Mechanical Engineering

International Review of

Mechanical Engineering (IREME) Contents

The Mechanism of Hydrogen Bubble Formation Caused by the Super Hydrophobic Characteristic of Taro Leaves by R. Subagyo, I. N. G. Wardana, A. Widodo, E. Siswanto

95

Flow Visualization of Flat and Curved Trapezoidal Winglet Vortex Generator in Fin-and-Tube Channel in Inline Arrangement by Mohd Fahmi M. Salleh, Mazlan A. Wahid, Aminuddin Saat, Natrah Kamaruzaman, Mohsin M. Sies, Ummikalsom Abidin

101

An Approximate Method for Stress Analysis in Butt Joints of FRP Pipelines by D. G. Pavlou

108

MATLAB Simulation and Validation of Fluid Properties in the Cross Flow Wet Cooling Tower by N. A. Rawabawale, S. N. Sapali

114

Attachment Probability of Particle on Bubble Surface and the Stability of Its Aggregate by Warjito, Harinaldi, Manus Setyantono, Sahala D. Siregar

121

Heat Carrier Vortex Motion Influence on the Hydrodynamics and Heat Exchange in the Pipes with Transverse Collars and Flow Core Energizers by D. Zhumadullayev, A. A. Volnenko, O. S. Balabekov, Zh. Serikuly, S. A. Kumisbekov, L. I. Ramatullayeva

127

An Investigation on Suppression of Vortex Instabilities for Flow Past a Bluff Body Using a Passive Device by Sunil A. S., Tide P. S.

132

Investigation on Failure Strength of Bolted Joints Woven Fabric Reinforced Hybrid Composite by D. Sivakumar, L. F. Ng, R. M. Chew, O. Bapokutty

138

CMM Path Planning, Position and Orientation Optimization Using a Hybrid Algorithm by Gabriel Mansour, Dimitrios Sagris, Αpostolos Tsagaris

144

Neural Network Model to Predict Exhaust Emissions on a Stationary Diesel Engine Operating with Castor-Oil-Plant Biodiesel Fuel by F. O. Narváez Argoty, A. Lyons Cerón, F. E. Sierra Vargas

151

Copyright © 2017 Praise Worthy Prize S.r.l. - All rights reserved

International Review of Mechanical Engineering (I.RE.M.E.), Vol. 11, N. 2 ISSN 1970 - 8734 February 2017

The Mechanism of Hydrogen Bubble Formation Caused by the Super Hydrophobic Characteristic of Taro Leaves R. Subagyo1,2, I. N. G. Wardana3, A. Widodo4, E. Siswanto5 Abstract – This study is aimed at uncovering the mechanism and role of the super hydrophobic characteristic of taro leaves on the process of hydrogen gas formation when there is a contact with a water droplet. The investigation was organized as: SEM-EDX analysis on the surface of taro leaf, observation on gas bubbles within a water droplet on the surface of taro leaves, and the detection of hydrogen gas production. The study result shows that the super hydrophobic characteristic of taro leaves caused the formation of great contact angle and high surface tension energy in droplets. A pointed-shaped nano texture caused the tension energy of the droplet surface to increase. As a result, particles randomly vibrate triggering the reaction between H2O droplets and Mg, K, and Ca on the surface of leaves producing hydrogen gas bubbles. Some gas was trapped in the nano grooves on the leaves surface and some with high pressure broke through the droplet and then were driven out by the Brownian motion. Copyright © 2017 Praise Worthy Prize S.r.l. - All rights reserved.

Keywords: Super Hydrophobic, Taro Leaves, Water Droplets, Hydrogen Bubbles

I.

A variety of superhydropobic surfaces has been produced in laboratory and even commercialized [3] – [5]. Some examples of self-cleaning products are paints, roof tiles, fabrics, and window glasses. The superhydropobic and hydropilic surface has been developed for the photoresponsive surface with inorganic oxyde and photoreactive organic molecules [6], copolymer film sensitive towards pH [7] or to electric field [8]. A leaf surface has the superhydropobic characteristic when it has a static contact angle above 1500. The examples of leaves surface being superhydropobic are lotus and talas leaves. Both of them have the highest contact angle with water [9] – [14]. The contact angle formed on taro leaves when it touches the water droplet is 165° [15]. The water droplet can roll on the surface of taro leaves with a very little drag coefficient because of the chemical composition and the surface topography of superhydropobic leaves [16]. The microstructure of taro leaves consists of a polygonal epidermic cell with a micro-bump scale in the size of 15-30 μm and a papilla in the central part of every cell. The whole surface is covered by epicuticular crystal wax cells with nano-size consists of aliphatic compound [17], [18]. The wax crystal is shaped like some little white hair [19]. The wax layer plays an important role in the self-cleaning mechanism of taro leaves [20]. The existence of the wax layer reduces the adhesive force and drag coefficient [19]. The result of XPS analysis (X-Ray Photoelectron Spectroscopy) of the wax layer on taro leaves shows that the atomic carbon concentration is 98,21% and atomic oxygen is 1.79% [21].

Introduction

The behavior of animals and plants is very interesting to observe, especially when they are adaptating to their environment. The anatomical form, together with the surface structure, has a uniqueness that still is the secret of nature. Animals and plants on earth provide a variety of material examples, surface forms and structures (replica) in which the features are replicable for practical application [1]. Based on some research, it shows that there are several plants having typical characteristics, such as: being able to convert chemical energy, possessing superhydropobic, hydrophilic, adhesive characteristics, and being able to show responsive movements when there is some stimulation [2]. Many experts on biology and materials started researching on plants superhydropobic characteristic. A uniqueness owned by the superhydropobic characteristic of plants is the ability to clean their own self up. The self-cleaning ability of superhydropobic leaves is influenced by the contact angle of hysteresis. The low contact angle of hysteresis plays an important role in self-cleaning or obstacle reduction process in fluid flow. The contact angle of hysteresis is the standard of dissipation energy during the flow of drops on the solid surface. At the low value of hysteresis contact angle, it is more likely for the drops to slide and roll wiping out the particle contaminant. At a less than 100 of contact angle of hysteresis, it is generally mentioned as surface able to perform the self-cleaning. The coarseness of the superhydropobic material surface and the self-cleaning ability on its surface bring up some inspiration in various applications.


https://doi.org/10.15866/ireme.v11i2.10621

95

R. Subagyo, I. G. N. Wardana, A. Widodo, E. Siswanto

The hydropobic characteristic of lotus leaves when it is in contact with water has been studied by [22]. The study result shows that there is some gas trapped in the surface of lotus leaves having an imporant role in the superhydropobic characteristic of lotus leaves [22]. This study concludes that gas trapped is air that is trapped due to the leaf surface roughness. The hydropobic characteristic of taro leaves has been previously observed by many researchers. But, the study on the mechanism of hydrogen gas formation in water droplets when in contact with taro leaves is still rarely considered.

II.

II.2.

II.2.1. The Measurement of Droplet Volume and Contact Angle The procedure for making droplets and measuring the contact angle is shown in Fig. 2. The droplets (5) were made using the syringe pump (1) with a particular volume and expelled drop by drop on the surface of taro leaves (6). Then, the contact angle of the droplet was measured using a digital microscope with the specification of: Image Sensor: 2.0 MP, focus range: 0100mm, magnification ratio: 1000× from the position (4) using the measurement software provided in the digital microscope.

Materials and Methods II.1.

Research Procedure

Material

II.2.2. The Observation of Hydrogen Bubbles

The material studied was fresh taro leaves as shown in Figure 1(a). The taro plant belongs to the kingdom: plantae, division: magnoliophyta, class: liliopsida, ordo: arales, family: araceae, genus: xanthosoma, and species: xanthosoma roseum. The taro leaves were cleansed from dust or faces clung on their surface. The composition and topography of taro leaves surface were examined using SEM-EDX with 1000, 5000, 10000, and 20000x magnification. The SEM instrument (Scanning electron microscope) used was Inspect S 50 FEI. Meanwhile, the experimental instrument of EDX (Energy Dispersive X-ray) used was Amatex-edax. Figure 1(b) is the surface topography and EDX experimental research of taro leaves of the entire surface: papilla part (the arrow sign) and mesophyll layer. The measurement results show that the surface of taro leaves contains: Carbon (C), Oxygen (O), Magnesium (Mg), Potassium (K), and Calcium (Ca). Metal elements, like Mg, K, and Ca, are highly reactive and strong oxidizing substances.

