j. al-rafidain engineering vol.18, no.3 (2010)

2

Al-Rafidain Engineering Vol. 18 No. 3 June 2010

ENGLISH SECTION

CONTENTS

No. Title PageNo.

1. Analysis And Simulation Of Coherent Antenna ElementsAnd Focuses.Khalil H. Saydmarie , Eanass U. T. Al-Shabkhoon.

1

2. The Effect of Grain Boundary on the electrical andphotoelectrical characteristics of Au/p-Si Schottky Diode.Khalid Khaleel Mohamed.

10

3. Improved Performance of ZnO/n-Si Solar CellsAkela M. Al-.Khalid Khaleel Mohamed.

19

4. Reactive Power Control of an Alternator with StaticExcitation System Connected to a NetworkOmar HazimDhiya Ali Al-Nimma , Majid Salim Matti

29

5. Design and Simulation of an Optical Gigabit Ethernet.Salah A. Jaro Alabady, Omar Ahmed Yousif

46

6. Segmentation of Conversational Speech Using ProbabilisticNeural Network.Ahmed Maamoon Alkababji

62

7. The Effect of ambient refractive index on the action ofLong Period Fiber Grating.Furat.y.Abdul-Razak.

71

8. Cylindrical Manipulator Path Planning Among StaticObstacles Using Artificial Potential Fields.Rawand Ehsan Jalal.

82

9. An Experimental Study of Parameters Affecting a Heat PipePerformance.Hussain H.Ahmad, Raqeeb H. Rajab

97

Saydmarie:Analysis And Simulation Of Coherent Antenna Elements And Focuses

1

ANALYSIS AND SIMULATION OF COHERENT ANTENNAElements and Focuses

A continued research in a part of “Focusing Feasibility in Fractal Array Antennas” [1]

Prof. Dr. Khalil H. Saydmarie Dr. Eanass U. T. Al-ShabkhoonDept. of Communication Engineering Dept. of Med. Instrumentation Eng.

College of Electronic Engineering Technical College / Mosul / Iraq University of Mosul / Iraq. Foundation of Technical Learning

Abstract

The current research is a result of the study in fractal dimension of antenna application, but itbecomes in out of line. The objective of study was to full-fill the space of near field region by acumulative electromagnetic pixel at focal point; as a result the space was full-fill by multi-pixels knownas coherent focuses.

Coherently between the rays reaching focal point leads to focusing. One of the studied results“Focusing Feasibility in Fractal Array Antennas”; is the relation between the quadratic distancesbetween the elements of focused array and the appearance of coherent focal points. The simulation,analysis, and proven include a deal with the essential mathematical relations which leads to concludea number of new forms likes, definite focal region, definite coherent focal region, predicate of co-herent focal positions.

"]"1[

..

/ / /

" " , -

.

":" .

.

Received 10/5/2009 Accepted 31/8/2009

Al-Rafidain Engineering Vol.18 No.3 June 2010

2

1. Introduction:

Electric field can be estimated as a function of current distribution across various shapesof conductive materials. Distribution of electric field is affected by antenna geometry, distancefrom antenna, physical properties of media, antenna material, and the signal of excitation. Fieldof point source is used to estimate electric field for an extended antenna or for an array antenna.The solution that is applied here to describe the variation of electric field at certain focal distancefrom a point source is given in the following equation:

and for array antenna of perpendicular polarization [Steinberg2] is:

where:E: Electric field [V/cm]

: represent phase constant [rad/cm].: represent attenuation constant [Neper/cm].

r: represent the distance between observation point and the isotropic source or space radius [cm].an: represent a product factor (usually known as element factor), this factor includes the effects ofexcitation, mutual coupling, physical properties of media such as impedance, and others.Solution like that is acceptable from scientist whom interest in electromagnetic theory since it iscompatible with physical truths, where the relation between instantaneous variation of currentand gradient field intensity had been found. The intensity of field is attenuated due to surface areaof vacuum and due to attenuation factor of medium.The geometrical optics solution is applied to solve problem of propagation. Solution is based onray tracing from source till observation point. Results are affected by previous distances andconstants.

Equation (2) is used to describe field variation for two regions:1. Near field region2. Far field regionThese two regions are defined according to the longest dimension of antenna L. Classification ofthe two main regions is based on describing the distance r by a convergence series [Balanis3]. Asa result, polynomial of phase includes: linear, quadratic, cubic,……etc phase factors. The higherdegree terms are usually be neglected when the sum of phase factors is less than /8. So thedifferences between describing the electric field in near field region and far field region isassuming as a linear curve varied with an element position , and assuming that the

variable r having a certain value assigned by ro where ro represents the central distance betweenthe original point of antenna layout and the observation point.


3

[Steinberg2] uses another way differs from [Balanis3] to define the cutoff plane between far fieldregion (Fraunhofer region) and near field region (Fresnel region). The first one take the valueof phase in term of distance that provide a value of one from (Cornu Spiral), then he takes thetwice value of distance to define far field region, so he reached the same result that [Balanis3]reached.Notice that [Balanis3] defined a third region which is known as an active region and bounded by( ).

Focal point will be observed as well as all radiation rays from different direction sourcesreaching it coherently [4], but there is a lack of knowledge about focal region limitation before.There is an ability of locating focal point in nearest region, but it cannot be moved toward zero asit was proved by a greater number of analyses [1]. All types of linear antenna fail to move focalpoint to zero. The reason of that failed will be discussed here in details.

2. Principle of Study:The research is based on locating virtual isotropic source at focal point, and it is based on

describing distances as in figure (1), in summery there are:1. Virtual geometrical distance, which is limited by a visible region of isotropic source with

respect to array antenna.2. Approximated distance which is based on resolved the actual variation form into first linear

and second quadratic “which it has the most significant influence in near field region”polynomial terms.

3. Complement distance, which is used to compensate the phase differences that are introducedfrom the distances differences between focal point and the element of array antenna.

Now the condition of hypothesis can be listed as follows:1. The length of array antenna cannot exceed a visible project of virtual source.2. The quadratic distance should not exceed the virtual distance from central point of array

to the virtual source.3. The conjugate method should be applied, since the assumption deals with virtual isotropic

source locating at focal point, and providing conjugate phase differences.

The conditions and assumptions are mentioned to define a limited range of the nearest focalpoint, where the third condition is for application, while the first and the second are competitiveconditions used to employ the following criterion:

Focal point cannot be observed for a distance shorter than the half length of straight lineantenna divided by square root of two.

Referring to far focal point that is defined according to [Stienbergs’2] note about integrating tofind near field from (Cornu Spiral). That note about fading after the distance (L2 ), losses afterthat distance is extremely steady so it is defined here as a far focal point. Notice that far fieldregion does not mean you must have a point of localization absolutely, it means you will have afocal plane in which electric field is distributed in form of lobes (main, minors, and side lobes). Itcan be said that far field region follows far focal region, in the other word:

Focal point can be observed at a distance not greater than the quadratic length of antennadivided by two.


4

Actually many types of linear antenna fail to move focal point to zero as was submittedby the studied [1] and as it is listed in table (1).

3. Design of Coherent Antenna (Elements and Focuses):

The concept of design coherent antenna is based here on selecting elements position thathaving a multi-wavelength of propagation rays; it means that the assum-ption of coherency canbe verified by selecting element distance at:

Locating the element of array at squared distance of dn will never change the principle ofconjugating where the squared value of integer wavelength leads directly to a 2pi’scomplementary phase. So the resulted radiation pattern in near field region looks like far fieldpattern as one can see in figure (2-1), color map is used to represent intensity level of normalizedarray factor across xy-plane.

ANTENNASTYPE

ELEMENTSPLACEMENT ALONG

X-AXIS

PHASEEXCITATION

FUNCTION

ANTENNALENGTH

NEARESTFOCALPOINT

Traditional ± [0, 2.5,5,7.5,10] Conjugate 20 7

Binary ± [ 0, 2.5, 5,7.5,10] Conjugate 20 7

Cantour ± [0, 2.5, 3.75, 5] [010 111 010]pi’s 10 3.5

Table (1): Nearest Focal Point for One Dimensional Array [1] (scale in cm)

-1.5 -1.25 -1 -0.75 -0.5 -0.25 0 0.25 0.5 0.75 1 1.25 1.51.50

0.25

0.5

0.75

1

1.25

1.5

Figure (1): The Description of Array Antenna Layout (scale in cm)

Geometrical Distance

x

virtual isotropic source

Quadratic Distance

Complement Distance

Fp

x

y

z

r

Coordinate

Observation point

visibleregion of the virtual

isotropic source

Elements Placements along x-axis

Nearest focal region


5

Focused antennas array of coherent placement is defined as:Focused Coherent Antenna.

Antennas array of coherent placement and focuses is defined as:Coherent Antennas Array.

The array antenna is designed to have 8-cm length, four elements which are located at thepositions of eq.(3), and a desired focal point located at 1-cm far from the antenna. Antennas arrayare excited uniformly by a 3GHz signal to generate far field distribution as mentioned previously.Far distribution is shown in figure (2-1) in a media of ( r=100). Bipolar phase excitation isapplied then to form coherent focal point. Array factor or normalized field intensity is foundalong on-axis as can be seen in figures (2-2), (2-3), and (2-4). Field distribution is found far from(2L2 ) as in figure (2-5). Note that coherent focal can be observed clearly in far region only. Theexpected reason of that is the decay factor (1/r), so it’s effect is separated and studied alone,where previously the effect of element placement was only discussed by [1], so coherent focalregion is defined in condition that (1/r) is considered to be steadies for acceptable error factor (v).The slope of such curve goes to zero as well as the area reach v, and as a solution:

Let , then the result of integral can be easily found as in the follows:

Field distribution is found for (v=0.1) and ( r=6cm) to observe coherent focal points as infigure (2-6) in which central focuses having lowest intensity with respect to side focal points.Intensity faded may be caused from the synchronization between the electrical length of rays andcos( /2).

4. Analysis of Results:

Analysis is basically depended on how one looks to the mathematical form of distancevariation and how one compares them with standard forms. Analysis is begin by finding theoriginal distance of element number n.Element distance rn can be found in term of element place Dn as follows:

Let ro refers to original distance(x2+y2), then coherent distance of elements rcn can be found interm of element place Dn as follows:


6

But

Final manipulation is considered as mod. additive where any displacement of phase by a multiplevalue of 2pis will not affect the absolute value of array factor. Antenna like that has a sinusoidalelectric field function and has focal points separation equal half wavelength along on-axis(central-axis). Antenna is considered to be symmetry about y-axis but out of phase, so it can beproven that:

0.05

0.1

0.15

0.2

0.25

Figure (2): The Normalized Field intensity of antennas array [-8 -3 0 3 8] where;

1- Electric field across xy-plane of uniform excitation function.2- Electric field along central axis of bipolar excitation function.

3- Zooming central axis of (2) for (y>F1{L2/ })4- Zooming central axis of (2) for (y>F1{2L2/ })

5- Electric field across xy-plane of bipolar excitation function.

y110

120

130

140

150

160

170

180

190

200

0.02

0.04

0.06

0.08

0.1

0.12

0.14

0.16

0.18

131

132

133

134

135

136

137

138

139

140

53

54

55

56

57

58

59

60

Fade

d

Cohe

rent

Foca

l

1 2

3 4

5 6 xx

y y

yy

E E

E

x

yE

EE


7

Applied equation (13) cannot be fully successful in near region while the results in farregion are identical with the result of the basic equation as can be seen in figure (3) and as can becompared with figure (2-2), where the possibility of accuracy error is neglected. Error like thatwill not cause to generate a new curve, but it cause curve distortion as can be seen in figure (4),where a random variable (equal 1% of wave-length) is added to the electrical length of rays.The differences between figure (3) and figure (2-2) are related to:

Now let us rewrite the condition of far focal region in term of phase shift ( ), the relatedphase of far focal distance is /2, so for all array elements this phase must exceed /2, thencritical case can be defined as follows:

Then faded point is observed. So one should retain again to the definition of distance r in term ofpolynomial equation to find the region of faded points:

The third term what is caused field intensity decreasing till zero for any point on central axis.Faded region is calculated according to:

The longest ro which Fd is obtained from it, is calculated for maximum value of n equal N/2.Faded region will not be observed by any why far (m2 ) which is equal 64-cm here.Referring to result, it can be found that the faded region had been displaced towards antenna,where field intensity in near region does not depend on the sinusoidal function only, but it isaffected by distance attenuation (1/r) also which causes the displacement.

40 40.5 41 41.5 420.69

0.7

0.71

0.72

0.73

0.74

0.75

0.76

0.77

Unr

ealiz

able

Regi

on

Figure (3): Normalized electric fieldintensity along the central axis andaccording to equation (13).

Figure (4): Field distortion due to amultiplicative random noise. Electricallength is affected by the error factor.

Distortion

y [cm] y [cm]

E [V/cm] E [V/cm]


8

5. Conclusions:The importance does not limited by the obtained coherent distribution, but it may be extended,for example distribution like that seems to be uniform distribution if the wavelength is shorterthan cross area section. The following notes are interested also:

1- The length of desired antenna should be longer than one wavelength. Desired focal pointshould be placed at a distance length equals one wavelength.

2- Field distribution will not be affected so much if a desired focal point is located at a onewavelength or at multiple wavelength distance, but the length of antenna could be shorter.

3- Field distribution may be considered as a coherent or a multi-point field according to thescanned area.

4- Coherent may be used for scanning by full-fill the vacuum by multi-small focal (pixels) orto generate a stamp or alarm signal for the base station of cellular system using a singlefrequency band.

5- Color distribution may differ slightly depending on the digital processor and the intensitylevel of color map.

6- Classification of distances should be taken into account, where if influence of (1/r) factoris neglected by any design procedure in near field, focal point will displace.

7- Focal point is important to initiate what is recently known as target capture.8- The approximate separation between two coherent focuses is half wavelength.9- The approximated distances of polynomial form seem to be preferable in order to sub-

dividing the curve of complex equation, then find a typical solution for each partaccording to following regions:a- Selectable focal region:

b- Expectable faded region:

c- Focal region:

d- Coherent focal region:

Appendix [Proven of nearest focal distance criteria]:

Virtual distance must equal Fp for a point or isotropic source (condition number one in sectionnumber two).Quadratic distance is defined by [Stienberg2] and it is applied here in term of focal distance as inthe follows:

Maximum value of quadratic distance must be given in term of antenna length:


9

Condition number two ( ) in section number two is applied here to comfort the reality ofdistance variation, as a result:

List of Symbols:

Acknowledgment: Thesis[1] discussion committee (2006-2007) for their invaluable suggestionand for who interested in the idea of research in technology culture in technical college (2007-2008).

References:[1] E. U. T. Al_Shabkhoon “Feasibility of Focused Fractal Array Antennas” Ph.D. Thesis,

College of Engineering, University of Mosul, (2006-2007).[2] B. D. Stienberg “Principle of Aperture and Array System Design”, John Wiley and

Sons, Inc, New York 1972, Chapter 1, Page (3-23). [3] C. A. Balanis “Antenna Theory Analysis and Design” Harper and Row, Publisher, Inc.,

1982, Chapter 4, Page (100-159)[4] J. Loane, H. Ling, B. F. Wang, and S.W. Lee “Experimental Investigated of Retro-

Focusing Microwave Hyperthermia Applicator: Conjugate-Field Matching Scheme”IEEE Transaction on Microwave Theory and Techniques, Vol MTT-34, No.5, May 1986.

SYMBOL

GREEK NAME PHYSICAL MEAN UNIT

an --- Element factor Vdn --- Coherent separation (first order) cmDn --- Coherent separation (second order) cmE --- Electric field V/cmFp --- Focal point or central distance cmL --- Antenna Length cmm --- Mode number constantn --- Index number of array elements constantN --- Total number of array elements constantro --- Radius of spherical coordinate cm

Alpha Attenuation constant Neper/cmBeta Phase constant rad./cm

Lambda Wavelength cm

The work was carried out at the University of Mosul

Mohamed: The Effect of Grain Boundary on the electrical and photoelectrical ------

10

The Effect of Grain Boundary on the electrical andphotoelectrical characteristics of Au/p-Si Schottky Diode

Dr. Khalid Khaleel Mohamed

Electrical Engineering

Mosul University, Iraq

Abstract

This paper is intended to study the influence of the grain boundaries on the electronicand optoelectronic behavior of Au/P-Si Schottky diode. These diodes were fabricated byevaporation of gold layers onto polycrystalline silicon wafers using vacuum evaporationtechnique. The current-voltage characteristics at different grains boundary and temperatures,spectral response were investigated. It is found that the Schottky barrier height for Au/P-Sidiode obtained form I-V and spectral response characteristics are depends mainly on thesurface grain boundary density and state density.

Keyword: Grain Boundary, Au/p-Si, Schottky Diode.

Au/p-Si

.

/

P-Si

Au/Si-P . P .

. Au/Si-P .



11

1- Introduction:

A metal / semiconductor (Au/P-Si) is a very interesting model for understanding themetal / silicide formation. It's properties are important for both the fundamental andtechnological points of view, especially as metal thin film deposited on Si at temperature wellbelow the processing temperature for Si devices. Also polycrystalline silicon is one of themost promising materials for the realization of low-cost solar cells for terrestrial applications.The physics of the polycrystalline grain boundaries has a great influence on the photovoltaicproperties of the solar cell may be assessed. Most of the researches performed using themetal-semiconductor (MS) and metal-insulator-semiconductor (MIS) do not relate theelectrical and optoelectrical behavior to the structural features of the substrate.

This paper focus the attention on the influence of grain boundaries on the experimentalelectrical and optoelectrical properties of Au/P-Si Schottky barriers. The schottky barrier maybe used as an experimental toll to study the nature of the grain boundaries, also schottky-barrier solar cells may be the best way to reduce cost in device fabrication.

In polycrystalline Schottky diodes, the average grain size of the substrate has adominant effect on the ultimate efficiency since the grain boundary contributes to minoritycarrier recombination that reduces the photo generated current. Also the current conductionmay change from Schottky barrier to bulk limited transport at smallgrain size [1]. For a Schottky barrier made on a single crystal with the energy band diagramshown in figure (1).

The interface state charge density is given by [2].BnogSSS qqEDqQ … (1)

Where:Ds: is the surface density of states.Eg: is the semiconductor energy bandgap.

o: is the neutral level and Bn is the barrier height.

o

m

x/q

Ohmic

contact

Egei

Bn

insulator

Ev

Ef

Ei

EcGold

inversion region neutral regiondepletion region

Qss

Silicon

eis

Figure(1): Schottky barrier band diagram for crystalline region


12

Neglecting the space charge in semiconductor, than the barrier height is givenby [2].

ocmBn qECxC /1 … (2)

S2

ii DqC … (3)

Where:m: is the metal work function.

x: is the electron affinity, andi: is the dielectric constant of the interfacial layer with a thickness .

When a Schottky barrier is formed on a Si surface, the grain boundary intersecting thesurface introduces a surface-state distribution DBS and neutral level BO as shown in Figure(2).

o S o B BS Bo S B BSD d D D d D … (4)The barrier height of Schottky diodes on polycrystalline silicon has the same form of

equation (2) but only with change of o and DS, therefore Bn is given by:

oqmBn qECxC /1 … (5)

and the total surface charge is given by

SS S g o on B BS g Bo Bnq D E q q q d D E q q … (6)

Where dB is the surface grain boundary density if a cubic grain is assumed [2] than:

SS S B BS g o Bnq D d D E q q … (7)

Where:

2 .i i BS B BSC g D d D … (8)

Bo

m

x/q

Ohmic

contact

Eg

B

n

insulator

Ev

Ef

Ei

EcGold

inversion neutraldepletion

QBs

Silicon

Figure (2): Schottky barrier band diagram for surface grain boundary


13

Therefore the incremental change of the barrier height due to the surface grain-boundarystate is given by (Fig. 2).

BnBnBn

2 2B BS i S B BS Bn q Boq d D q D d D E q … (9)

For small grain size 2B BS S id D D q , the Fermi level is pinned to the grain

boundary neutral level and the Schottky barrier is equal to Eq/q- Bo [3]. This means that theSchottky barrier is the same as the bulk potential spike due to the grain boundary for smallgrain devices, therefore the neutral level and the barrier height may depend on the grainboundary intersecting the Schottky barrier interface.

In this paper the Au/p-Si schottky diod were fabricated using vacuum evaporationtechnigue with different number of grain boundary. The electrical and optoelectricalcharacteristics are studied and the barrier height for different grain sizes were calculated.

2- The Au/P-Si Structure Fabrication:

The fabricated samples were prepared by vacuum evaporation technique using Balzerunit as a coating system. Small pieces 1.5 cm2 of P-polysilicon were cut from silicon waferswith different grain sizes. The silicon wafers are subjected to a rigorous cleaning cycle inthree steps, in order to reduce the pin hole formation [4]. A P-polysilicon wafer with thethickness of 300 m and resistivity of 4.5 ohm. cm were used. The samples were cleaned withethyl alcohol to remove organic residues. Then they were stored and protected fromatmospheric contamination in vacuum desiccators. Aluminum thin film (2000A ) weredeposited as a back contacts for the fabricated samples at pressure of 10-6 torr, the sampleswere heated under vacuum up to a temperature of 350 C for half an hour. This heat treatmentis necessary to obtain an ohmic contact between the aluminum and the wafers[5]. Thesamples were then coated with 500 A thick gold layer at a pressure of 10-6 torr at differenttemperature ranging from 100 – 500 C. The electrical measurements were performed usingconventional dc techniques, and the I-V characteristics for different samples were measured atroom temperature.

3- Results and Discussion:

The experimental I-V characteristic of the Au/P-Si Schottky diode structure as afunction of grain boundary is shown in figure (3).

It is clear that the grain density has a great effect on the value of barrier height of theAu/P-Si Schottky diode, the increasing of barrier height with the increasing of grain boundaryis attributed to the splitting of the quasi Fermi-levels, which lead to change the interface-stateoccupancy [5].The grain boundary contributes the minority carrier recombination ,whichreduces the generated current across the junction.

A more detailed display for forward charactristic is shown in figure(5) as a semi-logplot.

The forward current is given by [ 6].1KTnqvexpJJ o ….(10)KTqTAJ Boexp2 exp(qV / KT ) ….(11)


14

Where:

Jo: saturation current density and is given by:KTqexpTAJ Bo

2o ….(12)

KTqexpTAJ Bo2

o exp(qV) ….(13)KTqexpATJ Bo

2o ….(14)

n: is the ideality factor.a: is the Richardson constant.J: the actual current density.

0 0.2 0.4 0.6 0.8 1 1.2 1.40

0.2

0.4

0.6

0.8

1

1.2

1.4

VOLTAGE (V)

CU

RR

ENT

(mA

)

53050

Figure (3): I-V Characteristics as a function of grain boundary at

temperature (300) C

0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 1.1-6

-5.5

-5

-4.5

-4

-3.5

Voltage (V)

LOG

I(A

/cm

2)

53050

Figure (4):The forward I-V characteristics of Au/p-Si structure for grain boundary =5,30,50


15

S: the area of the diode Neglecting the series resistance then the resulting forward current is:

KTAqvJInJIn o ….(15)

The value of the barrier height is estimated from the forward characteristics and foundto be 0.74 , 0.7 , 0.64 eV for grain size boundary 5 , 30 , 50 respectively.The effect of annealing temperature of gold layer on the I-V charactristic of Au/p-polysiliconis shown in figure (4).

The increasing of current with the increasing of temperature can be attributed to theincrease of surface recombination velocity of the gold silicide samples, while the band -to-band recombination life time deceasing. The surface and bulk recombination process hasincreased and the schottky curves were observed this increase for samples annealed at 300Cand 500C. The schottky curves at these annealing temperatures were formed due to theformation of Au7Si silicide[ 5].Figure (6)shows the reverse I-V characteristic for Au/P-Si structure.

10-3 10-2 10-1 100 101-5.5

-5

-4.5

-4

-3.5

-3

-2.5

-2

Voltage (V)

LOG

I (A

/cm

-2)

a

b

c

Figure (6): The reverse I-V characteristics of Au/p-Si structure

0 0.2 0.4 0.6 0.8 1 1.2 1.40

0.2

0.4

0.6

0.8

1

1.2

1.4

1.6

VOLTAGE (V)

CU

RR

ENT

(mA

)

500300100

Figure (5): The effect of annealing temperature on the I-V characteristics at grainboundary =30.


16

It is noticed that the reverse I-V characteristic has three regions, (a, b, c) a linear region(a) for small voltages, and a region showing a tendency toward saturation but with someincrease of current, which can be attributed to the effect of generation-recombination currentand volume generated current[7]. . According to equations 12 , 13 and 14 , the barrier heightcan be found from the plot of Ln(J0/T2) against 1/T as shown in figure (7) , the plot is astraight line with the slope directly yielding the mean barrier height of thesample.

Figure(7): The plot of Ln J0/T2 against 1/T

It is found that the barrier height is about 0.76 , 0.72 , 0.65 eV for grain boundary 5 ,30 , 50 respectively. The estimated value of the barrier height from the forward and reversecharacteristics agree fairly well with each other.

4- Photo Measurements:

The barrier height of the Au/P-Si structure can be found by the measurement of thephoto response of the cell, usually a graph of the square root of the relative photo responseplotted against photon energy will gives a straight line. The intercept of the straight line onthe photon axis gives the metal-semiconductor work function. The spectral response of thephotocurrent has been measured as a function of wavelength in the range (0.2 < < 1.3) mas shown in figure ( 8).

The barrier height was found from the long wavelength side of the response curve byplotting the square root of the photo responsivity (R)1/2 against photon energy hv. , which canbe given by [8]

whereqhvBR ( )2 ….(15)

R is the photo responsivity and B is a constant.An extrapolation of the linear portion of this curve is called Fowler plot [8]and(R)1/2 =0 gives the barrier height.

1.6 1.8 2 2.2 2.4 2.6 2.8 3 3.2 3.4 3.6-11

-10.5

-10

-9.5

-9

-8.5

1000/T

Ln (

J0/T

2)

GRAIN SIZE=50GRAIN SIZE=30GRAIN SIZE=5

Ln(J

o/T2 )


17

The spectral response is normalized to it's maximum value after correction for spectralresponse distribution of the illumination setup. The barrier heights is determined from theintercepts of the obtained straight line with x-axis. Its found that the barrier height is about0.81 , 0.78 , 0.76 eV for grain size boundary 5 , 30 , 50 respectively. The contribution of grainboundary states is clearly seen in grain boundary density dB and state density DBS. Thereforethe neutral level and barrier height may depend strongly on the grain boundary intersectingthe Schottky barrier interface. Also it is noted that the barrier height calculated from the I-Vcharacteristics is lower then that calculated from photo measurement and this is due to thethick front top contact during the I-V measurement , which prevent the light penetration intothe silicon wafer directly under the contact and there will be a reduction of barrier potentialdue to splitting of the guasi Fermi level.

5-Conclusions:

The fabrication of Au/P-polycrystalline silicon Schottky diodes is performed, usingvacuum evaporation technique. The surface features of the fabricated diode has a greatinfluence on it's I-V characteristic. It is found that the variation of Schottky barrier height onpolycrystalline silicon depends on the surface-grain-boundary density and state density.

6- References:

1- NEAMEN, Semiconductor Physics and Devices, (C) Richard D. Irwin, INC., 1992.2- Shewchun J. S., Singh R. and Green M. A., "Theory of Metal-Insulator-

Semiconductors Solar Cells", J. Applied Physics. Vol. 48, No. 2, (USA), (1977).3- Todorovic D. M. and Smiljanic M., "Theory of Photoacoustic Effect in Metal-

Semiconductor System", Institute for Chemistry, Technology and Metallurgy,Njegoseva 12, 1100. Belgrade, Yugoslavia, (2001).

4- Djoko I. and Hartano H., "Ohmic Contact Schottky Barrier", University of Indonesia,Vol. 18, pp. 74, (USA), (2004).

0.5 1 1.5 2 2.5 30

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1

Photon Eneregy (eV)

Sgua

re ro

ot o

f Res

pons

ivity

53050

Figure(8):The Photo response of the Au/p-Si Structure at 5,30,50 Grain Boundaries

R1/2


18

5- Yap Siew Hong, Carrier Transport and I-V Characteristic of Au/Si Silicodes UsingOpen Photo-acoustic Cell, Solid State Science and Technology. Vol. 13, No. 1 and 2,287-295, (2005).

6- D. Derkacs, S. H. and E. T. Yu, Improved Performance of Amorphous Silicon SolarCell Via Scattering from Surface, Applied Physics Letters 89, 093103 (2006).

7- Milnes A. G. and Feucht D. L., "Hetrojunction and Metal Semiconductor Junction",Academic Press, Vol. 49, pp. 133-149, (London), (1972)

8- Dieter K. Schroder “ Semiconductor material and codevice characterization “pyright (2006 ) by John wiley and sons.

9- C. Lanza and H. J. Hovel, IEEE Trans. Electron Devices, Vol. ED-24, 1977

The work was carried out at the college of Engg. University of Mosul

Mohamed:Improved Performance of ZnO/n-Si Solar Cell

19

Improved Performance of ZnO/n-Si Solar Cells

Dr. Khalid Khaleel Mohamed

Electrical EngineeringMosul-University, Iraq

Abstract

This research is intended to improve the performance of ZnO/n-Si solar cells. Thestructures were fabricated using thermal evaporation techniques. The indium dopant atsuitable heat treatment is used to enhance the electrical characteristics of ZnO layer resultingin reducing atmospheric condition to change the stiochiometry of ZnO layer. The electricproperties of the fabricated samples are dependent on many parameters such as annealingtemperature, ZnO layer thickness, Indium layer thickness and temperature. The indium layerwere deposited at different thickness (10-30) nm during the fabrication of the ZnO/n-Si solarcells. The resultant samples has been studied and the results obtained show an improvementin the efficiency of 0.4% compared with the standard ZnO/n-Si solar cell.

Keyword: ZnO/n-Si, Indium, Thermal Evaporation.

ZnO/n-Si

. /

ZnO/n-Si .

ZnO/n-Si .

.(10-30)nmZnO/n-Si In- In-ZnO/n-Si

0.4% .ZnO/n-Si



20

1-Introduction:

Compound semiconductor and their alloys are of considerable interest in the field ofelectronic and optoelectrical device. ZnO film is receiving increased attention for variousmicroelectronic applications. It has potential uses in photo detectors, solar cells and lightemitting diodes [1]. ZnO is a II-VI compound semiconductor with a wide direct band of 3.3ev (at room temperature) and has a hexagonal quartzite structure with cell parameters of a =0.325 nm, c = 0.5206 nm [2]. Zinc oxide has emerged as one of the most important windowmaterials due to its large bandgap and used in photovoltaic especially in large area solar cells[3]. The high production cost of the conventional solar cells require the search for cheapermethods suitable for solar energy conversion. ZnO thin film have been prepared by widevariety of techniques such as sputtering, evaporation, chemical vapor deposition (CVD) andspray pyrolsis [4].

The large excitation binding energy (60 mev) of ZnO leads to the existence and extremegrowth of high-quality ZnO film on Si. The main obstacles to get high-quality ZnO on Si areto overcome the large lattice mismatch and to avoid the amorphous SiO2 layer generated atthe Si surface prior to or during ZnO growth, since Si can be easily oxidized in the oxygenenvironments [4]. To avoid these problems, some efforts have been made to use ZnS, Znmetal layer, and nitridation of silicon surface [5].

In this paper ZnO films were deposited by thermal evaporation technique on the Sisurface, followed by deposition of indium thin layer with variable thickness. Then a thin layerof indium tin oxide is deposited to suppress oxygen diffusion to Si wafer. The effect of ZnOlayer thickness, indium layer thickness, annealing temperature, deposition rate are measuredand analyzed.

2-Experimental Work:

The fabricated samples were prepared by vacuum evaporation technique using Balzerunit as a coating system. Phosphorus doped two-inch diameter monocrystalline n-Si waferswith thickness of 300 m and resistivity of about 4.5 .cm, oriented in (100) plane were used.They were cut into relatively small segments (4cm2 ) to be used as base semiconductor for thefabricated solar cells samples. The samples were cleaned with ethylalcohol to remove organicresidues and then etched in buffered hydrofluoric acid for 2 minute to remove oxide films,then they were stored and protected from atmospheric contamination in vacuum desiccators.

