See discussions, stats, and author profiles for this publication at: [Link]

net/publication/311856963

Load Analysis

Chapter · January 2017

DOI: 10.1007/978-3-319-50829-0_5

CITATIONS READS

0 5,664

1 author:

Seongwoo Woo
Addis Ababa Science and Technology University
147 PUBLICATIONS   652 CITATIONS   

SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Reliability Growth in Applicance View project

All content following this page was uploaded by Seongwoo Woo on 12 October 2017.

The user has requested enhancement of the downloaded file.

Chapter 5

Load Analysis

Seong-woo Woo

Email: twinwoo@[Link]

Reliability Association of Korea, Seoul, Korea (Republic of)

Abstracts: This chapter will explain how to model the mechanical/civil systems – automobiles, aircraft,

satellites, rockets, space stations, ships, bridge, and building subjected to the random loading.

Product have their own particular structural loads in field. A typical pattern of repeated load or

overloading may cause structural failure in product lifetime. Such possibility should be assessed in the

design phase whether structure subjected to loads endures in its lifetime. A modeling is a

mathematical representation of the dynamics system to describe the real world used by traditional

system modeling method like Newtonian. Here, as alternative method, the bond-graph will be

introduced because it is easily applicable to the mechanical/civil systems. If products are modeled,

the time response of system simulation for (random) dynamic loads will obtain. As the time response

is simplified and counted as a sinusoidal input, the rain-flow counting method and miner’s rule can

assess the system damage. Because there are a lot of assumptions, this analytic methodology is

exact but complex to reproduce the reliability disasters due to the design failures. So we should

develop the final solutions – experimental method like parametric ALT that will be discussed in

Chapter 7. Load analysis will be helpful to figure out the failure of problematic parts and finally

discover them in the reliability-embedded design process..

Keywords: Load Analysis, Mathematical modeling, Bond-graph, Miner’s rule, Rain-flow counting,
5.1 Introduction

Loads cause stresses, deformations, and displacements in the structures of product. Assessment of

their effects can be implemented by the structural modeling and its analysis using finite element. In a

result repeated load or overloading may cause structural failure.

Two generic types of mechanical static or dynamic loading exist. A static load – tension or

compression can exhibit motion or permenent change like dislocation if repeated in a lifetime.

Eventually, they will be a permanent deformation. The examples of static loading are as following:

Structural load and deflection versus material stress and strain

Tension and compression loads

Torsion and bending loads

A dynamic load, sometimes also referred to as probabilistic loads, is a force exerted by a moving body

on a resisting member, usually in a relatively short period of time. Because such loads are usually

unstable, we can say the dynamic load. Dynamic loads involve motion and therefore are time varying

load conditions. The examples of dynamic loading are as following:

Impact, vibration and shock loads

Unbalanced inertia loads

An impact load is one whose time of application on a material is less than one-third of the natural

period of vibration of that material. A variety of cyclic loads on a structure can lead to fatigue damage,

cumulative damage, or fracture. These loads come from repeated loadings on a structure or can be

due to vibration.
5.2 Modeling of Mechanical System

5.2.1 Introduction

The modeling of mechanical product is a mathematical representation of the dynamics structural

systems to figure out their characteristics.

characteristics Typical modeling methods – Newton, Lagrange,

Hamiltonian mechanics, and D'Alembert's

D'A Law are commonly used in dynamic system. As an output,

models might describe the system behavior that can be represented in random variables (or state

space). In a result the state space are expressed as vectors and provides a convenient and compact

way to analyze systems with

ith multiple inputs and outputs.
outputs

Fig. 5.1 Operational loads due to the random vibration on road

When observed in most mechanical/civil components,

components loads in field follows a more or less random
curve that constant load amplitudes are quite seldom. For example, automobiles possess completely

random stochastic load curves due to the street roughness, car speed, and environmental conditions.

And for airplane, a mean load change repetitively occurs on the wing of a transportation airplane

when it takes off or lands (See Figure 5.1).

On the other hands, the load of the gas turbine blade in an airplane is to a large extent deterministic

that there is no randomness in the system states, though the load sequence is still variable. With

simple algorithms and fast processors an on-line load measurement for parts can be directly

measured during operation.

However, a measurement during operation is quite time-consuming and actually impossible to figure

out the whole transmitted loads in product lifetime. To do that, the engineer depends on the

mathematical modeling, analysis and response such as the Newtonian model that was develeoped for

long time ago (See Figure 5.2).

(a) Base random vibrations

 (b) A simplified modeling of the automobiles

Fig. 5.2. Random loads and modeling of the automobiles by Newtonian modeling

5.2.2 D'Alembert's modeling for

or automobile

Engineer uses D'Alembert's principle and free body diagram to model mechanical system. If there is

an automobile that is used for transportation,

transportation we can model a simple system with a mass that is

separated from a wall by a spring and a dashpot. The mass could represent an automobile, with the

spring and dashpot representing the automobile's bumper. If only

nly horizontal motion and forces are

considered, it is represented
sented in Figure 5.3.
5.3

(a) Typical automobile subjected to wind flow

 (b) Mass-spring-dashpot system

Fig. 5.3 Typical Mechanical Automobile Modeling

The free body diagram is a drawing method showing all external forces acting on a body. There is

only one position in this system defined by the variable "x" that is positive to the right. It is assumed

that x=0 when the spring is in its relaxed state. As seen in Figure 5.4, there are four forces to develop

a model from the free body diagram: 1) An external force (Fe) such as friction force and air resistance

force, 2) A spring force that will be a force from the spring, k·x, to the left, 3) A dashpot force that will

be a force from the dashpot, b·v, to the left, 4) Finally, there is the inertial force which is defined to be

opposed to the defined direction of motion. This is represented by m·a to the left.