The hydrogen gas bubbles were observed using the digital microscope from two positions, (2) and (3), as shown in Fig. 2. The position of the microscope (2) is vertical towards the droplet to observe the gas bubbles in the base of the droplet and position (3) is to observe the bubbles formed due to the reaction of water droplets and taro leaves. This position is used because the appearance of gas bubbles is clearer.

Fig. 2. Droplet measurement technique and photography done using digital microscope

II.2.3. Hydrogen Sensor and Calibration Technique Figure 3 shows the calibration technique for a MQ-8 hydrogen sensor with pure hydrogen injected into the container (1) through silicon nipples (2) using a syringe pump (3) for every 0.25 ml up to 2.5 ml. The measurement result of the hydrogen sensor was compared to the hydrogen concentration actually injected into the container. Every measurement was repeated a hundred times (100×) and then the result was averaged. The calibration result is shown in Fig. 4.

Figs. 1. (a) Taro leaves (xanthosoma roseum) and (b) The topography and EDX experimental result of taro leaves surface


International Review of Mechanical Engineering, Vol. 11, N. 2

96


II.2.5. The Hydrogen Production Verification With Gas Chromatography The production of hydrogen gas from tube container in Fig. 5 was also verified using the gas chromatography (GC). The gas sample from container (2) was collected in the tube collector for the GC. The GC instrument used in this experiment is GC-8A Simadzu single detector TCD.

III. Result and Discussion The form of taro leaves surface at the magnification of 15000x is very coarse. The coarseness of the papilla part on the surface (arrow sign) is more prominent compared to the part of epidermis cell which is polygonal (as shown in Fig. 6(a). The surface texture is indeed sharppointed as shown in Fig. 6(b). Papilla has the most sharppointed texture. The sharp-pointed texture consists of epicuticular wax crystal nano cells that results in a superhydrophobic characteristic. Based on the EDX experimental result, epicuticular wax crystal cell layers contain elements such as C, O, Mg, K and Ca as in Fig. 1(b).

H2 concentration in ppm

Fig. 3. The calibration setting of the hydrogen detector

1300 y = 1,09x - 133,95 R² = 0,9897

800 300 -200

0

500

1000

1500

Sensor measurement in ppm Fig. 4. The calibration result using pure hydrogen

II.2.4. Data Acquisition Using Hydrogen Detector The hydrogen gas originating from the contact between water and taro leaves was detected using the hydrogen sensor by data acquisition system as in Fig. 5. Taro leaves (1) with areas varied as 0.04 m2, 0.08 m2 and 0.12 m2 were each put into a closed container (2) filled with water (3) and the hydrogen gas production was released into the air section on its top. The hydrogen sensor (5) was put into the tube as shown in Fig. 5. The process of collecting data of hydrogen gas was done in the interval of 8 to 24 hours.

Figs. 6. The magnification of taro leaves surface (a). Polygonal epidermis cell and (b).Papilla in the centre of every cell

The sharp-pointed nanotexture causes a very high surface tension when in contact with water droplets. The contact angles formed between the droplets and the leaves surface are 1650, 1580 and 1480 for the droplet diameter of 1, 2, and 3 mm, respectively as shown in Fig. 7. The contact angle reduced as the droplet volume increased as shown in Fig. 8. The angle of 1650 shows that taro leaves have the superhydrophobic characteristic. The large contact area makes the droplet on the taro leaves have very high surface tension energy. When the water droplet is in contact with the taro leaves surface, the texture of taro leaves which is sharp-pointed in nanoscale, then, propped up the surface of the droplet. The contact angle formed between nano texture droplets and taro leaves became larger.

Fig. 5. Data acquisition system of hydrogen gas Fig. 7. The super hydrophobic characteristic in taro leaves with droplet



97

Contact angle (degrees)


Figure 11 shows the gas bubbles movement photographed from the front side. It is seen that the gas bubbles move from the bottom to the top near the droplet edge. This gas bubble is from gas trapped on the leaves surface whose pressure has exceeded the surface tension of the droplet.

170 165 160 155 150 145 1

2

3

Droplet volume (ml) Fig. 8. The relation between droplet volume and contact angle on taro leaves

As a result, the surface tension energy on each of the sharp points with nanotexture popping up the droplet became significantly higher. The very high surface tension energy causes reaction between H2O droplet and the elements of taro leaves surface. The reaction follows these equations: Mg(s) + 2H2O

→ Mg(OH)2 + H2

(1)

Ca(s) + 2H2O

→ Ca(OH)2 + H2

(2)

2K(s) + 2H2O

→ 2KOH

(3)

+ H2

Fig. 9. Gas bubbles on the taro leaves surface when in contact with the droplet in the magnification of 1000x

The reaction is confirmed by the formation of bubbles on the taro leaves surface as shown in Fig. 9. Gas bubbles spread across all part of taro leaves surface in contact with the droplet. According to the measurement result, the average diameter of the bubbles is 21,2μm. The bubbles diameter has the same size of polygonal epidermis cells with a micro-bump scale as shown in figure 6a (circle sign). It means that some of the hydrogen gas is trapped in the polygonal epidermis cell spaces as a result of reaction. Fig. 10 shows the movement of hydrogen gas bubbles (pointed by white arrows) in the water droplet photographed above. This movement of gas bubbles indicates two things. First, it indicates that there was a reaction between water and the taro leaves surface on the contact point with papilla. The contact point on nano sharp points of the leaves surface stores a very high surface tension energy that triggered the reaction to produce hydrogen gas. Gas was trapped in the sharp particle spaces of papilla (Fig. 9). Gas production continually undergoing caused an increase of the volume of gas trapped so that the pressure increased as well. After pressure exceeded the surface tension of the droplet, the gas bubbles breakthrough the droplet and moved in the droplet. Second, the circling bubble movement signifies that there was Brownian motion indicating that the water molecule vibrates more intensively due to the high surface tension of the droplet. The stronger vibration of water molecule is a source of energy that promotes a reaction between the elements of taro leaves surface (Ca, Mg, and K) and water producing H2 gas.

Fig. 10. The emergence of hydrogen gas bubbles

Fig. 11. Hydrogen gas formation in the droplet of taro leaves



98


Fig. 12 shows that the gas sample from the reaction between water droplets and the leaves surface examined with GC apparently contains hydrogen gas that is of 2,287% (22870 ppm). It assures that the water droplet has reacted with the chemical element contained in the taro leaves surface as a result of a very high surface tension called superhydrophobic. The measurement result of hydrogen gas using MQ8 sensor is depicted in Fig. 13 with the areas of taro leaves of 0.04m2, 0.08m2 and 0.12m2. It appears that the production rate of hydrogen gas fluctuates almost periodically. It is as a result of hydrogen gas formed and trapped in the nano particle spaces on the surface of taro leaves periodically penetrating the water droplet in the form of gas bubbles.

when a water droplet is in contact with taro leaves surface. The mechanism of how the hydrogen gas is formed on taro leaves can be concluded as follows: a) There is a reaction between water and taro leaves surface at the contact point with papilla. The contact point on the nano sharp points of leaves surface stores a very high surface tension energy triggering a reaction to produce hydrogen gas. b) The emergence of Brownian motion indicates that the water molecule vibrates more intensively due to the high surface tension of the droplet. The more intense vibration of water molecules triggers a reaction between the elements on taro leaves surface (Ca, Mg, and K) with water producing H2 gas. c) There are two gas phenomena forming on the taro leaves surface, which are: gas trapped in the nano spaces and gas emerged in the droplet coming from the gas trapped on the leaves surface. The gas bubbles emerging in the droplet is caused by the pressure exceeding the surface tension of the droplet so that it periodically penetrates the water droplet.

Acknowledgements To the Ministry of Research, Technology, and Higher Education through General Directorate of Higher Education providing Scholarship for Post-Graduate Education in Home Country (BPPDN).