A ZnO thin film (50-100) nm were deposited on the front surface of the silicon wafers,then indium thin film (10-30) nm were deposited on the front surface of the ZnO/n-Sistructure as shown in figure (1).

n-Si

Substrate

Indium Layer

ITO Layer

ZnO Layer

AL Contact meshLayer

AL Back ContactLayer Figure (1): The In-ZnO/n-Si Structure


21

The In-ZnO/n-Si structure were annealed in the temperature range of(200-500 C ) for 20 minute at pressure of 10-6 torr. A thin layer of indium tin oxide (ITO)was deposited on In-ZnO/n-Si structure to suppress oxygen diffusion from the environment,ITO is highly transparent in visible region and has high electrical conductivity and can beused as a front contact for the cell. The fabrication process steps of the In-ZnO/n-si solar isshown in figure(2).

Figure.(2): The general processing for the fabricated In-ZnO/n-Si solar cells

Si Wafercleaning

ZnO deposition thickness(50-100) nm

In layer deposition thickness 30-50 nm

Annealing in vacuum (200-500) C, 20 min.

ITO deposition50 nm

Removing SiO2 from siliconback

Front contact metallization

Back contact metallization

Test

ZnO thin filmtest

In-ZnO thin filmtest


22

The back side of the samples was exposed to 10% HF acid to remove the silicon dioxidelayer, the back contact was formed by the deposition of Aluminum layer (150 nm) at the backside of the fabricated samples at 10-6 torr and annealed at 400 C for an 15 minute. This heattreatment is necessary to obtain an ohmic contact between the Aluminum and the siliconwafers [6].

3-Results and Discussions:

3.1 The transmittance spectra of ZnO thin films:

In order to study the effect of ZnO layer thickness on the properties of the fabricated solarcell, the ZnO thin films were deposited on a glass slices by vacuum deposition so the opticalcharacteristic can be studied and correlated with preparation parameters to optimize the bestconditions fitting the solar cell performance. The transmittance spectra of the ZnO thin filmswere measured using a photo spectrometer as shown in figure (2).

Figure. (2) The transmittance spectra of ZnO thin films.

It is noticed that all obtained films exhibit a high transmission (85%) in the visibleregion with sharp absorption edge at 390 nm.

The energy band gap of the ZnO thin films is about (3.31) eV and the refractive index is1.8 as measured by the computerized photo spectrometer. The values enable the ZnO filmsacting as window and antireflection materials.

3.2 The transmittance spectra of In-ZnO thin films:

The optical characteristics of the In-ZnO thin films is also measured using acomputerized photo spectrometer with different indium thickness in order to obtain the bestperformance of the fabricated In-ZnO/n-Si solar cell. The transmission spectra of the In-ZnOthin films is shown in Fig. (3).

200 400 600 800 1000 1200 14000

10

20

30

40

50

60

70

80

90

100

Wave length (nm)

Tran

smitt

ance

( 0/

0)

50nm80nm100nm


23

It is found that the In-ZnO films exhibit also a high transmission 83% in the visibleregion with sharp absorption edge at 390 nm. The energy band gap obtained from thecomputerized photo spectrometer is about 3.19 and has not change much by the indium layerand the refractive index is 1.77, this value is comparable to the value obtained from the ZnOthin films and enable the In-ZnO films acting as window and antireflection coating materials.The I-V characteristics of the. In-ZnO/n-Si samples were measured under dark andillumination condition respectively as shown in Fig. (4).

-2 -1 0 1 2 3 4 5-1.5

-1

-0.5

0

0.5

1

1.5

2

VOLTAGE (V)

CU

RR

ENT

(mA

)

DARKILLUMINATED

Figure (4): I-V characteristics of the In-ZnO/n-Si samples under dark andillumination condition .

200 400 600 800 1000 1200 14000

10

20

30

40

50

60

70

80

90

100

Wave length ( nm )

Tran

smitt

ance

( 0/

0 )

ZnO onlyIn=30 nmIn=50 nm

Figure. (3) The transmittance spectra of In-ZnO thin films.


24

The forward current increases superlinearly as the applied voltage increases. When theapplied voltage is reversed, the reversing current rises slowly and linearly with increasingvoltage. The curve is similar to a typical I-V characteristic of semiconductor diodes. It isnoticed that the reverse current for the illuminated samples keeps about 30 A for zeroapplied voltage. The reverse current increases strongly for large reverse voltages and tends tobe saturated at about 400 A. The effect of ZnO layer thickness on the generated current solarcell at AM 1.5 is shown in figure Fig. (5).

It is noticed that the photo generated current is increased with the increasing of ZnO layerthickness, the photo generated current starts decreasing gradually at 80nm thickness for thesample annealed at 400Co, while the photo generated current for other samples startsdecreasing at 70nm thickness. The decreasing of the photo generated current can be attributedto the decreasing of the fill factor, the series resistance increased at large value of ZnOthickness layer and tend to decrease the photo generated current. Also it is clear that theannealing process of the ZnO layer has improve the surface roughness, and the growthcondition for buffer layer is optimized to have the best characteristics at 400 C .

The effect of indium layer thickness on the photo generated current is shown in fig.(6).

Figure (6): The effect of indium layer thickness on the

10 15 20 25 30 35 40 45 508

10

12

14

16

18

20

Indium layer Thickness (nm)

Gen

erat

ed C

urre

nt (m

A/c

m2)

400 c500c200c

20 30 40 50 60 70 80 90 1002

4

6

8

10

12

14

16

18

20

ZnO Layer Thickness (nm)

Gen

erat

ed C

urre

nt (

A/c

m2)

400 C500 C200 C

Figure (5):The effect of ZnO layer thickness on the generated current at AM 1.5


25

The maximum generated current is occurred at indium thickness of about 30 nm, at thisvalue of thickness the band gap of InZnO layer are large enough to be transparent to most ofthe useful solar spectrum and the resistivity is small enough to avoid series resistance effect.

The normalized photo current response of the InZnO/n-Si solar cell at differentannealing temperature and with ZnO layer thickness egual to 80nm and indium layerthickness egual to 30 nm is shown in Fig. (7). The upper curve shows the photo response ofIn-ZnO/n-Si solar cell at 400co ,while the second curve is the photo response of ZnO/n-Sisolar cell with out indium layer, so the indium layer has enhance the current response of the

Figure (7):Normalized photo response of the In-ZnO/n-Si solar cell

fabricated samples at wavelength 500-1000 nm.It is clearly found that the strong photo response is shown at 390 nm wavelength due to

the existence of ZnO film, the strongest response corresponds to very high energy of photons.This result indicate that a large number of photo electrons occur when the sample isilluminated by light with short wavelength. The strongest absorbing transition occurs at thislevels which generates a lot of electrons in conduction band. The photovoltaic parameter of In-ZnO/n-Si solar cell is shown in the table below.

Measurement DescriptionsVO.C(V)

IS.C(mA) F.F Barrier Voltage

( )(V)

Parameter taken after fabrication. 0.49 18.8 0.62 5.7 0.69

After one month. 0.489 18.78 0.619 5.68 0.68

After six month. 0.488 18.77 0.618 5.66 0.670

It is found that the fabricated In-ZnO/n-Si solar cell has an efficiency of 5.7 measured atAM 1.5, the cell output parameters stability is improved using the ITO layer at the front sideof the In-ZnO/n-Si solar cell structure. The oxygen tends to diffuse through the ZnO layer toSi substrate allowing more degrading of cell performance. The built in potential of the solar

200 300 400 500 600 700 800 900 10000.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1

Wavelength(nm)

Phot

o re

spon

s

400c with indium500c with indium200c with indium400c with out indium


26

cell is measured using the capacitance-voltage (CV) measurement. This CV characteristics areuseful to find the donor concentration (ND), the built in potential can be estimated from theslope and the intercept of 1/C2 versus reverse voltage. The barrier height b is related to builtin potential Vbi by the following formula:

fcbib EEVq … (1)

Ec-Ef is the difference between conduction band and Fermi energy.

D

cfc N

NInq

KTEE … (2)

Where:

Nc = Effective density of states in conduction band.K = Boltzmann constant.T = Temperature in Kelvin.q = Electronic charge.The silicon doping concentration ND is given by:

2

V2

osD

C1Aq

2N … (3)

Where s is the Relative permittivity of silicon., o is the Permittivity of free space and Ais the Solar cell area.

The barrier height can be found from the relation of built in potential and otherparameter using equations (1-3) as shown in figure (8).

Figure (8): The plot of 1/C2 versus reverse voltage for In-ZnO/n-Si solar cell.

-1 -0.8 -0.6 -0.4 -0.2 0 0.2 0.4 0.6 0.8 10

2

4

6

8

10

12

Reverse Voltage ( V )

[1/C

exp

2] (

nF e

xp-2

)

After FabricationAfter 3 monthsAfter 6 months


27

The decreasing of solar cell open circuit voltage (VO.C) is caused by lowering the energybarrier height from 0.69 to 0.67 (after 6 month) as shown in table (1). The fabricated In-ZnO/n-Si solar cell has a good performance stability, the deposition of the indium layer withthe ZnO layer play a dominant rule for minimizing cell degradation with time as well as theindium layer tends to improve the cell efficiency beyond the visible region (700–1000) nmwavelength, the overall efficiency improvement of the fabricated solar cell is 0.4%.

The degraded characteristic of In-ZnO/n-Si solar cell can be also minimized by thedeposition of thin oxide (ITO) layer on the front side of In-ZnO/n-Si solar cell, so that theITO layer will suppress oxygen diffusion from the environment. Also ITO is highlytransparent in visible region and has small sheet resistance.

The electrical characteristics of ZnO layer could be enhanced by a proper doping likeindium and a suitable heat treatment in order to reduce the atmospheric conditions whichchange the stiochiometry of ZnO layer or inert atmosphere at certain temperature for a certaintime to minimize the imperfections and increase the crystal size which allows an increase inthe generated current [7]. As well as the absorbed oxygen on the zinc oxide surface and thatdiffused into the bulk are evolved leaving behind a doner state which increase the free carriers[8]. It is observed that, heat treatment results in alarge reduction of In-ZnO film sheetresistance because of oxygen desorption from the surface pores and grain boundaries,resulting in creation of vacant sites which will act as donor states.

Figure (9) shows the effect of annealing time on the sheet resistance of In-ZnO/n-si solar cell.

It is noticed that the sheet resistance of the fabricated In-ZnO-n-si sample is reducedfrom 1.9 K / ( /cm2) for the as deposited film to a low value of about 211 / for theannealed samples, the solar cell output parameters are improved and the stability wasextended for longer time.

0 5 10 15 20 25 30 35 40 45 500

0.5

1

1.5

2

2.5

Annealing time ( min )

Shee

t res

ista

nce

(ohm

per

sgu

ar c

m )

Figure.(9): The sheet resistance of In-ZnO/n-Si cell at different annealing time.


28

4-Conclusions:

Indium-Zinc oxide/monocrystalline silicon solar cell is fabricated using vacuumevaporation technique at different annealing temperature. The electrical characteristics of ZnOlayer was enhanced by indium doping and with suitable heat treatment resulting in reducingatmospheric conditions to change the stiochiometry of ZnO layer. The fabricated solar cellparameter are, VO.C = 0.48, ISC = 18.8 mA, fill factor = 0.62, and efficiency = 5.7%. thefabricated solar cell shows a good performance stability compared to the standard ZnO/n-Sisolar cell. The In-ZnO/n-Si gives an efficiency improvement of about 0.4%.

5- References:

5- NEAMEN, Semiconductor Physics and Devices, (C) Richard D. Irwin, INC., 1992.6- Shewchun J. S., Singh R. and Green M. A., "Theory of Metal-Insulator-

Semiconductors Solar Cells", J. Applied Physics. Vol. 48, No. 2, (USA), (1977).7- Todorovic D. M. and Smiljanic M., "Theory of Photoacoustic Effect in Metal-

Semiconductor System", Institute for Chemistry, Technology and Metallurgy,Njegoseva 12, 1100. Belgrade, Yugoslavia, (2001).

8- Djoko I. and Hartano H., "Ohmic Contact Schottky Barrier", University of Indonesia,Vol. 18, pp. 74, (USA), (2004).

9- Yap Siew Hong, Carrier Transport and I-V Characteristic of Au/Si Silicodes UsingOpen Photo-acoustic Cell, Solid State Science and Technology. Vol. 13, No. 1 and 2,287-295, (2005).

10- D. Derkacs, S. H. and E. T. Yu, Improved Performance of Amorphous Silicon SolarCell Via Scattering from Surface, Applied Physics Letters 89, 093103 (2006).

11- D.G.baik,S.M. cho. Thin solid films, 354 ,227 .(1999)12- Milnes A. G. and Feucht D. L., "Hetrojunction and Metal Semiconductor Junction",

Academic Press, Vol. 49, pp. 133-149, (London), (1972).9-W.F.Mohammed “Infrared Response and Guantum Efficiency of In-Doped

Silicon (n) Structure” Renewable Energy 21 (2000) 323-331 .

10-H. C. Card and E. S. Yang, IEEE Trans. Electron Devices, Vol. ED-24, pp. 397,(1977).


Al-Nimma :Reactive Power Control of an Alternator with Static Excitation ------

29

Reactive Power Control of an Alternatorwith Static Excitation System Connected to a Network

Dr. Majid Salim Matti Dr. Dhiya Ali Al-NimmalecturerAssist. Prof.

Mosul UniversityMosul Unoversity

Abstract

In recent years, the scale of power systems has been expanding, and with that expansionsmooth power operation is becoming increasingly important. One of the solutions is to realizea practical high speed, highly reliable exciter system that is suitable for stable operation of apower system.

In this work, a model of a static excitation system of an alternator connected to a networkvia a transformer have been built using MATLAB-SIMULINK PSB. The parameters of themachine has been obtained from Mosul dam power station taking into account saturationeffects. A PI controller is used to control the output reactive power of the synchronousgenerator for both pure DC excitation and static excitation systems. A method based on stepresponse has been proposed and verified for tuning the parameters of the controller. In orderto validate the simulated results of the system with AVR, the results have been compared withpractical results of Mosul dam and a good agreement has been realized. However, in largegenerating units, undesirable oscillations in the active power and speed result as a side effectof the AVR control or due to outside disturbances.

KEY WORDS: Static Excitation, Reactive Power Control.

- -

.

.

(MATLAB –SIMULINK- PSB) .

. –(PI) (DC)(Static Excitation) .

. (AVR) .

(AVR).



30

IntroductionFor many years the exciters used in alternators were DC generators driven by either the

steam turbine on the same shaft of the generator or by an induction motor. In the last threedecades, static excitation systems are introduced. Old systems are being replaced by newsystem for many advantages (such as quick response, online maintenance and high fieldcurrent). The static systems consist of some form of controlled rectifiers or choppers suppliedby the ac bus of the alternator or from an auxiliary bus. The voltage regulator controls theoutput of the exciter so that the generated voltage and reactive power can be controlled. Theexcitation system must contribute to the effective voltage control and therefore enhance thesystem stability. It must be able to respond quickly to a disturbance, thereby enhancing thetransient stability as well as the small signal stability. In most modern systems the automaticvoltage regulator (AVR) is a controller that senses the generator output voltage and thecurrent or reactive power then it initiates corrective action by changing the exciter control tothe desired value. The excitation system controls the generated EMF of the generator andtherefore controls not only the output voltage but the reactive power as well.

The response of the AVR is of great interest in studying stability. It is difficult to makerapid changes in field current, because of the high inductance in the generator field winding.This introduces a considerable lag in the control function and is one of the major obstacles tobe overcome in designing a regulating system. The AVR must keep track of the generatoroutput reactive power all the time and under any working load conditions in order to keep thevoltage within pre-established limits. Based on this, it can be said that the AVR also controlspower factor of the machine once these variables are related to the generator excitation level.

The AVR quality influences the voltage level during steady state operation and alsoreduces the voltage oscillations during transient periods, affecting the overall system stability.

Most researchers on modeling and simulation of generating systems found in the literature[1-5] did not use detailed models for the generating units with their detailed excitation system.Moreover researchers who implemented PI and PID controller for AVR in their modelsignored a detailed procedure for determining controller parameters.

Figure 1 shows the block diagram of a typical excitation system of a large synchronousgenerator [6].

In this work, which is part of a Ph. D thesis [7], a static excitation system of an alternatorconnected to a network via a transformer have been modeled and simulated using MATLAB-SIMULINK PSB.

network

Limiters andprotection

Automaticvoltage

regulator

Terminalvoltage and

generator

Power systemstabilizer PSS

exciter

Figure 1 Block diagram of synchronous generator and

excitation system with AVR and PSS.


31

Types of excitation systemsBased on excitation power source, the excitation systems have taken many forms over the

years, namely, dynamic excitation and static excitation systems. In a dynamic excitationsystem the most parts are connected to the rotor, so that the carbonic brushes can be removed.It is sometimes called brushless excitation systems. This type uses some sort of rotatingmachines; thus their responses are poor besides the need for regular maintenance. In staticexcitation systems, on the other hand, all components are static or stationary. Static rectifier,supply the excitation current directly to the field of the synchronous generator through sliprings. The supply of the power to the rectifiers is from the main generator or via the stationauxiliary bus through a step down transformer.

Automatic Voltage Regulator AVR is the brain of the excitation system. Its responsibilityis to control current such that, building generator voltage at starting, regulating voltage andoutput reactive power after connecting the unit to a network. The AVR must have high gain tokeep the operational variations within prescribed limits, good open circuit response, minimumdead band and high speed of response [8].

The AVRs work on the principle of error detection. The alternator three phase outputvoltage obtained through a potential transformer is compared with a reference value. Whenthe alternator is connected to a network, and in order to control the output reactive power, thesignal delivered and compared are the output voltage and output current. From these twovariables the output reactive power is determined and compared with a reference signal inorder to determine the error used to suggest the increment or decrement of field voltage.

Power ConverterMostly, the power converter is a thyristor three-phase bridge. The power converter may be

controlled by manual channel or by AVR. All excitation power is normally derived eitherfrom the synchronous machine terminals or from auxiliary source through an excitationtransformer. The voltage regulator controls the thyristor converter through a pulse-triggeringunit. The power rectifying bridges are full converter, 6 pulse, inverting type and can providecurrents up to 10000 A DC and voltage up to 1400 V DC. Each rectifier bridge includesprotection circuitry such as snubbers and fuses. Depending on the rating of the system, therectifier may comprise a single stack or multiple units in parallel for higher power levels. Inmost redundant applications, each bridge is rated to the full excitation requirement for theparticular generator; however, during normal operation all bridges are put to work sharing theload. The benefits are that by sharing load the life expectancy of the SCR’s is extended whileat the same time providing a hot backup.

System DescriptionThe basic function of any excitation system is to provide direct current to the synchronous

machine field winding. The excitation system controls and protects essential functions of thepower system for satisfactory operation and performance. The control functions include thecontrol of the generator voltage, reactive power flow and the enhancement of system stability.The protective functions ensure that the capability limits of the synchronous machine,excitation system and other equipment are not exceeded.

The presented system used in this study consists of an alternator connected to an infinitebus via a transformer. Static excitation system is used for the generator. The Simulink modelfor the system under study is shown in Figure 2. The whole system has been modeled using


32

MATLAB SIMULINK and power system blockset (PSB), in which the machine model thatcan be operated as a motor or as a generator, has been represented by the sixth order statespace model [9].

The parameters of the most important block of the model, i.e. the synchronous machine,are presented in appendix A The static excitation system is a three phase controlled bridgeconverter. Using PSB the machine block accepts the excitation voltage Vf as an input signal.If the signal is abstracted from the Rf-Lf load of the bridge, no loading effects of the machinewill be imposed on the thyristor bridge and thus the simulation results would not be correct.To overcome this problem and to model the whole system as one network, the machine blockhas been modified as shown in Figure 2.

In order to validate the simulation results, parameters of the machine and the systemparameters of one generating unit in the Mosul dam power station have been adopted andused. The parameters are tabulated in appendix A. Company's test results of the generator areused for comparison.

AVR Control

In order to control the output reactive power, the field voltage must be changed in thedesired way. In this paper, methods for controlling the output reactive power are described.using conventional PI controller applied for pure DC supply as well as for static excitationsystem

PI Controller Design with Pure DC Excitation

The PI and PID controllers are widely used in industrial control systems because of thereduced number of parameters to be tuned. The most popular design technique isZiegler_Nichols method [10], in which its parameters can be obtained from the step responseof the system. This method is suitable for some types of step responses specially with timedelay, but if the step response of the system has no time delay, this method fails. The stepresponse has several values that are of importance in obtaining an approximate transferfunction for the system.

The relation between field voltage and output reactive power can be approximated by thefirst order transfer function [10].

T.F =1s

K (1)

Where, is the time constant of the system, K is the gain.

To obtain approximate transfer function, firstly, we find Yss1 and Yss2 which are thesteady-state values for the output before and after step change in the input. Secondly, wedetermine the area Ao in order to calculate the approximate time constant of the system asshown in Figure 3 where,

= Ao / (Yss2 - Y ss1) (2)

The simulink model used to determine Ao and is given in Figure 4.


33

Figu

re 2

SIM

ULI

NK

mod

el o

f the

sys

tem

und

er s

tudy

.

stat

ic e

xcite

r

[vfm

ean]

vfm

ean1

[vfm

ean]

vfm

ean

[vf] vf1

[vf]

vf

trigg

er a

ngle

term

inal

vol

tage

spee

d, q

, p, v

f, vf

mea

n

refe

ranc

e va

lue

act

ref

trig

ang

reac

tive

pow

er c

ontro

ller

Cont

inuo

us

ABC

load

2

ABC lo

ad1

limite

r trig

ger

field

vol

tage

mea

n va

lue

field

vol

tage

v+ - Vd

v+ -

Va

v+ -

VCA

v+ -

VBC

v+ -

VAB

g A B C

+ -

Thyr

isto

r Con

verte

r

A B C

a b c

Thre

e-ph

ase

Tran

sfor

mer

210

MVA

15

kV /

133

kV

A B C

a b c

Thre

e-Ph

ase

exci

tatio

n tra

nsfo

rmer

N

A B CThre

e-Ph

ase

Prog

ram

mab

leVo

ltage

Sou

rce

Pm Vf_

m A B C

Sync

hron

ous

Mac

hine

pu S

tand

ard

alph

a_de

g

AB BC CA Bloc

k

puls

es

Sync

hron

ized

6-Pu

lse

Gen

erat

orSt

ep1 In

Mea

n

Mea

n Va

lue

-K-

Gai

n1

1

Gai

n

f(u)

Fcn

s- +

Cont

rolle

d Vo

ltage

Sou

rce

0

A B C 10,0

00 M

VA, 1

33 k

Vso

urce

<Out

put r

eact

ive

pow

er Q

eo (p

u)>

<Mut

ual f

lux

phi

mq

(pu)

>

<Mut

ual f

lux

phi

md

(pu)

>

<Rot

or s

peed

wm

(pu

)>

<Out

put a

ctiv

e po

wer

Pe

o (p

u)>

Vd


34

The time constant of the approximate transfer function of the first order so obtained forthe system was 2.75 sec (although it may be changed due to non linearity of the synchronousmachine). Thirdly, we find the parameters of PI controller, which is sufficient for the firstorder system as explained below:

Figure 5 shows the system to be controlled and the PI controller with the parameters Kpand Ki .

Figure 5 System controlled by PI controller.

K

T.s+1system transfer function outputinput Subtract

Kp.s+Ki

sPI controller

Kg

Gain

output reactive power

Ao

field voltage

step change in field voltage

0.6632

Yss2-Yss1

0.1076

Yss2

Time constant

Step2

Scope2In1 Out1

S/G connected to a network

1s

Integrator1

Dot Product

Divide

Add2

Figure 4 SIMULINK model used to determine approximate time constant.

Figure 3 Ao, Yss1 and Yss2 in a step response

Out

put

of a

uni

t st

epfu

nctio

n


35

From the above system it can be seen that the system has a pole at s=-1/ and thecontroller has a zero at s= -Ki/Kp then getting a system with transfer function K/s if wechoose the value Ki/Kp=1/ which has a response as a unit step function without overshoot,and by changing the overall gain Kg*K to get optimal value of response by decreasing therising time. If we assume that Ki=1 then we can say that if Kp= we can get a responsewithout over shoot. If Kp< we get over damped response and if Kp> we get an underdamped response. After that and in order to prove the assumption we use the SISO (SingleInput Single Output ) MATLAB tools and GUI (graphics user interface ). Figure 6a shows theSIMULINK model to compare the step response of the system without controller and with PIcontroller. The parameters of the PI controller thus obtained were Kp=2.75 and Ki=1. In orderto decrease rising time, the over all gain must be increased.

Figure 6b shows a comparison between step responses of the close loop systemwithout PI controller and with PI controller with different gains (see the rising time).

Figure 7 shows a comparison between step responses for the system with PI controllerand (fixed gain and Ki=1) but variant Kp (Kp= , Kp< and Kp> ).

It must be noted that the time constant of the studied system varied from 2.75 sec to3.3 sec depending on the range of the reference change and the parameters of thetransformer. It is found that the parameters of the PI controller can be fixed with acceptableresponse at minimum =2.75.

In order to control the output reactive power of the alternator connected to a networkusing the suggested PI controller, the model was built using SIMULINK with pure DCexcitation as a first step.

Figure 8 shows the result obtained from the model when the set value of reactive powerchanges from 0.125 pu leading to 0.125 pu lagging at time=20 sec at constant input power0.25 pu. Figure 9 shows the result obtained from the model when the set value of reactivepower changes from 0.125 pu leading to 0.25 pu lagging at time=20 sec. It is found that thesettling time is 1.1 sec. in the response of the output reactive power. This is regarded asgood, however the

output active power and rotor speed both oscillate with a certain frequency ofapproximately 0.9 Hz or 6.6 rad/sec. which is regarded as undesirable. The reason of thisoscillation is due to the change in load angle which affects the output active power. Thisoscillation can be damped using power system stabilizer.


36

2.75s+11

system without controller

2.75s+11

systemStep1Scope2

actual

refference

output

PI controller

Figure 6 Comparison of step responses of a first order systemwith PI controller for different gains and witout PI controller.

(a) SIMULINK model. (b) The step responses

(a)

(b)

0 5 10 150

0.2

0.4

0.6

0.8

1

1.2

1.4

time sec.

outp

ut without controller

with controller gain=1

with controller gain=2

T=2.75 sec.Kp=2.75Ki=1

Out

put

of a

uni

t st

ep fu

ncti

on

Figure 7 Output step response of a first order systemwith PI controller for different Kp.

0 5 10 150

0.2

0.4

0.6

0.8

1

1.2

1.4

time sec.

outp

ut

Kp =1.75

Kp =3.75

Kp=2.75

Ki=1T=2.75 sec.

without controller

Kp=4.75

Out

put

of a

uni

t st

ep fu

ncti

on


37

Figure 8 Model response results for a reference step in the reactive power from 0.125pu lead. to 0.125 pu lag.

12 14 16 18 20 22 24 26 28 300.9995

1

1.0005sp

eed

pu

12 14 16 18 20 22 24 26 28 30

0.22

0.24

0.26

activ

e po

wer

pu

12 14 16 18 20 22 24 26 28 30-0.2

0

0.2

reac

tive

pow

er p

u

12 14 16 18 20 22 24 26 28 300

2

4

time sec

field

vol

tage

pu

12 14 16 18 20 22 24 26 28 301.3

1.4

1.5

1.6

1.7x 104

time sec

term

inal

vol

tage

vol

ts


38

1trigger angle

60 biosing signal

1s

Integrator

50

Gain4

-K-

Gain2ref

1act

Figure 10 PI controller for static excitation system.

Figure 9 Model response results for a reference step in thereactive power from 0.125 pu lead. to 0.25 pu lag.

20 22 24 26 28 30

0.9995

1

1.0005sp

eed

pu

20 22 24 26 28 300.22

0.24

0.26

0.28

activ

e po

wer

pu

20 22 24 26 28 30-0.1

0

0.1

0.2

reac

tive

pow

er p

u

20 22 24 26 28 300

2

4

field

vol

tage

pu

12 14 16 18 20 22 24 26 28 301.3

1.4

1.5

1.6

1.7x 104

time sec

term

inal

vol

tage

vol

ts


39

PI Controller Design with Static Excitation

The same steps presented in the previous section can be followed to design a PI controllerfor reactive power output of the alternator with static excitation system. An additional signalmay be added to the controller (biasing signal) in order to improve its response. Figure 10shows the PI controller modified for static excitation system. A biasing signal of 60 degree isused since it is near the normal operating point of the controller.

Figure 11 shows the results after changing the set value of reactive power from 0.125 puleading to 0.125 pu lagging. The figures show the unit speed, output active power, outputreactive power, field voltage and mean value of field voltage. It is clear from figures that theresponse in output reactive power has a good rising ( 0.6 sec) and settling time (1.1 sec), butstill there is an oscillation in output active power with frequency of about 7 rad/sec (whichmay affect the stability of the system), and in unit speed. This oscillation occurs as a result ofthe disturbance coming from the sudden change in the field voltage.

Figure 12 shows the practical results after changing the set value of reactive power from0.125 pu (30 MVAr) leading to 0.125 pu (30 MVAr) lagging. Figure shows the output activepower, output reactive power, mean value of field voltage. Let us examine the oscillation inoutput active power and compare it with the results obtained from the simulation (see Figure11). A comparison between the results shows acceptable (95%) between the SIMULINKmodel results compared with the practical ones by Toshiba (see Figure 12) [11].

In PSB, the machine block accepts the excitation voltage Vf as an input signal. If the signalis abstracted from an Rf-Lf load of the bridge, no loading effects of the machine will beimposed on the thyristor bridge, and thus the simulation results would not be correct speciallywhen the PI controller decides a value of trigger angle which makes the mean field voltagenegative. To overcome this problem and to model the whole system as one network, themachine block has been modified as shown in Fig.2.

Figures 13 and 14 show SIMULINK results when the set value of reactive power changesfrom 0.125 pu to -0.125 pu. The first figure shows the result without modification while thesecond shows the results after modification. The figures show that the mean field voltage canbe negative after modification which affects the field current to change faster than withoutmodification..

Figure 15 shows the output reactive power controlled by the PI when the time constant ischanged for 2.7 and 3.1 seconds..

Table.1 shows a comparison between rising time and settling for PI controller.

Table 1 comparison between responses of fixed parameters PI controller for different time constants

=2.7 SEC =3.1 SEC

rise time

(s)

settling time

(s)

rise time

(s)

settling time

(s)

0.9 1.2 1.2 1.5


40

Figure 11 Model results with reference step change in

reactive power from 0.125 pu lead. to 0.125 pu lag.

9 10 11 12 13 14 15 16

0.99940.99960.9998

11.00021.0004

spee

d pu

10 11 12 13 14 15 16 17

0.23

0.24

0.25

0.26

0.27

activ

e po

wer

pu

10 11 12 13 14 15 16 17-0.2

-0.1

0

0.1

0.2

reac

tive

pow

er p

u

10 11 12 13 14 15 16 17-5

0

5

field

vol

tage

pu

10 11 12 13 14 15 16 17

0

2

4

aver

. fie

ld v

olta

ge p

u

9 10 11 12 13 14 15 16

0

50

100

time sec.

trigg

er a

ngle

deg

.

9 10 11 12 13 14 15 16 171.3

1.4

1.5

1.6

1.7x 10

4

time sec

term

inal

vol

tage

vol

ts


41

Figu

re 1

2 Pr

acti

cal r

esul

ts w

ith

refe

renc

e st

ep c

hang

e in

reac

tive

pow

er fr

om 0

.125

pu

lead

. to

0.12

5 pu

lag.

(act

ive

pow

er, r

eact

ive

pow

er, f

ield

vol

tage

and

ter

min

al v

olta

ge).


42

Figure 13 Results obtained when the set value ofreactive power changed from 0.125 pu to -0.125 pu

without modifications.

9 10 11 12 13 14 15 16

-1

0

1

2

aver

. fie

ld v

olta

g pu

9 10 11 12 13 14 15 16406080

100120140

trigg

er a

ngle

deg

.

9 10 11 12 13 14 15 16-0.2

-0.1

0

0.1

reac

tive

pow

er p

u

9 10 11 12 13 14 15 16

0.24

0.25

0.26

time sec.

activ

e po

wer

pu

9

10

11

1213

1415

16

- 2

0

2

4

6

field

vol

tage

pu


43

Figure 15 Reactive power for two different timeconstants using fixed gain PI controller.