Fig. [Link] free body diagram for automobile modeling

Newton's second law states that an object accelerates in the direction of an applied force, and that

this acceleration is inversely proportional to the force, or

 ∑F = m⋅ a
all lexternal
(5-1)

Subtracting the right-hand side results in D'Alembert's principle,

∑F − m⋅a = 0
all lexternal
(5-2)

If we consider the m·a term to be inertia force (or D'Alembert's force), D'Alembert's law will be left

∑ F ⋅ δr = 0
all
(5-3)

To visualize this consider pushing against a mass (in the absence of friction) with your hand in the

positive direction. Your hand experiences a force in the direction opposite to that of the direction of the

force (this is the -m·a term). The inertial force is always in a direction opposite to the defined positive

direction. We sum all of these forces to zero and get

F e (t ) − ma (t ) − bv (t ) − k ⋅ x (t ) = 0 (5-4)

In other words, we can change

 d 2x
+ kx (t ) = F e (t )
dx
m 2
+ b (5-5)
dt dt

5.3 Bond graph modeling

5.3.1 Introduction

Bond Graph is an explicit graphical tool for modeling multidisciplinary dynamic systems including

components from different engineering areas – the mechanical/civil, the electrical, the thermal, and

the hydraulic system. When designing a new dynamic system, it is a good method to utilize a

graphical representation for communicating other engineers to express the dynamic modeling. In

engineering disciplines, linear graphs have long traditions among several graphical representation

means.

In 1959 Bond Graph method was developed by Professor Henry Payner and his former students at

MIT, who gave the revolutionary idea of portraying systems in terms of power bonds, connecting the

elements of the physical system to the so called junction structures which were manifestations of the

constraints. This power exchange portray of a system is called bond graph.

In 1961 the Paynter’s books were published as entitled “Analysis and Simulation of Simulation of

Multiport Systems.” In 2006, the three authors have published the fourth edition entitled as “System

Dynamics – Modeling and Simulation of Mechatronic Systems”. Now several disciplines of Bond

Graph have been widely accepted in the world as a modeling methodology. There are many

literatures about Bond Graph method and its applications to analyze dynamic systems.

In a result this method will give a brief description for analyzing loads applied to structure and

understanding its work.

 Fig. 5.5.A typical modeling
ing of the automobiles subjected to repetitive random vibrations

Fig. 5.6.
[Link] hydrostatic transmission modeling

5.3.2 Basic elements, Energy relations,

relations and Causality of Bond Graph

A Bond Graph is a graphical representation of a physical dynamic system. It is similar to the better

known block diagram and signal--flow graph. While the symbols in Bond Graph represent bi-directional
bi

exchange of physical energy, those in block diagrams and signal-flow

signal flow graphs represent uni-directional
uni

flow of information. Bond Graph also can be applicable in multi-energy

energy domain - mechanical/civil,

electrical, and hydraulic system.

The dynamic systems analysis is relatively simple when the steady state behavior or the few degrees

of freedom has.. In most of the cases, the main concern of engineers is to establish the mathematical

model that represents

ts the dynamic behavior of the system and how the different parameters influence
the system behavior, because the system dynamic equations are usually partial differential equations,

whose solutions require deep mathematical knowledge.

As the fundamental bases of the Bond Graph theory, energy flow is a basic element in a system. It

flows in from one or more sources, is temporarily stored in system components or partially dissipated

in resistances as heat, and finally arrives at “loads” where it produces some desired effects. Power is

the rate of energy flow without direction.

Fig. 5.7. Power flow in Bond Graph for electric-hydraulic system

Bond Graph represents this power flow between two systems. This flow is symbolized through an

arrow (Bond) as Figure 5.7 illustrated. Each bond represents the instantaneous energy flow or power.

The flow in each bond is denoted by a pair of variables called 'power variables' whose product is the

instantaneous power of the bond. Because power is not easy to measure directly, engineers can be

represented as two temporary variables - flow and effort. Every domain has a pair of effort and flow

variable. For example in mechanical system, flow represents the “velocity” and effort the “force”, in

electrical system, flow represents the “current” and effort the “voltage”. The product of both temporary

variables – power is represented as:

P = e(t ) ⋅ f (t ) (5-6)
The method makes possible the simulation of multiple physical domains, such as mechanical,

electrical, thermal, hydraulic, etc. Flows and efforts should be identified with a particular variable for

each specific physical domain which is working. Table 5.1 also shows the physical meanings of the

variables in different domains.

Table 5.1 Energy flow in the multi-port physical system

Modules Effort, e(t) Flow, f(t)

Mechanical translation Force, F(t) Velocity, V(t)

Mechanical rotation Torque, τ(t) Angular velocity, ω(t)

Compressor, Pump Pressure difference, ∆P(t) Volume flow rate, Q(t)

Electric Voltage, V(t) Current, i(t)

Thermal Temperature, T Entropy change rate, ds/dt

Chemical Chemical potential,µ Mole flow rate, dN/dt

Magnetic Magneto-motive force, em Magnetic flux, φ

The Bond Graph is composed of the "bonds" which link together "1-port", "2-port" and "3-port"

elements. Whether power in bond graph is continuous or not, every element is represented by a multi-

port. Ports are connected by bonds. The basic blocs of standard bond graph theory are listed in Table

5.2.