References [1]

Fig. 12. The experimental result of gas chromatography of taro leaves gas

Hydrogen product in (ppm/s)

0,16

[2]

0,12

[3]

0,08

[4]

0,04 [5]

0 1

101

201

301

401

[6]

501

Times (s) The areas of taro leaves (0.08m2) The areas of taro leaves (0.04m2) The areas of taro leaves (0.12m2)

[7]

Fig. 13. Hydrogen production in (ppm/s) [8]

IV.

Conclusion

This research aims at uncovering the mechanism of gas formation that is trapped or formed within a droplet and at identifying the substance contained in the gas

[9]


Dean B, Bhushan B, (2010), Shark-skin surfaces for fluid-drag reduction in turbulent flow: A review. Philosophical Transactions of the Royal Society A, 2010, 368, 4775–4806. DOI: 10.1098/rsta.2010.0201 Bhushan, B., (2009), Biomimetics: lesson from nature-an overview, Phil. Trans. R. Soc. A (2009) 367, 1445-1486. DOI: 10.1098/rsta.2009.0011 Bhushan, B., Jung, Y. C. & Koch, K. In press. Micro-, nano-, and hierarchical structures for superhydrophobicity, self-cleaning and low adhesion. Phil. Trans. R. Soc. A 367. DOI: 10.1098/rsta.2009.0014 Koch, K., Bhushan, B., Jung, Y. C. & Barthlott, W. In press b. Fabrication of artificial lotus leavesand significance of hierarchical structure for superhydrophobicity and low adhesion. Soft Matter. DOI: 10.1039/B818940D Roach, P., Shirtcliffe, N. J. & Newton, M. I. (2008) Progress in superhydrophobic surface development. Soft Matter 4, 224–240. DOI: 10.1039/b712575p Wang, S., Song, Y. & Jiang, L. (2007) Photoresponsive surfaces with controllable wettability. J. Photochem. Photobiol. C: Photochem. Rev. 8, 18–29. DOI: 10.1016/j.jphotochemrev.2007.03.001 Xia, F., et al, (2006) Dual-responsive surfaces that switch between superhydrophilicity and superhydrophobicity. Adv. Mater. 18, 432–436. DOI: 10.1002/adma.200501772 Bhushan, B. & Ling, X. (2008) Integrating electrowetting into micromanipulation of liquid. J. Phys. Condens. Matter 20, 485 009. DOI: 10.1088/0953-8984/20/48/485009 Barthlott, W. and Neinhuis, C., (1997), Purity of the Sacred Lotus, or Escape from Contamination in Biological Surfaces,Planta 202 (1997) 1-8. DOI: 10.1007/s004250050096


99


[10] Wagner, P., et al, (2003), Quantitative Assessment to the Structural Basis of Water Repellency in Natural and Technical Surfaces, J. Exper. Botany 54 (2003) 1295-1303. DOI: 10.1093/jxb/erg127 [11] Burton, Z. and Bhushan, B., (2006), Surface Characterization and Adhesion and Friction Properties of Hydrophobic Leaf Surfaces, Ultramicroscopy 106 (2006) 709-719. DOI: 10.1016/j.ultramic.2005.10.007 [12] Bhushan, B., (2009), Biomimetics: Lessons from Nature – An Overview, Phil. Trans. R. Soc. A367 (2009) 1445-1486. DOI: 10.1098/rsta.2009.0011 [13] Koch, K., Bhushan, B., and Barthlott, W., (2008), Diversity of Structure, Morphology, and Wetting of Plant Surfaces (invited), Soft Matter 4 (2008a) 1943-1963. DOI: 10.1039/B804854A [14] Koch, K., Bhushan, B., and Barthlott, W., (2009), Multifunctional Surface Structures of Plants: An Inspiration for Biomimetics (invited), Prog. Mater. Sci. 54 (2009a) 137-178. DOI: 10.1016/j.pmatsci.2008.07.003 [15] Hans J. Ensikat., et al., (2011), Superhydrophonicity in perfection: the outstanding properties of the lotus leaf, Beilstein J. Nanotechnol. 2011,2,152-161. DOI: 10.3762/bjnano.2.19 [16] R. Blossey, (2003), Self-cleaning surfaces- virtual realities, Nature Materials, 2 301 306. DOI: 10.1038/nmat856 [17] Bhushan, B. dan Jung, Y.C., (2006), Micro and Nanoscale Characteriztion of Hydrophobic and Hydrophilic Leaf Surfase, Nanotechnology 17 (2006) 2758-2772. DOI: http://dx.doi.org/10.1088/0957-4484/17/11/008 [18] Muler, F., et al., (2007), Self-Cleaning Surfaces Using the Lotus Effect, in Handbook for Cleaning/Decontamination of Surfaces. 2007, Elsevier Science B.V.: Amsterdam. p. 791-811. [19] Burton, Z. and Bhushan, B., (2005), Hydrophobicity, Adhesion, and Friction Properties of Nanopatterned Polymers and Scale Dependence for Micro- and Nanoelectromechanical Systems, Nano Lett. 5 (2005) 1607-1613. DOI: 10.1021/nl050861b [20] Koch, K. and H.-J. Ensikat, (2008), The hydrophobic coatings of plant surfaces: Epicuticular wax crystals and their morphologies, crystallinity and molecular self-assembly. Micron, 2008. 39(7): p. 759-772. DOI: 10.1016/j.micron.2007.11.010 [21] Jianwei Ma, (2010), Impact of surface topography on colloid and bacterial adhesion, A thesis submitted in partial fulfulment of the requirements for the degree master of science, Rice University, Houston,Texas. [22] Jiadao Wang., et al., (2009), Investigation on hydrophobicity of lotus leaf: Experiment and theory, Plant Science 176 (2009) 687695. DOI: 10.1016/j.plantsci.2009.02.013

Rachmat Subagyo was born in Balikpapan, Indonesia, on 5th of August 1976. He obtained her bachelor’s deegre inmechanical engineering energy conversionfrom the National College of Technology (STTNAS) Yogyakarta, Indonesia and the Master of engineering degree in Hasanuddin University, Makassar. He is currently a Ph.D. student under supervision of Prof. I.N.G. Wardana. His research is centered on Bioenergy, Renewable Energy and fluid mechanics. In 2008 he received an offer to join Mechanical Engineering Department, Lambung Mangkurat University as lecturer. I. N. G. Wardana is a professor of mechanical engineering, Brawijaya University. He was Born in, Denpasar 3th of July 1959. He received the B.E. degree in mechanical engineering from the University of Brawijaya, Malang, Indonesia, the M.Eng.and Ph.D degrees in mechanical engineering from Keio University, Yokohama – Japan. Prof. I.N.G. Wardana has more than 25 years experience in mechanical engineering, and is the author of over 50 peer-reviewed scientific publications, conference papers, books and patents. Agung Sugeng Widodo was born in Malang, Indonesia, on 21 of March 1971. He received the B.E. degree in mechanical engineering from the University of Brawijaya, Malang, Indonesia. The Master of engineering and Ph.D. deegre in mechanical engineering from Gajah Mada University Yogyakarta, Indonesia and University of Southern Queensland Australia, respetively.In 1998 he received an offer to join Mechanical Engineering Department, Brawijaya University as lecturer. He has published over 2 International publications. Eko Siswanto was born in Sidoarjo, Indonesia, on 17 of Oktober 1970.He received the B.E. and the Master of engineering degree in Brawijaya University, Malang, Indonesia. Ph.D. deegre inYamaguchi University, Japan.In 1998 he received an offer to join Mechanical Engineering Department, Brawijaya University as lecturer.At this time also served as head of the laboratory production process.He worked research in fluid mechanics, Heat and Mass Transfer, and Condensation. He has published over 2 International publications.

Authors’ information 1

PhD student of Mechanical Engineering Brawijaya University, Indonesia. 2

Department of Mechanical University, Indonesia.

Engineering

Lambung

Mangkurat

3,4,5

Department of Mechanical Engineering Brawijaya University, Indonesia.