9 10 11 12 13 14 15-0.2

-0.15

-0.1

-0.05

0

0.05

0.1

reac

tive

pow

er p

u

T=3.1 sec.

T=2.7 sec.

Figure 14 Results obtained when the set value of reactive power

changed from 0.125 pu to -0.125 pu with modifications.

9 10 11 12 13 14 15 16

-2

-1

0

1

aver

. fie

ld v

olta

ge p

u

9 10 11 12 13 14 15 166080

100120140

trigg

er a

ngle

deg

.

9 10 11 12 13 14 15 16-0.2

0

0.2

reac

tive

pow

er p

u

9 10 11 12 13 14 15 160.22

0.24

0.26

0.28

time sec.

activ

e po

wer

pu

9 10 11 12 13 14 15 16

-4

-2

0

2

4

time sec.

field

vol

tage

pu


44

Conclusion

A method for tuning the parameters of the controller has been proposed which depends onthe step response of the system.

The relation between field voltage and output reactive power can be approximated by firstorder transfer function for a certain range of field voltage (normal operating conditions).

It is found that the parameters of the PI controller can be obtained mainly from the timeconstant of the step response. But the time constant of the approximated system is not fixedfor all operating conditions; it varies from 2.75 sec to 3.3 sec, due to nonlinearity ofsynchronous machine and depending on the range of reference change and parameters of thetransformer.

It is found that the best ratio of the proportional gain to the integral gain (Kp/Ki) is equal tothe time constant of the system. The proposed method has no overshoot for normal operatingconditions, but it has small overshoot (5%) for other conditions and a small rising time whichcan be reduced by increasing the overall gain of the controller.

However, if the gain is increased it will affect the output active power, in such a way as toincrease the oscillation time and its maximum overshoot.

The above procedure has been applied to the system with pure DC excitation. Thesuggested method has been also applied to the generator with static excitation system. Theparameter of the PI controller in this case demands an additional biasing signal (30-90) deg.for the trigger angle. A value of 60 deg has been chosen which is very near to the operatingpoint at normal conditions. In this case the PI controller either increase or decrease the triggerangle without exceeding its boundary conditions.

The simulation results obtained are compared with the practical results obtained fromMosul Dam power station. This comparison shows that there is an acceptable agreementbetween these results (about 95%).

The suggested method of designing the PI reactive power controller is easy to implement witha straightforward design. The direct design method of the controller allows the excitationsystem designer to choose the parameters of controller and place the poles of the controller atthe location where it gives a desired performance. The time constant of the step response ofthe output reactive power can be varied (2.7 to 3.2 sec), it is found that the adjustment of thePI controller parameters is based on the smallest time constant rather than the maximum timeconstant.

References

1. A.S. Ibrahim, "Self tuning voltage regulators for a synchronous machine', IEEProceedings, Vol. 136, Pt. D. No. 5, September 1989.

2. Shigeyuki Funabiki and Atsumi Histsumoto, "Automatic voltage regulator for asynchronous generator with pole-assignment self-tuning regulator" Industrial Electronics,Control and Instrumentation, 1991, Proceedings IEE on industrial conference, page 1807-1811.

3. A. Godhwani and M.J.Basler,"A digital excitation control system for use on brushlessexcited synchronous generators", IEEE Transaction on energy conversion, Vol. 11, No. 3,Sept. 1996.


45

4. R. C. Schaefer, "Application of static excitation systems for rotating exciter replacement",IEEE Transaction on energy conversion 1997.

5. Vinko Casic and Zvonko Jurin, "Excitation system with microprocessor based twin-channel voltage regulator for synchronous machines", EPE-PEMC 2002 Dubrovnik &Cavtat.

6. Goran Andersson, “Dynamics and Control of Electric Power System”, Swiss FederalInstitute of Technology Zurch, 2006.

7. Matti M. S., "Modeling and Simulation of a Static Excitation System of an AlternatorConnected to a Network", Ph. D. thesis, Mosul University 2007.

8. Basilio J. C. and Matos S. R., “Design of PI and PID controllers with transientperformance specification”, IEEE Transaction on education, Vol. 45, No. 4, Nov 2002.

9. The Mathworks, Inc., ”MATLAB version 7 help”, copyright 2004.10. A.H.M.S. Ula and Abul R. Hasan, “Design and implementation of a personal computer

based automatic voltage regulator for a synchronous machine”, IEEE Transaction onenergy conversion, Vol. 7, No. 1, March 1992.

11. Toshiba company, "Static Excitation System ", Mosul Dam Documentation. 1990.

Appendix A

The parameters of the machine in MOSUL dam power station and the block parameters ofthe synchronous machine used in the system model in Fig.2. :

Where:

Xd , Xq are the direct and quadrature axis synchronous reactances respectively, Xd' , Xq' arethe direct and quadrature axis transient reactances respectively, Xd' , Xq' are the direct andquadrature axis subtransient reactances respectively, Tdo' is direct axis transient open circuittime constant, Tdo'' is direct axis sub transient open circuit time constant, Tq'' is quadratureaxis sub transient short circuit constant..

XD 0.92 P. U.Xq 0.66 p. u.Xd

' 0.35 p. u.Xd

'' 0.2 p. u.Xq

'' 0.27 p. u.Tdo 6.7 secTds 2.5 sec

Rated MVA 237 MVARated power 193 MWRated Voltage 15000 VNo. of phases 3Rated current 9123 A

Frequency 50 HzSpeed 120 rpm

Connection StarRated field voltage 362 VRated field current 2220 AField current at no

load at ratedvoltage

1149 A

Inertia constant 5 sec


Alabady: Design and Simulation of an Optical Gigabit Ethernet Network

46

Design and Simulation of an Optical Gigabit Ethernet Network

Omar Ahmed YousifSalah A. Jaro AlabadyCollege of EngineeringCollege of Engineering

Computer Engineering. Dept.Computer Engineering. [email protected][email protected]

Abstract

This paper deals with the design and simulation of an optical gigabit Ethernet usingOPTSIM 3.6 packet software. The main aim of the proposed design is to build a MANoptical network using one-gigabit Ethernet technique, and what are the necessaryrequirements to build these networks. As a case study, all states center are connected asStar – Bus topology using layer2 and layer3 optical switches. In addition, in this paperone-gigabit optical transmitter and receiver are designed to work as a node in thenetwork topology. Further more, the benefits of using L- Band wavelength fortransmission take in consider the linear and non-linear effects on fiber optic ispresented.

–– / /

: )OPTSIM 3.6 . (

.

. )1Gbps (

)L (.

Keyword: - Optical communications, Optical networks, Optical gigabit Ethernet


mailto:omar_hachum:@yahoo.com

mailto:salah_alabady:@yahoo.com


47

1- Introduction

Increasingly networks are being converted from copper to optical fiber. As a result,network design procedures need to be adjusted to address the specific requirements of fiberoptic cable as a communications transport media. The use and demand for optical fiber hadgrown tremendously and optical-fiber applications are numerous. Telecommunicationapplications are widespread, ranging from global networks to desktop computers. Theseinvolve the transmission of voice, data, or video over distances range from less than a meterto hundreds of kilometers, by using one of a few standard fiber optic designs instead ofseveral copper cable designs. All networks involve the same basic principles: information can be sent to, shared with,passed on, or bypassed within a number of computer stations (nodes) and a master computer(server). Network applications include LANs, MANs, WANs, SANs, intrabuilding andinterbuilding communications, broadcast distribution, intelligent transportation systems (ITS),telecommunications, etc. In addition to its advantages (i.e. bandwidth, durability, ease ofinstallation, immunity to EMI/RFI and harsh environmental conditions, long-term economies,etc.), optical fiber better accommodates today’s increasingly complex network architecturesthan copper alternatives [1-3].Gigabit Ethernet solutions have become a necessity with the accelerating growth of LAN andMAN traffic. The fundamental designing objectives in an optical network concern theoptimizations regarding the following two metrics: The first one is to minimize the networktotal cost. The second one is to maintain an acceptable quality of service (QoS), and goodperformance. This paper provides, detailed general guidance to design optical network usinggigabit Ethernet technique at longer wavelength, and give a case study to connect alluniversities in IRAQ.

2 - Overview of Optical Gigabit Ethernet

Optical gigabit Ethernet solutions have become a necessity with the accelerating growthof LAN traffic, pushing network administrators to look for higher speed network technologiesto meet the demand for more bandwidth [2,3]. While most copper systems will support Gigabit Ethernet, fiber optics provides a muchhigher degree of flexibly and future bandwidth/speed expansion compared with its coppercounterparts. Generally, copper will support Gigabit and multi-gigabit transmission rates, butonly for very short distances. Copper is affected by EMI (electromagnetic interference) andRFI (radio frequency interference). Fiber optics will support Gigabit and multi-gigabittransmission for both short and long distances, with immunity to EMI and RFI, making fibermore suitable solution for a number of applications. Depending on the fiber type and core size, Gigabit Ethernet applications supported byfiber optics are now transmitting signal reliably at 10Gbps, up to 80km using single modesystems, and well over that for Gigabit and multi-gigabit transmission rates. While fiberoptics provide some clear advantages over copper legacy technology, most systems usingfiber today also use copper at the end user point, creating a hybrid system for datatransmission. The advantages in utilizing a hybrid system exists by leveraging the bandwidthand EMI/RFI advantages of fiber for the longer length and main distribution lines of thenetwork, while using copper for the very short desktop and non-backbone connectivity,allowing a very easily routable and inexpensive connectivity solution without implementingmedia converters [3,4].


48

3 - System Design and Simulation of Optical Networks

The two key parameters, which affect the maximum cable length or maximum possiblerepeater spacing and the maximum possible bit rate, are attenuation and dispersion [5]. Signalattenuation in fiber is expressed in dB/km. The dB (Decibel ) is used for comparing twopower levels and is defined as the ratio of the output power (PO) to the input power (PI). Thepower budget for a data network is important because we have to ensure that we have enoughpower in the source to reach the farthest station without saturating the nearest station. (This isthe dynamic range of the source.)The following table 1 shows the typical Attenuation Loss for each component [6].

3.1 - Fiber Span Analysis

Span analysis is the calculation and verification of a fiber-optic system's operatingcharacteristics. This encompasses items such as fiber routing, electronics, wavelengths, fibertype, and circuit length. Attenuation and nonlinear considerations are the key parameters forloss-budget analysis. Before implementing or designing a fiber-optic system, a span analysisis recommended to make sure that the system must work over the proposed link. Both thepassive and active components of the system have to be included in the loss-budgetcalculation. Passive loss is made up of fiber loss, connector loss, splice loss, and lossesinvolved with couplers or splitters in the link. Active components are system gain, transmitterpower, receiver sensitivity, and dynamic range [2, 4]. The overall span loss, or link budget as it is sometimes called, can be determined by usingan optical meter to measure true loss or by computing the loss of the system components. Thelatter method considers the loss associated with span components, such as connectors, splices,patch panels, jumpers, and the optical safety margin. The safety margin sets aside 3 dB tocompensate for component aging and repair work in event of fiber cut. Adding all of thesefactors to make sure that their total sum is within the maximum attenuation figure, in order toensure that the system will operate satisfactorily. Allowances must also be made for the typeof splice, equipment, and the environment (including temperature variations). Nonlinear effects occur at high bit rates and power levels. These effects must be mitigatedusing compensators, and a suitable budget allocation must be made during calculations. Non-linear effects of fiber became apparent with specialized applications such as underseainstallations. Some of these effects are important to know when designing fiber opticssystems, include: Stimulated Brillouin Scattering (SBS), Stimulated Raman Scattering (SRS),Four Wave Mixing (FWM), Self-Phase Modulation (SPM), Cross-Phase Modulation (XPM),and inter-modulation (mixing). Non-linear effects limit the amount of data that can betransmitted on a single optic fiber. System designers must be aware of these limitations andthe steps that can be taken to minimize the detrimental effects of fiber non-linearities. Linear effects include Attenuation and Dispersion. A fiber with a lower attenuation willallow more power to reach its receiver than a fiber with higher attenuation. Figure (1) shows

COMPONENT LOSS VALUEAdapter 3dBSplice Joint 0.1 dBSingle mode Fiber Cable @ 1580nm 0.2 dB/kmSafe margin 3 dB

Table (1) : Typical Attenuation Loss


49

the BER versus wavelength, this figure demonstrates that the longest wavelength (1580nm inthis case) is the best for transmitted because the attenuation in this wavelength is lower thananother wavelength. Also the dispersion equal zero at this wavelength as shown in figure (2).We can calculate the bit error rate (BER) and plotted the wavelength versus BER usingequation (1)

221 QerfcBER ----------------------------------------------------------------------- (1) [7,8]

Where erfc ( ) is the complementary error functionThe Q-factor for a Gaussian shape pulse is given by:

LH

IIQ 01 ----------------------------------------------------------------------------- (2)

These currents, can be calculated using the idea of counting the number of electrons incidentWhere the 1I & 0I are the currents obtained for ‘one’ and ‘zero’ logic’s respectively. at

the photo detector when a logic ‘one’ or logic ‘zero’ is received.

0101

01 mmmm

mmQ ------------------------------------------------------ (3)

Where:

dmbmm 00 , the average number of electrons representing the ‘zero’ symbol

dmbmm 11 , the average number of electrons representing the ‘one’ symbol

qTdi

dm , the average number of electrons correspond the dark current di during the

symbol interval [0,T].

Tphcbm 00

Tphcbm 11

where: efficiency of photo detector wavelength of photo transmitter

h Blank constant (6.625*10-34 J.S) c light speed (3*10+8 m/s) q electron charge (1.602*10-19 C ) id dark current po optical power representing the ‘zero’ symbol p1 optical power representing the ‘one’ symbol

Dispersion will have an impact on the proposed system transmission because it will increasethe pulse broadening of the transmitted data, which leads to pulse distortion.


50

Chromatic dispersion for CorningLEAF_submarine fiber can is characterized by.[9]

000 ln..SD -------------------------------------------------------------(4)

Where:S0 is the zero-dispersion wavelength= 0.11 ps/km.nm2 value .Aeff m2for CorningLEAF_submarine fiber.

The effective length (Leff) is characterized by [10]

0.6 0.7 0.8 0.9 1 1.1 1.2 1.3 1.4 1.5 1.6

x 10-6

10-11

10-10

10-9

10-8

10-7

10-6

10-5

10-4

10-3

Wavelength

BER

Wavelength Vs. BER

at 1Gbps, 100km

Figure (1): Wavelength versus BER

Figure (2): Dispersion of Corning LEAF_Submarine Fiber


51

Leff eL 11

-------------------------------------------------------------------(5)

Where:L : is the actual fiber length,

: is the fiber attenuation constant in 1/km characterized by the equation [10]

eAlog10

-----------------------------------------------------------------------------(6)

Where:A = 0.2 : is maximum fiber attenuation in dB/km at the 1580nm.

Dispersion for CorningLEAF_submarine fiber versus transmitted wavelength can be plottedusing equation (4) as shown in Figures (2). The analysis in Figure (2) shows the dispersion ofCorning LEAF_submarine fiber, in the best case, will required to send data at wavelength1580 nm.

3.2 - Stimulated Raman Scattering (SRS)

When light propagates through a medium, the photons interact with silica moleculesduring propagation. The photons also interact with themselves and cause scattering effects,such as stimulated Raman scattering (SRS), in the forward and reverse directions ofpropagation along the fiber. This results in a sporadic distribution of energy in a randomdirection. SRS refers to shorter wavelengths pumping up the amplitude of longerwavelengths, which results in the longer wavelengths suppressing signals from the shorterwavelengths. One way to mitigate the effects of SRS is done by decreasing the input power[2]. In SRS, a short wavelength wave called Stoke's wave is generated due to the scattering ofenergy. This wave amplifies the higher wavelengths. The gain obtained by using suchwaveforms is considered in the Raman amplification basis. The Raman gain can extend mostof the operating bands (C- and L-bands) for WDM networks. SRS is pronounced at high bitrates and high power levels. The margin design requirement to account for SRS/SBS is 0.5 dB[2, 3].

The 3dB power threshold for SRS can be calculated by the following formula [10].

effR

effth Lg

ASRSP 16------------------------------------------------------ (7)

Where: SRS gain gR=1x10-13m/W.Aeff m2for CorningLEAF_submarine fiber.Leff can be calculated using equation (5)

The SRS power threshold verse fiber length can be plotted using equation (7) as shown inFigures (3). The SRS threshold power for the farthest distance between two cites (fiber length250 km) The Pth(SRS)=27.1 dBm (523mW) Therefore SRS will not be a concern for theproposed system.


52

3.3 - Stimulated Brillouin Scattering (SBS)

Stimulated Brillouin scattering (SBS) is due to the acoustic properties of photon interactionwith the medium [2]. Stimulated Brillouin scattering (SBS) is a nonlinear process that canoccur in optical fibers at input power levels much lower than those needed for stimulatedRaman scattering (SRS). It manifests through the generation of a backward- propagatingStokes wave that carries most of the input energy, once the Brillouin threshold is reached.Stimulated Brillouin scattering is typically harmful for optical communication systems. At thesame time, it can be useful for making fiber-based Brillouin lasers and amplifiers [4]. It issimilar to SRS in as much as it manifests through the generation of a Stokes wave whosefrequency is downshifted from that of the incident light by an amount set by the nonlinearmedium. However, major differences exist between SBS and SRS. For example, the Stokeswave propagates backward when SBS occurs in a single-mode optical fiber, in contrast toSRS that can occur in both directions. The Stokes shift (~10 GHz) is smaller by three ordersof magnitude for SBS compared with that of SRS. The threshold pump power for SBSdepends on the spectral width associated with the pump wave. It can be as low as 1 mW for acontinuous-wave (CW) pump or when the pumping is in the form of relatively wide pumppulses (width > 1 s). In contrast, SBS nearly ceases to occur for short pump pulses (width<10 ns). All of these differences stem from single fundamental change acoustical phononsparticipate in SBS whereas optical phonons are involved in the case of SRS [4].Rayleigh scattering, a major source of fiber losses, is an example of elastic scattering in whichthe frequency of scattered light remains unchanged. In contrast, the frequency is shifteddownward during inelastic scattering. Raman and Brillouin processes provide two examplesof inelastic scattering.To calculate the 3dB threshold power for SBS, the effective area (Aeff) and effective length(Leff) must be known. The effective area (Aeff) can often be found in the manufacturer’s datasheet.The SBS 3dB power threshold follows the equation [10].

Figure (3): Stimulated Raman Scattering (SRS) power threshold versusfiber length


53

effB

effth Lg

ASBSP 21------------------------------------------------------ (8)

where :SBS gain gB= 5x10-11m/WAeff m2for CorningLEAF_submarine fiber.Leff can be calculated using equation (5).

The SBS power threshold versus fiber length can be plotted using equation (8) as shown inFigures (4) a The SBS threshold power for the farthest distance between two cites (fiber length250 km) Pth(SBS)= 1.3dBm (1.3mW). This means we can assume the scattered power due toSBS will occur in the proposed system above Pth(SBS)=1.3dBm , we should limit launchpower to less than 1.3dBm.

3.4 - Self-phase Modulation

Phase modulation of an optical signal by itself is known as self-phase modulation (SPM).SPM is primarily due to the self-modulation of the pulses [11]. Generally, SPM occurs insingle-wavelength systems. At high bit rates, however, SPM tends to cancel dispersion. SPMincreases with high signal power levels. In fiber plant design, a strong input signal helpsovercome linear attenuation and dispersion losses. However, consideration must be given toreceiver saturation and to nonlinear effects such as SPM, which occurs with high signallevels. SPM results in phase shift and a nonlinear pulse spread. As the pulses spread, theytend to overlap and are no longer distinguishable by the receiver. The NRZ encoded in opticaltransmitter used in system suggested therefore , SPM phase shift in a NRZ digital systembecomes significant when phase shift ( 2 ) [4].

Before the phase shift of the pulse can be calculated, the nonlinear propagation coefficientmust be known. The nonlinear propagation coefficient is estimated by [10]

Figure (4): Stimulated Brillouin Scattering (SBS) power threshold

versus fiber length


54

---------------------------------------------------------------- (9)

Where:The nonlinear index coefficient n=2.5x10-20m2/W,

=1580nm,Aeff= m2for CorningLEAF_submarine fiber.

The 3dB power threshold for SRS can be calculated by the following formula [10].

effin LPSPM ---------------------------------------------------------------------- (10)

Where:Leff can be calculated using equation (4) when the real fiber length = 250km.Aeff=71mm2for CorningLEAF_submarine fiber.

= 2.6x10-3/Wm for MetroCor.

According to Figure (5) Self-phase Modulation in CorningLEAF_submarine fiber becomesignificant if our launch power is 50mW (17dBm), so we can assume SPM will beinsignificant for the proposed system because the maximum launched power will not exceed1mW (0 dBm).

4 - Fiber Optics System

Basically, in a fiber optic system, information (such as voice, data or video) is transmittedover fiber in the following way: once encoded into electrical signals, these get converted intolight signals that travel down the fiber until they reach a “detector” which then changes thelight signals back into electrical signals. Finally, the electrical signals are decoded intoinformation in the form of voice, data, or video [12,13]. OPTSIM 3.6 packet software wasused to simulate the optical gigabit Ethernet (transmitter and receiver). The Architecture of

Figure (5): SPM of CorningLEAF_submarine fiber


55

transmitter and receiver fiber optic system is shown in figure (6). This figure shows theoptical transmitter in one node connected with the optical receiver in another node. Whilefigure (7) show the transmitter and receiver connected between two nodes.

The main components of a fiber optic system include:1) Transmitter. A transmitter converts information – such as voice, data, or video, encodedinto electrical signals to light signals. The transmitter receives a modulated electrical signaland converts it into a modulated light signal, after which it sends the light signal into the fiberoptic cable. A Light Emitting Diode (LED) or a Laser can be used for generating the lightsignals.

2) Fiber optic cable – the medium which carries the signal.

3) Receiver. At the other end of the fiber optic cable from the transmitter is the receiver whichuses a photo detector to convert the incoming light signal back into an electrical signal. Thewavelength designation of the receiver must match that of the transmitter. Importantcharacteristics of receivers are System Performance, which is the Bit Error Rate (BER) fordigital systems or Signal to Noise Ratio (SNR) for analog, Saturation and Sensitivity [14].This accepts the light signal and converts it back into a modulated electrical signal. Table (2)shows the Characteristic of Optical System Design.

5 - Simulation and Analysis Results

At the beginning, the place of the center states was explored, and the distance betweenthese states was calculated using the google earth software. Moreover, to be on the safe side, a50km distance was added to each distance between the centers of states. Then created threemain star topology, the first to connect the north states groups, the second connect the middlestates groups, and the last to connect the south states groups. Figure (8) shows the distributedthree network groups as a star – bus topology. One city in each main star topology selected asa center depended on the short distances between the cities. Also to get the maximum fiberlength not greater than 250km, cities (5- KIRKUK),(1- BAGHDAD ) and (11- NASIRIYA)were selected as main centers for the north, middle, and south topology groups respectively,and in these cities the optical switch layer three is used to connect the optical switches layertwo in another cities. Because the non-linear effectives (SBS), the fiber length limited to(250km), and the maximum allowed transmitted power is (1mW) as shown in figure (3). Soswitch in the city (7- TIKRIT) used to connect between the center of the north and center ofthe middle network groups, and for the same reason the switch in city (10- DIWANIYA)used to connect between the center of the middle and the center of the south network groups.The reason for selecting the KIRKUK city as center in the north network topology, is themaximum distances between KIRKUK and DOHUK cities is 250km as shown in table (3),and this is the maximum distance allowed to prevent the effects of SRS and SBS. On the otherhand, the distance between KIRKUK -center of north network topology- and BAGHDAD -center of middle network topology- is (290km) it is greater than allowed distance (250km),for this reason we select TIKRIT city to connect between the north and middle networkgroups, while the distance between TIKRIT and BAGHDAD is (200km) as shown in table(4). To the same reason, we select DIWANIYA city to connect between center of the middleand center of the south network groups. The distance between the BAGHDAD -center ofmiddle network topology- and NASIRIYA - center of south network topology is (360km).While the distance between DIWANIYA and BAGHDAD is (210km) as shown in table (4).


56

Figure (6): The Architecture of transmitter and receiver fiber optic system

Figure (7): The Architecture of Two Nodes Fiber Optic System


57

BIT RATE 1 GBITS/SECONDNumber of Channels 1 channel

Channel Wavelength 1580nmBit Error Rate 10-10

Digital SNR (Q) 7Maximum, fiber length 250 kmMinimum fiber Length 100 km

Maximum transmitted power 0dBmMaximum Attenuation 58.4 dBMinimum Attenuation 18.6 dB

Maximum Launched power 0 dBTransmitter Laser Line Width (FWHM) 10 Mhz

Receiver Quantum Efficiency 0.7Receiver Responsivity (at reference frequency) 0.892041905713 A/W

Channel Coding NRZReceiver Single-Pole Electrical Filtering

(-3dB Bandwidth)2 GHz

Receiver Dark Current 0.1 nAReceiver Reference Wavelength 1580 nm

Fiber attenuation 0.2 dBFiber Chromatic Dispersion at 1580 0 ps/nm/km

Fiber Chromatic Dispersion Slop at 1580 0.11 ps/nm2/kmFiber Mode Field Diameter 71m2

Polarization Mode Dispersion <0.1 ps/km1/2

TIKRIT

SULAIMANIYA

ERBILMOSUL

DOHUK

CITY

115 km97 km87 km149 km200km

Real distanceto KIRKUK

170 km150 km140 km200 km250km

Approximatedistance toKIRKUK

-16dBm

-20 dBm-

21.67dBm

-10dBm0dBmPowertransmitter

Table (2): Characteristic of Optical System Design

Table (3): Fiber length and quantity power transmitted to get the BER equal


58

Table (5): Fiber length and quantity power transmitted to get the BER equal(1e-10) at 1Gbps, and 1580nm wavelength, in south network groups

In the figure (8) , there are tow type of switches the first (green and blue color) islayer 2 switch that offer a frame forwarding service based on the physical addresses that areavailable as part of Layer 2 (i.e., the MAC address of the destination) as well as performingthe signal regeneration functions of a repeater. this type of switch will configure to work withmulti VLAN. The second type is switch layer 3 (red color) this type of switch work as routerthat offer a packet forwarding service based on the logical destination addresses (IP address)that are available as part of Layer 3 (as well as providing bridging and repeating functions).This type of switch use Layer 3 header information (packet header) to make selectiveforwarding decisions, allowing broadcast and multicast traffic to be suppressed. Multipleactive paths between communicating hosts can be supported. Conversions between differenttypes and speeds of data link (such as LAN-WAN conversions) are supported by routers.Switch layer 3 are also intelligent enough to provide functions such as access controls basedon the type of application protocol being carried and various forms of packet filtering. Thereality is that a Layer 3 Switch is simply a class of high performance router that is optimizedfor the campus LAN or intranet ,incorporate routing functionality which allows the switch toperform inter VLAN routing, used to connect between two networks at the network layer.

DIWANIYA

NAJAFKUTHILLA

KARBALA

RAMADIBAQUBA

TIKRIT

CITY

157 km146 km160 km93 km89 km104 km50 km150km

Realdistance toBAGHDA

D

210 km200 km210 km150

km140 km155 km100 km200

kmApproximate distance

toBAGHDA

D-8.24dBm-10dBm-8.24dBm-

20dBm-22dBm-

18.86dBm-30dBm-

10dBm

Powertransmitter

BASARHAMARASAMAWADIWANIYA

CITY

160 km125 km100 km160 kmRealdistance to

NASIRIYA210 km175km150km210kmApproxima

te distanceto

NASIRIYA-8324dBm-15.23dBm-20dBm-8.24dBmPower

transmitter

Table (4): Fiber length and quantity power transmitted required to get the BER equal(1e-10) at 1Gbps, and 1580nm wavelength, in middle network groups


59

The defining characteristics of these devices is their ability to work with switched LANs, tosupport Gigabit Ethernet speeds, and to handle more complex traffic patterns efficiently, andthis is the one of goals for using this type of switch in this proposed network system.Each layer 2 switch in the state center will be configure to work with multi VLAN as staticVLAN to connect different offices in the state, on the other hand the layer 3 switch (router)used to connect between the networks (north, middle and south), inter VLANs and filteringthe packets.

From the simulation results, figures (9, 10) show the optical spectrum signal at thetransmitter and receiver respectively at 1Gbps to achieve BER (1e-10) at wavelength 1580nm.These figures clearly show the maximum power transmitter and receiver at 1580nmwavelength. In addition, the transmitted power is leases than (0dBm) therefore the SRS andSBS effects not find on the system. Figure (11) shows the fiber cable length versus powertransmitted at bit rate 1Gbps , 1580nm wavelength to achieve BER (1e-10), the figure showwe must increase the transmitted power when the fiber optic length increase, but also we musttaking into account the maximum transmitted power not pass (1mW) to prevent the effects ofSBS and SRS in this system. Figure (12) shows the Eye diagram at the receiver, this figuregives indicator for the systems performance.

6 - Conclusion

This paper demonstrates the design and simulation of optical gigabit Ethernet, to connectbetween the states in IRAQ as case study, using optical system software (OPTSIM 3.6).The results obtained from the proposed design indicate that one can prevent or minimize thelinear effects including attenuation and dispersion, and non-linear effects includingStimulated Brillonin Scattering (SBS) ,Stimulated Raman Scattering (SRS), four wave mixing(FWM), self-phase modulation (SPM), cross-phase modulation (XPM). In addition, one canget the best performance when using longer wavelength (L- band) 1580nm, the effect ofattenuation and dispersion in this wavelength decrease to the minimum. The effect of (SBS),(SRS) and (SPM) will be insignificant by using lower power transmitter 0dBm.

References

[1]: Chinlon Lin, "Broadband Optical Access Networks and Fiber-to-the-Home Systemstechnologies and Deployment Strategies ", John Wiley & Sons Ltd, 2006[2]: Vivek Alwayn, " Optical Network Design and Implementation", Cisco Press, 2004. [3]: Mohammad Ilyas, Hussein T. Mouftah, "The handbook of optical communicationnetworks", CRC Press LLC, 2003 [4]: Govind P. Agrawal, "Nonlinear Fiber Optics", Third Edition, Academic Press, 2001 [5]: Eric J. Mitchell, " Simulation of an Optical Network System for a Space Based HighPerformance Computer System ",Submitted to the Department of Electrical Engineering andComputer Science in Partial Fulfillment of the requirements for the Degree of Master ofEngineering in Electrical Engineering and Computer Science at the Massachusetts Institute ofTechnology May 22, 2002[6]: Thomas Sims, "Designing Optical Transmission Networks Principles and Approaches",16thMarch 2006.(Available at http://www.webarchive.ja.net/services/events/calendar/2006/optical-networking/Verizon.pdf )

http://:@www.webarchive.ja.net/services/events/calendar/2006/optical-


60

[7]:E.G. Shapiro, M.P.Fedoruk, and S.K.Turitsyn,” Numerical estimate of BER in opticalsystems with strong patterning effects”, Electronics letters. Vol. 37,No. 19,pp. 1179-1181.2001.[8]: G.Bosco, A.carena, V.curri, R.Gandino, P.poggiolini, S.Benedetto.”A Novel AnalyticalMethod for the BER Evaluation in Optical Systems Affected by Parametric Gainn”, IEEEPhotonics Technology letters,Vol. 12,No. 2,February 2000.[9]: Corning Incorporated, "Single-Mode Dispersion Measurement Method ",ISO 9001Registered, Supersedes: August 2000, Issued: September 2001.[10]: Tom Baldwin, Steven Durand, " IF Fiber Selection Criteria" EVLA Memorandum No.32 Project, National Radio Astronomy Observatory, Washington, D.C., November, 2001.(Available at http://www.aoc.nrao.edu/evla/geninfo/memoseries/evlamemo32.pdf) [11]: Agrawal, G. P., "Fiber-Optic Communication Systems", John Wiley & Sons, NewYork, 1997.[12]: Yongyut , Prof. P. L. Chu, "Simple Star Multihop Optical Network", IEEE Journal ofLightwave Technology, Vol.19, No.4, pp. 425-432, April 2001.[13]: Yongyut , Prof. P. L. Chu, "A New Multihop Optical Network: Simple Star", OpticalNetwork Magazine, Vol. 3, Issue 4, July/August 2002. [14]: P. Green, "Progress in optical networking", IEEE Communications Magazine, Vol. 39,pp. 54–61, Jan.2001.