For 1-ports there are effort sources, flow sources, C-type elements, I-type Elements, and R-type

Elements that can connect power discontinuously. For 2-ports, there are Transformer and Gyrator

that can connect power continuously. For 3-ports, there are 0-junction and 1-junction that can make

up the network.
Power bonds may join at one of two kinds of junctions: a “0” junction and a “1” junction. In a “0”

junction, the flow and the efforts satisfy Eqs (5.7) - (5.8):

∑ flow iinput = ∑ flowoutput (5-7)

effort 1 = effort 2 = ...... = effort n (5-8)

This corresponds to a node in an electrical circuit (where Kirchhoff's current law applies). In a “1”

junction, the flow and the efforts satisfy Eqs. (5.9) - (5.10):

∑ effort iinput = ∑ effort output (5-9)

flow1 = flow 2 = ...... = flow n (5-10)

 Table 5.2 Basic Elements of Bond Graph

Elements Symbol Relation equations

Effort Se
 = 
1-Port
Elements
(Sources)
Flow Sf
 = 

 1
C-type Elements C = 

 

1-Port
 1
I-type Elements I = 
Elements


R-type Elements R  =  ∙ 

10 20   =  ∙  
Transformer TF 1
  = ∙ 
 
2-Port
Elements
10 20   =  ∙  
Gyrator GY 1
  = ∙  
 

10 20
0-junction 0   =  
3-Port
junction
elements 10 20
1-junction 1   =  
This corresponds to force balance at a mass in a system. An example of a “1” junction is a resistor in

series. In junction, the premise of energy conservation is assumed, no lost is allowed. There are two

additional variables, important in the description of dynamic systems.

For any element with a bond with power variables – effort and flow, the energy variation from t0 to t

can be expressed by:

H (t ) − H (t 0 ) = ∫ e (τ ) f (τ )d τ
t
(5-11)
t0

For C-type elements, e (effort) is a function of q (displacement). If displacement is differentiated, flow

is obtained as

q (t ) = ∫ f (t )dt = f (t )
dq
⇒ (5-12)
dt

If Eq. (5.11) is changing variables from t to q, the linear case can be expressed as:

H (q ) − H (q 0 ) =
1
2C
(
q 2 − q 02 ) (5-13)

For I-type elements, f (flow) is a function of p (momentum). If momentum is differentiated, effort is

obtained as
p (t ) = ∫ e (t )dt ⇒ = e (t )
dp (5-14)
dt

If Eq. (5.11) is changing variables from t to p, the linear case can be expressed as:

H ( p )− H ( p0 ) =
1
2I
(
p 2 − p 02 ) (5-15)

Resistor elements represent situations where energy dissipates - electrical resistor, mechanical

damper, and coulomb frictions. In these sorts of elements there is a relationship between flow and

effort as the Eq. (5.16) shows. The value of “R” can be constant or function of any system parameter

including time.

e(t) = R⋅ f (t) (5-16)

Compliance elements represent the situations where energy stores - electrical capacitors, mechanical

springs, etc. In these sorts of elements there is a relationship between effort and displacement

variable as the Eq. (5.17) shows. The value of “K” can be constant or function of any system

parameter including time.

e (t ) = K ⋅ q (t ) (5-17)

Inertia elements represent the relationship between the “flow” and Momentum (electrical coil, mass,

moment of inertia, etc.) as the Eq. (5.18) shows. The value of “I” tends to be constant
p(t ) = I ⋅ f (t ) (5-18)

A transformer adds no power but transforms it, such as an electrical transformer or a lever.

Transformers represent those physical phenomena that are variation of the values of output flow and

effort on the values of input flow and effort. If the transformation ratio is given by the “TF” value, then

the relationship between input and output is shown in Eqs. (5.19) - (5.20).

eoutput (t ) = TF ⋅ einput (t ) (5-19)

f output (t ) = ⋅ f input (t )
1
(5-20)
TF

Fig. 5.8."half-arrow" sign convention and meaning of the causal stroke

One is the "half-arrow" sign convention. This defines the assumed direction of positive energy flow. As

with electrical circuit diagrams and free-body diagrams, the choice of positive direction is arbitrary,

with the caveat that the analyst must be consistent throughout with the chosen definition. The other
feature is the "causal stroke". This is a vertical bar placed on only one end of the bond. It is not

arbitrary (Figure 5.8).

On each Bond, one of the variables must be the cause and the other one the effect. This can be

deduced by the relationship indicated by the arrow direction. Effort and flow causalities always act in

opposite directions in a Bond. The causality assignment procedure chooses who sets what for each

bond. Causality assignment is necessary to transform the bond graph into computable code.

Any port (single, double or multi) attached to the bond shall specify either "effort" or "flow" by its

causal stroke, but not both. The port attached to the end of the bond with the "causal stroke" specifies

the "flow" of the bond. And the bond imposes "effort" upon that port. Equivalently, the port on the end

without the "causal stroke" imposes "effort" to the bond, while the bond imposes "flow" to that port.

Once the system is represented in the form of Bond-graph, the state equations that govern its

behavior can be obtained directly as a first order differential equations in terms of generalized

variables defined above, using simple and standardized procedures, regardless of the physical

domain to which it belongs, even when interrelated across domains.