100


Flow Visualization of Flat and Curved Trapezoidal Winglet Vortex Generator in Fin-and-Tube Channel in Inline Arrangement Mohd Fahmi M. Salleh1, Mazlan A. Wahid2, Aminuddin Saat2, Natrah Kamaruzaman2, Mohsin M. Sies2, Ummikalsom Abidin2 Abstract – In this study, a flow visualization experiment using the dye injection technique was conducted. Visual observation of the flow characteristics across the scaled up fin-and-tube heat exchanger (FTHE) channel model with and without trapezoidal winglet vortex generator (TWVG) were presented. Four cases of FTHE channel in inline tube arrangement were examined with TWVG at different geometries and arrangements, as well as a case without the vortex generator. The two TWVG geometries used in this study were a flat trapezoidal winglet (FTW) and a curved trapezoidal winglet (CTW) either arranged in common flow up (CFU) or common flow down (CFD) compositions. The Reynolds number (Re) used ranged from 500 to 2500. It was found that with the introduction of TWVG in the FTHE channel, the wake region size behind tubes reduced compared to the baseline case. However, the formation of an additional wake region behind the TWVG was also observed. Copyright © 2017 Praise Worthy Prize S.r.l. - All rights reserved.

Keywords: Trapezoidal Winglet, Vortex Generator, Fin-And-Tube Heat Exchanger

Fig. 1. Vortex generator common geometry and orientation

It was noted that the wing type vortex generator is classified as the transverse vortex generator (TVG) and the winglet type is classified as the longitudinal vortex generator (LVG) [2]. The LVG was found to be more effective than the TVG when heat transfer and pressure drop is considered according to Biswas et al. [3] and Tigglebeck et al. [4]. From the study done by Torii et al. [5] and Yanigahara et al. [6] on a pair of delta shape vortex generators mounted on a flat plate, the winglet pair which produced counter rotating vortices was found to provide the highest heat transfer enhancement compared to the wing pair. The LVG can be arranged in two different compositions known as common flow up (CFU) and common flow down (CFD) [7]. The CFD arrangement is when the distance between the trailing edges of the winglet pair is longer than the leading edges. While the CFU arrangement is when the leading edges of the winglet pair are more than the trailing edges. Besides the common vortex generator geometry, other geometries such as the block type winglet and the annular vortex generator have been investigated by previous researchers [8],[9]. However, a simple geometry such as delta winglet is more encouraging because of its efficiency, low maintenance and low operating costs [10]. Another simple vortex generator geometry that has the potential to improve the FTHE performance is the trapezoidal winglet vortex generator (TWVG) as reported by Jalil et al. [11], Lotfi et al. [12], Lu et al. [13] and Zhou et al. [14]. According to Jalil et al. [11], the TWVG was found to enhance heat transfer more efficiently than the rectangular winglet and delta winglet. Differently, the vortex generator in curved geometry was also studied by



I.

Introduction

Vortex generator as an enhancement technique in plain FTHE captured the attention of researchers in the past decades. The application of a vortex generator to the FTHE is purposely designed to overcome the high thermal resistance that occurs at the air side of the FTHE by creating the secondary flow, disrupting the growth of the flow boundary layer, creating fluid swirling and disturbing the flow [1]. Thus, it resulted in a better flow mixing and further enhanced the heat transfer across the plain FTHE. A vortex generator can be mounted on the fin in various geometries. Fig. 1 shows the common vortex generator geometry and the orientations studied by previous researchers. They are delta wing, rectangular wing, delta winglet and rectangular winglet.

101

Mohd Fahmi M. Salleh Salleh,, et al.

Gong et al. [15] and Lin et al. [16 [16]. ]. It was found that the curved geometry vortex generator significantly increases heat transfer performance across the FTHE as compared to the reference plain FTHE. The TWVG has shown to provide the optimum heat transfer enhancement across the FTHE than other simple vortex generator geometries. geometries. However, a few studies have been conducted on the effect of the TWVG in different geometr geometries ies and arrangement arrangements rrangements on the flow characteristic across the FTHE channel. Therefore, a flow visualization experiment using the dye injection technique is performed to visualize the flow pattern especially in the wake region and secondary vortices created across the FTHE channel with and without the presence of TWVG. This paper aims at obtain obtaining ing a better understanding on the flow characteristic across the FTHE when different TWVG geometr geometries ies and arrangements arrangements are applied behind the tubes. The result results showed the flow characteristic across the FTHE especially on the wake characteristic region formation behind the tubes and other effects effect of applying the TWVG such as the additional vortices created. Although no quantitative data is presented, it is important to gain an in-depth in depth understandi understanding ng on the flow structure and beha behavior vior for different TWVG geometries and arrangement arrangements across the FTHE since they are related to the thermal hydraulic performance performance of an exchanger [17,18 [17,18]. ]. Therefore, the outcome can be used to further improve the FTHE performance performance using TWVG.

II. II.1.

within the range of the contra contraction ction ratio area suggested by Mehta et al. [19 [19] which is 6 to 99.. The contraction section iiss connected to the honeycomb section purposely to remove the swirl and the lateral mean velocity variations of the flow before entering the inlet section. The inlet section iss applied to guarantee that the fully develop developed flow enters enter the test section. At the test section, the the 3 to 1 sscaled caled up FTHE channel model iiss placed an andd the flow visualization experiment using the dye injection technique is performed. The test section iiss connected connected to the exit section which iiss the last section of the water tunnel. Water will flow from the tank to th thee exit section before the water flow can be recirculat recirculated ed back to the tank using a pipeline. A 3--phase phase centrifugal pump with a flow rate capacity of 20 to 120 liter/minute is used to drive the water flow across the water tunnel experimental rig rig.. In order to control the speed of the centrifugal pump, an inverter (Fuji Inverter 60 Hz Hz)) unit with the smallest increment of 0.01 Hz is connected to the centrifugal pump. The he water temperature of the tank is measured by using a temperature sensor in order to calculat calculatee the thermodynamic property of water during the experiment. Generally, stainless steel and acrylic were used to fabricate the water tunnel experimental rig. The tank, the contraction section, the honeycomb section, the entrance section and the exit sectio section n were made of stainless stainless st steel.. The The test section was made using transparent acrylic as main material and supported by a stainless steel plate placed at the edges of the acrylic structure. The dye is introduced in the test section using gravity feed. The dy dyee injection rate is controlled by using a drip valve placed in the dye injection tube in order to obtain the correct dye velocity respect to the water flow velocity. Three locations of the dye probe were used in this study; study; one of the location locations is directed to the center stagnation point of the tube while the other two locations are 7 mm apart from the dye probe placed at the center location. The flow characteristic across the FTHE channel with and without a vortex vortex generator is then recorded using using Sony HDR HDR XR500V that has a resolution of 1920 × 1080 and a frame rate ability of 30 frames per second.

Experimental Setup Experimental Rig And Procedure

The fflow low visualization experiment using dye injection technique requires a water tunnel experimental rig, a dye injection system, an illumination system and a recording device in the setup. The water tunnel exp experimental erimental rig used comprises a tank, a contraction section, a honeycomb section, an entrance section, a test section and an exit section section,, as shown in Fig. 2. The experimental rig tank is filled with water for the flow visualization experiment. experiment i linked to a contraction section eriment. The tank is where the chosen contraction ratio area is 66.. This is

Fig. 2. Experimental Experimental facility (1) Tank (2) Contraction section (3) Honeycomb section (4) Entrance section (5) Test section section (6) Exit section (7) Pump (8) R Recording ecording equipment (9) Dye injection set

Copyright © 2017 Praise Worthy Prize S.r.l. - All rig rights hts reserved

International Review of Mecha Mechanical nical Engineering, Vol. 11, N. 2

102


II.2.