Figure (8): The distributed three networks groups as a star – bus topology

http://:@www.aoc.nrao.edu/evla/geninfo/memoseries/evlamemo32.pdf)


61

Figure (9) : Optical power spectrum at the transmitterLevel length (mm)

Figure (10) : Optical power spectrum at the

Level length (mm)

Figure (11): Cable length Vs. powertransmitted at 1Gbps ,

50 100 150 200 250

-40

-35

-30

-25

-20

-15

-10

-5

0

Cable length (km)

Pow

er T

x (d

Bm

)

Cable length Vs. Power Tx

at 1Gbps , BER = 1e-10Wavelength = 1580nm

Figure (12): Eye diagram at the


Alkababji: Segmentation of Conversational Speech Using Probabilistic Neural ------

62

Segmentation of Conversational Speech Using ProbabilisticNeural Network

Dr. Ahmed Maamoon AlkababjiLecturer

Computer Engineering Department, Collage of Engineering,University of Mosul, Mosul, Iraq.

AbstractAutomatic segmentation of audio streams according to speaker identities, environmentaland channel conditions has become an important preprocessing step for speechprocessing, speaker recognition and audio mining. This paper presents an automaticspeech segmentation system where the performance of the probabilistic neural network(PNN)(which is the main part of the system) is examined and then enhanced in the areaof segmentation of conversational speech. The results show that a percentage falsesegmentation (PFS) of 18% can be achieved. PFS is dropped to 6.1% enhancing thesystem. The experiments were carried out on a dataset created by concatenatingspeakers from the TIMIT database.Keywords: Speech segmentation, PNN, Probabilistic neural network.

.

, , , ,

.

18% ,6,1 % .TIMIT.



63

1.IntroductionAutomatic speech segmentation aims to find the speaker change points in an audio

stream. It is a preprocessing task for audio indexing, speaker identification - verification -tracking, automatic transcription, information extraction, topic detection, speechsummarization and retrieval[1].

Several techniques have been used for speech segmentation. Among them are thosebased on Bayesian Information Criterion(BIC) [1]. In [2] the (BIC) is compared with theCumulative Sum (CuSum) algorithm for automatic segmentation. The Use of an adaptiveVowel/ Constant/ Pause (V/C/P) classification method [3] attempt to segment speech withoutspeech recognition. The use of Artificial Neural Network (ANN) was for more than a decadeas in [4] for automatic speech segmentation and its performance was compared to that ofHidden Markov Models (HMMs).

In the present work a special type of neural network is used. This neural networkuses a kernel-based approximation to form an estimation of the Probability Density Functions(PDFs) of categories in a classification problem; this is the Probabilistic Neural Network(PNN). This particular type of ANN provides a general solution to pattern classificationproblem by following the probabilistic approach based on the Bayes decision theory [5].

The rest of the paper is organized as follows. The overview of our system includingfeature extraction is described in Section 2. Section 3 presents a brief description of the database used for the evaluation of the proposed segmentation system. Section 4. demonstrates theheart of the proposed system (PNN) and its training as well as the results of testing the systemby the selected data base. In addition to that a new technique is described and tested on thesame data base showing an advancement in the segmentation results of the proposed system.Finally, in section 5, some concluding remarks of this work is given.

2.System OverviewThe proposed segmentation system relays on the neural network to make the decision

to which speaker, a currently examined segment belong. Thus the system has two phases; thetraining phase and the testing or segmentation phase. In both phases the speech sample mustbe preprocessed. The preprocessing includes framing, windowing, feature extraction. This isachieved by the stages described in the following subsection. Figure 1 shows the phases of thesystem and the stages of each phase.

Figure 1: The phases and stages of the proposed segmentation system.


64

2.1 Framing and WindowingThis is the front end of the system and the first stage of the preprocessing that is to be

carried out to prepare the input speech signal to the next stage (the feature extraction stage).In this stage the speech signal is segmented into 20ms long frames with an overlap of 10ms.Then each frame is multiplied by a window function to reduce the effect of the spectralartifacts that result from the framing process. The Hamming window is used in this system forsuch aim.

The selection of the frame length is a crucial parameter for successful spectralanalysis, due to the trade-off between the time and frequency resolutions. The window shouldbe long enough to adequate frequency resolution, but on the other hand, it should be shortenough so that it would capture the local spectral properties. Typically, a frame length of 10-30 milliseconds is used [6]. Usually adjacent frames are overlapping by some amount. Atypical frame overlap is around 30 to 50 % of the frame size. The purpose of the overlappinganalysis is that each speech sound of the input sequence would be approximately centered atsome frame [7, 8].

2.2 Feature extraction stageIn the feature extraction stage Linear Predictive Coding (LPC) is used. The model

based representation of speech gives rise to Linear Prediction Coefficients model (LPC). LPCis a very important spectral estimation technique because it provides an estimate of the pole ofthe vocal tract transfer function. The LPC algorithm is an nth order predictor which attemptsto predict the value of any point in a time varying linear system based on the values of theprevious n samples. The rationale in linear prediction (LP) analysis is that adjacent samples ofthe speech waveform are highly correlated and thus, the signal behavior can be predicted tocertain extent based on the past samples. According to [9] an LPC predictor larger than 15 issufficient to represent the features of a speech segment. Therefore a 16 LPC predictor is used.This 16 LP coefficients c[n]are then converted to its corresponding LPC Cepstrumcoefficients (LPCC) using [10]

….(1)

A noticeable thing is that although there are finite number (p) of LP coefficients, theLPC cepstrum sequence c[n] is infinite. However, the magnitudes of |c[n]| 0 fast with n,and thus relatively small number of coefficients is needed to model the spectrum[10].

At this point each 20ms frame of the input signal is represented by a vector of 16LPCC coefficients this vector is called code vector. Due to the 50% (10ms) overlap thenumber of code vectors (N) for a speech signal of duration D can be calculated as follows:

Number of code vectors (N)= [(D-mod(D,10ms))/10ms]-1 ….(2)

For an example, for a 1 second input speech signal there is (1000/20=50) 20msframes. Due to the overlap the number of frames is multiplied by 2. Therefore, to find thenumber of frames which equals the number of code vectors, the duration of the speech signalis divided by 10ms. Before division the duration is made a precise multiples of 10ms. It mustbe noticed that the last frame can not be overlapped. Therefore, the total number of frames(code vectors) must be decreased by 1.

1

11

1],[][][

],[][][

n

kn

pnk

pnknakcnkna

pnknakcnk

nc


65

Therefore, the input speech signal is represented by an 16*N matrix (16 LPCCcoefficients by N code vectors), this matrix is called codebook. Refer to Figure 1 whichabstract the derivations of the codebook from the speech signal.

3. Speech CorpusThe speech corpus is used to examine the performance of this proposed speaker

segmentation system. It is the standard American English TIMIT provided by Linguistic DataConsortium [11]. TIMIT is an acoustic-phonetic database including 6300 sentences and 630speakers who speak English. The audio format is PCM, the audio samples are quantized in 16bit, the recordings are single-channel, the mean duration is 3.28 sec and the standard deviation(st. dev.) is 1.52sec. From all the available data in the TIMIT corpus two arbitrary subsets ofspeakers are used in this work. The male speaker's subset contained 70 speakers and thefemale speaker's subset contained 70 speakers too. There are 10 speech files for each speaker;two of the files have the same linguistic content for all speakers, whereas the remaining eightfiles are phonetically diverse.

For each speaker, a codebook is built using the following process: Three of the tenfiles available for each speaker are used including the two of the phonetically identical file.As in section 2.1 and 2.2 for each file, a codebook is created. The three codebooks are pooledresulting one large codebook, then the k-mean clustering algorithm was used to cluster thislarge codebook to obtain an (16*128) codebook (an overall sum of 140 codebooks, 70 for themale speakers and 70 for the female speakers). These codebooks are then used to train theconversation segmentation system.

In a conversation there are three probabilities for the speakers participating in it: amale-male conversation, female-female conversation and male-female conversation. For thetesting phase of the conversation segmentation system 70 conversations are created for eachof the three probabilities mentioned above by concatenating the remaining seven speech filesof the speakers in the selected subsets (a total of 210 conversations). For an example aconversation between speaker 1 and speaker 2 is created as follows:((1st file of speaker 1,1st

file of speaker 2,2nd file of speaker 1, 2nd file of speaker 2…7th file of speaker 2).

4. The Probabilistic Neural NetworkThe main stage of this conversation segmentation system is the probabilistic neural

network. This neural network has been used in many speaker recognition systems as in [12-14]. In this work the performance of the probabilistic neural network as a conversationsegmentation tool is to be investigated.

A useful interpretation of the network outputs under certain circumstances is toestimate the probability of class membership, in which case the network is actually learning toestimate a probability density function (PDF). This is the case of the probabilistic neuralnetwork (PNN).

The network paradigm basically uses the Parzen-Cacoulos estimator to obtain thecorresponding PDF of the classification categories. PNN uses a supervised training set todevelop probability density functions within a pattern layer [5]. The PNN implements theParzen window estimator by using a mixture of Gaussian basis functions. If a PNN forclassification in K classes is considered, the probability density function fi(xp) of each class Kiis defined by:

iMM

jijp

Tijp

iddpi Ki

MMf

122/ ,...,2,1)),()(

21exp(1

)2(1)( xxxxx ….(3)

where xij is the j-th training vector from class Ki , xp is the p-th input vector, d is thedimension of the speech feature vectors, and MMi is the number of training patterns in class


66

Ki. Each training vector xij is assumed to be a centre of a kernel function, and consequentlythe number of pattern units in the first hidden layer of the neural network is given as a sum ofthe pattern units for all the classes. The variance acts as a smoothing factor, which softensthe surface defined by the multiple Gaussian functions.

For each conversation, which in this work contains the speech of two speaker only, aPNN is designed to decide if the input segment (10 msec. long) belongs to the first or to thesecond speaker. Both speakers are represented by codebooks which are used in training thenetwork. Therefore the problem is reduced to classifying the input test vector to one of twoclasses (K=2).Figure 2 shows the architecture of the probabilistic neural network used in thissegmentation system, the two hidden-layers of the PNN used in this system are shown [15].The Radial Basis layer is defined as ai1 = radbas(||i IW1,1 - pN || bi1) …(4)

and the Competitive layer is given bya2 = compet(LW2,1a1) …(5)

Figure 2: Architecture of the Probabilistic Neural Network, after [5].whereIW1,1 are the first layer input weights, set to be equal to the transpose of the matrix formed from the QN training vector pairs. LW2,1 are the second layer weights, set to be equal to the matrix of target vectors.|| . || the Euclidian distancei ith element of a1 & b1,the ith row of the IW1,1.pN the input feature vectorb1 the bias for the Radial Basis layer, defined as [5]:

/ln(0.5)-1b …(6)a2 the binary output of the PNN second layer

The transfer function of the Competitive layer employs the winner-take-all rule. Thebiggest weighted sum of probabilities from the first layer is granted a ‘1’, while the othersreceive zeros.

In the test phase (segmentation phase), the PNN classifier decides whether the inputtrial belongs to the first speaker or to the second speaker. In order to do this, the pre-trained


67

PNN is tested by the feature vectors extracted from the input speech. The degree of similarityof the input feature vector to either of the two speaker’s model is estimated by computingtheir corresponding distances. For each input vector, a binary decision is made: output ‘1’means it belongs to the first speaker, while ‘0’ is produced when the feature vector is moresimilar to the second speaker model.

Although the PNN network is used in the designing of this speaker segmentationsystem (which is known with its complexity and memory requirement) it has been found thatthe speaker segmentation system described above is capable of working in real time oncommon personal computers.

4.1 The training phaseFor the purpose of segmentation, the two speakers participating in the conversation

which will be segmented are known in prior. Therefore, in the training phase (which isperformed before the segmentation or the testing phase) a PNN is trained by the two codebooks of the speakers participating in the conversation (refer to section 3). These two codebooks are pooled to get a (16*256) matrix which is used for training. A target matrix is builttoo, this target matrix contains one row (128 “1” followed by 128 “0”). As a result of theabove, 210 PNN’s are trained in this work to segment the 210 conversation used to test thissegmentation system (refer to section 3).

4.2 The testing phaseIn the testing phase, the PNN’s that are obtained from the previous section are tested

in three different categories, (male-male, female-female, male-female) each category has 70conversations. To evaluate the system performance, the percentage of the false segmentationsof the input segments (FS) to the total number of segments in the conversation under test (N)is used as a figure of merit referred to as (PFS).PFS(%)=FS/N * 100% ….(7)4.3 Segmentation results

For the three categories described above the segmentation results are demonstrated inTable 1. Three results are shown for each category. From the 70 segmentation trialsperformed for every category the maximum value of the percentage of the false segmentation(PFS) of a conversation, the minimum value of (PFS), and the average of all (PFS) for the 70segmentation trials are found and they are shown in Table 1.

Table 1: The results for the segmentation systemCATEGORY MAX. PFS(%) MIN. PFS(%) AVERAGE

PFS(%)

Male-Male 31.8309 11.4971 22.8088

Female-Female 31.0738 16.6656 23.7910

Male-Female 24.2105 12.0422 18.0936

It is noticed form examining the segmentation results that there are some errors in thesegmentation which are due to the silence periods between two word of the same speaker or afalse segmentation of one or two segments during the speech of a certain speaker resulting ina "1" or two among a string of zeros or a "0" or two among a string of ones. Therefore, a newtechnique is added to the segmentation process to overcome this problem and take advantage


68

of these false segmentation to enhance the overall performance of the segmentation system asshown in the next subsection.

4.4 The enhancement of the segmentation system and its resultsAs mentioned earlier there are some errors in the segmentation which are due to the

silence periods between two word of the same speaker or a false segmentation of one or twosegments during the speech of a certain speaker. For an example the output of the neuralnetwork for an input phrase belonging to one of the speakers must be a stream of ones. But apattern like the following could be found at the output of the neural network(111111011111100001111111100011111111). To enhance the proposed segmentationsystem and to overcome the problem of some of false segmented segments a sliding windowof a finite length is moved along all the output of the neural network. If the number of 1's inthis window is greater than the number of 0's a single "1" will represent this window and viceversa. The size of the window must be an odd number to avoid a situation where the numberof 1's equal the number of 0's. Starting with the first two odd values after one, the values 3and 5 are chosen for the size of this window. After moving the window over the output of theneural network, it is found by observing the results that there are still some errors that aresimilar to the first case (a "1" or two among a string of zero's or vice versa). Therefore, thesliding window is moved for a second time to overcome these errors and enhance thesegmentation results. It must be noticed that each output of the neural network represent a10ms frame of the input. Therefore, merging more than one output of the neural network inone (by using the technique described above) is not an infinite process. The reason behind thisis that a speaker phrase can be as short as 200-250ms (for example the phrase "yes" or 'no").Adding the silence periods after and before, a speaker can have a maximum length phrase of350ms. Therefore, we stopped at a maximum window size of 150ms (the 3/5 or the 5/3 casewindow). The result of this enhancement step is illustrated in Tables 2,3,4.

Table 2: The results for the segmentation system for the male-male category

SIZE OF THE FIRSTWINDOW / SIZE OF

THE SECONDWINDOW

MAX. PFS(%) MIN. PFS(%) AVERAGE PFS(%)

0/0 31.8309 11.4971 22.8088

3/3 23.7504 4.6567 14.4969

3/5 23.9521 2.8031 10.9496

5/3 21.8584 2.9509 11.4644


69

Table 3: The results for the segmentation system for the female-female category


THE SECONDWINDOW


0/0 31.0738 16.6656 23.7910

3/3 22.7051 9.9061 15.1781

3/5 20.0984 2.6208 11.3113

5/3 20.0045 3.4088 11.5922

Table 4: The results for the segmentation system for the male-female category


THE SECONDWINDOW


0/0 24.2105 12.0422 18.0936

3/3 15.3839 3.1090 9.7224

3/5 13.2927 1.8487 6.1389

5/3 13.1810 0.6825 6.2379

5. ConclusionsThe performance of a probabilistic neural network based segmentation system has

been examined and enhanced in this paper. The system has been evaluated on the TIMITdatabase. The system has been examined in three different categories. It is found that for thesituation where the two speaker participating in a conversation have different sex the systemhad its best performance with a minimum average percentage of false segmentation (PFS)equal to 18.0936%. To enhance the performance of the system a sliding window has beenmoved along all the output of the neural network twice. This step enhanced the performanceof the segmentation system by reducing the value of PFS by approximately 66% for thedifferent sex speaker system and by approximately 52% for the same sex speaker system. Inaddition it was found that a sliding window with a size of 3 for the first time and 5 for thesecond time gave the best enhancement for most of the results. Noting that the worst case(female-female) PFS is 11.3%. Comparing that with the best result of [1] which is (MissDetection Rate) MDR =18.8% and (False Alarm Rate) FAR=21.8%. WhereMDR=MD/GTFAR=FA/(GT+FA)MD the number of miss detectionsGT the actual number of speaker turns, i,e. ground truthFA false alarmsPFS MDR + FAR.


70

6.References[1] M. Kotti, L.G.P.M. Martins, E. Benetos, J.S. Cardoso, C. Kotropoulos, "Automaticspeaker segmentation using multiple features and distance measures: a comparison of threeapproaches" ICME 2006 - IEEE 2006 International Conference on Multimedia & Expo,Toronto, Canada, July, 2006, pp. 1101-1104.[2] M. K. Omar, U. Chaudhari, and G. N. Ramaswamy "Blind Change Detection for AudioSegmentation", Proc. of ICASSP-05, Philadelphia, Pennsylvania, March 2005, pp. 501-504.[3] D. Wang, L. Lu, H.J. Zhang, "Speech segmentation without speech recognition",Proceedings of the 2003 IEEE International Conference on Acoustics, Speech and SignalProcessing, 2003, pp. 468–471.[4] M. Sharma, R. Mammone, "Automatic speech segmentation using neural tree networks"Proceedings of the 1995 IEEE Workshop on Neural Networks for Signal Processing, 1995,pp. 282-290.[5] F. Gorunescu, "Benchmarking Probabilistic Neural Network Algorithms", InternationalConference on Artificial Intelligence and Digital Communication, Research Center forArtificial Intelligence, (2006).[6] Kinnunen T., "Spectral Features for Automatic Text-Independent Speaker Recognition",Licentiate’s Thesis, University of Joensuu, Finland (2003).[7] Douglas A ., Thomas F., Robert B., "Speaker Verification Using Adapted GaussianMixture Models", Digital Signal Processing, Vol. 10, No. 1-3, pp. 19-51, (2000).[8] Adami A., Hermansky H., "Segmentation of Speech for Speaker and LanguageRecognition Conditions", In Proc. 8th European Conference on Speech Communication andTechnology (Eurospeech 2003), pp. 841–844, (Geneva, Switzerland, 2003).[9] Kinnunen T., "Optimizing Spectral Feature Based Text-Independent SpeakerRecognition", Ph.D. thesis, University of Joensuu, Finland, (2005).[10] Kinnunen T., "Spectral Features for Automatic Text-Independent Speaker Recognition",Licentiate’s Thesis, University of Joensuu, Finland (2003).[11] Garofolo J., Lamel L., Fisher W., "Darpa TIMIT Acoustic-Phonetic Continuous SpeechCorpus CD-ROM Manual", National Institute of Standards and Technology (NIST), (1993).[12] Ganchev T., Fakotakis N., Kokkinakis G., "One-speaker Detection – Limited Data: TheWCL-1 System", NIST 2003 Speaker Recognition Workshop, College Park, MD, USA,(June24-25 2003).[13] Ganchev T., Fakotakis N., Kokkinakis G., "Impostor Modeling Techniques for SpeakerVerification Based on Probabilistic Neural Networks" Signal Processing, Pattern Recognition,and Applications (SPPRA 2003), Rhodes, Greece, (6/30/2003 - 7/2/2003).[14] Ganchev T., Fakotakis N., Kokkinakis G., "Text Independent Speaker Verification Basedon Probabilistic Neural Networks", In Proceedings of the Acoustics 2002,Patras, Greece,pp.159-166(2002).[15] Demuth H., Beale M., "Neural Network Toolbox User’s Guide", Version 4,MATLABCD-ROM documentation, MathWorks Inc, pp. 7.12-7.20,( July, 2002).


Abdul-Razak: The Effect of ambient refractive index on the action of Long Period Fiber Grating

71

The Effect of ambient refractive index on the action ofLong Period Fiber Grating

Furat.y.Abdul-Razak

[email protected] Software Eng., University of Mosul,Iraq

ABSTRACT In this paper,we introduce a new model, of long-period fiber grating, by taking the value ofambient refractive index greater than the refractive cladding index, where the difference betweenthem must equal( 0.2).The results show the change in power attenuation coefficient, where itincreases with the increase in refractive ambient index. The power attenuation coefficient shiftshows a dramatic change of a sharp increase from 0.00 dB to 0.03788 dB.

Key Word: Long Period Fiber Grating, WDM, Erbium Doped Fiber Amplifier,

[email protected].

, ,)0.2(.

)dB0.00dB

0.03788.(


mailto:Furat_al2006:@yahoo.com

mailto:_al2006:@yahoo.com


72

1-Introduction:

Fiber Optic Gratings or Fiber Bragg Gratings (FBG) were first reported in 1978 by Hill etal[1]. However, such devices only attracted the researcher’s attention in1989, when newproduction techniques allowed their use with optical communication wavelengths. Severalmethods exist for FBG production, among them the direct phase mask writing and phase maskinterferometers stand out[1]. A new class of fiber grating called Long Period Fiber Grating (LPFG) was demonstrated byVengsarkar et al in 1996[1,2].The name is due to the refractive index change periodicity from100µm to 700µm, about 100 times larger than the values employed for FBG formation. Thisdifference makes possible the use of amplitude masks instead of phase masks, resulting in lowermanufacturing costs when compared to the FBG production’s costs. Besides this, the LPGpresents other surpassing characteristics such as low insertion losses, low back reflection,relatively simple fabrication, and a high sensitivity to changes in physical external parameters.These features made the LPG outstanding devices for application such as band rejection filtersand gain equalizing filters in optical communication, beyond its wide applicability as a fiber opticsensor[1,2]. A long-period fiber grating, which couples light from a fundamental guided core mode intoco-propagating cladding modes at various wavelengths. the LPG have also been used as gain-flattening filter and as optical fiber polarizer [2]. Unlike FBG the cladding mode configuration ofthe LPG is extremely sensitive to the refractive index of the medium surrounding the cladding,thus allowing it to be used as an ambient index sensor. The wavelength at which the couplingfrom core to cladding modes takes place is directly dependent on the difference between the coreand cladding indices, the dimensions of the core and cladding and grating period, and any changein these values can shift the transmission spectral profile [2]. Erbium Doped Fiber Amplifiers (EDFA) are indispensable tools for providing opticalamplification in wavelength-division-multiplexed (WDM) systems. However, it is difficult totransmit and amplify many WDM channels using EDFA’s since the gain profile is wavelengthdependent (nonuniform), while the transmission medium loss is, to first order, wavelengthindependent. This creates significant differences in the signal-to-noise ratios among the differentamplified WDM channels which may, depending on the system power budget or dynamic rangeof the receiver, cause system impairments and degrade performance[3]. Although gain-flattenedEDFA’s (uniform gain profile) have been fabricated, due to possible changes in operatingcondition and to network reconfiguration operations such as channel add/drop, variations can stillexist among the power levels of the WDM channels amplifier by an EDFA [3].

2-Wavelength Division Multiplexer (WDM):

A powerful aspect of an optical communication link is that many different wavelengths can besent along a single fiber simultaneously in the 1300nm-to-1600nm spectral band. The technologyof combining number of wavelength onto the same fiber is known as wavelength-division-multiplexing or WDM. Conceptually, the WDM scheme is the same as frequency-division-multiplexing (FDM) used inmicrowave radio and satellite systems. Just as in FDM, the wavelengths (or optical frequencies)


73

in WDM must be properly spaced to avoid interchannel interference. The key system features ofWDM are as follows[4,5]:

Capacity upgrade: The classical application of WDM has been to upgrade the capacity ofexisting point-to-point fiber optic transmission links. If each wavelength supports an independentnetwork signal of perhaps a few gigabits per second, then WDM can increase the capacity of afiber network dramatically.

Transparency: An important aspect of WDM is that each optical channel can carry anytransmission format. Thus, using different wavelength, fast or slow asynchronous andsynchronous digital data and analog information can be sent simultaneously, and independently,over the same fiber, without the need for a common signal structure.

Wavelength Switching: Whereas wavelength-routed networks are based on a rigid fiberinfrastructure, wavelength-switched architectures allow reconfiguration of the optical layer. Keycomponents for implementing these networks include optical add/drop multiplexers, optical crossconnects, and wavelength converters.

Wavelength Routing: In addition to using multiple wavelengths to increase link capacity andflexibility, the use of wavelength-sensitive optical routing devices makes it possible to usewavelength as another dimension, in addition to time and apace, in designing communicationnetworks and switches. Figure (1) shows the use of such components in a typical WDM link containing various typesof optical amplifiers[5].

3-Long-Period Fiber Grating Theory: Long period gratings are fiber optics based devices made up of periodic changes in core’srefractive index. Photo induced long-period fiber gratings (LPG) with periods

mLLPG32 1010 . LPG are transmission grating in which the coupling is between forward

propagation core and cladding modes, propagation in the same direction. Typically in a singlemode fiber(see figure 2) an LPG couples the fundamental guided core mode to a co-propagatingcladding mode at a coupling (or resonance) wavelength[2].

Fig.1 Implementation of a typical WDM network containing various types ofoptical amplifiers.

Span

1

2

N

1

2

N

Optical fiber

preamplifierIn- line amplifierpost amplifier

MUX

DMUX


74

The excited cladding mode attenuates in the coated fiber part after the grating, which results inthe appearance of resonance loss in the transmission spectrum. The interaction of one mode of a fiber with other modes is commonly described with the helpof coupled-mode theory in which only two modes are supposed to be nearly phase-matched andcapable of resonant coupling. Based on this theory, quantitative information about the couplingcoefficients and spectral properties of fiber grating can be obtained [10,11]. Two modes arecoupled by a grating with period L , if their propagation constants b1 and b2 satisfy the phasematching condition[6]:

Lpkbb /212 (1)where P and k is an integer describing the order of the grating and, in which the modecoupling occurs. Calculation methods of spectral characteristics of LPG’s can be found inpapers[12,13].Below will be considered the most important relation describing the gratingproperties. Equation (1) for the resonant coupling of the fundamental mode and one of thecladding modes can be rewritten as[6]:

LPGLPGcladeff

coreeff ILnn (2)

wherecoreeffn and

cladeffn are effective refractive indexes of the core and cladding modes,

respectively, and LPGI is the resonance coupling wavelength. In order to get a complete set of modes HEIM and EHIM ( I and M are azimuthally and radialorders of the mode, respectively ), the wave equation for a dielectric cylinder with a certain radialindex distribution should be solved. In single-mode fiber, only HE11 mode is guided by the fibercore at > C ( where C is the cutoff wavelength)[6,7]. Normally, a large quantity of modes ( N~104 at next =1) can be guided by the cladding (strippedfiber with 125µm cladding diameter). Nevertheless, only some of them have a significant overlapintegral I with the fundamental core mode. The integral should be taken in the fiber cross-sectionregion, where modulation of the refractive index has been induced (for photo induced gratings,the integration region usually coincides with fiber core)[6,7]:

LPGCladding

Fundamental

Grating length

Outer layer Namjacket

Core

Fig.2 Long-period fiber grating


75

0

2

0

*

0

2

0

*

0

2

0

*

rdrdEErdrdEE

rdrdEEI

cladcladcorecore

a

corecore

(3)

where a is the core radius Ecore and Eclad are the amplitude of the electrical field of the core andcladding modes, respectively, r is radial and azimuthal coordinates. The overlap integral Idefines the efficiency of inter-modal conversion. Its value is large only for HE1m (m>1) claddingmodes, because only these modes have a sufficiently great electric field component in the fibercore. Fig. (3) shows the energy-normalized radial distribution of the electric field for some HE1mcladding modes. These modes are linearly polarized, their intensity distributions are axiallysymmetric, and the number of zeroes in the radial direction is m-1[6,7]. The overlap integral increases with increasing the radial mode number up to m~10, which isaccompanied by an increase in the inter-modal coupling intensity. The latter can be seen from thetransmission spectra of LPGs (Fig.4).Starting with a certain value of m, the overlap integraldecreases to zero and thereafter oscillates with m, the amplitude of the oscillation tending tozero[6,7,8].

The solution of coupled mode equations in the approximation of two interaction modestraveling in the same direction and in the assumption of small amplitude of induced indexmodulation in comparison with the silica glass index, gives the following energy exchange law( for initial condition R(0)=1, S(0)=0)[9,10,11]:

)/())((sin))((cos)( 222/122222/1222 dhdhzddhzzR (4)

)/())((sin)( 222/12222 dhdhzhzS (5)

Fig.3 Radial distributions of the electric field

amplitude of the cladding modes , HE , HEFig.4.A typical transmission

spectrum of a LPG.


76

where )(zR and )(zS are the normalized energies of the core and cladding modes, respectively,considered as a function of z-coordinate along the fiber axis (the beginning of the gratingcorresponds to z=0). The normalized frequency d is:

)/1)(/()/(1 2LPGLPGeff IDLPIDpDnd (6)

the normalized frequency, describes the deviation from the exact synchronism; h is thecoupling coefficient defined by a relation:

LPGIICpDnh /mod (7)

modDn is the induced index modulation amplitude of the fiber core, related with the total inducedindex change indDn via relation 2/mod indDnDn ,C is a constant equal to the first coefficient inthe Fourier transform of the grating pitch shape. If the index profile is sinusoidal, this constant isequal to unity. For a rectangular profile, which is more typical for LPG, pLpxC LPG /)/sin(4 ,where p and x is the size of the irradiated part of the fiber within one grating period[12,13].

4-Results and discussion:

Here, we use matlab program version 6.5 to simulate and design the transmissioncharacteristics of an LPFG coupler. At the exact resonance (d=0), equations (4&5) gives asinusoidal law of the energy exchange, showing a possibility of mutual energy transfer from onemode to another:

)(cos)( 2 hzzR (8)

)(sin)( 2 hzzS (9)

Figure (5) shows the work style of LPFG. We can notice that transmission spectrum of thisdevice when there is one input signal whose wavelength is 1554nm, The effective indices of thecladding modes are dependent on the cladding index and the index of the surrounding (outer)ambient environment ( amn )(see fig.2). This mean that the nth cladding mode couplingwavelength will change as the index of the surrounding environment changes. Cladding modesare most accurately calculated using three-layer modes for case of clad

cladeffam nnn and

irradiated part within one grating period =0.5e-3. In this paper we focus on modeling of the leakyconfiguration (i.e. when nam> nclad ). The output signal whose of Erbium Amplifier, as shown in fig.(6),consists of two parts (theactivated emission signal and the spontaneous emission signal). This device removes part of thissignal when the resonance condition is realized (the resonance condition is 1557nm for thissignal) so as to equalizer (flatting) the gain when the gain is changed in accordance with usingseveral input signal to be amplified simultaneously (when there is a variety in the input power ofthis signal ).


77

The two-layer model, The power attenuation coefficient of the first order cladding mode beingcalculated over a wide range of nam . Figure (7) shows that the most sensitive regions are whennam is close to nclad and when nam changes from 1.889 to 2.113 (i.e. 0.2).

The power attenuation coefficient shift shows a dramatic change from a sharp increase of 0.00dB to a sharp of 0.03788 dB. This “ switching property “, in this sensitive region, has a potential

Fig.(5) Transmission spectrum of LPFG vs.wavelength at input signal (1554).