5.3.2 Case Study: Hydrostatic Transmission (HST) in sea-borne winch

The winch structure is designed for launching, owing, and handling the cable and array in ship. The

operation conditions of sea-borne winch can be varied such as operation conditions – sea state, ship

speed, and towing cable length. Because its operation requires high tension, sea-borne winch is

commonly used by the hydrostatic transmission (HST). It consists of electric motor, pump, piping,

hydraulic motor, and loads. Tension and the response characteristics under the states of launching,

towing, and hauling should be known before the design of HST. Tension data can be obtained from

tension experiment. However, as an experiment, obtained the exact time response characteristics has

many difficulties. And many previous design methods for HST involve extensive calculations because

energy type of HST changes from mechanical to hydraulic, and then mechanical system. Bond Graph

can easily model HST system and the dynamic response (Figure 5.8).
Fig. [Link]-driven winch system in ship-borne
Fig. [Link] Graph and derivation
erivation of the state equations of the Hydrostatic transmission in sea-

borne winch
HST as shown in Figure 5.9 is commonly divided into electric motor, hydraulic pump, piping system,

safety switches, and hydraulic motor. A rotating electric motor operates a hydraulic pump, which

supplies oil to pipe system.

As cylinders in a hydraulic motor are filled with oil, shat rotates load. Therefore, HST is a kind of the

closed-loop power transmission. The effort and flow in the rotating mechanical/civil system are torque

and angular velocity, respectively. If two elements are integrated, they became momentum and

volume. No matter what systems in HST may be, power does not change.

Bond Graph of electric motor and hydraulic pump is shown in Figure 5.11. Source flow SF11 indicates

an electric motor with constant angular velocity. It is assumed that a 10% among total torque perishes

out by resistance element R12. Transducer element MTF11 represents the capacity of a variable piston

pump which can control capacity with swivel angle αo.

Fig. [Link] motor and hydraulic pump modeling

A bulk modulus B with implies oil compressibility chooses 10,000 bar among 6,000 ~ 12,000 bar. Fluid

condensers C23 = C21 are described as V/B. Fluid inertia I24 represents oil mass. Using the least

square method, resistance R22 and R26 are calculated from the pump and motor leakage. Because
pipe flow is laminar, fluid resistance R25 can be calculated. Motor capacity TF3128 is determined from

the number of filling cylinders. Moment of inertia of drum and flange I33 can be calculated. It is

assumed that torque loss of flange R32 is about 10%. When Bond Graph is drawn from top and

bottom – starting with the electric motor and ending with the load, a total Bond Graph and derivation

of the state equations of a HST is represented as:

To obtain non-dimensional state equations, non-dimensional variables are introduced as

~ P Q ~ t 1
p= , q~ = , t = −1 = (5-21)
IQ or ω CP ωn IC

dQ d (CP q~ ) d (CP q~ ) d t
~
= = ~ = CP ωn q~& (5-22)
dt dt dt dt

dP d (IQ ~
p ) d (IQ ~
p) dt
~
= = ~ = IQ ωn ~
p& (5-23)
dt dt d t dt

where P,Q and t are dimensional integral of pressure, volume and time p , q and t̃ are non-

dimensional integral of pressure, volume and time, respectively. Therefore non-dimensional state

equations are derived as:

C 23ω n P ~& P Q
q 23 + q~23 + ~
p 24 = 1 (5-24)
SF11 MTF2113 R23 SF11 MTF2113 SF11 MTF2113

I 24ω n ~& P ~ P ~
p 24 − q 23 + ~
p 24 + q 27 = 0 (5-25)
R25 R25 Q R25 Q
 Q R26 ~ ~ ωTF 3128R26 ~
C27ωn R26q~&27 − p24 + q27 + p33 = 0 (5-26)
P P

I 33ωω n ~& P TF3128 ~ R ω

p33 − q 27 + 32 ~p33 = 0 (5-27)
SE34 SE34 SE34

To investigate the dynamic stability of the system, simple asymptotic approach can be used and

perturbations around stable points are expressed as:

( )
q~23 = q~230 + ε 1q~231 + O ε 2 (5-28)

~
p24 = ~
p240 + ε 1 ~
p241 + O ε 2( ) (5-29)

( )
q~27 = q~270 + ε 1q~271 + O ε 2 (5-30)

~
p33 = ~
p330 + ε 1 ~
p331 + O ε 2( ) (5-31)

whereε is very small value.

1

Substitute Eq. (5.24) to (5.27) into (5.28) to (5.31), then the terms of ε is yield
0
 P
(q~230 ) + Q ~
p 240 = 1 (5-32)
R22 SF11 MTF2113 SF11 MTF2113

P ~ P ~
− q230 + ~
p240 + q270 = 0 (5-33)
R25Q R25Q

Q R26 ~ ωTF 3128R26 ~

− p240 + q~270 + p330 = 0 (5-34)
P P

P TF3128 ~ R ω
− q270 + 32 ~p330 = −1 (5-35)
SE34 SE34

And then the terms of ε is yield

1

C 23ω n P ~& P Q
q 231 + q~231 + ~
p 241 = 0 (5-36)
SF11 MTF2113 R23 SF11 MTF2113 SF11 MTF2113

I 24ω n ~& P ~ P ~
p 241 − q231 + ~
p 241 + q 271 = 0 (5-37)
R25 R25Q R25Q

Q R26 ~ ωTF 3128R26 ~

C27ωn R26q~&271 − p241 + q~271 + p331 = 0 (5-38)
P P
I 33ωω n ~& P TF3128 ~ R ω
p331 − q 271 + 32 ~p331 = 0 (5-39)
SE34 SE34 SE34

If the perturbed Eqs. (5.36) to (5.39) are expressed as state space form
 ⁄
  =  , then