Test Model

The FTHE channel models model used in this study were made of transparent acrylic purposely to visualize the flow characteristic using the dye injection technique. The FTHE ch channel annel and the TWVG dimension were developed based on the study by Gholami et al. [20], Lotfi et al. [12 [12], ], Z Zhou hou et al. [14] [1 ] aand nd Jalil et al. [11]. [11 Table I shows the detail dimension for the FTHE used in this study. In order to obtain sufficient resolution for the flow visualization experimental experimental result and due to the limitation on the test section, a 3 to 1 scaled up FTHE channel model was used. TABLE I FTHE CHANNEL DETAIL DIMENSION Dimension Parameter Symbol (1 : 1) Tube row number N 4 Outer tube diameter D0 10.55 mm Tube collar diameter Dc 10.95 mm Fin thickness δ 0.20 mm Fin pitch Fp 3.00 mm Transversal tube pitch Pt 25.40 mm Longitudinal tube pitch Pl 25.40 mm Characteristic length L 101.60 mm

(a)

Dimension (3 : 1) 4 31.65 mm 32.85 mm 0.6 mm 9 mm 76.2 mm 76.2 mm 304.8 mm

(b)

As shown iin n Fig. 3(a) 3(a),, the FTHE channel model consist consists of 2 flat plates represent representing ing the top and bottom fin surface of the heat exchanger. A ttotal otal of 12 tubes arranged in inline arrangement slotted through the holes in the flat plates. While the TWVG is placed behin behind d the tubes. The bottom plate of the FTHE channel model is design designed to have the ability to mount different TWVG geometries and arrangements arrangement using the same bottom plate. Therefore, the bottom plate is especially specially designed to have a series of platform port ports with respect to the center of the vortex generator based on a study by Gholami et al. [20 [20]. Then, the vortex generator mount mounted ed on the vortex generator platform was connected to the platform port according to the desired TWVG geometry and arrangement arrangement,, as sshown hown in Fig. 3(b) 3(b).. The FTHE channel with and without the presence of the TWVG was assembled using the socket model purposely to hold the model inside the test section. The model socket is also used to control the fin pitch between the plates according to the the standard dimension used. The detail dimension for the TWVG used is presented in Fig Figs.. 4 and Table II. Note that there are two TWVG geometries used in this study. They are the flat trapezoidal winglet (FTW) and the curved trapezoidal winglet (CTW) as sho shown wn in Fig. 4(a) and Fig. 4(b), 4(b) respectively respectively.. The CTW are designed with respect to the tube center with the dimension similar to FTW construction. The TWVG arrangement behind the tubes is shown in Fig. 5. There are two arrangement FTW, arrangementss used for the FTW which are the CFD and the CFU arrangement. While the CTW is arranged in a similar way to the FTW in CFU arrangement.

Figss. 3. FTHE channel (a) assembly an and d (b) bottom plate construction

Parameter Thickness Baselength Height Toplength Angle of attack

TABLE II TWVG DETAIL DIMENSION Dimension (1 Symbol : 1) δ 0.20 mm a / a’ 6 mm b / b’ 3 mm c / c’ 3 mm β 30°

(a)

Dimension (3 : 1) 0.6 mm 18 mm 9 mm 9 mm 30° 30

(b)

Figss.. 4. Schematic Schematic of (a) FTW and (b) CTW

Fig. 5. The TWVG arrangement

III. Result and Discussion III.1. Case without without Vortex Generator The injected dye in front of the first tube row at Re = 500 was found to hit the round tube before it twisted back a little forming horseshoe vortex. The swirl was created and it flows surrounding the round tube. Then, it



103


was separated ffrom rom the tube and the dye travelled travelle towards the second tube row. It was observed that, after the dye hit the front portion of the second tube row, it turned back to the first tube row forming a huge recirculation zone, as shown in Fig. 6(a). A similar recirculation zone or wake region was also found behind recirculation the next tube row further downstream. It was also observed that the top line and the bottom line of the injected dye formed a straight streakline as it travelled further downstream. This scenario showed llow ow flow mixing across the baseline case occurred at Re = 500 and might affect the overall heat transfer. A ffurther urther increment of the Reynolds number showed that the horseshoe vortices were also formed and a nearly similar flow trajectory such as in Re = 500 was found.

(a) Re=500

The recirculation zone was observed to be slightly smaller than the Re = 500 case with the recirculation center shift towards the rear portion of the first tube. This situation might be due to flow fluctuation and flow mixing in the surrounding surrounding of the tube junction led by the horseshoe vortices [1]. The dye injected at the top line and the bottom line was found to become more unstable as the Reynolds number increased, as shown in Fig. 6(b) and Fig. 6(c). This unstable flow might increase the overall heat transfer due to better flow mixing across the FTHE channel together with smaller wake regions behind the tube observed as the Reynolds number increased.

(b) Re=1500

(c) Re=2500

Figs. Fig . 6. Wake region behind tub tubes es and dye trajectory of the top line and the bottom line for baseline case

The result results obtained for the plain FTHE channel was found to be similar to the previous studies studies. As shown in Fig. 77, a similar flow structure was obtained at Re = 600 in the st udy y by Lemouedda et al. [21 [21], especially focused stud on the wake structures formed behind the tubes. While the horseshoe vortices previously previously described by Mon and Gross [2 2] and Wang et al. [9 [9]] had been observed in the the [22] current experimental study. The horseshoe vvortices ortices were found to be weak in this study at low Reynolds number numbers and became more sub substantial stantial as the Reynolds number increased.

compared to the baseline case. Then, the flow that separated from the tube was directed by the TWVG to the rear region of the first tube row and directed near the front stagnation point of the second tube row. This resulted in a smaller recirculation region found formin formingg behind the tube. A nnearly early similar dye trajectory was observed further downstream. As shown in Figs. Fig . 8 8,, a further increment of the Reynolds number resulted in a more unstable flow trajecto trajectory ry across the FTHE channe channell with the presence of TWVG TWVG. It was also observed observed, at Re = 1500, that no significant flow structures can be ttraced raced further downstream starting from the second tube row rear region aass shown in Figs Figs.. 88(b),, (d)) and (f). ( However, the recirculation region behin behind d the tube was observed to form behind the first tube row for both Re = 500 and Re = 1500 cases. The FTW in CFD arrangement as in Fig. 8(a) and Fig. 8(b) 8(b) was observed to provide the smallest wake region formed behind the tube. While the wake region behind tube for the CTW was found to be slightly smaller than the wake region for the FTW in CFU arrangement. Although the size of the wake region behind the tubes was reduced with the introduction of the TWVG as compared to the baseline case, an additional wake region was found formed in the rear re region gion of the vortex generator. Figs. Fig . 9 show the wake region formation behind the TWVG obtained obtained from the experimental study. The biggest size of the wake region behind the vortex generator was observed at the FTW in CFD arrangement compared to the FTW in C CFU FU arrangement and CTW.

Fig. 7. Flow structure across the plain FTHE channel

III.2. Case with ith Vortex Generator For all FTHE channels channel with TWVG cases, th the horseshoe vortices were observed to form in the first tube row before the flow followed the circular tube contour and separate separated d from the first tube surface. The TWVG was observed to delay the flow separation point in the first tube row with respect to the the front stagnation point as



104


Figss. 8. Flow visualization result for β=30° and λ=2 =2: (a) (a – (b) (b) FTW in CFD arrangement, (c (c)) – (d) d) FTW in CFU arrangement and (e) (e – (f) f) CTW

Figss. 9. Wake rregion egion formation behind the TWVG

While the wake region size behind the vortex generator for CTW was found to be slightly larger than the case of FTW in CFU arrangement. A further observation on the flow characteristic across the FTHE with TWVG found the secondary vortices generated by the vortex generator, as shown in Fig Figss.. 10. 10 For all TWVG geometries and arra arrangements, ngements, the main vortices observed generated from the vortex generator, as shown in Fig Figs.. 10 10. The main vortex was generated due to the effect of flow passing the sharp edges of the TWVG

geometry. While for the FTW in CFU arrangement and the CTW case, corn corner er vortices were found and directed towards the rear region of the first tube, as shown in Fig. 10(b) and Fig. 10 10(c). (c). The corner vortex was created after the flow hit the TWVG area facing the main flow. For the CTW case, the direction and the intensity of corner vortices towards the first tube rear region was observed to be stronger than the FTW in CFU arrangement. Note that the TWVG area facing the main flow for



105


CTW winglet is wider than the FTW in CFU arrangement. Therefore, it can be said that the corn corner er vortices generated are dependent on the TWVG area facing the main flow with respect to the similar arrangement.

observed to be directed near to the front stagnation point of the second tube row. 3) The effect of implementing the vortex generator on the fin and tube channel was observed to provide swirling motion and encourag encouragee flow destabilization to the flow field which may cause an intense mixing between the main flow and the flow in the wall region. This phenomena may result in heat transfer enhancement. 4) However, it was noticed that an additional wake region was formed behi behind nd the vortex generator.