Fig.(6) Transmission spectrum of EDFAvs. wavelengths

Fig.(7) Effect the surrounding medium Fig.(8) Effect the surrounding medium


78

application in optical communications and optical sensors(i.e. work region of our LPFG modelwill be only in sensitive region). Fig.(8) shows the same effect result but with change to the core refractive index, claddingrefractive index and surrounding refractive index (nam>nclad ).From the above equations(from 1to 9), we can derive the equation that shows the relationship between input signal wavelengthand index ambient (i.e. choose the input signals wavelength that is flatting):-

cladamLPGwavelength NNLA (10)

Where A is fiber core area(1-10)µm2. From the above equation we can design our the LPFGmodeling in order to use it in flatting gain to EDFA when there are multi signals amplified onthe link. Our model consists of a range from (1.4 to 1.48) to index ambient, that is larger thancladding index (i.e. 0.2 ). Figs(9,10,11,12) show our model, where this figs show effectiveindex ambient on power attenuation coefficient, we notice a dramatic change and sensitive regionand power attenuation coefficient value at eff

ambN 1.61.

Fig.(9) Modeling index ambient Fig.(10) Modeling index ambient

Fig.(12) Modeling index ambientFig.(11) Modeling index ambient


79

)(nmeff cladN rangN ambeffambN>cladN n

1523 1.4 1.4213-1.61852 1.618524689 0.21852461533 1.42 1.4365-1.61767 1.617679101 0.19767911543 1.43 1.4472-1.60636 1.606366659 0.17636661553 1.44 1.4568-1.61602 1.616022964 0.1760229

From the figs and the table above notice the difference between cladding refractive index andsurrounding refractive index must be 0.2, i.e. working region will be in sensitive region( eff

ambN > cladN ). we will use our model in gain flatting to output signals for EDFA. Fig.(13) illustrates the output power of the EDFA with the wavelength when using two inputsignals to be amplified simultaneously( the value of the first one is -60dBm, and the value of thesecond -43dBm). We can detect the change that takes place in the amplifier output power withevery input signal ( that means there is a change in the amplifier’s gain when using more than oneinput signal with various powers to be amplified simultaneously(i.e. WDM technique )). Fig.(14) presents the flatting process for the change in the output power of the amplifier as wasshown in Fig.(13). The LPFG was fixed behind the amplifier so as to flat the out coming power as is made clearby Fig.(13),( we can compensate for the power loss by way of fixing another amplifier justbehind the LPFG). Fig.(15) illustrates the output power of the EDFA with the wavelength when using five inputsignals to be amplified simultaneously (wavelength used in modeling). It can be noticed a changein the amplifier’s output power at every input signal. Fig.(16) shows the flatting process of the change in the output power in the amplifier’s to thatwhich is shown in Fig.(15) using our modeling.

Table (1) Wavelength(flatting) used in modeling

Fig.(13) Output power to EDFA vs.wavelength with two signals.

Fig.(14) Output power to LPFG vs.wavelength with two input signals


80

5- Conclusions:

In this paper, We have presented the LPFG model by changing the ambient refractive indexhigher than cladding refractive index (nam>nclad ). The results may come up with many points:

1. The model works over a wide range (1.4-1.6).i.e. many inputs at the same time areflatting it (i.e. we have flat gain for the output power of EDFA over a wide range (1520-1565)nm).

2. is presented effective ambient index to LPFG working where the difference betweencladding refractive index and surrounding refractive index in our modeling must be equalto n 0.2.

3. The model may be used as an index sensor (when coated with a suitable material such aswater).

4. In this modeling, the best working point, is when power attenuation coefficient is of ahigher value 0.03788 dB (see fig.9,10,11,12).

5. In long-haul amplified WDM optical links, the characteristics of the amplifiedspontaneous emission (ASE) noise introduced by the in-line Erbium Doped FiberAmplifiers (EDFA’s) may be modified by fiber nonlinear phenomena such as parametricgain (PG). therefore, the ASE noise affecting the signal at the receiver may be a non-white random process, and may present a correlation between the in-phase andquadrature components. For the above reasons, LPFG’s were suggested to reduce thiseffect(i.e. to avoid gain saturation in EDFA’s that introduced by ASE noise in opticallinks. therefore, put LPFG’s in link to flat ASE noise to reduced effect PG ).

6-Referance:

1- Rosane Falate, Jose Luis Fabris, Marcia Muller and Hypolito Jose Kalinowski “Long PeriodGrating Sensor to monitor Fuel Quality”. Asian Journal of Physics.Av.Sete deSetembro,3165-80230-901. Curitiba,Brazil 2002.

2- Jinho Bae, Jun Kye Base,Joohwan Chun, ” Analysis for long period fiber gratings usingthermal kernel function”, Optical Society of America 2004.

3-Lawrence R. Chen, David J.F. Cooper, and Pwtwr W. E. Smith, ” Transmission Edge FiltersFor Power Equalization of EDFAs”, IEEE PHOTONICS TECHNOLOGYLETTERS.VOL.12,NO. 7, JULY 2000.

4-Jean-Pierre Laude,“ DWDM fundamentals, components, and applications” ,685 CantonStreet Norwood,MA 020682, Artech House Boston. London 2002. www.artechhouse.com.

5-Gerd Keiser,” Optical Fiber Communications”, McGraw-Hill Higher Education,InternationalEditions 2000,www.mhhe.com.

6-A.M. Vengsarkar, P.J.Lemaire,“ Long-period grating theory”,Glass photosensitivity andFiber Gratings, Fiber Research Center GPI RAN,2002.

7-O.Frazao,G.Rego,M.Lima,A.Teixeira.el,” EDFA Gain Flattening Using Long- Period FiberGratings Based on the Electric Arc Technique”, Electronics and TelecommunicationsDept.,University of Averiro,3810-193 Aveiro,Portugal.2002.

http://:@www.artechhouse.com./


81

8-Omer Mermer,” EDFA Gain Flattening by Using Optical Fiber Grating Techniques”,Department of Electrical and Electronics Engineering, Ege University, Bornova, Izmir,Thesis Advisor: Asst. Prof. Gokalp Kahraman.2003.

9-Kin Seng Chiang, “ Analysis of Two Parallel Long-Period Fiber Gratings”, Journal ofLightwave Technology, VOL.22,NO.5, MAY 2004.10-T.Tamir,ed.,” Integrated Optical”, Vol.7 of Topics in Applied physics, Springer-Verlag,1975.11-T.Erdogan, “Fiber grating spectra”,J.Lightwave Technol., Vol 15,pp. 1277-1294,1997.12-S.A.Vasiliev,E.M.Dianov,A.S.Kurkov,O.I.Medvedkov,V.N.protopopov, “photoinduced in-fiber refractive-index gratings for core-cladding mode coupling”,Quantum Electron., Vol 27,pp.146-149,1997.13-T.Erdogan,”Cladding-mode resonances in short-and long-period fiber grating filters”,J.Opt.Soc.Am.A, Vol 14, pp. 1760-1773, 1997.


Jalal :Cylindrical Manipulator Path Planning Among Static Obstacles Using Artificial Potential

82

Cylindrical Manipulator Path Planning Among Static ObstaclesUsing Artificial Potential Fields

Rawand Ehsan Jalal(Assistant Lecturer)

College of Engineering – Kirkuk University

Abstract In this paper, path planning for cylindrical manipulator of 4-DOF is studied. Another view point is

presented for using so called ‘artificial potential fields’ which is used as the base of searching for new andsafe points in the manipulator’s workspace among static obstacles. Three vectors are used for safemanipulator navigation. The first vector is determined between the end-effector and the goal points and it isused to attract the end-effector to the goal point. While the second and third vectors are computed frompoints defined on the obstacle and the end-effector. These two vectors are used to repel the end-effector andthe arm from the boundaries of the obstacles. In this work, the obstacles are suggested to have a cylindricalshape with different sizes. Displacement detections between the manipulator (its end-effector and arm) andthe obstacles are used as sensors for collisions impending. A random movement is suggested for joint two toavoid contacting between the arm and obstacles. At the off – line path mode, all path points are determinedby the presented method and some of them are updated, if an arm collision is detected, then joint variablesare calculated at each point. In real mode, these joint variables are fed to a simple real control system tomake the manipulator tracks the found path. The method gives a safe path for undertaken manipulator. Anexperimental work is also presented.Keywords Path planning, cylindrical manipulator, static obstacles, artificial potential fields,configurations, collision detections, line parametric equations, control system.

)(–

. " "

. .

. . .) (

..

. . ..



83

1- Introduction:

A robotic manipulator needs to be able to operate safely in cluttered, 3D environments. Inorder to achieve a manipulation task, a manipulator must be able to plan a path through itsworkspace to a goal location, while avoiding obstacles. Therefore, the robot path planningproblem can be defined as the problem of finding a collision-free path between two specifiedconfigurations among obstacles. Motion or path planning has received much attention for the pasttwo decades from robotics in order to:1- Automatically generate the movements of mobile robots and their arms;2- Automatically plan and program the motions of manufacturing robots and mechanical parts in

assembling products.Automation of path planning offers a number of advantages over the existing alternatives. Itrelives human workers of the continual burden of detailed motion design and collision avoidance,and allows them to concentrate on the robotic tasks at a supervisory level. Robots with anautomatic motion planner can accomplish tasks with fewer and higher-level operative commands. Although humans have a superb capability to plan motions of their body and limbs effortlessly,the path planning problem turns out to be a high complexity problem[1]. A great number ofdifferent techniques has been and are still developed in order to carry out efficient robot pathplanning. One of the most popular path planning method is based on the potential functionsutilization which is initially developed by Oussama Khatib in 1980. The basic idea behind thepotential field approach is to treat the robot as a point particle in the configuration space underthe influence of an artificial potential field U. The field U is constructed so that the robot isattracted to the final configuration while being repulsed from the obstacles. The negative gradientof the generated global potential field is interpreted as an artificial force acting on the robot andcausing variation on its movement. Nevertheless, as a main presented drawback, these methodscan result to a trapped robot in local minima generated by the same potential functions[2]. Thislimits the applicability of the artificial potential approach. Looking for a potential field withoutlocal minima has become a central concern in this approach. Harmonic functions are solutions toLaplace’s equation. Such functions can be used as advantage for potential-field path planning,since they do not exhibit spurious local minima. This is proposed by Connolly C. and GrupenA.[3]. Harmonic functions that are presented in their work have a number of properties which areessential to robotics applications and the derived paths are generally smooth. They also showhow these functions can be used as the basis for a reactive admittance control. Such schemesallow incremental updating of the environment model and respond well to sensed changes in theenvironment. However, the computational cost of the Laplace potential method will grow as anexponential function if the degree of freedom (DOF) of the robotic arms becomes larger[4]. Anew methodology named Evolutionary Artificial Potential Field (EAPF) is introduced byVadakkepat P., Tan K., and Liang W.[5] for real-time robot path planning. The suggestedartificial potential field method is combined with genetic algorithms, to derive optimal potentialfield functions. To avoid local minima associated with EAPF, they produce an algorithm namedescape-force. In their work, the potential field functions for obstacles contain tunable parametersand multi-objective evolutionary algorithm is also utilized to identify the optimal potential fieldfunctions. The results showed that the proposed methodology is efficient and robust for robotpath planning with non-stationery goals and obstacles. Wang Y. and Chirikjian G.[6] present a new artificial potential method. The model simulatessteady state heat transfer with variable thermal conductivity, and then the optimal path problem isthe same as a heat flow with minimal thermal resistance. The novelty of the presented technique


84

is that the thermal resistance in the configuration space for all different orientations of the robotcan be superimposed and this reduce a search on RnxSO(n) to one on Rn followed by a search onSO(n). Artificial potential field is not applied only to planning paths for serial manipulators; it has beenapplied to collision avoidance for humanoid robot arms by Sahara A, and Anzai[7] named“CAHRA”. There is a difficulty when robot developers make motions of humanoid robot arms,because the right and left arms may collide each other. CAHRA uses a potential method withvery small computational cost and can avoid collision between its arms in 97% without beingnervous. Shimoda S. and Iagnemma K.[8] propose a potential field-based method for high speednavigation of unmanned ground vehicles (UGVs) on uneven terrain. The potential field isgenerated in the two-dimensional “trajectory space” of the UGV path curvature and longitudinalvelocity. The proposed method is subjected to local maximum problems, rather than localminimum. The presented method succeeded in navigating a UGV between pre-defined waypointsat high speed, while avoiding unknown hazards.

Pervious approaches have made the path planning with artificial potential fields highlycomplex by employing distance functions that cause many local minima in the search methodsfor global minimum (the goal point). In this paper instead of using functions, vector methodshave been used for searching for new and safe points in the manipulator’s workspace andavoiding collision with the obstacles that are found in the workspace. Three vectors aresuggested. The first one is calculated between the end-effector and the goal points and it is usedto attract the end-effector to the goal point, the second one is produced between the probablecontacting point on the obstacle’s surface in the direction of the first vector and the end-effectorpoint and it is used to repel the end-effector from the obstacle boundaries. Finally, third vector isfound between the knowing (reference) point on the obstacle and the end-effector and it is used torepel the arm from the obstacle boundaries. Collision detection is a vital part of any path planner.Furthermore, because path planners spend most of their time on collision or distance queries, theefficiency of the collision detection algorithm will greatly affect the overall efficiency of theplanner[9]. In this work, the collision detection between the manipulator (its end-effector andarm) and the surrounding obstacles is sensed by determining the distances between them, and thisis done by finding the nearest point on the obstacle’s surface that collision would occur with it inthe direction of the first vector and find the distance between it and the end-effector. For the arm,the third vector is used to find the distance perpendicular to the arm from the obstacle’s referencepoint. For multi-obstacles, the distances between the hand and each obstacle are found in thedirection of the first vector and to reduce the computations, the above procedure is done for thecloser one. In the beginning of the presented method, the straight line from the starting and goalconfigurations is checked, and if a part of the line is found to be outside the manipulator’sworkspace, a new goal point is proposed and named a virtual goal point that connects the startand goal points with straight lines within the workspace.

This paper is organized as follows: kinematics and workspace of the manipulator are presentedfirst while method principles and vector computations are introduced secondly. Then, collisiondetections and virtual goal point are explained. Focus on how joint variable computations andobstacle representations are done followed by Experimental works and results. Finally,conclusions are given.


85

2- Manipulator Forward Kinematics and Workspace:

2-1 Manipulator Forward Kinematics: The coordinate frame assignment is depicted in Fig (1). Details about definitions of thecoordinates and Denavit-Hartenberg (DH) parameters can be found in [10]. The currentcylindrical manipulator has a4 = 0 which differs from that is in [10] and this is done to reducesome computations. Following the DH methodology, the general transformation matrix (forwardkinematics), which expresses the position and orientation of the gripper with respect to the arm’sbase (frame 0), is given as:

10000 21344

12131414141

1.2131414141

04 qdaSqCq

SqaCqqCqdCqCqSqSqSqCqaSqqSqdSqCqCqSqCq

T

and the DH matrix that gives the position and orientation of the frame 2 with respect to the arm’sbase is:

1000010

00

21

1211

1211

02 qd

SqaCqSqCqaSqCq

T ..(2)

which is useful in pervious sections. In matrices (1) and (2), Cqi & Sqi denote cos(qi) and sin(qi)respectively, and a2, a3, d1, d4 are constants and depend on manipulator dimensions andgeometry. The joint variables are: the relative angles between links one and two q1, and betweenlinks three and four q4 and links two and three extensions q2, & q3.

..(1)

x0

y0

z0

x1

y1

z1y2

x2

z2

x3

z3y3

o2

o0

o1 o3

o4

d1

q2

q3

q1

q4

a2

d4

a3

Prh

o4 = Ph

o2 = Po2

y4z4

Fig (1) Reference Coordinate Assignments and the Built in-House 4-link 4-DOF Cylindrical Manipulator

The built in-house cylindrical manipulatorEnd-effector(gripper) x4


86

2-2 Workspace:

The total workspace of a manipulator is defined as the space reachable by a reference point inthe hand[11]. The forward kinematics DH conversion can be written as:

10

00 qR

T nn ..(3)

where 0nR is a (3x3) rotational matrix and (q) is a (3x1) translational vector. The position

vector of a point on the end-effector of the manipulator can be written in the terms of jointcoordinates as:

X = (q) ..(4)where X is a position vector in Cartesian space, q Rn, q is a vector of joint variables and n is thenumber of DOF. For the current cylindrical manipulator, Eq (4) can be written as:

213

1.21314

1.21314

)(qda

SqaCqqCqdCqaSqqSqd

q

where q = [q1 q2 q3]T and its constraints are 00 q1 1900, 0 mm q2 431.8 mm and 0 mm q3 584.2 mm. To draw the manipulator workspace, each two joint variables are assumed to be in

its limits and combined to produce a set as [ itj

iti qq limlim , ], for i, j: 1 n; i j. By substituting each

set into Eq (5) and varying the third one, apart of the workspace boundary is drawn. Fig (2)shows the drawn workspace.

..(5)

Frame (0)

x0

y0

z0

x0

y0

Top View3D

Fig (2) The undertaken cylindrical manipulator workspace


87

3- Method Principles:In traditional artificial potential field methods, potential field U is assumed around the robot.

When the robot is close to obstacles, the potential becomes large, and when the robot is close togoal point, the potential becomes smaller. Then the robot moves to the direction F which is thesum of an attraction vector Fattract (x) to the goal and repulsion vector(s) Frep (x) from obstacles atthe current position (see Fig (3)). In other wards, the robot moves to the direction where thepotential U is smaller. In this work, instead of using potential function which can be a function ofdistance (for Fattract & Frep), a vector is used so that there will not be need of an optimizationsearch method that can easily trap into local minima. The vector that attracts the end-effector tothe goal configuration is computed between the end-effector point (the origin point of frame 4)and the goal point, it always directs to goal point, as following:

kzjyixPP hghghggh )()()( ..(6)where [ X]hg = Xg – Xh, Xg & Xh are coordinate vectors of the goal and current end-effectorpoints. To repel the gripper from obstacle’s boundaries, a second vector is used. This vector isdefined between the end-effector point and the probable collision point on the obstacle’s surfacein the direction of the first vector and it directs from the obstacle to the gripper. Eq.(7) gives therepel vector.

kzjyixPP OgOgOggO )()()( ..(7)where [ X]Og = Xh – XO, PO is the probable contacting point and XO is its coordinate vector. Eq(7) is computed and starts affecting the solution when a certain distance between the two pointsreaches a given value. That is:

0)7(Eq

hO PP ..(8)

where Da is the distance between PO & Ph and is given by Eq.(9) and is the influencing distanceof the obstacle.

)()()( OhOhOha zzyyxxD ..(9)For each obstacle found in the manipulator’s workspace and satisfies Eq (8), there will be a repelvector. After each vector is computed (including the third vector in the next section), the resultantof Eqs (6) & (7) is calculated and a new and save point can be found in the direction of theresultant as shown:

hOgh PPPPR ..(10)and

RRP inewh ) ..(11)

where R is the resultant vector, Ph)new is the new point (position) of the gripper, and i is a scalardetermines the step size. It is important that i be small enough that the robot is not allowed to

if aDif aD


88

jump into obstacles while being large enough that it does not require excessive computation time.Therefore, i is considered to have variable values depending on the distance to the nearestobstacle (Eq (9)) and empirical basis.

a

ai D

D;2.1;2.0

..(12)

4- Collision Detections and Virtual Goal Point:

4-1 Collision Detections:

A collision occurs when the robot contacts an obstacle in its workspace. If this workspace, thatis, the Cartesian space in which the robot moves, is denoted by W and the obstacle region, the setof all points in W that belong to an obstacle i, is denoted by i, i.e. i W. Also, denoting theCartesian space by X, the manipulator by and the sub set of the workspace that is occupied bythe end-effector by (X). The set of Cartesian space for which the robot collides with an obstacleis referred to as the space obstacle and it is defined by:

X = {X W (X ) 0} ..(13)

in which = i. The remaining portion of the Cartesian space is called the set of collision-freespace and is simply the set:

X free = X \ X ..(14)

Note that the definition of space obstacle in Eq (13) includes arm-collisions without self-collisions, since; the undertaken cylindrical manipulator does not self-collide. The collisiondetections that are presented below are true if the obstacles, which have cylindrical shapes, standwith its base parallel to the x-y plane of manipulator reference frame (frame 0).

Fig (3) Evolutionary Artificial Potential Field and Resultant Direction

O

G

R

O

FattractFrep

Direction of Safe movement

F G

O obstacle

G goal


89

4-2 End-Effector Collision Detection:The parametric equations of the line that joints Pg & Ph are calculated as:

[X] = [Xh] + [ X]hg. t, 0 t 1 ..(15)where t is a parameter. From Fig (4), the following computations can be done:

hg

CghgC

CO

Cg

Oggh

Oggh

OgOgOgOg

hghghghg

DD

XXP

DDDD

PPPP

PPPP

zzyyxxD

zzyyxxD

)(

)sin()cos(

)(cos

)()()(

)()()(

1

222

222

where Dhg, DOg are distances between Pg & Ph and Pg & PrO respectively, PrO is the obstacle’sreference point, is the angle between Oggh PPPP & . DCg & DCO are the distance between a

point (PC) on hg PP to Pg and the smallest distance from PC to the obstacle’s ( i) reference pointrespectively and PC is the coordinates of nearest point to PrO. Now, if PC i, this means thatthe end-effector is going to have collision with i in its direction toward Pg. Therefore, the closerpoint on the i surface to the end-effector must be found. This is done by finding the parametricequations of the line that extends between PC & Ph:

[X] = [XC] + [ X]hC. t, 0 t 1 ..(17)

where [ X]hC = Xg – XC. By changing t in step of (0.01), a new point P(k) on the hC PP isdetermined and if P(k) i, a new point P(k+1) is found and so on until a point P(k+m) iscomputed that i, then, P(k+m-1) is an approximation of the closer point which is denoted byPO .

..(16)

Ph

Pg

PrO

PC

Dhg

DOg

DCO

DCg

Top View

FrontView

PC

i

End-effector

PC

Fig (4) Two Cases of End-Effector Collision Detections


90

4-3 Arm Collision Detection: From a point above Ph directly at distance of a3 (Fig (5)), denoted by Prh and PrO on i a vectoris computed as:

kzjyixPP rOrhrOrhrOrhrOrh ... )()()( ..(18)

where [ X]rh.rO = XrO – Xrh., and from Prh and the original point of frame (2), Po2, another vectoris noted as:

kzjyixPP orhorhorhorh 2.2.2.2 )()()( ..(19)

where [ X]rh.o2 = Xo2 – Xrh. Po2 is given by the first three elements in last column in Eq (2). Fig(5) shows the above and the additional following computations:

22.

..

222.

2

21

)(

)sin()()()(

)(cos

rOhRrOreal

hOrRrO

rOrhrOrhrOrhhrO

orhrOrh

orhrOrh

zzDD

DDzzyyxxD

PPPP

PPPP

where is the angle between rOrh PP & 2orh PP , DrO.h is the distances between Prh & PrO, DrO.R

is the perpendicular distance between PrO and the arm and Dreal is the projection of DrO.R on thex0-y0 plane. After the shortest distance between the obstacle reference point and the manipulatorarm is determined, it must be ensured that the point which locates above or down PrO withdistance equal to z = (zh - zrO), i.e. (xrO, yrO, z), i, Fig (5), if it does, and Dreal be ROvi, q2must be actuated to avoid collision of the arm, where ROvi = radius of i + F and F is a scalar.The movement of q2 is suggested to be randomly as follows:A random step from q2 is obtained by randomly adding and subtracting a small fixed constant uto q2 at same time:

q2)random-step = q2 ± i.u ..(21)

where i = 1,2,…,m. At each direction, the arm collision is detected and if there is detection, Eq(21) is repeated with a step of 2u and so on until a collision in one direction is not detected. Then,that side is taken as a recommended movement for q2.

The above procedures for collision detections are repeated for every obstacle found inside theworkspace. For multi-obstacles, the collision computations become very expensive in time andsome computations are useless. To reduce the computations and save time, it has been suggestedto have collision calculations in the direction of the first vector gh PP for end-effector detections

..(20)


91

by assuming a cone around the vector with half angle of h, Fig (6a). The angle between gh PP

& rOih PP for each obstacle i is found as:

)(cos1

rOihgh

rOihghi PPPP

PPPP ..(22)

and if hi that i is undertaken. h has a critical value since it must reduce the computationsand in same time it must not ignore obstacles. If a collision is detected for more that one obstacle,DrO.h is projected on gh PP for each i, Eq (22), then obstacle with shortest distance is used forcollision detection.

2)( gh

OrhghOrhPP PP

PPPPPPproj

gh ..(23)

For arm collision, Fig (6) shows how obstacles are chosen. Every i makes angle i that00 900 i is neglected.

The Arm

(xrO, yrO, z) i

Po2

Ph

PrO

Prha3

DOr.h

DrO.R

The Armi

(xrO, yrO, z) i

Dreal

DrO.Rz

= (z

h- z

rO)

i

ROvi

Fig (5) The Arm Collision Detection and Checking For Point (xrO, yrO, z)

PrO2

PrO5

a- Only obstacles 1 and 2 are used for

end- effector collision detections.

b- Only obstacles 1, 2 and 6 are used

for arm collision detections.

Prh

PhPrO3

PrO4 PrO5

PrO6PrO1

Pg

h

PrO2

PrO1

PrO3

PrO4

PrO6Pg

Ph Po2

Fig (6) The Choice of Obstacles for Collision Detections


92

4-4 Virtual Goal Point:When a part of the line that joining Ph & Pg , gh PP lies outside the manipulator workspace,

the presented path planning approach fails to give a path because the points on the produced pathW , and to avoid that, a so called virtual goal point is suggested. This point is connected to Ph

from side and with Pg from other side (Fig (7)). The lines that join the three points must be insidethe workspace. For the current cylindrical manipulator, to find out if a part of gh PP lies outsidethe workspace, Eq (15) is used by eliminating the z- coordinate part and changing t at step of 0.1,then if 22 )]([)]([)( tytxtr is < rib a part of the line lies outside the workspace where rib is theradius of the inside workspace boundary. From Fig (7), the following computations can be done:

)2)(( gh XXm ..(24)

where m is the mid point of gh PP . The parametric equations of lines that connect points: B &m, Ph & Vg, Pg & Vg are:

[XB] = [Xm] + [ X]. tm, 0 tm 1 [Xh.vg] = [Xh] + [ X]. t, 0 t 1 ..(25) [Xg.vg] = [Xg] + [ X]. t, 0 t 1

where B is a point on the outside workspace boundary, Vg is the virtual goal point that producedafter each changing in tm, and X = XB – Xm, X = XVg – Xh, X = Xvg – Xg respectively. At eachchanging of tm, a new Vg is determined at which the lines gh VP and gg VP must be checked if

they are W and this is done as with gh PP using the last two parametric equations in Eq (24).In addition, Vg must i, therefore, Vg will be proper if:

gggh VPVP & ..(26)

If conditions (25) are not satisfied for each value of tm, then point m is replaced by:))(( dgh nXXm ..(27)

where 0 < nd < 1 is a random quantity. When the virtual goal point is presented, the path planningis divided into two parts; first part is from initial Ph point to Vg (Pg)secondary = Vg) where Vg playsthe turn of goal point, and second part from Vg to Pg (Pg)original) where Vg represents the Ph point.

W and Vg i

Fig (7) The Computations of Virtual Goal Point

O0

Ph

Pg

B x0

y0

mVg

Outside boundaryof the workspace

Inside boundarywith radius of rib

Top view


93

5- Joint Variables and Obstacle Space:5-1 Joint Variables Computations: The aim of any path planning algorithm is finding the manipulator configurations (joint variables) alongthe determined path. In the current implementation, three joint variables might be calculated (q1, q2, andq3) since q4 is used for changing the orientation of the end-effector only as can be seen from Eq (1). Theinverse kinematics (IK) is the inverse problem of finding joint variables in terms of the end-effector’sposition and orientation. The presented approach gives only the end-effect’ position (Eq (11)); therefore,using IK is not possible. To solve this problem, the following simple methods have been used:For q1: After Ph)new has been found, two vectors can be drawn from frame (0) to Ph and Ph)new denoted by

ho PP 1 & hno PP 1 which can simply dotted to find q1 without z- coordinates:

)(cos11

111

1

hoho

hnoho

PPPP

PPPPq ..(28)

For q2: Movement in the z0-direction is the responsibility of joint two only. Therefore, its value can befound from Eq (5) as:

132 dazq hn ..(29)For q3: Since q1 and q2 have been found, the second (or first) element in Eq (5) can be used to find q3.

)sin()cos()sin(

or)cos(

)sin()cos(

1

12143

1

12143

qqaqdxq

qqaqdyq

hn

hn

..(30)

Always 00 q1 < 900 ( i is small) and if there is a division by zero (q1 = 00) switching in using of Eq (29)is suggested.5-2 Obstacle Space:

In last sections, X free is determined among obstacles that are assumed to have cylindrical shapes withdifferent sizes and are located in the manipulator’s workspace with their bases parallel to x0-y0 plane. Allpoints that W and i, must be defined to the path planner so that X be known. In this work, i issimply represented by a circle equation with z- coordinate for height as:

ii

iiiii

HzRuygx 222 )()(

..(31)

where (gi, ui) and Ri & Hi are the reference point, radius and height of i, respectively. (gi, ui) is alwaysassumed to be located at the base of i and the sign is used to ensure that all point i be represented.

6- Experimental Work and Results:

In this work, the manipulator movement from an initial configuration to final (goal) one is done amongtwo modes. In the off – line path mode, a path is planned (if a one exists) through the given obstacles. Ateach point path, the three joint variables (manipulator configuration) are determined and saved in sets.In real-time mode, each saved set is sequentially fed as command signals into the control system so thatthe end-effector can track the path. Joint actuators, optical sensors and a PC (as a controller) form thecontrol system. More details about the control system and interfacing can be found in [10]. The overall

i


94

path planning approaches are illustrated as a flow chart and diagram in Fig (8). Two different sets ofinitial and goal configurations are used as inputs to the path planner. In set one, calculation of virtual goalpoint is needed and in set two it is not. The resulted manipulator actions can be seen in Fig (9).

7- Conclusions:

This paper presents a path planning for cylindrical manipulator of 4-DOF based on artificial potentialfield. Instead of using functions for safe path searching, vectors have been used with distancecomputations for collision detections. The suggested vector based potential field does not need anoptimization search method for searching of a goal point which can easily fall into local minima. It ispossible to build X free through the suggested with non parallel base to the x0-y0 plane (arbitrary) byintroducing the Euler rotation matrix with respect to frame (0). The proposed collision detections (withlittle improving) can be used for other obstacle shapes. If the orientation of the end-effector is known(desired), it will be possible to use the proposed path planner with IK or any other improved solutions toplan a path for robotic arms with more DOF. For not very complex mediums of obstacle shapes andnumbers, the vector based potential field can be used successfully for path planning.

Diagram of the two modes

Flow chart of off-line path mode

Fig (8) The Two modes of path finding and tracking

Input: Initial and goalconfigurations and

Forward Kinematics Eq (1)Ph and Pg)original

IsNo

Calculate the virtual goalpoint Eqs (23) through (26)

Yes

Calculate the attractivevector Eq (6)

Pg)secondary = Vg Calculate the angle Eq (22)

For each i, calculate the distancealong PP Eq (23)

Calculate the repel vector Eq (7)

Yes Are there morethan one i

Is a collisiondetected?

Yes

No

No No

The repel vector = 0

Calculate the resultant vector Eq (10)

Calculate the new safe point Ph)new Eq (11)

Are arm collisionsdetected?

YesCalculate q2 Eq(21)

No

Save each point in setsYes

Is Ph - Vg ?

Vg = Ph

YesIs Ph - Pg ?

No

End

W ?

Is Vg existed?

Yes

No

For i with shortest

distance, is a collisiondetected?

Start

PC in off-line path mode works as a pathfinder

PC in real-time mode works as acontroller

Path Planner

Computed Point Sets

Computing Joint VariablesEqs (28-30)

Joint actuators

Joint displacements

PC


95

Path planningamong sixobstacleswithout Vg

calculation

Manipulatornavigation withVg calculation

Fig (9) Results of experimental works


96

References:

[1] Pang C. Chen and Yong K. Hwang “SANDROS: A Dynamic Graph Search Algorithm forMotion Planning” IEEE Transaction on Robot and Automation, Vol. 14, No. 3 June 1998,390-403.

[2] Planas R.M., Fuertes J.M. and Martinez A.B. “Qualitative Approach for Mobile Robot PathPlanning Based on Potential Field Method” Automatic Control Dept. Technical University ofCatalonia, Span, 2002.