 1 Q 
− R C ω −
C23ωn P
0 0 
 q~& 231    ~
22 23 n
−P
− 25
R
−
P   q231 
 ~&    ~ p241 
0
 p241  =  Q I 24ωn I 24ωn Q ωn I 24 
 q~& 271   1 1 ωTF3128   q~271  (5-40)
 ~&   0 − −
P C27ωn C27ωn R26 P C27ωn   ~ 
 331  
p

p331 
− P TF3128 R
 0 0 − 32 
 I 33ωωn I 33ωn 

To investigate the dynamic stability of the non-dimensional state Eq. (5.40), eigen-value of The Bond

Graph can be represented as a state equation form | − # |  = 0. The system is unstable if eigen-

value are λ > 0 and the system is stable λ < 0. When the state equations are represented as state

space form of

 dQ23  − 
1 1
− 0 0 
 dt   C 23 R 22 I 24
 Q   MTF 
 dP   1 R 1 0
0 
23
− 25 −
2113
 24    
  P24   0   0
 dt  =  C 23 I 24 C 27
+ [
TF3128  Q27   0  11  0  34
SF ] +  [SE ] (5-41)
 dQ27   1 1
 dt   0 − − 
I 24 C 27 R26 I 33   P33   0   
− 1
 dP  
 33   TF3128 R32 
0 0 − 
 dt   C 27 I 33 

When Eq. (5.41) is integrated, the pump pressure and motor pressure are obtained as
 Ppump  1 C 23 0  Q 23 
 = (5-42)
 Pmotor   0 1 C 27  Q 27 

HST simulations are classified as models of low speed, high, and maximum tension. The tension

values might be obtained by the drag force analysis of cable. A steady solution of ε equation and
0

eigen-values from high speed, low speed, and maximum tension are calculated as stable. The values

of (a) perturbed state Q23 (b) perturbed state P24 (c) perturbed state Q27 (d) perturbed state P33 from

high speed mode are shown in Figure [Link] variables are converged after they perturbed around

steady state value ε . It can figure out that simulations results with a big overshoot reach a stead state
0

value (Figure 5.13).

(a) Perturbed state%

 23
(b) Perturbed state&
24

(c) Perturbed state%

 27
 (d) Perturbed state&
33

Fig. [Link] state %23, &24, %27 and &33

Fig. 5.13. Simulation results for hydrostatic transmission

5.3.2 Case Study: Failure analysis and redesign of a helix upper dispenser

The mechanical icemaker system in a side-by-side (SBS) refrigerator with a dispenser system

consists of many structural parts. Depending on the customer usage conditions, these parts receive a

variety of mechanical loads in the ice making process. Ice making involves several mechanical

processes: (1) the filtered water is pumped through a tap line supplying the tray; (2) the cold air in the

heat exchanger chills the water tray; and (3) after ice is made, the cubes are harvested, stocking the

bucket until it is full. When the customer pushes the lever by force, cubed or crushed ice is dispensed.

In the United States, the customer typically requires an SBS refrigerator to produce 10 cubes per use

and up to 200 cubes a day. Ice production may be influenced by uncontrollable customer usage

conditions such as water pressure, ice consumption, refrigerator notch settings, and the number of

times the door is opened. When the refrigerator is plugged in, the cubed ice mode is automatically

selected. A crusher breaks the cubed ice in the crushed mode. Normally, the mechanical load of the

icemaker is low because it is operated without fused or webbed ice.

However, for Asian customers, fused or webbed ice will frequently form in the tray because they

dispense ice in cubed mode infrequently. When ice is dispensed under these conditions, a serious

mechanical overload occurs in the ice crusher. However, in the United States or Europe, the icemaker

system operates continuously as it is repetitively used in both cubed and crushed ice modes. This can

produce a mechanical/civil overload.

Figure 5.14 overviews the schematic of the ice maker. Figure5.15 and Figure5.16 show a schematic

diagram of the mechanical/civil load transfer in the ice bucket assembly and its bond-graphs. An AC

auger motor generates enough torque to crush the ice. Motor power is transferred through the gear

system to the ice bucket assembly – that is, to the helix upper dispenser, the blade dispenser and the

ice crusher.
 Fig.5.14. Robust design schematic of ice maker

Fig.5.15. Schematic diagram for mechanical ice bucket assembly

 [Link] Graph of ice bucket assembly

The Bond Graph can be represented as a state equation form, that is,

dfE 2 dt = 1 L a × eE 2 (5-43)

dfM 2 dt = 1 J × eM 2 (5-44)

The junction from Eq. (5-43)

eE 2 = e a − eE 3 (5-45a)

eE 3 = R a × fE 3 (5-45b)

The junction from Eq. (5-44)

eM 2 = eM 1 − eM 3 (5-46a)

eM 1 = (K a × i ) − T Pulse (5-46b)

eM 3 = B × fM 3 (5-46c)

Because fM 1 = fM 2 = fM 3 = ω and i = fE 1 = fE 2 = fE 3 = i a ,
From Equation (5-45)

eE 2 = e a − R a × fE 3 (5-47)

fE 2 = fE 3 = i a (5-48)

If substituting Eqs. (5-47) and (5-48) into (5-43), then

di a dt = 1 La × (e a − R a × i a ) (5-49)

And from Eq. (5.46) we can obtain

eM 2 = [(K a × i ) − TPulse ] − B × fM 3 (5-50a)

i = ia (5-50b)

fM 3 = fM 2 = ω (5-50c)

If substituting Eq. (5-50) into (5-44), then

dω dt = 1 J × [(K a × i ) − TPulse ] − B × ω (5-51)

So the state equation can be obtained from Eq. (5.49) and (5.51) as following

 di a dt   − R a La 0  i a  1 La   1 
 dω dt  =  mk    +
− B J  ω   0   ea +  T Pulse (5-52)
   a − 1 J 