Acknowledgements The authors would like to thank Universiti Teknologi Malaysia for supporting this research work under the Research U University niversity Grant (RUG Vote 13H43) and FRGS Vote 4F777.

References [1]

[2]

[3] Figs. Secondary vortices created by the TWVG Fig . 10. Secondary [4]

IV. Conclusion A total of 4 FTHE channel models were tested in this study that includes the CTW and FTW in CFD and CFU arrangements. The experiment was done at Reynolds number ranging from 500 to 2500 for all cases. The test was performed using a water tunnel for the flow visualization method using the dye injection technique. Note that this pape paperr only presents the quantitative data on the flow characteristic across the FTHE with the introduction of the TWVG. Although no quantitative data was reported in this study, the flow pattern obtained shows promising prospective towards heat transfer enhanc ement using the vortex generator. The major enhancement conclusions of the present study are summarized as follows follows: 1) The case without vortex generator was observed to have the biggest recirculation region behind tubes. As supported by the previous literature, the recirculation recirculation region provides low mixing which resulted in poor overall heat transfer of the heat exchanger. 2) By implementing the vortex generator, the size of the wake region behind the tubes was found to be reduced compared to the case without vortex generator generator.. The flow from the first tube row was also

[5]

[6]

[7]

[8]

[9]


Sahin, B., Akkoca, A., Ozturk, N N.. A., Akilli, Akilli, H., Investigations of flow Characteristics in a Plate Fin and Tube Heat Exchanger Model Composed of Single Cylinder, (2006) International Model Journal of Heat and Fluid Flow Flow,, 27 (3) (3),, pp. 522–530 522 530. doi:: 10.1016/j.ijheatfluidflow.2005.11.005 Fiebig, M., Embedded Vortices in Internal Flow - HeatFiebig, Heat-Transfe Transferr and Pressure Loss Enhancement, (1995) International Symposium on Turbulence, Heat and Mass Transfer Transfer, Lisbon, Portugal: Butterworth-Heinemann Butterworth Heinemann, Heinemann pp. 376 376–388 388. doi: 10.1016/0140-6701(96)87924 10.1016/0140 6701(96)87924 6701(96)87924-0 Biswas, G., Deb, P., Biswas S., Generation off Longitudinal Streamwise Vorties - A Device for or Improving Heat Heat-Exchanger Exchanger Design, (1994) Journal off Heat Transfer - Transactions of the Design, ASME 116 (3), pp. 588– ASME, 588–597 597. doi:10.1115/1.2910910 doi:10.1115/1.2910910 Tiggelbeck, S., Mitra, N.K., Fiebig Fiebig,, M., M Comparison of of WingWingType Vortex Generators for for Heat Transfer Enhancement inn Channel Flows Flows,, (1994) Journal off Heat Transfer, Transfer 116, pp. 880– 880– 885 885. doi:10.1115/1.2911462 doi:10.1115/1.2911462 Torii, K., Yanagihara, J.I., Nagai, Nagai, Y., Heat Transfer Enhancement by Vortex Generators Generators,, (1991) (1991).. Proceedings of the ASME/JSME Thermal Engineering Joint Conference Conference,, New York: York: J.R. Lloyd and Y. Kurosaki, pp. 77– –83. doi: http://id.ndl.go.jp/bib/000002226653 Yanagihara Yanagihara,, J. I., Torii, Torii, K K., Heat Transfer Augmentation by Longitudinal Vortices Rows. (199 (1993) 3) Proceedings of the Third World Conference on Experimental Heat Transfer, Fluid Mechanics, and Thermodynamics Thermodynamics, 1, Honolulu, Honolulu, Hawaii, USA: M. D. Kelleher, pp. 560–567 560 567. doi: http://dx.doi.org/10.1016/B978 http://dx.doi.org/10.1016/B978-00-444 444-81619 81619--1.50065 1.50065-4 1.50065 Jacobi A. M., Shah, Shah, R. K., Heat Hea Tr Transfer ansfer Surface Enhancement Through the he Use off Longitudinal Vortices: A Review of of Recent Progress Progress,, (1995) Experimental Thermal aand nd Fluid Science, Science 11(3), 11(3), pp. 295–309 295–309. doi: 10.1016/0894-1777(95)00066 10.1016/0894 1777(95)00066 1777(95)00066-U U Leu, J.S., Wu, Wu, Y.H., Jang J.Y., Heat Transfer and Flu Fluid id Flow Analysis in Plate Plate-Fin Fin and Tube Heat Exchangers with a Pair ooff Block Shape Vortex Generators, (2004) International Jo Journal urnal of Heat and Mass Transfer, Transfer, 47, pp. 4327-4338 4327 4338. doi: 10.1016/j.ijheatmasstransfer.2004.04.031 Wang, C.C., Lo, J., Lin, Lin, Y.T., Liu Liu,, M.S., Flow Visualization off Wave Type Vortex Generators Having Inline Fin Wave-Type Fin-Tube Tube Arrangement, Arrangement, (2002) International Journal off Heat an nd d Mass Transfer Transfer, 45, 45, pp. 1933-1944 1933 1944.. doi: 10.1016/S001710.1016/S0017-9310(01)00289 9310(01)00289 9310(01)00289-77


106

Mohd Fahmi M. Salleh, et al.

[10] Salviano, L. O., Optimization of Vortex Generators Positions and Angles in Fin-Tube Compact Heat Exchanger at Low Reynolds Number, (2014) Ph.D, Thesis, Sao Paulo State University. Available at www.teses.usp.br/teses/disponiveis/3/3150/tde.../THESIS_Leandr o_Salviano.pdf [11] Jalil, J.M., Abdulla H.K., Yousif, A.H., Effect of Winglet Shape on the Heat Transfer from Heated Cylinder in Cross Flow, (2006) JKAU: Engineering Science, 17 (2), pp. 119-130. doi: 10.4197/Eng.17-2.6 [12] Lotfi, B., Sunden, B., Wang, Q., An Investigation of the ThermosHydraulic Performance of the Smooth Wavy Fin-and-Elliptical Tube Heat Exchangers Utilizing New Type Vortex Generators, (2016) Applied Energy, 162, pp. 1282-1302. doi: http://dx.doi.org/10.1016/j.apenergy.2015.07.065 [13] Lu, G., Zhou, G., Numerical Simulation on Performances of Plane and Curved Winglet-Pair Vortex Generators in a Rectangular Channel and Field Synergy Analysis, (2016) International Journal of Thermal Sciences, 109, pp. 323-333. doi: 10.1016/j.ijthermalsci.2016.06.024 [14] Zhou, G., Ye, Q., Experimental Investigations of Thermal and Flow Characteristics of Curved Trapezoidal Winglet Type Vortex Generators, (2012) Applied Thermal Engineering, 37, pp. 241248. doi: 10.1016/j.applthermaleng.2011.11.024 [15] Gong, B., Wang, L.B., Lin, Z.M., Heat Transfer Characteristics of a Circular Tube Bank Fin Heat Exchanger with Fins Punched Curve Rectangular Vortex Generators in the Wake Regions of the Tubes, (2015) Applied Thermal Engineering, 75, pp. 224-238. doi: 10.1016/j.applthermaleng.2014.09.043 [16] Lin, Z.M., Liu, C.P., Wang, L.B., Numerical Study of Flow and Heat Transfer Enhancement of Circular Tube Bank Fin Exchanger with Curved Delta-Winglet Vortex Generators, (2015) Applied Thermal Engineering, 88, pp. 198-210. doi: 10.1016/j.applthermaleng.2014.11.079 [17] Huisseune, H., Performance Evaluation of Louvered Fin Compact Heat Exchanges with Vortex Generators, (2011) Ph.D. Thesis, Gent University. doi: http://hdl.handle.net/1854/LU-1968989 [18] Onal, S., Investigations of Flow Characteristics around a Single and Staggered Slotted Cylinders, (2010) Ph.D. Thesis, Cukurova University. doi: http://library.cu.edu.tr/tezler/8052.pdf [19] Mehta, R.D., Bradshaw, P., Design Rules for Small Low Speed Wind Tunnels, (1979) Aeronautical Journal of the Royal Aeronautical Society, pp. 443–449. doi: https://doi.org/10.1017/S0001924000031985 [20] Gholami, A.A., Wahid, M.A., Mohammed, H.A., Heat Transfer Enhancement and Pressure Drop for Fin-and-Tube Compact Heat Exchangers with Wavy Rectangular Winglet-Type Vortex Generators, (2014) International Communications in Heat Transfer, 54, pp. 132-140. doi: 10.1016/j.icheatmasstransfer.2014.02.016 [21] Lemouedda, A., Breuer, M., Franz, E., Botsch, T., Delgado, A., Optimization of the Angle of Attack of Delta-Winglet Vortex Generators in a Plate-Fin-and-Tube Heat Exchanger, (2010) International Journal of Heat and Mass Transfer, 53 (23-24), pp. 5386–5399. doi: 10.1016/j.ijheatmasstransfer.2010.07.017 [22] Mon, M.S., Gross, U., Numerical Study of Fin-Spacing Effects in Annular Finned Tube Heat Exchangers, (2003) International Journal of Heat and Mass Transfer, 47, pp. 1953–1964. doi: 10.1016/j.ijheatmasstransfer.2003.09.034