[3] Christopher I. Connolly and Roderic A. Grupen “On the Applications of Harmonic Functionsto Robotics” Journal of Robotic System, Vol. 10, No. 7, 1993, 931-946.

[4] Carlos Vazqez znd Jan Rosell “Use of Path Planning Techniques Based on HarmonicFunctions for the Haptic Guidance of Teleoperated Assembly Tasks” Mechatronics androbotics Dept. University of Politecnica, Germany, 2007.

[5] Prahlad Vadakkepat, and Wang Ming-Liang “Evolutionary Artificial Potential Fields andTheir Applications in Real Time Robot Path Planning” Dept. of Electrical Eng. NationalUniversity of Singapore, Singapore, 1999.

[6] Yunfeng W. and Gregory S. Chirikjian “A new Potential Field Method for Robot PathPlanning” International Conference on Robotics & Automation, San Francisco, USA, April,2000, 977-982.

[7] Akiyoshi Sahara and Yuichiro Anzai “CAHRA: Collision Avoidance System for HumanoidRobot Arms with Potential Field” Faculty of Science and Technology, Kieo University,Japan, 2004.

[8] Shingo S. and Karl I. “Potential Field Navigation of High Speed Unmanned Ground Vehicleson Uneven Terrain” International Conference on Robotics & Automation, Barcelona, Spain,April, 2005, 2839-2844.

[9] Morten S. “Robot Path Planning: an Object-Oriented Approach” PhD thesis, Dept. OfSignals, Sensors and Systems, Royal Institute of Technology, Sweden 2004.

[10] Rawand E. J. “Displacement Detection Application of HDNS A2051 Optical Sensor forControlling a 4-DOF Cylindrical Manipulator” will be published in Journal of Al-RafidainEng. College of Eng. Mosul University, 2008.

[11] Kishna C. “Mechanics and Control of Robots” Springer, USA, 1997, 171p.

Ahmad: An Experimental Study of Parameters Affecting a Heat Pipe Performance

97

An Experimental Study of Parameters Affecting a Heat PipePerformance

Dr. Hussain H.Ahmad Raqeeb H. Rajab College of Engineering College of Agriculture and forestry

University of Mosul University of MosulAbstract

An experimental test rig was designed and manufactured to investigate theperformance of a heat pipe(HP). The heat pipe consists of a stainless steel pipe lined with athree-layer stainless steel mesh wick. The evaporator section of the heat pipe wassurrounded by three heaters representing the heat source. The condenser was jacketed withgalvanized cylinder to accommodate the cooling water flow. The entire HP was insulated.Different affecting parameters were investigated experimentally in this study including thepower input the filling charge of the working fluid(water) represented by a volumetric ratiowith respect to evaporator volume and the inclination angle with a horizontal line. All testswere carried out at a pressure around the atmospheric pressure during steady stateconditions. The experimental results showed that the conductivity was about (2060) timesthat of the solid piece of the stainless steel (the material of the HP).A comparison betweenthe present work results with empirical and theoretical correlations of other researchersshowed a good agreement.

Key words: heat pipe , wick , filling ratio

.

/ /

.) .(

, . . ,) (

.)2060 ( .

.



98

Nomenclature

Symbol Meaning UnitM2AREAA

J/kg. CSpecific heat capacityCp---Constant, determined from experimental dataCsfmDiameterD%Filling ratioFR

m/s2Gravitation accelerationgW/m2. CHeat transfer coefficienth

J/kgLatent heat of vaporizationhfg

---Heat pipeHPm2PermeabilityK

W/m. CThermal conductivitykmPipe lengthL

N/m2PressureP- - --Prandtl number = kpCPr

WHeat transfer rateQW/m2Heat fluxq

m ResistanceRCTemperatureT

WPowerW

Pumping head N/m2

Viscosity kg/m.sDensity kg/m3

Surface tension N/mAngle of inclination degreeWet angle degree

Subscriptsa Adiabatic ent Entrainment o Outerav. Average eff. Effective v Vaporbo. Boiling exp. Experimental vis Viscousc Condenser in Inner s Soniccap. Capillary l Liquid. w Walle Evaporator t Total ws Wick structure


99

1: Introduction

The heat pipe is a device that efficiently transports heat from one end to another. Itutilizes the latent heat of the vaporized working fluid instead of the sensible heat. The two-phaseheat transfer mechanism results in heat transfer capabilities from one hundred to several thousandtimes that of an equivalent well known conductor such as copper[1,7]. The heat pipe operates ona closed two phase cycle. Figure(A) shows a schematic of a typical heat pipe. Inside the heat pipethere is liquid-vapor equilibrium and when the heat is supplied to the evaporator, this equilibriumbreaks down as the working fluid evaporates. The vapor, which then has a higher pressure, movesinside the HP to the condenser section where it condenses. Thus, the vapor gives up the latentheat of vaporization and transfers heat from the input to the output end of the heat pipe. Thecapillary pressure created by the menisci in the wick pumps the condensed fluid back to theevaporator section. Therefore, the heat pipe can continuously transport the latent heat ofvaporization from the evaporator to the condenser section. This process will continue as long asthere is a sufficient capillary pressure to drive the condensate back to the evaporator. For the lastfour decades, two-phase passive heat transfer devices like heat pipes were used. A considerableexperimental and theoretical works have been done on the application and design modificationfor improving heat pipes performance. Yahya [1] carried out an experimental study concerned thedevelopment of an indirect type of solar cooker using a heat pipe for transporting energy fromthe focal spot. Frank[2] conducted an experimental investigation of improved injection lanceswith heat pipe having two wraps of stainless steel mesh as a wick structure in the heat pipe.Kempers et al.[3] carried out an experimental study to determine the effect of the number of meshlayers and amount of working fluid on the heat transfer performance of copper–water heat pipeswith screen mesh wicks. Jianlin et al.[4] described a proposed capillary pumped loop(CPL) usingthe multi-layer copper mesh as the capillary structure in the evaporator. Shwin-Chung et al. [5] intheir work presented visualization of the evaporation/boiling process and thermal measurementsof operating horizontal transparent heat pipes.

The most obvious pointer to the success of the heat pipe is the wide range of applicationswhere its unique properties have proved to be beneficial. Some of these applications of the heatpipe have played an important role in a variety of engineering heat transfer systems. It would be adifficult task to list all the applications of heat pipes; therefore, only a few important industrialapplications are given in this section. In the aerospace industry, heat pipes have been usedsuccessfully in controlling the temperature different component. Heat pipes have been applied incooling different electronics devices. Other applications include cooling of turbine blades,generators, motors. Also heat pipe are used in heat collection from exhaust gases, solar andgeothermal energy. In general, heat pipes have advantages over many traditional heat-exchangedevices when heat has to be transferred isothermally over relatively short distances, low weight isessential and low maintenance is mandatory[6].

There has been a potential consideration of using heat pipes recently because of the widerange of applications specially in cooling electronic systems. The present work investigatesexperimentally heat transfer characteristics performance of HP by designing and construction ofan experimental test rig studying the affecting parameters. The experimental results of this workwere compared with well known theoretical and empirical correlations of Rohsenow andImura[6,7,13].


100

Figure(A):Schematic diagram of the HP and its principles of operation[9].

Figure(A):Schematic diagram of the HP and its principles of operation[9].

2: Transport LimitationsThe heat input to the heat pipe can be limited to a certain value beyond which the heatpipe failure or works at low performance. So it is essential for a designer to examine all

types of limitations to be sure that it works perfectly at high level of performance.Limitations of the maximum heat input that may be transported by a heat pipe can be

divided into two primary categories: limits that result in heat pipe failure and limits that donot[9]. For the limitations resulting in heat pipe failure, all are characterized by insufficient liquidflow to the evaporator for a given heat input, thus resulting in dry out of the evaporator wickstructure. The two categories and basic phenomena for each limit may be summarized asfollows:

2.1: Limitations(Failure): limitations include:2.1.1:Capillary Limit: The ability of a capillary structure to provide the circulation for a givenworking fluid is called capillary limit or hydrodynamic limit. It occurs when the pumping rate isnot sufficient to provide enough liquid to the evaporator section[6].To understand this heattransfer limit, it is essential to know the capillary action and the phenomenon that governs it. Formost heat pipes the maximum heat transfer due to the capillary limitation can be expressed asfollows[6,10,12].

cos)(2

.max, tgL

l

l

capr

effLws

l

fghllcapQ (1)

Where; effL is the effective length of the HP =0.5(Le+2La+Lc)

2.1.2:Boiling Limit: When the radial heat flux in the evaporator section becomes too high, theliquid in the evaporator wick boils and the wall temperature becomes excessively high. The vaporbubbles that form in the wick prevent the liquid from wetting the pipe wall, causing hot spots.


101

Severe case of this phenomenon is complete evaporator dry out. This is known as the boilinglimit.

An expression for the heat flux beyond which bubble growth will occur can be writtenas[6,10,12]:

max,2

ln

)(2

.max, capnrvrirvfgh

vT

wseffkeL

boQ (2)

2.1.3:Entrainment Limit: In a heat pipe, liquid and vapor move in opposite direction. A shearforce exists at the liquid-vapor interface. High vapor velocity may cause some droplets of liquidto be carried away with the vapor, back to condenser. This inhibits the return of the liquid to theevaporator. A method to determine the entrainment limitation using Weber's number criterion(the Weber number is defined as the ratio of viscous shear force to the surface tension) as follows[6,12].

5.0

2.max, caprv

fghventQ (3)

2.2:Limitations (Nonfailure):2.2.1:Sonic Limit: The evaporator and condenser sections of a heat pipe represent a vapor flowchannel with mass addition and extraction due to evaporation and condensation, respectively.Sonic limitation is analogous to a converging-diverging nozzle with a constant mass flow rate.The vapor velocity increases along the evaporator and reaches a maximum at the end of theevaporator section[12]. An expression for this limit derived from one dimensional vapor flowtheory[6,9,10,14]with a final form:

2/1)(474.0.max, vvfghvsoQ (4)

2.2.2:Viscous Limit: At low operating temperature, the vapor pressure difference betweenevaporator and condenser regions of a heat pipe may be extremely small. The viscous forceswithin the vapor region may be dominant over the pressure gradient because of the temperature.In this condition, the pressure gradient may not be sufficient to generate flow and the vapor maystagnate. Mathematically, this limit can be expressed as[6,9,12]

effLv

vvfghrv,vis.

Q16

2

max (5)

2.2.3:Condenser Limit: The heat transfer rate in the condenser section is governed by thecoupling of the condenser with the system heat sink. At steady state, the heat rejection rate in thecondenser must equal the heat addition rate in the evaporator. Typically, the condenser couplingis either by convection and/or radiation[9].To reach the condenser limit it can be low convectiveheat transfer coefficients (e.g., natural convection), low surface emissive, or limited surface area.The heat transfer (outlet heat) from the condenser, cooled by water, is determined by:

oQ)TiTo(pC.mQc

(6)

Additionally, the capillary, viscous, entrainment, and sonic limits are axial heat fluxlimits, that is, functions of the axial heat transport capacity along the heat pipe. However, theboiling limit is a radial heat flux limit occurring in the evaporator. The maximum theoretical


102

power input limitations calculated from equations 1,2 ,3,4and 5 are shown in table(1).Thesevalues are essential to indicate the maximum input power limit, which were considered in thedesign of the present experimental rig.

Table(1): Summarize the maximum theoretical power input limitations.QIN(W)

QMAX,BO.(W)

QMAX,CA

P.(W)

QMAX,EN

T.(W)

QMAX,SO.(W)

QMAX,VIS

.(W)

1500 2396 4629 20429 32639 10986

3: Experimental ApparatusThe heat pipe body , was made from stainless steel pipe with length of (1200)mm, outside

and inside diameter of (48)mm and (43)mm respectively. The heat pipe is heated by electricalcoils clamped on the evaporator and it is cooled by water flowing through a jacket along thecondenser. Between the evaporator and condenser there is adiabatic zone. The pipe lined withcapillary structure of three layer stainless steel screen wire mesh (2×150 and 1×80).

Twenty two calibrated thermocouples (chromel – alumel ; type K) were used intemperature measurement that distributed along and around the entire length of HP. Thethermocouples were embedded, along (110)mm equally spaced, in 1.5mm depth groovesmachined on the outer surface of the wall. An electronic reader [model:E5C4,range (0 - 400 ºC),type K],with a resolution of (1ºC) was used to display temperature readings directly. The sheathsof the thermocouples were fully insulated.

As shown schematically in figure (B) the evaporator has a length of (360)mm, heatedelectrically by three clamped heaters. The end of the evaporator was sealed with a cup equippedwith a drain valve, and another valve connected to a glass level measurement to determine theworking fluid level in the evaporator. In order to measure the average temperature of theevaporator, eight thermocouples were distributed along and around it.

The condenser has a length of (500)mm, cooled by water flowing through a jacketof(120)mm diameter and (510)mm in length, fabricated from a galvanized plate of (2)mmthickness. To measure the average temperature of the condenser, six equally spacedthermocouples distributed along and around the external circumference of the condenser.

The segment between the evaporator and the condenser is normally referred to as theadiabatic section with a length of (340)mm. To measure the average temperature of the adiabaticsection, six thermocouples were distributed along and around it.

The power was supplied to the evaporator by electrical coil heaters (500 Watt each )mounted between two layers of mica . The length of each heater is (100)mm and (4)mmthickness. The heaters were fixed tightly around the outside surface of the evaporator to insuregood contact with its outer surface. surrounding by (20)mm thick asbestos. In order to reduceheat losses to the surrounding, the whole length of the heat pipe was wrapped by two layers offiber glass.

An accurate wattmeter covering the anticipated power range, and a variac wereincorporated in the heater electrical circuit to record the exact power supplied, as shown infigure(B).The heat output of the condenser was calculated from the amount of cooling waterflowing through the condenser and the temperature difference between the inlet and outlet ofcooling water. The water mass flow rate was measured by an accurate rotameter.


103

Thermocouple board

Water jacket

Variac

Electric source

A

v

Ameter

Voltmeter

Wattmeter

Water out

Water in

Flowmeter

Water source

Pump

200

Condenser section Adiabatic section Evaporator section

110110 110110110

water out

10

T1T2T3T4

11011020

43 Heater1Heater2Heater3Dranage valve

liquid level gauge

Notes :

1- Thermocouple position.2- T1,T2, T3 and T4 represents the average temperature ofcircumference thermocouples in the indicated levels respectively3- All dimensions in mm.

Outlet waterthermocouple

Inlet water thermocouple

Thermocouple position

Thermocouple position

Figure(B):The experimental test rig diagram showing locations heaters and thermocouples.

4: Result and Discussion4.1: Start-up operation

Heat pipe operation processes can be divided into three stages : start-up, transition andsteady state. The first stage is startup, and HP startup tests are the most critical for evaluating thecapillary evaporator reliability[10].The startup time for low heat loads was longer than that forhigh heat loads. Figure(1) shows the distribution of temperatures of the evaporator surface atdifferent positions and different time intervals when the heat input to the evaporator was 250Wand 1500W, the filling ratio was 75%,and angle of orientation was 90o. At low heat load, thestartup time takes about 90 min. to complete, while at high heat load the startup time takes about60 min.. The startup HP needs some time in order to discharge excess liquid and to stabilize thevapor-liquid interface at the top. At the end of the startup process a transition stage begins, andone can observe the scattering in wall temperatures. The steady process is indicated by thestability of the HP pressure at nearly one atmosphere, and the stability of the wall temperatures.

4.2:Heat Pipe Thermal Resistance and Thermal Conductivity:Mathematical correlations were used to calculate each thermal resistance. The effective thermalconductivity of the heat pipe is defined as the heat transfer rate divided by the temperaturedifference between the heat source and heat sink. Under normal operating conditions, the totalthermal resistance is relatively small according to the low value of the difference between thetemperature of heat source in the evaporator and the heat sink in the condenser represented by

T). Thus, the effective thermal conductivity in a heat pipe can be very large[14]. The quantity


104

kAL . is equivalent to a thermal resistance R of the HP wall against the flow of heat byconduction; kALR . .The reciprocal of the thermal resistance R1 , refers to the thermalconductance ( LkA. )[15]

av.QT

tR

.QavL

Ak

A tR t

L tHPeffk )( (7)

From equation (7) effk was found to be ( 35638 W/m.oC )Compared to the conductivity of sold stainless steel (16.4 W/m.oC),the maximum thermalconductivity ( ..ststeff kk ) = 2060

Figure(1):Temperature variation of the HP wall surface during the startup process for watercharge and different heat input (Refer to Fig.B).

T i m e ( m i n . )

0 3 0 6 0 9 0 1 2 0 1 5 0 1 8 0 2 1 0 2 4 0 2 7 0 3 0 0 3 3 0 3 6 0 3 9 0

Tem

pera

ture

(o C)

1 0

2 0

3 0

4 0

5 0

6 0

7 0

8 0

9 0

1 0 0

1 1 0

1 2 0

1 3 0

1 4 0

1 5 0

W a l l s u r f a c e t e m p e r a t u r e a t t h e e v a p o r a t o r e n d c u p ( T 1 )W a ll s u r f a c e t e m p e r a tu r e a t th e e v a p o r a to r to p (T 4 )C o n d e n s e r t e m p e r a t u r e .

T i m e ( m i n . )

0 3 0 6 0 9 0 1 2 0 1 5 0 1 8 0 2 1 0 2 4 0 2 7 0 3 0 0 3 3 0 3 6 0 3 9 0

Tem

pera

ture

(o C)

1 0

2 0

3 0

4 0

5 0

6 0

7 0

8 0

9 0

1 0 0

1 1 0

1 2 0

1 3 0

1 4 0

1 5 0

W a l l s u r f a c e t e m p e r a t u r e a t t h e e v a p o r a t o r e n d c u p ( T 1 )W a l l s u r f a c e t e m p e r a t u r e a t t h e e v a p o r a t o r t o p ( T 4 )

C o n d e n s e r t e m p e r a t u r e .

F R = 7 5 % w a t e r Q i n = 1 5 0 0 W

= 9 0 o

F R = 7 5 % w a t e r Q i n = 2 5 0 W

= 9 0 o

( b )

( a )


105

4.3: Effects of Filling Ratio and Power input on the temperature distributionThe amount of liquid charge is governed by two considerations; too small quantity can

lead to dry out, while an excess of liquid can lead to a condition where an amount of the liquidbeing carried up to the condenser causing a blockage of surface preventing condensation in thecondenser. In this study the filling ratio of the working fluid in the HP is defined as the ratio ofthe working fluid volume to the entire volume of the evaporator,(i.e. fluid charge rated to the totalvolume of the evaporator). Three different filling ratio(25%,50% and 75%)were considered. Theinventory volume must be able to accommodate at least the liquid volume swing and densitychanges between the hot section and the cold section of the HP [5]and it must be enough tosaturate the wick. At any operating mode of the heat pipe, both liquid and vapor phases have tocoexist in it. Hence, it is worthy to study the filling ratio effect on the heat pipe performance.Figures (2)and(3) show the variation of the temperature along the heat pipe for three differentfilling ratios with heating load rates of 250 and 1500 W at a given inclination angle. Thetemperature starts to increase along the evaporator to a maximum value and decreases until theadiabatic zone.

Figure(2):Variation of average wall temperature along the HP with different water filling ratiosfor different inclination angles.

Q in= 25 0W=2 5o

D ista nce(m m ) a lo ng H P.

0 20 0 4 00 60 0 8 00 10 00 12 00

(Tw

in) av

.[o C]

7 0

8 0

9 0

10 0

11 0

12 0

Q in= 250 W=9 0o

D istance(m m ) alo ng H P .

0 200 40 0 6 00 80 0 10 00 1 20 0

(Tw

in) av

.[o C]

7 0

8 0

9 0

10 0

11 0

12 0

F R= 7 5%F R= 5 0%F R= 2 5%

F R = 75 %F R = 50 %F R = 25 %

Q in= 25 0W

=5 0o

D ista nce(m m ) a lo ng H P.

0 20 0 4 00 60 0 8 00 10 00 12 00

(Tw

in) av

.[o C]

7 0

8 0

9 0

10 0

11 0

12 0

F R= 5 0%F R= 7 5%

F R= 2 5%

Q in=2 50W= 75 o

D istance(m m ) along H P .

0 20 0 4 00 60 0 8 00 1 00 0 12 00

(Tw

in) av

.[o C]

70

80

90

1 00

1 10

1 20

F R = 75%F R = 50%F R = 25%


106

Then the temperature decreases sharply(due to cooling) when vapor enters the condenserand stays constant with very little variation. At all heat transfer rates, however, the temperature atthe end of the evaporator is always lower than the other evaporator temperatures. The reason ofthis lower value is not clear[7], but probably due to that the end of the evaporator was notcompletely insulated, resulting in a heat leak from this point. Also from the figures above, ingeneral, the temperature is the highest when the filling ratio is(25%). At all input powersfigures(4)and(5) show that as the power input increases the temperature also increases regardlessof the filling ratio at a given inclination angle. The increase of the temperature along theevaporator section of the heat pipe is due to the phase change of the working fluid from liquid inthe end cup of the evaporator to mixture (vapor and liquid) and then to single phase (vapor)where the maximum temperature difference occurs, due to returned liquid from condenser to theevaporator ( because of the capillary effect) the temperature decrease again.

Qin=1500W

Distance(mm) along HP.

0 200 400 600 800 1000 1200

(Tw

in) av

.[o C]

80

100

120

140

160


0 200 400 600 800 1000 1200

(Tw

in) av

.[o C]

80

100

120

140

160

FR=75%FR=50%FR=25%

Qin=1500W


0 200 400 600 800 1000 1200

(Tw

in) av

.[o C]

80

100

120

140

160

Qin=1500W


0 200 400 600 800 1000 1200

(Tw

in) av

.[o C]

80

100

120

140

160

Qin=1500W

FR=75%FR=50%FR=25%

FR=75%FR=50%FR=25%

FR=75%FR=50%FR=25%

Figure(3):Variation of average wall temperature along the HP at different water filling ratios fordifferent inclination angles.


107

Distance along HP(mm)

0 200 400 600 800 1000 1200

(Tw

in) av

.[o C]

80

100

120

140

160


0 200 400 600 800 1000 1200

(Tw

in) av

.[o C]

80

100

120

140

160


0 200 400 600 800 1000 1200

(Tw

in) av

.[o C]

80

100

120

140

160


0 200 400 600 800 1000 1200

(Tw

in) av

.[o C]

80

100

120

140

160


0 200 400 600 800 1000 1200

(Tw

in) av

.[o C]

80

100

120

140

160


0 200 400 600 800 1000 1200

(Tw

in) av

.[o C]

80

100

120

140

160

FR.=50%=90o

FR.=25%=90o

FR.=75%

FR.=50%=75o

FR.=75%o

FR.=25%=75o

Qin=1500WQin=1250WQin=1000WQin=750WQin=500WQin=250W

Qin=1500WQin=1250WQin=1000W




Qin=1500W

Qin=1250WQin=1000W



Figure(4):Variation of average wall temperature along the HP at different water filling ratios fordifferent powers.


108


0 200 400 600 800 1000 1200

(Tw

in) av

.[o C]

80

100

120

140

160

180

FR.=75%=25o


0 200 400 600 800 1000 1200

(Tw

in) av

.[o C]

80

100

120

140

160

180

FR.=50%=25o

0 200 400 600 800 1000 1200

(Tw

in) av

.[o C]

80

100

120

140

160

180

FR.=25%=25o

Qin. =1500WQin. =1250WQin. =1000WQin. =750WQin. =500WQin. =250W




0 200 400 600 800 1000 1200

(Tw

in) av

.[o C]

80

100

120

140

160

180


0 200 400 600 800 1000 1200

(Tw

in) av

.[o C]

80

100

120

140

160

180

0 200 400 600 800 1000 1200

(Tw

in) av

.[o C]

80

100

120

140

160

180

FR.=50%=50o

FR.=75%=50o

FR.=25%=50o




Figure(5):Variation of average wall temperature along the HP at different water filling ratios fordifferent powers.

4.4: Effect of the Inclination AngleThe heat transfer by boiling in the evaporator section shows larger local differences and

depends on inclination. The intention here to gain better insight into the transport processes of HPby observing local phenomena in different parts of the device. especially the effects of the


109

inclination of HP. This behavior is observed in the experimental HP as shown in figures (6) and(7) show the temperature against the distance along HP for (75%and50%) filling ratios. Thus,they show that the temperature of the evaporator is higher when the inclination angle( ) is 25oanddecreases when equals 50oand 75o, and starts to increase when the angle becomes 90o. Whilefor the filling ratio 25%, figure(8) the temperature decreases as the inclination angle increasesfrom 25o to 90o. As the angle of inclination further decreases (i.e. as the heat pipe approaches thehorizontal position) then plug flow boiling dominates the entire region of the evaporator. Boilingchanges from nucleate boiling (in the liquid film region and the submerged region at verticalposition)to nucleate boiling (in wetted region and dry region).

D ista n ce (m m )

0 2 0 0 4 0 0 6 0 0 8 0 0 1 0 0 0 1 2 0 0

(Tw

in) av

.[o C]

8 0

1 0 0

1 2 0

1 4 0

1 6 0

D ista n ce (m m )

0 2 0 0 4 0 0 6 0 0 8 0 0 1 0 0 0 1 2 0 0

(Tw

in) av

.[o C]

8 0

1 0 0

1 2 0

1 4 0

1 6 0

D ista n c e(m m )

0 2 0 0 4 0 0 6 0 0 8 0 0 1 0 0 0 1 2 0 0

(Tw

in) av

.[o C]

8 0

1 0 0

1 2 0

1 4 0

1 6 0

D ista n c e(m m )

0 2 0 0 4 0 0 6 0 0 8 0 0 1 0 0 0 1 2 0 0

(Tw

in) av

.[o C]

8 0

1 0 0

1 2 0

1 4 0

1 6 0

D ista n ce (m m )

0 2 0 0 4 0 0 6 0 0 8 0 0 1 0 0 0 1 2 0 0

(Tw

in) av

.[o C]

8 0

1 0 0

1 2 0

1 4 0

1 6 0

P = 7 5 0 WF R = 7 5 % W ate r

D ista n c e(m m )

0 2 0 0 4 0 0 6 0 0 8 0 0 1 0 0 0 1 2 0 0

(Tw

in) av

.[o C]

8 0

1 0 0

1 2 0

1 4 0

1 6 0

P = 1 2 5 0 WF R =7 5 % W ater

P = 1 5 0 0 WF R = 7 5 % W a ter

P = 1 0 0 0 WF R =7 5 % W ater

P = 5 0 0 WF R = 7 5 % W a te r

P = 2 5 0 WF R = 7 5 % W a te r

F ig .(5 -1 0 ) :V a ria tio n o f av erag e w al l tem p era tu re a lo n g th e H P a t d ifferen t in cl in a tio n a n g les .

9 0o

7 5 o

5 0 o

2 5 o

9 0o

7 5 o

5 0 o

2 5 o

9 0o

7 5 o

5 0 o

2 5 o

9 0o

7 5 o

5 0 o

2 5 o

9 0o

7 5 o

5 0 o

2 5 o

9 0o

7 5 o

5 0 o

2 5 o

Figure(6): Variation of average wall temperature along the HP at different inclination angles andpower input.


110

Distance(mm)

0 200 400 600 800 1000 1200

(Tw

in) av

.[o C]

80

100

120

140

160

Distance(mm)

0 200 400 600 800 1000 1200

(Tw

in) av

.[o C]

80

100

120

140

160

Qin=1500WFR= 50% Water


Distance(mm)

0 200 400 600 800 1000 1200

(Tw

in) av

.[o C]

80

100

120

140

160


Distance(mm)

0 200 400 600 800 1000 1200

(Tw

in) av

.[o C]

80

100

120

140

160


Distance(mm)

0 200 400 600 800 1000 1200

(Tw

in) av

.[o C]

80

100

120

140

160


Distance(mm)

0 200 400 600 800 1000 1200

(Tw

in) av

.[o C]

80

100

120

140

160


Figure(7):Variation of average wall temperature along the HP at different inclination angles andpower input.


111

Distance(mm)

0 200 400 600 800 1000 1200

(Tw

in) av

.[o C]

60

80

100

120

140

160

180

Distance(mm)

0 200 400 600 800 1000 1200

(Tw

in) av

.[o C]

60

80

100

120

140

160

180

Qin=1500WFR=25% Water

Distance(mm)

0 200 400 600 800 1000 1200

(Tw

in) av

.[o C]

60

80

100

120

140

160

180


Distance(mm)

0 200 400 600 800 1000 1200

(Tw

in) av

.[o C]

60

80

100

120

140

160

180


Distance(mm)

0 200 400 600 800 1000 1200(T

win

) av.[o C

]

60

80

100

120

140

160

180


Distance(mm)

0 200 400 600 800 1000 1200

(Tw

in) av

.[o C]

60

80

100

120

140

160

180



Figure(8):Variation of average wall temperature along the HP at different inclination angles andpower input.


112

4.5: Effects of Filling Ratio on the Heat Transfer CoefficientFigure(9) represents the average heat (Qav.) against the heat transfer coefficient for

different filling ratio charges at a given inclination angle. The heat transfer coefficient (hexp) canbe calculated by using the following correlation[7,15]:

).(

..exp

vTwTeliD

Qavh (8)

Where;

2oin

avQQQ

The large value of the heat transfer coefficient, which can be seen from figures, is at 50% charge,for all power inputs at angles(90o, 75o and 50o).While when the HP was tilted by an angle 25o, asshown, the higher values of heat transfer coefficient are at75% filling ratio. As the HP inclinestowards the horizontal level ( =25o) the heat transfer coefficient increases with the increase of thefilling ratio due to the increase of the heat transfer area between the wall and the working fluid.For other inclination angles (50o , 75o and 90o), the maximum heat transfer area found to be at50% filling ratio.

Qav.[W ]

0 200 400 600 800 1000 1200 1400 1600

h exp.

[W/m

2 o C

]

0

1000

2000

3000

4000

5000

Qav.[W ]

0 200 400 600 800 1000 1200 1400 1600

h exp.

[W/m

2 o C

]

0

1000

2000

3000

4000

5000

FR=75%FR=50%FR=25%

Qav.[W ]

0 200 400 600 800 1000 1200 1400 1600

h exp.

[W/m

2 o C

]

0

1000

2000

3000

4000

5000

FR=75%FR=50%FR=25%

FR=75%FR=50%FR=25%

Qav.[W ]

0 200 400 600 800 1000 1200 1400 1600

h exp.

[W/m

2 o C

]

0

1000

2000

3000

4000

5000

FR=75%FR=50%FR=25%

Figure(9):Heat transfer coefficient vs. the average heat input with different filling ratio anddifferent inclination angles.


113

4.6: Effect of Inclination Angle on the Heat Transfer CoefficientFigure(10)shows that the maximum heat transfer coefficient was obtained when the

inclination angle is 50o and the filling ratio is 50% and 75%, while the maximum heat transfercoefficient at 25% filling ratio is when the HP is at vertical position. The minimum heat transfercoefficient occurs when the inclination angle is 75o and the filling ratio is 25%.

In general the highest heat transfer coefficient for all experimental tests was found at theinclination angle 50o and the filling ratio 50%.

As the HP tilted towards the horizontal position the heat transfer decreases because of theaccumulation of bubbles on the inside upper surface of the evaporator due to the buoyancyforces. On the other side when the HP approaches 50o inclination angle, the effect of thebuoyancy force seems to be vanished and the maximum heat transfer coefficient decreases.

Figure(10):Heat transfer coefficient vs. the average heat input at different filling ratios fordifferent inclination angles.