When Eq.(5-52) is integrated, the angular velocity of the ice bucket mechanical assembly is obtained

as

i 
y p = [0 1] a  (5-53)
ω 
5.4 Load Spectrum and Rain-flow Counting

5.4.1 Introduction

As seen in previous sections, we know that product subjected to a variety of loads can be simulated

through dynamics modeling like bond-graph. On the other hands, to experimentally measure the load

over time, strain gage type transducers are attached to the critical areas of the component. The

acquired data from the transducers are usually recorded and stored by a computer or by other

devices. After the recorded data is filtered to isolate the primary loads from noise, the recorded data

converted from the strain values to torque are counted by rain-flow counting methods. After

simplifying the fatigue damage computations, we can apply the Miner's rule (See Figure 5.17).

With the measurement data over time, we can perform a peak and trough detection to find the turning

points in the data. This is known as rain-flow counting. The output of this calculation is called the

torque count statistics. Some engineers stop at this point and define the rain-flow data as the load

spectrum, however it is not. Using the rain-flow data, it is then possible to calculate the histogram.

This histogram is the load spectrum. This load spectrum is very import during the design phase or a

refinement phase. The information from the load spectrum can be used with test rigs or simulation

software to reduce, but not remove, the need for field tests.

Realistic representation of loads is a key ingredient to successful fatigue analysis & design. It will

accurately measure the applied loads on an existing product and predict loads on a component or

structure that does not yet exist. Historically, complex load histories are often replaced by more

simplified loadings. The rain-flow cycle counting is a method for counting fatigue cycles from a time

history. The fatigue cycles are stress-reversals. The rain-flow method allows the application of Miner's

rule in order to assess the fatigue life of a structure subject to complex loading. And rain-flow counting

method may enable cumulative damage or the fatigue effects of loading events. The term "spectrum"

in fatigue often means a series of fatigue loading events other than uniformly repeated cycles.

Sometimes spectrum means a listing, ordered by size, of components of irregular sequences.

Maximum and minimum loads are also used to define the classifications in which the counts of cycles

are listed.
Fig.5.17 Classification and counting of the dynamic load
5.4.2 Rain-flow Counting

With the load-time, stress-time, or strain-time history, rain flowing down a roof can be represented by

the history of peaks and valleys. Rain-flow counting is a concept developed in Japan by Tatsuo Endo

and M. Matsuishi in 1968 [1] and in the USA for the segmentation of any arbitrary stress curve into

complete oscillation cycles. Rain-flow counting counts closed hysteresis loops in a load-time-function,

which are decisive for the damage of metal materials.

The following assumptions are valid for rain flow counting

Cyclic stable material behavior, that means that the cyclic stress-strain curve remains

constant, thus no hardening or softening of the material takes place.

Validity of the masing hypothesis, which means that the form of the hysteresis loop branches

correspond to the double of the initial load curve.

Memory behavior of the material which means that after a closed hysteresis loop, a

previously not yet completely closed hysteresis loop follows the same σ, ε path.

As seen in Figure 5.18, the tips of the largest hysteresis loop are at the largest tensile and

compressive loads in the load history (points 1 and 4). The notch strain-time history (Figure 5.18(c)) is

quite different from the corresponding notch stress-time history (Figure 5.18(e)). During each segment

of the loading the material “remembers” its prior deformation (called material memory). The damage

from each counted cycle can be computed from the strain amplitude and mean stress for that cycle as

soon as it has been identified in the counting procedure. The corresponding reversal points can then

be discarded.

That sequence clearly has 10 cycles of amplitude 10 MPa and a structure's life can be estimated from

a simple application of the relevant S-N curve.

.
Fig.5.18. Rain-flow counting method
An advantage of rain-flow counting is when it is used with notch strain analysis. The rain-flow counting

results in closed hysteresis loops, which representing a counted cycle. Therefore, the closed

hysteresis loops can also be used to obtain the cycle counting. If the dynamic load-time behaviors are

acquired in Figure 5.19., they can be summarized by rain-flow counting as Table 5.3.

[Link] of the dynamic load-time behavior with rain-flow counting algorithm

• Half-cycle starts at (A) and terminates opposite a greater tensile stress, peak (B); its range is

100MPa.

• Half-cycle starts at tensile peak (B), flow through (C), and terminates a greater tensile stress,

peak (E); its range is 200MPa.

Consequently, as seen in Table 5.3, we can count two cycles for 50 MPa range, two cycles for 100

MPa range, one cycle for 150 MPa range, and one cycle for 200 MPa range. Since calculated lifetime

estimations are afflicted with large uncertainties, it is desired to reconstruct the stochastic load-time

functions out of the load spectrums, in order to carry out experimental lifetime proofs with servo-

hydraulic facilities.

However, the reconstruction of a representative load-time function is not possible with the load

spectra alone. Two parametric rain-flow counting method is the most suitable method for the

acquisition of the local stress-strain hysteresis curves and influences the result of lifetime estimation.

Table 5.3 Summary of the dynamic loads by using rain-flow counting

Path From [MPa] To [MPa] Range [MPa] Cycles

A-B 0 100 100 0.5

B-E 100 -100 200 0.5

C-D -50 50 100 0.5

D-C 50 -50 100 0.5

E-F -100 100 200 0.5

F-I 75 -75 150 0.5

G-H 0 50 50 0.5

H-G 50 0 50 0.5

K-J -50 0 50 0.5

J-K 0 50 50 0.5

I-F -75 75 150 0.5

L-M 100 0 100 0.5

5.4.3 Goodman Relation

In the presence of a steady stress superimposed on the cyclic loading, the Goodman relation [2] can

be used to estimate a failure condition. It plots stress amplitude against mean stress with the fatigue

limit and the ultimate tensile strength of the material as the two extremes.