Mohd Fahmi Md Salleh received his B.Eng. from Universiti Teknologi Mara, Malaysia, in 2013 and M.Phil. from Universiti Teknologi Malaysia, Malaysia in 2017. Currently he serves as a lecturer at the Faculty of Mechanical Engineering, Universiti Teknologi MARA, Pulau Pinang, Malaysia.

Mazlan Abdul Wahid obtained his Ph.D. from State University of New York at Buffalo, New York in 2002, M.Sc. from University of Leeds, United Kingdom and B.Sc. from Embry-Riddle Aeronautical University, Florida. He is currently a Professor at the Faculty of Mechanical Engineering in Universiti Teknologi Malaysia, Malaysia. His main research interests are highspeed combustion phenomena, thermodynamics, energy studies and heat transfer enhancement. Currently, he leads the High Speed Reacting Flow (HiREF) Laboratory research group. Aminuddin Saat received his Ph.D. in 2011 from University of Leeds, United Kingdom, M.Sc. from Coventry University, United Kingdom in 2002 and B.Eng. from Universiti Teknologi Malaysia, Malaysia in 1999. Currently, he is a senior lecturer at the Faculty of Mechanical Engineering in Universiti Teknologi Malaysia, Malaysia. Mainly, his research interests are combustion and flame studies, burning rate and flame propagation, flame instabilities, combustion in aerosols and sprays, optical diagnostics, thermodynamics and heat transfer studies. Natrah Kamaruzaman is a dynamic and motivated young researcher with 3 years working experience in industry. She first graduated from University of The Ryukyus, Okinawa Japan in Mechanical Engineering System and finished Master and Phd in Mechanical Engineering at Universiti Teknologi Malaysia, Malaysia. She is currently teaching thermodynamics and heat transfer and working on various fluid dynamics and heat transfer in microchannel projects. She has deep interest in solving industrial problems and at the same time helping the community to achieve energy sustainability. Mohsin Mohd Sies is currently is a senior lecturer at Faculty of Chemical and Energy Engineering in Universiti Teknologi Malaysia, Malaysia. He obtained his M.Sc. in Mechanical Engineering from University of Michigan, Ann Arbor in 1995 and B.Sc. in Nuclear Engineering from Rensselaer Polytechnic Institute in 1989. His main research interests are on combustion, supersonic reacting flows, heat transfer and sustainable engineering. Mr. Mohsin is a member of Society of Automotive Engineers, Combustion Institute and Institute of Marine Engineering Science and Technology and a Chartered Marine Engineer. Ummikalsom Abidin currently serves as a senior lecturer at Faculty of Mechanical Engineering, Universiti Teknologi Malaysia, Malaysia. She obtained her Ph.D. from Universiti Kebangsaan Malaysia in 2016, M.Eng. from Universiti Teknologi Malaysia in 2006 and B.Eng. from Universiti Tenaga Malaysia in 2002. Her research interest are in the area of Lab-on-Chip (LOC), microfluidics, microelectromechanical system (MEMS) and microchannel heat transfer.

Authors’ information 1

Faculty of Mechanical Engineering, Universiti Teknologi MARA (Pulau Pinang), Jalan Permatang Pauh, 13500 Permatang Pauh, Pulau Pinang, Malaysia. 2

Department of Thermofluids, Faculty of Mechanical Engineering, Universiti Teknologi Malaysia, 81310 UTM Skudai, Johor Bahru, Malaysia.



107


An Approximate Method for Stress Analysis in Butt Joints of FRP Pipelines D. G. Pavlou Abstract - Butt joint is a standard connection type for FRP pipelines. Shear stresses are taking place on the interfaces between the wrapping layer and the pipeline. An analytical model for calculating above stresses for joints subjected to axial loading is presented. The model is based on the equilibrium equations of an adhesive element and the bonded pipe elements, as well as the material properties of the anisotropic pipe wall, the wrapping element and the adhesive. Numerical results for typical FRP joints are derived and commented. Copyright © 2017 Praise Worthy Prize S.r.l. - All rights reserved.

Keywords: Butt Joint, FRP Materials, Pipelines, Adhesive, Shear Stress

yield a significant reduction in operation and installation cost. However, the joining methods of FRP pipes are different [3], [4] than the steel ones [5], [6]. A typical joining method consists of a wrapping layer [7] of fibers impregnated with a catalyst resin over the butt joint (Fig. 1).

Nomenclature dx B D E1, E2 Ex Fx G12 Gg L tg t N εα εb γ ν12 σx τ τm τu Kh Kt ϑ

Length of the elementary adhesive layer Width of an elementary strip of the pipe material Exterior diameter of the pipe Modulus of elasticity in the principal directions of the lamina Extensional modulus of elasticity in the xdirection of the laminate Axial force of the pipeline Shear modulus in the plane 1-2 of the lamina Shear modulus of the adhesive layer Length of the bonded materials Thickness of the elementary adhesive layer Thickness of an elementary strip of the pipe material Number of laminae composing the laminate Normal strain on the adhesive layer Normal strain on the adhesive-pipe interface Shear strain on the adhesive layer Poisson’s ratio in the plane 1-2 of the lamina Normal stress acting on the pipe elements Shear stress on the adhesive layer Nominal shear stress Allowable shear stress Stress concentration factor Normalized stress concentration factor Fiber orientation angle

I.

Fig. 1. Typical joint of FRP pipes

During the installation and operation period, the joint should have the capacity to carry the shear stresses taking place on the interface between the wrapping layer and the pipes due to axial forces and bending moments of the pipeline. Numerical methods for stress analysis based on complex algorithms (e.g. FEM or BEM [8]) provide approximate results. The accuracy of any numerical method depends on the adopted assumptions for the material modelling as well as on the mesh density for the modeling of the geometry. In any case, the above design strategy is not practical for design engineers [9, 10]. Alternative to the numerical modelling, analytical methods yield functions or design curves that can be handled easily in the design practice [4], [11] – [13]. In the present work, an analytical procedure for stress analysis in butt joints of FRP pipelines subjected to axial loads is proposed. Efficient analytic formulae are provided and implemented on typical examples.

Introduction

Since FRP materials provide high strength to weight ratio [1], FRP pipelines represent a challeging solution compared to steel pipelines currently in use. Moreover, their stability in corrosive environment along with the excellent mechanical behavior in fatigue conditions [2]



108

D. G. Pavlou

II.

of the shear strain γ. The value of the above displacements DD1 and CC1 is equal to γtg. Apart from the effect of the shear strain, the normal strains εb and εα move the points B and C1 to the final locations B1 and C4 respectively. The value of the displacements BB1 and C1C4 is εbdx and εαdx respectively. It is convenient to analyze the total shear strain at the and of the segment dx in two components, namely γ and dγ. Assuming that the angle γ is small, the following equation can be obtained from the geometry of the deformed adhesive element:

Formulation of the Problem

The consequence of the action of axial forces on pipelines is the development of shear stresses on the adhesive between the wrapping layer and the connected pipes. To determine the shear stress distribution along the above interface of the connected pipes and the wrapping layer, the mechanical model shown in Fig. 2 is adopted.