Q a v . [ W ]

0 2 0 0 4 0 0 6 0 0 8 0 0 1 0 0 0 1 2 0 0 1 4 0 0 1 6 0 0

h exp

. [W

/m2 .o C

]

0

1 0 0 0

2 0 0 0

3 0 0 0

4 0 0 0

5 0 0 0

Q a v . [ W ]

0 2 0 0 4 0 0 6 0 0 8 0 0 1 0 0 0 1 2 0 0 1 4 0 0 1 6 0 0

h exp

. [W

/m2 .o C

]

0

1 0 0 0

2 0 0 0

3 0 0 0

4 0 0 0

5 0 0 0

Q a v . [ W ]

0 2 0 0 4 0 0 6 0 0 8 0 0 1 0 0 0 1 2 0 0 1 4 0 0 1 6 0 0

h exp

. [W

/m2 .o C

]

0

1 0 0 0

2 0 0 0

3 0 0 0

4 0 0 0

5 0 0 0

F R . = 7 5 % w a t e r

F R . = 5 0 % w a t e r

F R = 2 5 % w a t e r


114

5: Comparison of the Experimental Results with the Theoretical and EmpiricalCorrelations

The experimental heat transfer coefficients of the present work calculated by Equation (8)were compared with that predicted by Rohsenow and Imura in Equations (9)and (10) respectivelyis illustrated in Figure (11) for tilt angle 90o and ( 25%,50% and 75% ). It is clear from the figurethat the experimental results of heat transfer coefficient are well agreed with that calculated bythe theoretical (Rohsenow) and empirical (Imura) correlations specially when the filling ratio is50% .

Figure(11):Heat transfer coefficient vs. with average heat input compared with theoretical andempirical correlations for different filling ratios.

F R = 5 0 % W a t e r

Q a v . [ W ]

0 2 0 0 4 0 0 6 0 0 8 0 0 1 0 0 0 1 2 0 0 1 4 0 0 1 6 0 0

h evap

.[W/m

2 .o C]

0

1 0 0 0

2 0 0 0

3 0 0 0

4 0 0 0

5 0 0 0

F R = 7 5 % W a t e r

Q a v . [ W ]

0 2 0 0 4 0 0 6 0 0 8 0 0 1 0 0 0 1 2 0 0 1 4 0 0 1 6 0 0

h evap

.[W/m

2 .o C]

0

1 0 0 0

2 0 0 0

3 0 0 0

4 0 0 0

5 0 0 0h e x p .h R o h s e n o w ( t h e o r e t i c a l )h I m u r a ( e m p i r i c a l )

F R = 2 5 % W a t e r

Q a v . [ W ]

0 2 0 0 4 0 0 6 0 0 8 0 0 1 0 0 0 1 2 0 0 1 4 0 0 1 6 0 0

h evap

.[W/m

2 .o C]

0

1 0 0 0

2 0 0 0

3 0 0 0

4 0 0 0

5 0 0 0h e x p .

h R o h s e n o w ( t h e o r e t i c a l )h I m u r a ( e m p i r i c a l )

h e x p .h R o h s e n o w ( t h e o r e t i c a l )h I m u r a ( e m p i r i c a l )


115

7.1

315.0

)(

1

,

32

rvlg

l

lfghlpCfghsfC

eqRohsenowh (9)

3.0

.

.1.04.025.0

4.02.07.03.065.0

32.0ImatmP

vP

lfghv

eqgl

pClklurah (10)

6: ConclusionsFrom the present work, the following conclusions can be extracted:

1-The temperature distribution along the HP wall in the evaporator section is almost isothermal.The measured temperature along the condenser showed lower values. This drop of temperature isexpected because of the internal resistances due to boiling and condensation.2- For all heat inputs it was found that the average outside temperature of evaporator section islow when the filling ratio is 50% and the inclination angle is 50o.3-The experimental results indicate that the filling ratio and the heat input have the importanteffects on the heat transfer performance. The optimal performance of the HP was found when thefilling ratio ranged between 50–75%, at 50o inclination angle, while the minimum performancewas found when the filling ratio was 25% and 25o inclination angle.4- Maximum thermal conductivity of the HP was found to be about 2060 times that of a stainlesssteel piece of the same size.5-The experimental heat transfer coefficient results agreed well with the empirical andmathematical models for HP.

REFERENCES

1. Yahya, A. A., "Design and Development of An Indirect Type of Solar Cooker Using aHeat Pipe," M.Sc. Thesis, Mosul University, College of Engineering ,(1980).

2. Frank M., "Improving Injection Lances With Heat Pipe Technology," McGill University,(2000)

3. Kempers, R., Ewing,D. and Ching,C.Y.,"Effect of Number of Mesh Layers and FluidLoading of Screen Mesh Wicked Heat Pipe," Journal of Applied Thermal Engineering,(2006).

4. Jianlin Yu, Hua Chen, Hua Zhoa, and Yanzhong Li, "An Experimental Investigation onCapillary Pumped Loop With the Meshes Wick," International Journal of Heat and MassTransfer 50 (2007) 4503–4507.

5. Shwin-Chung Wong and Yi-Huan Kao, "Visualization and Performance Measurement ofOperating Mesh-wicked Heat Pipes," International Journal of Heat and Mass Transfer(2008).

6. Reay, D.A. and Kew ,p.," Heat Pipes Theory, Design and Applications,"5rd. Edition,Britain,(2006).


116

7. Noie, S.H., "Heat Transfer Characteristics of a Two-phase Closed Thermosyphon,"Journal of Applied Thermal Engineering 25(2005) 495–506.

8. Shiraish, M., Kikuchi, K. and Yamanishi,T., "Investigation of Heat TransferCharacteristics of a Two-phase Closed Thermosyphon," Japane(1981).

9. Bejan, A. and Kraus A.D., "Heat Transfer Handbook," Published by John Wiley & Sons,Inc., Hoboken, New Jersey,(2003).

10. Peterson, G.P. ,"An Introduction to Heat Pipes Modeling, Testing and Applications,"Newyork,(1994).

11. Sumana, B., "Design of Heat Pipe," Thesis, Indian Institute of Technology, MAY (2003).12. Ninad D. Sathayea "Incorporation of Heat Pipe into Engine air Pre-cooling Study," M.Sc.

Thesis, B. E., University of Pune,2000. Kansas state university Manhattan, Kansas(2003).13. Reay, D.A., "Advances in Heat Pipe Technology", Pergammon,1981.14. Kreith,F.,Boehm,R.F.,Raithby,G.D. and Hollands,K.G., "Heat and Mass Transfer

Handbook, " CRC Press LLC,(2000).15. Kreith, F., "Principles of Heat Transfer,"3rd Edition, Interstate Educational Publishers,

U.S.A. (1997).


:

. .

1. : ) ( .

2. : .

3. : ) ( :

.A4 . .35 . .

.

. . (Windows 95)

Time New RomanArabic Transparant . )16 ( )14 ( )12 . (

single space .

..4. )3- ( 15

.5.

.

)3- . (6. .

)3- . (

7. :) ( .

(ABSTRACT) 150 . .

. .

.) (

. [ ]

: ) (

.

. .

. 3.5

.

)30000 ( / /.

.

)3- ( )20000 (

) ( )2000 ( )4 ( .

13 . :

/-

http://www.alrafidain.engineering-coll-mosul.comE-mail: alrafidain@ engineering-coll-mosul.com

http://:@www.alrafidain.engineering-coll-mosul.com/

1832010

1 1.

13.

2.

25

.

3.

40 4.

56.

.5

85 .6

:

1

– /

) (

.

..

Effect of Liquids type on some Engineering Propertiesof Limestone Rock from Eski-Mosul

Thamer M. Nuri Ahmed M. Nejm Al-DeenAssis.Prof. Assis. Lecturer

Civil Dept. Engg.College-Mosul University Technical College-Mosul

Abstract This research studied the effect of liquids on the Engineering properties of the Limestone

rocks. The liquids used were: crude oil from Ain Zala, crude oil from Kirkuk, natural groundWater.

The Limestone used was brought from Eski Mosul situated at 45 Km north west of Mosul.Both the compressive and tensile tests were conducted on the Limestone rock specimens inthe dry and saturated case using;Uniaxial, Triaxial Compression and Bending tests.

The tests showed that saturation with any one of the liquids lead to a decrease of bothcompressive and tensile strength. The greatest decrease occurred when ground water wasused to saturate the specimens, and it was also noted that the effect of saturation gavegreater percentage of decrease on the compressive than the tensile strength.

The study also showed that liquids lead to decrease in cohesion ( c ) and the Friction angle) obtained from triaxial test. The liquids used for saturation didn't have clear effect on the

type and mode of failure on all specimens tested .Keyword: c:Compressive strength t: Tensile strength

2009/2/152009/8/16


2

1.

..

..

) (

(Cohesion Force)(Angle of Internal Friction).

(Modulus of Elasticity).Parate)1(Al-Mahdawi)2(Ali, &Noori)3((4) Por, & Galamrth

(90%).

2 .1.2 :

451.2 .

).(ISRM)5(

)1(.)1 (.

*(gm/cm3)2.060.11

(Gs)2.660.08(n%)23.421.88

.

(X-Raydiffraction)Calcite (CaCO3)Dolomite

(MgCO3).Quartz (SiO2))Calcite Limestone(ASTM(6).

2.2 :

) .()2 (.

:

3

)2 (22° C.

.3

1.3:

(5.47 cm)(11 cm)(L/D=2) ASTM(6))2-2.5.(

(Axial strain)(Lateral strain)(Volumetric Strain)

(strain gauges) ASTM(6)

)1(

)1 (

(Millipoises)2201428.85

H

V

H

V

D

L

V :

H:

)(

)(

H


4

.32:

.24x4x2 cm))xx (Elizzi(7) .

.)105º C (24.

4 .

100% Hawkes &Mellor(8) Vutukuri(9)Ali(10)Ali, & Noori(3)

)Dessicator (

.)72 ( .(Epoxy)

.

.5

(strain-meter))0.2 (

..

6 . :

1.6 :

) (( )(C).

Franklin and Hoeck(11)

.(0.75 N/mm2/sec) )5-10 (ASTM(6)

2.6:

(1500 kN)(0.1 kN)(0.75N/mm2/sec)ISRM)5(

.:

AP

c

:

5

:c :(N/ mm2).P :(N).A :(mm2).

3.6 :)2 (

Ali(10)Noori(12)Al-Mahdawi(2) .(0.75

N/mm2/sec)(0.6 mm/min.) .(Strain- meter)

Elizzi(7)Ali(10) Noori(12):

:

t :(N/mm²).)(3

2t

ctt bd

M

M :(N-mm) .t : .c :.

b :(mm) .d :(mm).

7 .

1.7:

)3((5,10,15 N/mm²)( : Angle of internal friction)(c: cohesion) .

)3 (.

)2 (.

P


6

)3 (.)1(N/mm²

)3(N/mm² )3-1(N/mm²

2852341.71031.7551540

24.75519.75371027

49.351534.3522.15517.1533.751023.75

45153019.15514.1529.391019.3939.651524.65

)3-1()3(

) < < <()3.(

)3 (.

Hart, & wang(13)Noori(14)Lockner, & Stanchists(15)

.Ballivy, et al.(16).

:(9).

0

5

10

15

20

25

30

35

40

45

0 5 10 15 20

Shea

r St

reng

th(

1–

3 )at

failu

re2

Confining Pressure N/mm2

:

7

)3 .()µ ( .

)17.(

)1 ()3 ()C ( .)4-7 ((Mohr Circles)

)4 ( )8 .(

)4 ( )c () (.

)C ( N/mm²)°(4.727

4.11253.44233.3319.6

)4 ()c () (

)c(,GHan(18).

2.7 :)5 (

(20.86 %)(33.45 %)

(41.7 %)

)5(.

N/mm2

(%)13.9_______1120.86

9.2533.458.141.7


8

Ali, & Noori(3)(30 % , 17 % ) .

Hadizadah & Law)19 ((45 %).

.)9-A(

(2) (3).)9-B(

.ISRM(5):

2 lav

:v : .a : .l :.

)4 (

:

9

2.7:

3.7

(Uniform Bending Moment) .)6.(

)6 (

(N/mm²)(%)

4.61----3.8716.053.524.08

3.2230.15

)9 (

] )A (

)B (

[.

0 2 4 6 8 10 12 14 16050

100150

200250

Axial Stress N/mm2

Axial Stress N/mm2

Volum

etric Strain,

0 2 4 6 8 10 12 14 16-1000-500

0500

10001500

Lateral Strain,M

icrostrainA

xial Strain,M

icrostrain)B

( )

A(


10

)%30.15 .(

Al-Mahdawi(2)Thabet, et al.(20)%)30 (

Ali(10) %)32-30 ( .Ali, & Noori(3)

%7 ,%15.

0

0.5

1

1.5

2

2.5

3

3.5

4

4.5

5

0 50 100 150 200 250 300 350

Strain in Tension , Microstrain

Tens

ile S

tress

, N

/mm

2

)10 (.

0

0.5

1

1.5

2

2.5

3

3.5

4

4.5

5

0 50 100 150 200 250

Strain in Compression , M icrostrain

Tens

ile S

tress

, N

/mm

2

)11 (

Strain in Compression, Microstrain

:

11

)10-11(.

.

(Neutral Axis).

.

8 .:

) (

( c)a)l).

(c)( ).

.

..

9.:

Parate, N. S.," Influence of Water on The Strength of Limestone.", Transactions of AIME, Vol. 254 , PP 127-131,(1973).

AL-Mahdawi, S.K., " Effect of Specimen Sizes and Water Saturation on Strength Properties of Jeribe Limestone",M.Sc. Thesis, civil Engg. Dept., College of Engineering, University of Mosul,(1985)..

Ali, S.A. & Noori, T.M"Effect of Oils on The Strength and Deformation Properties of Limestone ", Proce. Of TheSixth Sci. Conference for Foundation Tech. Institutes. Baghdad – Iraq, pp. 226-232, (1998)..

Por, L. & Galamrth, " Effect of Water Content on The Mechanical Behavior of Fine – Grained Sedimentary Rock ",Pizohishy Danishka, No.48,(2003).

ISRM., , "Suggested Methods for Determining Water Content, Porosity, Density, Absorption and RelatedProperties.", ISRM. Committee on Standardization of Laboratory Tests, Int. J. Rock Mech. Min. Min. Sci., Vol.16 pp.143-156,(1979).

ASTM, Standards, " Soil and Rock "American Society for Testing and Material, Vol.04-08,(1989).Elizzi, M.A, " time – Dependent Behavior of Some Evaporite Rocks ",Ph.D. Thesis, University of Sheffield, (1976).


12

Hawkes, I. & Mellor, M, "Uniaxial Testing in Rock Mechanics Laboratory ", Eng. Geol., Vol. 4, No. 3, pp.177-285,(1970).

Vutukuri, V.S., " The effect of the Liquids on the Tensile Strength of Limestone." Int. J. Rock Mech. Min. Sci. ,Vol.11 pp. 27-29,(1974).

Ali, S.A., " Creep Properties of Evaporite Rocks with Particular Reference to Gypsum ", Ph.D. Thesis, University ofSheffield,(1979)..

Franklin, J.A. &Hoeck, E., "Developments in Triaxial Testing Technique" Rock Mechanics 2 , pp. 223-228,(1970).

Noori, T.m., " Study of The Long – Term Strength of Gypsum ", M.Sc. Thesis, civil Engg. Dept., College ofEngineering, University of Mosul,(1989),

Hart, D.J., & Wang, H.F., " Laboratory measurements of a complete set of poroelastic moduli for Berea sandstoneand Indiana limestone ", J. Geophys. Res., No. 100, pp.741 – 751,(1995).

Noori, T.M., " Effect of Anisotropy on The Shear Strength Of Sandstone Rock in Triaxial Compression Test ",Raffidain Engineering Maga., No. 1 (1996),.

Lockner, D.A., & Stanchits, S.A., " Undrained poroelastic response of sandstones to deviatoric stress change ", J.Geophys. Res., No. 107,(2002).

Ballivy, G., Ladanyi, B., & Gill, D.E., " Effect of Water Saturation History on The Strength of Low – PorosityRocks ", ASTM. STP 599, pp. 4 – 20,(1976).

Hellmann, R., Renders, P.J., Gratier, J., &Guiguet, R"Experimental Pressure Solution Compaction of Chalk inAqueous Solutions, Part 1. Deformation Behavior and Chemistry ", The Geochemical Society, SpecialPublication, No. 7, (2002) .

Han, G., " Rock Stability under Different Fluid Flow Conditions ", Ph.D., Thesis, chemical Engg. Dept., College ofEngineering, University of Waterloo , Ontario, Canada,(2003).

Hadizadeh, J., & Law. R.D., " Water-weakening of sandstone and quartzite deformed at various stress and strainrates ", Int. J. Rock Mech. Min. Sci. & Geomech. Vol. 5, pp. 431 – 439,(1991).

Thabet, K.M., Khattab, S.I., & Al – Azzo, S.I., , "Geotechnical Characteristics of Some Limestone in Nineveh, Iraq", Confidential Report No. SM/SRC 10/1995,University of Mosul,(1995).

–

:

13

/ /

) ( .

:- ) )COD ()T.S ( , , , )Ec ( , )pH ( , , (

. ,)80 (/)66% ()COD ()73% ()T.S ()87% ()400(/)58% ()COD ()67% ()T.S ()50% (

.) ()89.28-97.8. (

:- .

PHYSIOCHEMICAL TREATMENT OF SEVERALHOSPITALS WASTEWATER IN MOSUL CITY

HALLA NABEEL ELEA

DEPT OF CIVIL ENGINEERING/COLLEGE OF ENGINEERING/UNIVERCITY OF MOSUL

ABSTRACT

This study aimed to treat the wastewater of several hospitals (Jamhory Hospital , Ibn-Sina Hospital ,Batool Hospital , Hazem Al-Hafez Hospital) in Mosul city, by using the method of coagulation and flocculationand by using the Jar_Test as a laboratory scale. Alum and Lime were used as coagulants in wastewatertreatment. The efficiency had been calculated by the following characteristics :- ( Chemical Oxygen Demand(COD), Total Solids Matter (T.S), Nitrate (NO3) , Phosphate (PO4) , (pH) , Electrical Conductivity (Ec) , Chloride ,Turbidity ) . At the optimum dose of alum (80) mg/l removal efficiencies were (66%) for (COD), (73%) for (T.S)and (87%) for Turbidity . the Nitrate removal efficiency was (65%) at the optimum dose (60) mg/l .The optimumdose of Lime was (400) mg/l for removal (COD) , (T.S) and Nitrate , the removal efficiency of (COD) was ( 58%) ,(67%) for (T.S) and (50%) for Nitrate . The Alum was better than Lime in removal efficiencies of pollutants fromhospitals wastewater. The heavy metals removal efficiency at the optimum doses were ranged from ( 89.28-97.8)% .Key words:- Physiochemical Treatment, Hospitals Wastewater , Heavy Metals , Alum, Lime, Coagulation.

2009/1/112009/8/17


14

:-

.

.

.

: -1-

,.

2-: - ))COD ()T.S (

)Ec()pH. ( (3-: -

) , , . (4-

.

:-]1[

.]2[) , ,(

,

,

.)Gautam,et al. (]8[

.)Adam , et al.(]5[

)–– ( .

.)Kugelman et. al. (]10[

.

:

15

) (]4[ ) (

)Al-Rawi et al. (]7[

. )Randtke (]12[

.

:-

)300-1000 (L/patient/day (]9[.

]14[.

:-)

( )4066 (

)15 (/ )30 (/.

)composite sample ()8.5–2.5 (. )2008–2008 (

)12 (.

)Mixers ()Paddles ()2×2 (

.)180 ( / )5 (

)30 ( )60 ( / . )G×T (

)30000-60000 ( )10000-100000 (]11[. )30 (

.

:-1- )T.S: ( -

]3[.

2- )COD ()ClosedReflux,Titrimetric Method(]13[.


16

3-)Ultraviolet Spectrophotometer Screening Method (]13[.4- )Stannous Chloride Method] (131.[5-]3[ .6-)pH ()Ec (.7-

)Atomic Absorption Spectrophotometer. (

:-: -

)1 (.

)1 (

]3 [1)T.S(530-980)/(---2)COD(200-835)/(1003)PO4(6.2-11.4)/(34 )NO3(0.41-7.8)/(505 )Cl(38-53)/(2006 )Ec(640-1200)/(-----7 )pH(6.6-7.99.5-6837-68 )NTU(---9)pb(0.1-0.33)/(0.1

10)cd(0.04-0.24)/(0.0111)Zn(1.2-4.5)/(212)Cu(0.36-1.87)/(0.2

: -.

)1 ()350,400,450,500,550,600,650,700 ( / )COD ()COD ()400 (

/)58%. (

)1 ()COD (

0

10

20

30

40

50

60

70

300 350 400 450 500 550 600 650 700 750

( / )

%(C

OD

)

0

10

20

30

40

50

60

70

80

300 350 400 450 500 550 600 650 700 750

( / )

%(T

.S)

)2 ()T.S (

:

17

)67% ()400 (/ )2. (

)85% ()600 (/)3.(

)4 ()50% ( )400 (/.

)5 ()78% ()550(/)76% (

)450 (/)6. (

)37% ()Ec ()400 (/)7(

0

10

20

30

40

50

60

300 350 400 450 500 550 600 650 700 750

( / )

%

40

50

60

70

80

90

100

300 350 400 450 500 550 600 650 700 750

( / ) %

)3 ()PO4 (

)4 ( )NO3 (

0

10

20

30

40

50

60

7080

90

300 350 400 450 500 550 600 650 700 750

( / )

%

30

40

50

60

70

80

300 350 400 450 500 550 600 650 700 750

( / )

%

)5 ()6 (


18

9.2

9.4

9.6

9.8

10

10.2

10.4

10.6

300 350 400 450 500 550 600 650 700 750

( / )

(pH)

0

5

10

15

20

25

30

35

40

300 350 400 450 500 550 600 650 700 750

( / )

%(E

c)

)8 ()pH ()7.62 ()350 (/)9.43 (

)10.5()700(/.

. ]6[.

Ca(OH)2 + Ca(HCO3)2 2CaCO3 + 2H2O

.

)2 (

350/

400/

450/

500/

550/

600/

650/

700/

COD/

22011292.4127.6136.4167178187204.6

T.S/

540281178248270345.6378421470

/7.53.52.41.871.71.41.11.652

/0.70.40.350.3780.480.50.540.570.6

/4229.424.719129101619

NTU552617.61316202324.733

Ec/

730547460511533591620657679

pH7.629.439.59.69.81010.210.310.5

)7 ()Ec(

)8 ()pH(

:

19

: -

)9 ()COD ()66% ()80(/)73% ()T.S ()10 (

0

10

20

30

40

50

60

70

80

40 50 60 70 80 90 100 110

( / )

%(T

.S)

0

10

20

30

40

50

60

70

40 50 60 70 80 90 100 110

( / )

%(C

OD)

)11 ( )12 ()13 ()90 (/)91% ()81% (

)65%()60(/.

50

55

60

65

70

75

80

85

90

95

40 50 60 70 80 90 100 110

( / )

%

0

10

20

30

40

50

60

70

40 50 60 70 80 90 100 110

( / )

%

)14 ()80(/)87%. (

)15 ()Ec ()60(/)43% ()Ec. (

)16()pH ()7.62 ()7.55 (

)7.2 ()100 (/.

)9 ()COD (

)10 ()T.S (

)11()12 ()NO3 (


20

0

10

20

30

40

50

60

70

80

90

40 50 60 70 80 90 100 110

( / )

%

40

50

60

70

80

90

40 50 60 70 80 90 100 110

( / )

%

7.15

7.2

7.25

7.3

7.35

7.4

7.45

7.57.55

7.6

40 50 60 70 80 90 100 110

( / )

(pH)

05

101520253035404550

40 50 60 70 80 90 100 110

( / )

%

Al(OH)3

]9[.)3 (

)3 (

5060708090100

COD/220158.4 12192.474.8121154T.S/540356.4270178.2145.8205.2286

/7.52.552.252.020.9750.6751.125/0.70.3570.2450.2940.4480.5390.567/4229.823.941811813.8NTU55242017712.6521.45

Ec/730709416445489540606pH7.627.557.427.377.37.267.2

)13 ()14 (

)15 ()Ec (

)16 ()pH (

:

21

: -)4 (

)4(

)400 (/)400 (

/

)80 (/

)80 (/

pb)/(

0.280.0389.28%0.00897%

Cd)/(

0.1570.00895%0.003397.8%

Zn)/(

3.14130.1994%0.08397.35%

Cu)/(

1.2740.10591.75%0.04496.54%

: -

)5 ( .

.)5(

%

)COD) (/(35218447.7

)T.S ()/(

96071225.8

)PO4) (/(6.56.16.15)NO3) (/(3.22.99.37

)/(4733.429)NTU(6239.336.6

)Ec ()/(

113878930.66

)pH(7.847.9---)pb) (/(0.280.1739)Cd) (/(0.1570.09738)Zn) (/(3.14131.88740)Cu) (/(1.2740.68346.4

: -


22

)6 (: -

)6(

)400(/)80(/ ]1[

)COD(

58%66%)50-89%(47.7%

)T.S(67%73%---25.8%

)PO4(68%87%25%6.15%)NO3(50%36%---9.37%

41%73.8%---29%68%87.27%---36.6%

)Ec(37%33%---30.6%

)pb(89.28%97%---39%)Cd(95%97.8%---38%)Zn(94%97.35%---40%)Cu(91.75%96.54%---46.4%

.

: -

)30 (/.: -

=30/. =400/.

=400/ ×30 /. ) =400 ×1000 ×30 (\ )1000 ×1000(

=12\. =12 ×24 =288\.

) =80 (\ ) =80 ×1000 ×30 (\ )1000 ×1000( =2.4\.

=2.4 ×24 =57.6\.

: -1-

.2-)COD ()66% (

)80 (/ )400 (/

:

23

)58%. (3-)80 (/)T.S ()73% (

)67% ()400 (/.4-)65% ()60 (/

)91% ( )90 ( .)50% ()400 (/ )85% ()600 (/.

5-)87% ()80 (/)76% ()450 (/.

6-)90 (/)81% ()550 (/)78%. (

7-)80 (/)400 (/ ) ( )89 – 97.8. % (

8-

.

: -

1-) (.

2-.

: -" "

)2002.(

" ")2005.(

"– ")1990.(

""8)1:(33-41)1997. (

Adams,C. ; ASCE,M. ; Wang, Y. ; Loftin,K. and Meyer, M. ; "Removal of Antibiotics fromSurface and Distilled Water in Conventional Water Treatment Processes " , J. Envir. Eng. ,Vol. 128 , 3 , pp. 253-260 (March 2002) .

Al-Layla, M.A. ; Ahmad, Sh. And Middlebrooks, E.J., "Handbook of wastewater collectionand treatment ", Garland STPM Press, New York and London (1980).

Al-Rawi, S.M.; Hana, G.Kh. and Ali, A.R., "Performance of Two Hospital WastewaterTreatment Plants in Removing Various Pollutants ", Al-Muhandis , 123: 17-24 (1997)


24

Gautam, AJAY K. ; Sunsil , Kumar ; P.C. , Sabumon ; "Preliminary Study of Physio-Chemical Treatment Options for Hospital Wastewater " , J. Environment Management , Vol.83 , No.3 , pp.(298-306) , (2007) .

Kiley, G., "Environmental Engineering ", McGraw- Hill Published Company, England(1997). Kugelman, I.J. and Carty, P.L. , " Cation toxicity and simulation in anaerobic waste treatment" , 19th , Industrial Waste Conference , USA , 667 ( 1974) .Metcalf and Eddy, "Wastewater Engineering", 2nd ed., Mc Graw-Hill, Inc. New York,USA. (1979).

Randtke, J.S., " Organic contaminant Removal by Coagulation and Related ProcessCompination", J.American Water Works Association, 80, 5, 40(1988).

Standard Method for the Examination of Water and Wastewater, 16th ed., APHA, AWWA,WPCF, New York (1985).

World Health Organization, "Mangment of Wastes from Health Care Activities, Geneva(1998).

–

:

25

.

--

, .

, .

. ,

, . ,

.

: , , ,.

The Influence of Carbon Contents on the Corrosion Resistance ofPlain-Carbon Steels in the Water Environments

Sobhi.I.Ibrahim Yasir.A.AbdullahAss. Prof Engineer

Mech. Eng. Dept. / College of Engineering / Mosul University

AbstractPlain-carbon steels are considered to be the most widely used materials in engineering andindustrial applications. In this study wide range of carbon steels are used in the mostcommon corrosion environments which are salt water and drinking water. The weight lossmethod is used, then corrosion rate is found to correlate with carbon percentage of steels.The results indicate good correlation which are related to the microstructure where higherpearlite show higher corrosion rate and pearlitic steel represents the maximum corrosionrate. This is found in both environments used where higher corrosion rate is thecharacteristic of salt water. It is also found that the longer exposure time the lower corrosion

rate while weight loss is still continuing.

Keyword: Plain-carbon steels,Carbon content, Corrosion rates, Water environments.

2009/1/202009/6/17


26

(Corrosion) , :(Corrosion environment) ,

(Chemical reaction)(Electrochemical) .(Wet Corrosion)

(Electrolyte) , .

(Galvanic Corrosion Cell)(Anode)(Cathode) ,

,[1] . : , ,(Single phase)

(Two phases)(Mechanicalproperties) ,(Tensile strength)(Hardness) ,

(Microgalvanic corrosion cells)[4,3,2] . .

: . , ,

(Electrolyte) , .

, .

,

[6,5] .

(Hypo-eutectoid Steels) ,(Eutectoid Steels)(Hyper-eutectoid Steels)(Corrosive media)(Salt water) ,(Drinking water) .

.)( .

.3.5 % .

,

)Localized breakdown ([7]. .

, ,pH ,.

[9,1,8].3.5%) (.

:

27

:(Han 2008)[10](0.13%C, 0.23%C) ,60 ,

(Melchers 2003)[11](0.12% C, 0.18% C) ,(Pillai 1982)[12]

(0.03% C, 0.13% C) ,

.(Tomlinsion 1983)[13]

, ,.

( Miroslav 2002 )[14]

(0.8%C, 0.5%C) . .(Raja

2002) [15]( 0.5%, 0.14%, 0.05%)5% .

.)Takasaki 2007 ([16] Cl--2, SO4

(Mild steels)(Drinking waters) .(Larson and

Skold 1958)SO4-2 ,Cl--

HCO3Cl- ,SO4-2HCO3

-.(Garcia 2008 )[17] (Cl-)(

Weight loss )(Conductivity ) ,NaCl(0.005, 0.01, 0.1, 0.6) ,

(0.12% C)NaCl. :(Takasaki 2007)[16]

(Mild steels) ,(Macdonald1978)[18]

(0.14%C)660 ,

.(Moller 2007)[19]Ca+2

(0.038% C) ,CaCo3) (

.(Corvo 2005)[20]

)(Cl-

, .


28

,.

,.

: -

:)10 ( , ,

,

. :(Equilibrium

heat- treatments)

.

:)320 ,600 ,800 ,1000 ()Polishing (

)Etching ()2% Nital ()2% Nitric acid and 98% alcohol ([21] ,(1).

: ,

)Close condition ( , , , ,2008 .

: -)Salt water :(3.5g %(NaCl)96.5g %

.)Drinking water :( / .

(1).

:

29

(1) :.8,7,6,5,4,3,2,1(X150).

(0.8% C)(X500), (X150).(X150).

1) 0.1% C Steel. 2) 0.2% C Steel. 3) 0.3% C Steel. 4) 0.35% C Steel.

5) 0.4% C Steel. 6) 0.5% C Steel. 7) 0.6% C Steel. 8) 0.65% C Steel.(a)

11) 1.0% C Steel. 12) 1.2% C Steel.(c)

9) 0.8% C Steel.

(b)

10) 0.8% C Steel.


30

(1) :.

DRINKINGWATER

SALT WATERCOMPOUND

2301700Total hardness mg/L120260Calcium hardness mg/L361596(Cl-)mg/L24.64322(Mg+2)48.1104.2(Ca+2)492775Electrical conductivity8.398.35PH-70.2-36.9Electrode potential

(Volt)

(2) :.

: , , ,

( 500 ml hydrochloric acid and 3.5g hexamethylene tetramine distilled water to make1000ml)

[22].

: .

.(2).

CHEMICAL COMPOSITIONS (% WEIGHT)FeAlMoNiCrSPMnSiC

STEELSAMPLE

.NObalance.003.012.044.107.016.008.510.238.120S1balance.020.003.007.015.009.022.9502.01.190S2balance.008.007.023.049.006.014.6701.74.341S3balance.019.062.220.230.036.013.8701.99.388S4balance.008.026.081.437.033.0051.19.230.473S5balance.013.008.066.117.028.016.660.281.586S6balance.038.042.091.810.035.002.790.192.681S7balance.011.037.770.610.017.004.790.271.702S8balance.009.043.1011.02.020.0081.09.251.813S9balance.008.007.023.049.006.014.6701.741.08S10balance.008.026.081.437.033.0051.190.231.13S11

:

31

:)Corrosion rate ( /)Mils Per Year ()W ( )T(

)) (D()A .([1,8]: -

….….( 1 )DATW534

=Corrosion rate ( mpy)

{a,b (2)} , .{a

(2) } ,

, )( ,

.{b (2)} .