 σm 
σ a = σ e' × 1 − 
σ u' 
(5-54)


where '( effective alternating stress at failure for a lifetime of Nf cycles, '′* is ultimate stress.

A very substantial amount of testing is required to obtain as S-N curve for the simple case of fully

reversed loading, and it will usually be impractical to determine whole families of curves for every

combination of mean and alternating stress. There are a number of strategems for finessing this

difficulty, one common one being the “Augmented” Modified-Goodman diagram, shown in Figure 5.20.

Fig.5.20. “Augmented” Modified-Goodman diagram

Here a graph is constructed with mean stress as the abscissa and alternating stress as the ordinate,
and a straight “lifeline” is drawn from ' on the '+ axis to the ultimate tensile stress ' on the ',

axis. Then for any given stress, the endurance limit (or fatigue limit) – the value of alternating stress at

which fatigue facture never occurs - can be read directly as the ordinate of the lifeline at line is drawn

from the origin with a slope equal to that ratio. Its intersection with the lifeline then gives the effective

endurance limit for that combination of ' and ',

5.4.4 Palmgren-Miner’s law for cumulative damage

Fatigue properties of a material (S-N curves) are tested in rotating-bending tests in fatigue testing

apparatus. The S-N curve is required as a description of the material behavior for the calculation of

fatigue strength and operational fatigue strength. Well before a micro-structural understanding of

fatigue processes was developed, engineers had developed empirical means of quantifying the

fatigue process and designing against it. Perhaps the most important concept is the S-N diagram,

such as those shown in Figure 5.21, in which a constant cyclic stress amplitude S is applied to a

specimen and the number of loading cycles N until the specimen fails. Millions of cycles might be

required to cause failure at lower loading levels, so the abscissa in usually plotted logarithmically.

Fig.5.21. S - N diagram for aluminum and low-carbon steel

There are three zones to distinguish between in the double logarithmic representation of S – N curve

3

Low cycle fatigue: high loads, plastic and elastic deformation, N = 10 ~ 10 cycles (1 stage)

High cycle fatigue: fatigue strength, the zone of the sloped lines, until the corner load cycles
6 7
to failure ND = 10 ~ 10 (2 stage)

Fatigue limit (endurance limit), zone of the horizontal lines starting from N > ND. However,

several materials such as austenite steels dot possess a distinct endurance strength (3

stage)

In this case, the S—N curve becomes horizontal at large N. The fatigue limit is maximum stress

amplitude below which the material never fails, no matter how large the number of cycles is. In most

alloys, S decreases continuously with N. In this case the fatigue properties are described by fatigue
7
strength at which fracture occurs after a specified number of cycles (e.g. 10 ). Fatigue life is number

of cycles to fail at a specified stress level.

Fatigue failure has three stages:1) crack initiation in the areas of stress concentration or near stress

raisers, 2) incremental crack propagation, and 3) final rapid crack propagation after crack reaches

critical size. The total number of cycles to failure is the sum of cycles at the first and the second

stages. That:

N f = Ni + N p (5-55)

where Nf number of cycles to failure, Ni Number of cycles for crack initiation, Np Number of cycles for

crack propagation

In the fatigue strength zone, the S – N curve can be described by the following equation if
represented in the double logarithmic form.

−k
σ 
N = N D ⋅  a  (5-56)
σ D 

If possible, the determination of the S – N curve for operational fatigue strength calculation should be

carried out on real parts. Often, however, due to cost and time limitations, the calculations are only

carried out on special test samples.

The resulting load cycles to failure are random variables, which mean that they lie scattered around

the mean value. Today, the transformation of results won from a tension/compression trial onto a real

component is difficult. Thus, the exact determination of a notch over the entire load cycle zone is still

not possible today. Therefore, one is forced to rely on tests and trials.

In some materials, notably ferrous alloys, the S – N curve flattens out eventually, so that below a

certain fatigue limit ' failure does not occur no matter how long the loads are cycled. Obviously, the

designer will size the structure to keep the stresses below ' by a suitable safety factor if cyclic loads

are to be withstood. For some other materials such as aluminum, no fatigue limit exists and the

designer will size the structure to keep the stresses below ' by a suitable safety factor if cyclic loads

are to be withstood. For some other materials such as aluminum, no fatigue limit exists and the

designer must arrange for the planned lifetime of the structure to be less than the fatigue point on the

S - N diagram.

Statistical variability is troublesome in fatigue testing; it is necessary to measure the lifetimes of

perhaps twenty specimens at each of ten or so load levels to define the S-N diagram with statistical

confidence. It is generally impossible to cycle the specimen at more than approximately 10Hz and at
7
that speed it takes 11.6 days to reach 10 cycles of loading. Obtaining a full S-N curve is obviously a

tedious and expensive procedure.

At first glance, the scatter in measured lifetimes seems enormous, especially given the logarithmic

scale of the abscissa. If the coefficient of variability in conventional tensile testing is usually only a few
percent, why do the fatigue lifetimes vary over orders of magnitude? It must be remembered that in

tensile testing, we are measuring the variability in cycles at a given number of cycles, while in fatigue

we are measuring the variability in cycles at a given stress. State differently, in tensile testing we are

generating vertical scatters bars, but in fatigue they are horizontal. Note that we must expect more

variability in the lifetimes as the S-N curve becomes flatter, so that materials that are less prone to

fatigue damage require more specimens to provide a given confidence limit on lifetime.

