6 6 .7 1 0 0 .0 50 .0 50 .0 0 .0 -1 0 2.6 -1 6 4.1 0 .0
CO 50 .0 33.3 -51 .3 0 .0 25.0 25.0 0 .0 -8 2.1
eEM 50 .0 20 0 .0 -26 7 .9 50 .0 25.0 1 8 9.1 -1 6 4.1 -8 2.1
net -50 .0 1 7 .9 -25.0 -25.0 8 2.1
dist -50 .0 9.0 4.5 4.5 0 .0 -9.6 -1 5.4 0 .0
CO 4.5 -25.0 -4.8 0 .0 2.2 2.2 0 .0 -7 .7
eEM 4.5 1 8 4.0 -26 8 .3 54.5 27 .2 1 8 1 .7 -1 7 9.5 -8 9.7
net -4.5 29.8 -27 .2 -2.2 8 9.7
dist -4.5 1 4.9 7 .5 7 .5 0 .0 -0 .9 -1 .4 0 .0
CO
e
ne
dist eE tM 7 .5
-7
-7 .5
.5 -2.2
1 961 .3.6
2.7 -0
-26
0 .7 .4
1 .2 601.0
.9 03.7
31
-31 .0.0 18 3.7
4.6
-3.7 0
8 .0
-1-2.30 .9 -0
-90
90 .7
.4.4
nCO
dist eE tM 0 .7
-0
-0 .7 1 -3.7
94.2
4.4 -0
-261 .101.7
.3 6012.6
.7
.0 31
-31.0
.3
.3
.3 -1
1 -0
8 .4
03.5
.3
.3 08.0
-1-0 3.2 0-1.0
-91
91 .1.6
.6
CO
e
ne
dist eE tM
E -1
-1 .1.7
1 .1.1 1-02.2
96
0 .2
0
.3.1
.4 -0
-2600
.1
.1.1.3
.1
6003.7
.0
.1
0
31
-31
0
.0
.6
.0.9
.9 -0
0
1 -0
8
-0
.6
3.9.1
.6
.2
08.0
-1-0 .2
3.4
.3
0
-0
-91
91 .0
0 .0
.1
.7.7
nCO
dist
CO e tM 0
-0
-0 .1 .1
.1 1 -0
10
95.8
0 .0.6
.7
.3 -0
-26
0 .2 .3 603.8
.0
.2 31
-31 .9
.1.9
00.0 180
0
0 .0
3.8
.0
.0 -108.03.7 -0
-91
91 .2
.9
.9
e
nCO
dist
ne
eE tM
eE tM
0 .2
-0
-0
0 .0
0 .0.2
.2 096
0.0
-0
1 960
.1.1.1
.1 .0 0 .0
-26
-26 .0 0 .1
0 .1
600 .0
4.0
.0
6 4.0
32.0
-32.0
0 .0
32.0
-32.0
18
180
0
0 .1
3.8
-0.0.1
3.8
.0
-10 8
-1 -0
0
.0
3.7
8 .0 .1
3.8
0
-91
91
0
-91
91
.0
.0.9
.9
.9
.9
dist
CO 0 .0.0 0
0 .0 0
e
ne
dist
nCO
eE
eE tM
tM
0
0
0 .0 1 96
0
1 96 .0.0
.0 .0 0 .0
-26
-26 .0 0 .1
0 .1
60 .0
4.0
604.0
.0 0
0 .0
32.0
-32.0
.0
32.0
18
18
0
0 .0
0 3.8
.0
.0
3.8
-1 0
8 .0
0 3.8
.0
-1 8 3.8
0
-91
91 .0
0 .0
-91
.9
.9
.9
dist
CO
e
neeE tM
E 0 .0
0 .0
.0
0
0
1 96
.0
0 .0
.0.0 0 .0
-26 0 .1 604.0
.0 -32.0
0 .0
32.0
-32.0 18
0
0
0
.0
.0
3.8
.0 -1 0
8 .0
3.8
91
0
-91
91
.9
.0.9
.9
dist
CO
nCO
e tM 0
0 .0 0
1 96 .0 .0 0 .0
-26 0 .1 0 .0
6 4.0 0 .0
32.0 180 .0
3.8 -1 0
8 .0
3.8 0 .0
-91 .9
dist 0 .0.0 0 .0
0 .0 0 .0 0 .0 -32.0
0 .0 0
0 .0
.0 0 .0 91
0 .0.9
eEM 0 .0 1 96 .0 -26 0 .1 6 4.0 32.0 1 8 3.8 -1 8 3.8 -91 .9

F21 y 59.6 F1 2y 20 .4 F42x 5 F53x -22

F42y 1 43.4 F53y 8 3.8
F23y 8 3.8 F32y 7 6 .2