(Microstructure)

. ,

[23,8] . ) + (

) + ([8] ,

(Primary ferrite)

, ) + ([24] .

(Solid Solution)(Intermetallic compound)

(Microgalvanic corrosion)

(w) : )(mg.(D) :(g/cm3).(A) :(in2).(T) :(hours).

(mil) :0.001inch.


32

(Different potentials)[21] ,100%

) + ((Two phase structure) ,

.(Nital)

0

1

2

3

4

5

6

0 0.2 0.4 0.6 0.8 1 1.2 1.4

% Carbon

Cor

rosi

on r

ate

(mpy

)

1 month2 months3 months4 months5 months

100% Fe

Salt water

0

0.0254

0.0508

0.0762

0.1016

0.127

0.1524C

orro

sion

rat

e (m

m/y

)

(a)

0

1

2

3

4

5

6

0 0.2 0.4 0.6 0.8 1 1.2 1.4

% Carbon

Cor

rosi

on r

ate

(mpy

)

1 month2 months3 months4 months5 months

100% Fe

Drinking water0

0.0254

0.0508

0.0762

0.1016

0.127

0.1524

Cor

rosio

n ra

te (m

m/y

)

(b)

(2) :.)a () .b (.

:

33

[8] ,{a,b (3)}

.:(Pillai 1982)[12] ,(Melchers 2003)[11] ,

(Han 2008)[10] ,.

{ a,b (2) } .

.

,(Corrosion products layers)

(Oxides protection layers) .

,[25] .:

(Macdonald 1978)[18].

(Azzerri 1981)[26].

(Takasaki 2007)[16].

{a,b,c,d,e (4)} ,

[27] . . ,

.

.

[28].(Yunping 2002)[29]

.(3.5% Nacl)

(Nacl)

(Yunping 2002)[29] .


34

0

1

2

3

4

5

6

Cor

rosio

n ra

te (m

py)

Salt water

Drinking water

5 Months

0

0.0254

0.0508

0.0762

0.1016

0.127

0.1524

Cor

rosi

on r

ate

(mm

/y)

%Pe

arlit

e

100% %(a)

0

5

10

15

20

25

30

35

Salt water

Drinking water

5 Months

Wei

ght

loss

%Pe

arlit

e

100% Fe % Carbon(b)

(3) :.(a) .(b).

:

35

0

1

2

3

4

5

6

0 0.2 0.4 0.6 0.8 1 1.2 1.4

% Carbon

Cor

rosio

n ra

te (m

py)

Salt water

Drinking water

100% Fe

1 Month

0

0.0254

0.0508

0.0762

0.1016

0.127

0.1524

Cor

rosi

on r

ate

(mm

/y)

(a)

0

1

2

3

4

5

6

0 0.2 0.4 0.6 0.8 1 1.2 1.4

% Carbon

Cor

rosio

n ra

te (m

py)

Salt water

Drinking water

100% Fe

2 Months0

0.0254

0.0508

0.0762

0.1016

0.127

0.1524

Cor

rosi

on r

ate

(mm

/y)

(b)


36

0

1

2

3

4

5

6

0 0.2 0.4 0.6 0.8 1 1.2 1.4

% Carbon

Cor

rosi

on r

ate

(mpy

)

Salt water

Drinking water

100% Fe

5 Months0

0.0254

0.0508

0.0762

0.1016

0.127

0.1524

Cor

rosi

on r

ate

(mm

/y)

(e)

0

1

2

3

4

5

6

0 0.2 0.4 0.6 0.8 1 1.2 1.4

% Carbon

Cor

rosi

on r

ate

(mpy

)

Salt water

Drinking water

100% Fe

3 Months

0

0.0254

0.0508

0.0762

0.1016

0.127

0.1524C

orro

sion

rat

e (m

m/y

)

(c)

0

1

2

3

4

5

6

0 0.2 0.4 0.6 0.8 1 1.2 1.4

% Carbon

Cor

rosio

n ra

te (m

py)

Salt water

Drinking water

100% Fe

4 Months

0

0.0254

0.0508

0.0762

0.1016

0.127

0.1524

Cor

rosio

n ra

te (m

m/y

)

(d)

(4) :.(a).(b).(c).(d).

(e).

:

37

(1) ,

(1).

:

.

.

..

.

.

Fontana G. and Green, D., “Corrosion Engineering”, 3rd Edition, McGraw-Hill International,U.S.A, pp. 138-172, (1986).

,New York,United States of America,”Hand Book of Corrosion Engineering“,.R.Roberge Ppp. 25-88, (2000).

Van Vlack H., “Elements of Materials Science and Engineering”, 5th Edition, WesleyPublishing Company, Inc., U.S.A, pp. 53-79, (1985).

Third,Corrosion Control,2.Vol,”Corrosion“,.T.and Burstein G,.A.Jarman R,.L.Shreir Ledition, Great Britain, pp. 43-67, (2000).

. ,"" , , ,16-72 ,)1989.( . , . ," - -" ,,31-53 ,)1987.(

Francis R., “Bimetallic Corrosion”, National Corrosion Service, London, U.K., pp. 28-67,(2000).

Winston R., “Uhlig’s Corrosion Handbook”, John Wiley & Sons, Inc., Canada, pp. 47-78,(2000).

Salem A.H., and Saber T.M., “Inhibition of The Corrosion of Steel Pipes Carrying PotableWater”, Elsevier Science Publishers B.V., Amsterdam, Vol.93, pp.461-471, (1993).

Han L., and Song S., “Ameasurement System Based on Electrochemical FrequencyModulation Technique for Monitoring the Early Corrosion of Mild Steel in Sea Water”, pp. 1-19, (2008)


38

Melchers R. E., “Effect on Marine Immersion Corrosion of Carbon Content of Low AlloySteels”, Journal of Corrosion Science, Vol. 45, pp. 2609-2625, (2003).

Pillai K. C., and Narayan R., “The Corrosion of Mild Steel in 0.01M, 1M, 3M HCL”, Journal ofCorrosion Science, Vol. 22, No. 1, pp. 13-19, (1982).

Tomlinsion W. J., and Giles K., “The Microstructures and Corrosion of 0.79C Steel Temperedin the Range 100-700 C”, Journal of Corrosion Science, Vol. 23, pp. 1353-1359, (1983).

Mirosalv H., and Mirosalv K., “Influnce of Carbides Over Some Steel Corrosion”, ActaUniversity, Vol. 41, pp. 45-55, (2002).

Raja V.S., Baligidad R.G., and Shankar Rao V., “Effect of Carbon on Corrosion Behaviour ofFe3Al Intermetallics in 0.5 N Sulphuric Acid”, Journal of Corrosion Science, Vol. 44, pp. 521-533, (2002).Takasaki S., and Yamada Y., “Effects of Temperature and Aggressive Anions on Corrosion ofCarbon Steel in Potable Water”, Journal of Corrosion Science, Vol. 49, pp. 240-247, (2007).

Garcia K.E., and etal, “Lost Iron and Iron Converted in to Rust in Steels Submitted to Dry-WetCorrosion Process”, Journal of Corrosion Science, Vol. 50, pp. 763-772, (2008).

Macdonald D. D., and etal, “Corrosion of Carbon steel during Cyclical Exposure to WetElemental Sulphur and the Atmosphere”, Journal of Corrosion Science, Vol. 18, pp. 499-501,(1978).

Moller H., “The Influence of Mg2+ on the Formation of Calcareous Deposits on a FreelyCorroding Low Carbon Steel in Sea Water”, Journal of Corrosion Science, Vol. 49, pp. 1992-2001, (2007).

Corvo F., and Minotas J., “Changes in Atmospheric Corrosion Rate Caused by Chloride IonsDepending on Rain Regime”, Journal of Corrosion Science, Vol. 47, pp. 883-892, (2005).

Higgins R.A., “Engineering Metallurgy”, The English University Press.Ltd, London, U.K, pp.152-160, (1999).

ASTM G 1-90: Standard Practice for Preparing, Cleaning, and Evaluating Corrosion TestSpecimens, ASTM, Washington, USA, pp. 1-8, (1999).

Tomlinsion W. J., and Giles K., “The Microstructures and Corrosion of 0.79C Steel Temperedin the Range 100-700 C”, Journal of Corrosion Science, Vol. 23, pp. 1353-1359, (1983).

Elsevier Science and,”Principles of Corrosion Engineering and Corrosion Control“,.Ahmad ZTechnology Books, pp. 9-17, (2006).

:

39

.pp,London,Second edition,”lSteel Work Corrosion Contro“,.H.and Deacon D,.A.Bayliss D60-70, (2002).

Azzerri N., and etal, “Assessment of Corrosion Rate of Steel in Sea Water by PolarizationResistance Technique”, Journal of Corrosion Science, Italy Vol. 21, No. 11, pp.781-787, (1981).

Louis F., Jack T., and Henry L., “Prediction of Corrosion Defect Growth on OperatingPipelines”, Trans. Canada Pipelines Ltd., Canada, (2004).

Salem A.H., and Saber T.M., “Inhibition of The Corrosion of Steel Pipes Carrying PotableWater”, Elsevier Science Publishers B.V., Amsterdam, Vol.93, pp.461-471, (1993).

Yunping X., and Zhaohui X., “Corrosion Effect of Magnesium Chloride and Sodium Chlorideon Automobile Components”, University of Colorado, U.S.A, Report No. CDOT-DTD- R-2002-4,(2002).

–

:

40

. . .

- -

Adaptation) (

) .:

.

:

The Adaptation of type in Architecture

Dr. Ahmad Abdul wahid Thanoon P.Dr. Miqdad Haidar AL-jawadi

Many Architectural studies deals with concept of adaptation in architecture with differentways various according to the trend of each study, this show’s the importance of studying theconcept of adaptation in the architectural field in general. This research tray to focus on theconcept of the adaptation of type in architecture because it’s important in the generation of thenew architectural models and the reiteration of the typological chain. Reviewing previousstudies focusing on this concept show’s the absent of a theoretical frame witch separate aspecific items of the process of adaptation of the architectural types through different naturaland cultural effects. Thus, the problem her was the absence of a specific imagination of theprocedures and mechanisms for achieving adaptation of the architectural types throughdifferent natural and cultural effects. To solve this problem, the researcher adopted theapproach of building a theoretical framework for the adaptation of type in architecture, Witchincludes the items: The target of adaptation, the kind of effect which cues adaptation, theprocedures and mechanisms to achieve adaptation. The achievement of these items dependedon the help of the previous studies.Key words: Adaptation, type, Typological Transformation, alteration

2009/1/52009/6/9


41

1-:Adaptation) (

.

:

.

2-:)Adaptation(

:

2-1 ::

)Theadaptation and Growth of the Bungalow in India)

Bungalow)(planformstructure

]1108-104 .[)1 .(

)Architecture and Identity Towards a global eco-culture()Adaptation (

Penang]2156[

)2 .() (

.

CroweNorman)Nature And The Idea Of A Man –Made World ()The

Classical Temple (Megaron House)(

]342-44[)3( .

.

:

42

]348[

]4125 .[

CroweNorman)4 (]361 -60[ .

]4177 .[

.

1-1-2 ::) (

David Kincaid)Adapting Buildings ForChanging Uses () (

)ReuseThe ()()

() () (]52 .[

) (

S.Sophia]628 [

)5(.

.

2-2 :: ) (

) Poetics of Architecture Theory of Design (Antoniades

)6](765[

]866.[


43

)Renaissance (

) () ](766.[

2-3 ::

)Easy Changing(

)](955 -48.[

SustainabilityDavid Kincaid

]5101 .[

) .(

)) ((

.)

(

)Transformation(Typological.

:(.

3-:

.]10139 .[

.

]10139 .[

Argan’s]11240[Argan’s

:

44

De Quincy”s )

](11243[ .

Bernard TschumiMichael BrawnHistorical Sequence

]12154-153[) (

.]10141 .[

) (]1329[

.Rowe

]1486 .[

) (

.

4-:

) ((Prototype)

)Trans-form (

:

]15936 .[

) () .(

]16112[:

4 -1 ::


45

) (]16113[)Scaling ()7 (

]16113 .[

TranslationRotation

Reflection]16114][175 ][184 [)8 .(

.

Dimensional Transformation

Subtractive TransformationAdditive Transformation]1948 ][178.[)9(

.

4 -2: :

SmashingBurningDismantling]16112[Stretching

.ShearPerspectiveTransformation]16116[)10( .

) ( .

)()(

.

5-:

:)

(.

.:

:

46

:

5-1 ::5 -2 : :5 -3 : :5 -4::

5-1 : :

:

):](20158 . [Argan

FormalArchitectural typologies

:Complete Configuration of building).(Major structural elements

).(3 (surface treatment

)..(PlanStructural system

Surface treatment]21244 .[Baker

)

](2275 -6343-26.[

)][

(.

5 -2: :

:


47

5 -2-1:):(

]2327][2155[]52[

]765.[

5-2-2:):(

]766.[

5-2 -3 ::

) ()"]("628[)The adaptation and

Growth of the Bungalow in India)Bungalow) (]1108 -104.[

)()(

.

5 -3 ::

:]2420-17 .[

]4231[

:

48

]22233[

]4233[

]4233[]4235 .[

Antoniades

]766 .[ :

)](766[

Antoniades]766.[

]765.[Gelernter

]253[ ) (

]2511[ .

Laseau Context :

]2686.[

]27193 .[

:

1 ( :)( .

2 (:)

.(


49

5 - :4:

) ( :

Relocation) (Transformation]289[

:

5-4-1 : :

]26120][16113][1810[:

TranslationRotation

Inversion & ReflectionScaling

6-4 -2: ):(

]1948][178[:

Dimensional TransformationTransformationSubtractive

TransformationAdditive )(

)(:

TranslationRotation Inversion & ReflectionScaling .)(:

Dimensional TransformationTransformation SubtractiveTransformationAdditive.

7-:

:

50

:

::-

---

::- :

1-2-3-.- :1-2-3-4-.

::- :

------:---

::---

:1.Desai, Miki, Madhavi Desai, “ The Adaptation and Growth of the Bungalow in India“,

Article has been published at the International Workshop on the Architectural Heritage ofAsia and Oceania (UIA Workgroup on Heritage) at the Rizvi College of Architecture,Bombay, in December 1995.

2.Abel, Chris,” Architecture and Identity Towards a global eco-culture”, First Ed,Architectural Press, London, 1997.


51

3.Crowe, Norman,” Nature and The Idea of A Man – Made World”, First Ed, MassachusettsInstitute of Technology, 1997.

4 . ""1970.

5.Kincaid, David, “Adapting Buildings For Changing Uses Guidelines for Change of UseRefurbishment ”, First Ed, Spon Press, London, 2002.

6. ""21971.

7. Antoniades, C. Anthony, “ Poetics of Architecture Theory of Design”, John Wiley & Sons,Inc , New York, 1992.

8. ""2005.

9.Nijaidi, H.R, “ Flexibility In The Design of Buildings “, thesis, Oxford Polytechnic, InCollaboration with Bartlett School of Architectural & Planning Univ. Collage, London,Oxford, 1982, P 106.

10 . ""2000.

11.Nesbitt, Kate, “Theorizing a New Agenda For Architecture “, Princeton ArchitecturalPress, New York, 1996.

12.Tschumi,Bernard, “Architecture & Disjunction “, MIT Press, Cambridge, London,1994.13.Schulz, Christian Norberg, “ The Concept of Dwelling “, Rizzoli International

Publications, New York, 1985.14. Row, Peter, “Design Thinking”, MIT Press, Cambridge, London, 1988.15.A.S.Hornby, “Oxford Advanced Learner‘s Dictionary of Current English”, Oxford

University Press, 3Ed, London, 1974.16.Mitchell, William J., “ The Logic Of Architecture, Design, Computation, and Cognition “,

3rd Ed, The MIT Press, London, 1992.171.Chase, S. C., “ Modeling Designs With Shape Algebra’s and Formal Logic “, Ph.D.

dissertation, University of California, Los Angeles, 1996.18.Gero, J. S., “ Shape pattern recognition using a computable shape pattern representation “,

Artificial Intelligence in Design, Kluwer, Dordrecht, 1998.19.Ching, Francis, D.K., “ Architecture, Form, Space, and Order”, 2nd Ed, John Wiley &

Sons, inc, U.S.A, 1996.20.Frankl, Paul,” Principles Architectural History –The Four Phases of Architecture Style:

1420-1900”, 1968.21.Argan, Guilio Carlo, “ On Typology of Architecture “, Architectural Design, Vol 33,

No.11/12, London, 1963.22.Baker, Geoffrey H., “ Design strategies in Architecture- an approach to the analysis of

form”, 2nd ED, St. Edmundsbury Press Ltd, Great Britain, 1996.23.Broadbent, Geofrey ,“Design in Architecture, Architecture and the Human Sciences”,

David Fulton Publishers, London, 1988.24.Kaissi, S., “ The Role of Nature & Cultural Environment 0n the fabric of the city”, Vol. 1,

Ph.D. Thesis, University of Sheffield, 1983.25.Gelernter, Mark, “ Sources of Architectural Form”, Manchester University Press,

Manchester & new York, 1995.26.Laseau, Paul,” Graphic Thinking for Architects & Designers”, 3rd Ed, John Wiley & Sons,

Inc, New York, 2001.27. ""1992.

:

52

28 . ""

20

)2(]2158[

)1 (Bungalow]1108-104[


53

)3(

)The Classical Temple (Megaron House)(

]342-44[

)4 (

]361-60[

)5 (S.Sophia

]12/5/2007[

:

54

)7 (

]16113[

)8 () (]10114[

)9 ( ]178[

)6 (]765 [


55

)10 (StretchingShear

]16116[

–

:

56

/ /

. .

.

. .

Analogy in the architectural design process in the academic mediumAnalytical study of students’ projects

Imad. M. ALBakri Assistant Professor

AbstractThe current research tackles the study of analogy depended on in designing products ofstudents of academic architectural study as it is one of the important designing strategies inthe process of architectural designing because it is closely related with the stage of synthesisand the derivations of architectural concepts within the designing process. The researchdiscusses the importance of this concept in order to explore the particular problem representedby the unclarity of analogy role in the designing products of students of academicarchitectural study over their different grades. Therefore the problem of the research wascrystallized and its objective and methodology were identified by studying analogy by twoaxes, included the analogy and the designer, the analogy and designing process, reaching thedetermination the theoretical framework, that involved four main items, which are; the item ofanalogy cause, analogy sources, the paradigm of dealing with the analogy source and thenature of the designing element relevant to analogy, firstly. And then applying the theoreticalframework on selected projects of students of academic architectural study in the second,third and fourth grades in Mosul University, secondly, in order to explore the change ofanalogy application paradigms adopted by the students in those grades, thirdly. The resultsshowed that the differences in students’ trends throughout different academic grades led to theemergence of similar differences in the nature of practicing analogy in those grades. Keywords: Analogy, Architectural design process, academic medium.

Bayda Hanna Saffo Assistant Lecturer

2008/10/32009/9/2


57

1-

. .

.

) ( .

2-

) (

.

3-

:

.:

.

.

4-) (

/ .

.

.

:

58

5-5 -1Analogy

.McGinty"

"[Mc Ginty, 1979, p.223].

.McGinty MetaphorSimile

.[McGinty, 1979, p.228].

Fowler . "

." " :

. "Fowler :LogicGrammarMathematics

Rhetoric[Broadbent, 1978, p.329]. .

.Broadbent

" .[Broadbent, 1978, p.329] "

.]1985493[ . ..

. ") (

. "

5-2

.

.

5-2 -1 :

.Popper :"


59

."[Cross, 1984, p.259] .GelernterHillier "

" ": ...

."[Gelernter,1995,p.274] .

.Broadbent "

: ."[Broadbent , 1978, p.340]

Sydney Opera HouseJohn Utzon"

.Utzon

. "]1996171-176[ .

Analogy .

) () . (Popper

) (Gelernter . .

-

. .Dowing "

Mode.......

] "Dowing, 1992, p.65 .[ Broadbent

" .

)engrams ( . .

EcclesSubconscious operationengrams .

. . "[Broadbent, 1978,p.341]

:

60

Lang Perception .

. [Lang , 1987,p.84]. :]199710 [.

: Unconscious .

. :

. .

.]199711[ .

) (

.

Guides .

-

Creative Techniques .

[Broadbent, 1978, p.421] .Synectics .

Synectics

[Broadbent , 1978, p.349- 353].Gorden :[Laseau, 1980,p.118-119]

-Symbolic Analogy : .

-Direct Analogy :

.-Personal Analogy : .

.-Fantasy Analogy :

.

.


61

5 -2-2 :

.

McGintyAnalogiesMetaphorsEssencesProgrammatic ConceptsIdeals

.[McGinty , 1979, p.223].Jencks

Analogical Models .

.]199618[ .

KuhnKubler :

KoestlerSchon

.[ Abel , 1997,p.141].Broadbent

:) (Pragmatic Design

[Broadbent , 1978, p.25-29] .IconicDesign

[Broadbent, 1978, p.30] .Analogical Design .

.[Broadbent , 1978,p.30-35] .Canonic Design

.[Broadbent, 1978, p.35-37] .Broadbent

. :

..]1998[..

Gelernter:

[Gelernter, 1995,p.7].

... .[Gelernter, 1995,p.15]

.

:

62

6-6 -1

:

6 -1-1 : .

) ( .

.

.) ().(

Sydney Opera HouseJohn Utzon"

]. "1996173 .[) (

.Green

Joyce]200061 .[

.

6 -1-2 :

.

-

.Broadbent "

.LeCorbusierRonchamp

WrightJohnson Wax Factory ] "Broadbent, 1978,p.423 .[.

.

1 -

. .

.

.Ham Common)1958 (Broadbent"

GownandStirlingLeCorbusierJaoulHouses"Paris)1956-1954 (

]Broadbent,1978.p.423 .[


63

.

2 -

. .

.Dowing

PrototypePrecedentsPrinciplesFormal ]Downing, 1992,p.67.[

) ( .Wayne O. AttoMathematical Analogy

Biological AnalogyOrganicBiomorphicLingwistic AnalogyMechanical Analogy

Dramaturgical Analogy]Atto,1979, p.23- 32 [

.Laseau

Physical Structural MechanicalControlOrganicPlanAnimal .Cultural

ManSocietySymbolic]Laseau, 1980, p.120- 121 .[

.JencksUtzon

" ] "Jencks, 1991,p.40 .[ "

] "200049 .[.

-

.Laseau

.

6 -1-3 :

.

.RonchampLe CorbusierBroadbent"

Sequence of direct Analogies Le Corbusier

:

64

SketchingStirling "Aschia ]Broadbent, 1978,p.342 .[Broadbent

.

.JencksRonchamp

".Coded Over

] "1996200 .[.

6 -1-4 :

. .

Ronchamp.

Twa TerminalNew YorkSaarinenJencks"] "Jencks, 1991,p.44 . [

Face HouseYamashitaJencks "] "Jencks, 1993,p.183 [

.

) . (

. .

]6-1.[7 -

.

:

.


65

.

. :

)(

1

)()(

2

3

4

A-1 -A1]7-1.[

]1-6 [

:

66

2 - A2]7-2.[3 -A3]7-3. [4 -A4]7-4.[B-1 - B1]7-5.[2 - B2]7-6.[3 - B3]7-7.[4 -B4]7-8.[C-1 -C1]7-9.[2 - C2]7 -10.[3 - C3]7 -11.[4 - C4]7-12.[

.

]8-1 [.

8-8 -1

.]8 -[

8 -1-1

- .

)50 (%)33.33 (%)16.67 (% .

)40 (%)20 (% .

)44.45 (%)11.11 (%.

- .)75 (%)25 (%

.)57.15 (% )42.85 (%

.)80 (%)20 (%

.


67

:

68

8-1-2 -

)60 (%)20 (%)20 (% .

)77.8 (%)22.2 (%.

)87.5 (% )12.5 (%.

1 -

)60 (%)20 (% .)60 (%

)20 (%)20 (%.

)60 (%)40 (%.

)75 (%)25 (%.

2 -)60 (%

)20 (%)20 (% .)25 (%)33.3 (%)16.7. (%

-

)100 (%)50 (%.

8 -1-3

)100 .(% .

)57.1 (%)42.9 (% .)66.7 (%

)33.3 (%.

8 -1-4.

)100 .(%

]1-8 [.


69

)42.2 (%)14.2 (% .

.)37.5 (%)12.5 (%)25 (%.

8 -2

.

8 -2-1-:

) (.

- .

.

.-

.

8 -2-2

.

:1 -

.

.

.

. .

.

.

:

70

2 -

. .

.

-

.

.

.-

.

..-

..

.-

. .

. .

..

3-

. .

..

4 - .

. .

.


71

..

5 - .

.

.

.

.

9-

. .

.

.

.

10-

.

.

.


73


75


77


79


81


83

:

84

" "–1997 ." "–1996

. " "–1998 .

" "–1996 . . " "–

1985 ." "–

. ": "

-2000.

Abel, Chris, " Architecture and Identity" , U.K., Architectural press. Oxford, 1997.Atto, Wayne, " Theory, Criticism, and History of Architecture" in James c. Snyder andAnthony J. Catanese, " Introduction to Architecture" , U.S.A, Mc Graw- Hill Book company ,1979.Broadbent, Geoffrey, " Design in Architecture- Architecture and the Human Sciences",London, John Wiley & Sons , 1978.Cross, Nigel, " Developments in Design Methodology ", London, John Wiley & Sons Ltd.,1984.Dowing , F.," The Role of Place and Event Imagery in the Act of Design" ,The Journal ofArchitectural and Planning Research , a: 1 Spring, USA, Locke Science Publishing Company, Inc, 1992.Gelernter, Mark, " Sources of Architectural form" , U.S.A, New York, Manchester Universitypress, 1995. Jencks, Charles, " The Language of Post- modern Architecture",16TH edition, London,Academy Editions, 1991.Jencks, Charles, " Architecture Today ", London, Academy Editions, 1993.Lang, Jon," Greating Architectural Theory - The Role of the Behavioral Sciences inEnvironmental Design", New York, Van Nostrand Reinhold company, 1987.Laseau , Paul," Graphic Thinking for Architects and Designers" , New York , Van NostrandReinhold company, 1980.McGinty , Tim, " Concepts in Architecture ", in James C. Snyder and Anthony J. Catanese,"Introduction to Architecture", U.S.A, Mc Graw- Hill Book company, 1979.

–

:

85

. .

/– /

. .

" . )Doorenbos and Al-Kassam ( ) ( .

. )0.10.222.50.30.1 ( .

.

. .

:

Estimating Yield Response Factors for Maize Crop in JensenModel

Dr. Ahmed Yousif Hachum Dr. Eman Hazim Sheet

Water Resources Department,College of Engineering, University of Mosul , Mosul, Iraq

ABSTRACTAmong the important proposed water-dependent crop production function is that of Jensen's.The application of this model requires knowledge of power parameters that reflect thesensitivity of each growth stage to water deficit. However, the values of these parameters areavailable only for a limited number of crops. Among the production models to whichsensitivity parameters are quite available is that of Doorenbos and Al-Kassam. The objectiveof this study is to relate these two models. The results of the analysis showed that the yieldresponse parameters for maize crop in Jensen model were found to be (0.1, 0.22, 2.5, 0.3, 0.1)respectively. Also, a polynomial equation is developed to relate predicted yield responseparameters in Jensen model to yield response factors, valid for all growth stages, inDoorenbos and Al-Kassam model. A comparison was made between the predicted parameterswith those obtained from previous works indicating acceptable agreement.Keywords: crop production function, yield response factor, Jensen model, maize.

2008/3/112009/6/29


86

– (Jensen,1968; Stewart and Hagan,1973;

Doorenbos and Al-Kassam, 1979) . . )1971 (Yaron

.)1978 ( Hexam and Heddy

, , , ,

, – )Doorenbos and Al-Kassam,1979 ( - .

, - .

)1984 (Martin et al.)(Simulation model . -

.

)1989(Martin et al..)2002 ( Kipkorir et al.

1.211.28 .)2002 (Liu. et al.- ,

- .

.

) Doorenbos and Al-

Kassam, 1979()1 (.

)1:()ky ( ) Doorenbosand Al-Kassam, 1979(

)(ky).(210.2260.4121.5250.4140.2

120

:

:

87

jjj

ns

j

CPETAETYY )(1

max ……….(1)

:-: Y

: maxY- : jAET

jCPET:-j

j :jj :

ns :5 j -

- )0.2 0.4 0.6 0.8 1.0 ( ) ( 55 =3125 .

) 2 ( 3125 )Ymax/Y ( ")1 :(

))1(1(1max j

jns

jj CPET

AETky

YY

………………. (2)

kyj

) Doorenbos and Al-Kassam, 1979 .(CPETAET

i .j

)1 ( :

5544332211 XXXXXZ ………………(3)

: = Z)Ymax/Y(

X =)CPETAET

(

12345 .3125) (ZX1X2X3X4X5)3 (j

) (j

:

)4......(iiiiiiiiiii XXXXXXXXXXY 5154143132122111


)6......(iiiiiiiiii XXXXXiXXXXXY 535434233223113 3


88


)8......(2555445335225115 iiiiiiiiiii XXXXXXXXXXY

i13125 .

:

…………..……..(9)54321 4.2004.2008.1264.2002.3524.473

54321 4.2004.2008.1268.3524.2002.493 ……………..….(10)

54321 8.1268.1266.1358.1268.1267.428 ……….…….…...(11)

54321 4.2008.3528.1264.2004.200505 …….………….….(12)

54321 8.3524.2008.1264.2004.2004.473 …….…………...(13)

)(MATLAB i )2:(

)2 : (j

123450.10.222.50.30.1

)0.992.(

iii kyky 29.09172.0 2 …………………….. (14)

i iky i

)Kipkorir ()3:()3 :(

0.150.10.320.221.922.50.420.30.1530.1

)3 (

. - )4(

.

)4 :(--y/ymax

:

89

12345

0.41.01.01.01.00.910.87

1.00.41.01.01.00.820.74

1.01.00.41.01.00.100.17

1.01.01.00.41.00.760.68

1.01.01.01.00.40.910.87

1. Doorenbos, J. and A. H. Kassam (1979). “Yield response to water”. FAO Irrigationand Drainage Paper No. 33.

2. Hexam, R. W. and E. O. Heady (1978). “Water production function for irrigatedagriculture”. Enter for Agriculture and Rural Development, Iowa State University press,Ames, IA.

3. Jensen, M.E. (1968).“ Water consumption by agriculture plants ”.Water Deficits and PlantGrowth , T T. Kozlowski, 1st ed., vol. 2,Academic press, New York, pp. 1-22.

4. Kipkorir, E. C. , D. Raes and B. Massawe (2002).”Seasonal water production functionsand yield response factors for maize and onion in Perkerra, Kenya”. Agricultural WaterManagement, 56: 229-240.

5. Liu, W. Z. , D. J. Hunsaker, Y. S. Li, X. Q. Xie, and G. W. Wall (2002).“Interrelations of yield, evapotranspiration, and water use efficiency from marginalanalysis of water production functions”. Agricultural Water Management, 56: 143-151.

6. Martin, D. L ., D. G. Watts and J. R. Jilley (1984). “Model and production functionfor irrigation management”. Journal of the Irrigation and Drainage Division, ASCE110(2): 149-164.

7. Martin, D. L. , J. R. Gilley and R. J. Supalla (1989). “ Evaluation of irrigationplanning decisions” Journal of the Irrigation and Drainage Division, ASCE, 115(1): 58-77.

8. Stewart, J. I. and R. M. Hagan (1973). “Function to predict effects of crop waterdeficits”. Journal of the Irrigation and Drainage Division, ASCE, 99(4): 421-439

9. Yaron, D. (1971). “Estimation and use of water production function in crops”.Journal of the Irrigation and Drainage Division, ASCE, 97(2): 291-303.

–

j. al-rafidain engineering vol.18, no.3 (2010)

Documents