Numerous different researchers have occupied themselves with the damage accumulation hypothesis

in fatigue failure, so that currently several variations exist. In general, the variations only distinguish

themselves by the fundamental S-N curve used: either fictitiously extrapolated or the real curve itself.

Oscillating loads cause an effect in materials, this is often referred to as “Damage” as soon as this

load surpasses a certain limit. It is assumed that this damage accumulates from the individual load

cycles and leads to a material fatigue. For an exact calculation this damage must be collected and

recorded quantitatively. This, however, has not yet been achieved with success.

Despite this fact, in order to gather information concerning the lifetime L out of the results of Wöhler

trials with irregular load cycle effects, around the year 1920, Palmgren [3] developed the fundamental

idea of linear accumulation, specific for roll bearing calculation. In 1945, Miner published the same

idea in a general form.

[Link] representation of Miner’s cumulative damage summation

Miner assumes that a part absorbs work during the fatigue process (Figure 5.22). The ratio of already

absorbed work to the maximal work which can be absorbed is a measurement for the current damage.

Thus, the ratio of the load cycle number n to the load cycles to failure N, which is determined in the

single-stage zone with the corresponding amplitude, is equal to the ratio of absorbed work w to

absorbed work W. This is denoted as the damage portion

w n
= (5-57)
W N

When the cycle load level varies during the fatigue process, a cumulative damage model is often

hypothesized. By definition of the S-N curve, take the lifetime to - cycles at stress level
 and -

at
 . If damage is assumed to accumulate at a constant rate during fatigue and a number of cycles

. is applied at stress
 , where . <- , then the fraction of lifetime consumed will be . ⁄- .

The Palmgren-Miner hypothesis asserts that the damage fraction at any level Si is linearly

proportional to the ratio of number of cycles of operation to the total number of cycles that would

produce failure at that stress level; that is

nj
Di = (5-58)
Nj

The limiting condition of strength happens when the absorbed work and absorbable work are the

same. That is, the prerequisite that the absorbed fracture work W is the same for all occurring load

sizes, allows the addition of the individual damage portions for load cycles of different sizes
w1 + w2 + L + wm
=1 (5-59)
W

So failure is predicted as follows,

w1 w 2 w n n n
D1 + D 2 + ... + D m = + +L+ m = 1 + 2 +L+ m =1 (5-60)
W W W N1 N 2 Nm

The generalization of this approach is called Palmgren-Miner’s Law, and can be written

nj
∑N ≤1 (5-61)
j

where./ is the number of cycles applied at a load corresponding to a lifetime of -/

Miner confined the applicability of this equation by the following conditions

Sinus formed load curve

No hardening or softening appearances in the material

The begin of a crack is considered as an incipient damage

Some loads lie above the endurance strength

Minor’s law should be viewed like many other material laws that might be accurate enough to use in

design. But damage accumulation in fatigue is usually a complicated mixture of several different
mechanisms, and the assumption of linear damage accumulation inherent in Miner’s law should be

viewed skeptically. If portions of the material’s microstructure become unable to bear load as fatigue

progresses, the stress must be carried by the surviving micro-structural elements. The rate of damage

accumulation could drop during some part of the material’s lifetime. Miner’s law ignores such effects,

and often fails to capture the essential physics of the fatigue process

With knowledge of the load spectrum and the tolerable material load in the form of the S – N curve, a

lifetime prediction can be made for a mechanical/civil system with the help of a damage accumulation

hypothesis. Here it should be considered, that this prediction can only be made with a certain

probability, since among other things the load spectrum as well as the load capacity expressed in the

form of S – N curve are random variables. Likewise, the damage accumulation hypotheses known

today have only been proven empirically in material science. Therefore, a practical lifetime prediction

requires balance field tests, test stand trials, calculation and a careful assessment and evaluation of

the data, if the prediction should be able to serve as an effective tool for the designer.

4 3
Example 5.1 Stress σ1 has lifetime N1 = 10 cycles, and a more rigorous stress σ2 has lifetime N2 =10

cycles. If 700 cycles at stress σ2 is operated, when will it stop to operate at stress σ1?

Solution) From Palmgren-Miner’s Law Eq. (5-61), we can calculate the cycles to fail.

700 x
+ =1
1000 10000

So the expected failure cycle is x = 3000 cycles

Example 5.2 A part is subjected to a fatigue environment where 10% of its life is spent at an

alternating stress level, σ1, 30% is spent at a level σ2, and 60% at a level σ3. How many cycles, n, can

the part undergo before failure?

 If, from the S-N diagram for this material the number of cycles to failure at σI (i=1, 2, 3), then from the

Palmgren-Miner rule failure occurs when:

0.1n 0.3n 0.6n

+ + =1
N1 N 2 N 3

so solving for n gives

1
n=
0.1 0.3 0.6
+ +
N1 N 2 N 3

If N1, N2, N3 are 103,104, and 105, the time to failure n will be 7353 cycles

REFERENCES

[1]. Matsuishi M & Endo T (1968) Fatigue of metals subjected to varying stress. Japan Soc. Mech.

Engineering.

th
[2]. Mott & Robert L. (2004) Machine elements in mechanical design (4 ed.) Upper Saddle River NJ:

Pearson Prentice Hall pp. 190–192

[3]. Palmgren AG (1924) Die Lebensdauer von Kugellagern Zeitschrift des Vereines Deutscher

Ingenieure 68(14): 339–341

View publication stats