   CC angle C2 B1C4   2 4   C2 B1

(1)

or:

  d 

 tg  dx   dx    dx   b dx  tg

Fig. 2. Basic geometric parameters of an adhesive joint

Since the longitudinal stresses in the two pipes are uniformly distributed along their perimeter, the model consists of two thin strips with a small width B and thickness t (material α). The wrapping layer (material b) has same thickness t, and the thickness of the adhesive is tg. Each pipe element is bonded to the wrapping layer along the length L. The coordinate center for the axis x is located in the middle of L. Aim of the analysis is the determination of the shear stress distribution along the length L as well as the maximum value.

(2)

After some simple algebraic manipulations, above equation yields:

d     b  dx tg

(3)

1  d  d  b   0 t g  dx dx 

(4)

or:

d 2 dx

III. Stress and Strain Analysis

2



With the aid of the Hooke’s law for the adhesive:

A free body diagram of an adhesive element will be considered (Fig. 3) for stress and strain analysis.

  Gg 

(5)

equation (4) can be written in terms of shear stress τ:

d 2 dx

2



Gg  d   d  b  t g  dx dx

 0 

(6)

Fig. 3. Free body diagram of an adhesive layer of elementary length dx

The un-deformed shape of the adhesive element to be analyzed is ABCD. The length and the thickness of the element are dx and tg respectively. The normal stress σx acting on the pipe elements (Fig. 2) causes: (i) shear strain γ on the adhesive layer, (ii) normal strain εb on the surface of the adhesive layer, which is in contact with the wrapping material, and (iii) normal strain εα on the adhesive layer which is in contact with the pipe wall. The displacement of the points D and C of the adhesive layer to the locations D1 and C1 respectively is a consequence

Fig. 4. Free body diagram of an elementary strip of a pipe material bonded to the wrapping material

The free body diagram for an elementary strip (Fig. 2) of the pipe material bonded to the wrapping material is demonstrated in Fig. 4. The shear forces equilibrate the axial force difference:



109

D. G. Pavlou

 Bdx   x Bt   x  d x  Bt

d 2

(7)

dx

 m 2  0

2

(13)

or: where: d  t x dx

(8) m2 

2Gg Ex N t t g

(14)

The general solution [15] of the ordinary differential equation (13) is:

  x   A1 cosh  mx   A2 sinh  mx 

(15)

The unknown parameters A1, A2 can be determined by the following conditions:

b  L / 2  0 Fig. 5. Principal (1-2) and global (x-y) coordinate systems

 a  L / 2 

Assuming that the fiber orientation angle for a single lamina is  (Fig. 5) the extensional modulus of elasticity in the x-direction is given (e.g. [14]) by the following equation:

Ex 

E1  E  cos 4    1  2 12  sin 2  cos 2   G12  E  1 sin 4  E2

 a  0  b  0  0

(16)

x NE x

(17)

 symmetry 

(18)

Denoting by Fx the axial force of the pipeline, the axial stress can be expressed as:

(9)

x 

Fx  Dt

(19)

Then, the condition (17) yields: where E1, E2 are the values of the elasticity modulus in the principal directions 1-2, and G12, ν12 are the values of the shear modulus and the Poisson’s ratio in the plane 12 respectively. Assuming that the filament wound laminated material α is composed by N laminae with symmetric and balanced winding (fiber orientation  ), the Hooke’s law can be expressed as:

 x   NEx  

  L / 2  

d dx

(10)

0

(21)

x 0

Combining above equation with the Hooke’s law for the adhesive given by eq. (5), the following formula can be obtained: (11)

d dx

Following the same procedure for the wrapping material b, the following equation can be obtained:

db 1   dx E x Nt

(20)

Taking into account eq. (18), equation (3) yields:

Therefore, eq. (8) yields:

d  1   dx E x Nt

Fx  D t N Ex

0

(22)

x 0

With the aid of above equation, the parameter A2 can be determined by the general solution (15):

(12)

A2  0

With the aid of eqs (11), (12), equation (6) can be written as:

(23)

Then, eq. (15) can be simplified as:



110

D. G. Pavlou

  x   A1 cosh  mx 

(24)

Kh 

The parameter A1 can now be determined with the aid of eqs (3), (16), (20). Combining these equations, the following formula can be obtained:

d dx

xL / 2

Fx  t g  D t N Ex

 max m

(31)

where τm is the nominal shear stress:

m 

(25)

Fx  DL

(32)

Therefore: With the aid of Hooke’s law for the adhesive, above equation can be written as: d dx

 xL / 2

Gg Fx t g  D t N Ex

Kh 

Gg L t g t N E x m tanh  mL / 2 

(26) The condition for adhesive failure is:

 max   u

Taking into account eq. (24) above equation yields: A1 

Gg Fx t g  D t N E x m sinh  mL / 2 

Fmax 

(28)

Gg Fx t g  D t N E x m sinh  mL / 2 

(35)

(29)

Fig. 7. Maximum axial force versus the length of the adhesive layer

Fig. 6. Schematic demonstration of the shear stress distribution along the adhesive layer in a butt joint of two pipes subjected to axial force Fx

From eq. (35) it can be concluded that the parameter tanh(mL/2) affects the value of Fmax. Therefore, the Fmax takes the maximum value for a length L satisfying the condition:

With the aid of the final solution (28), a schematic demonstration of the shear stress distribution is shown in Fig. 6. This figure indicates that the ends ( x   L / 2 ) of each segment of the bonded area are subjected to maximum shear stresses. For x   L / 2 the solution (30) yields:

 max  

u t g t  D N E x m tanh  mL / 2  Gg

Therefore, the length L of the adhesive can affect the maximum axial force. Schematic representation of Fmax versus L is shown in Fig. 7.

where the constant parameter λ is given by the following formula:



(34)

where τu is the allowable shear stress of the adhesive. Using above equation, the maximum axial load that the joint can carry is:

(27)

Since the constants A1, A2 have now been determined, the final solution for the shear stress distribution along the adhesive layer can be expressed by eq. (15):

  x    cosh  mx 

(33)

tanh  mL / 2   1

(36)

The value of Fmax corresponding to above condition is: (30)

Fo 

The shear stress concentration factor at the ends of the bonded segments can be calculated by the following equation: Copyright © 2017 Praise Worthy Prize S.r.l. - All rights reserved

u t g  D t N Ex m Gg

(37)


111

D. G. Pavlou

IV.

For pipes with nominal diameter D=1.0 m and adhesive with allowable stress τu=42 MPa the maximum axial force Fmax versus length L for ϑ=  30o, 45o, 60o can be derived by eq (35). The above results are demonstrated in Fig. 10. Figure 10 indicates that the optimum orientation for best axial load capacity of the joint is ϑ=30o. The effective length for the three cases 30o, 45o, 60o is 0.079 m, 0.088 m and 0.107 m respectively.

Implementation of the Model

A representative butt joint of multi-layered filament wound pipes made by glass fiber reinforced polymeric material is considered for the implementation of the proposed model. The shear modulus and the thickness of the adhesive layer is Gg=906 MPa and tg=0.3 mm respectively. The laminae composing the pipe wall have thickness tl=0.15 mm. A number of N=34 symmetric and balanced laminae with fiber orientation angle  is considered to compose the wall, yielding a thickness of t=5.00 mm. The material properties of the FRP material with respect to the principal coordinate system are E1=40 MPa, E2=8.8 MPa, G12=3.9 MPa, ν12=0.29. Taking into account the above data, the parameter m versus the orientation angle ϑ can be derived by eqs (9), (14) and demonstrated in Fig. 8.

Fig. 10. Maximum axial force versus length for ϑ=  30o, 45o, 60o

V.

1. An analytical procedure for approximate calculation of shear stress distribution in the adhesive layer of butt joint of FRP pipelines is presented. The model takes into account the equilibrium equations of an elementary volume of the adhesive layer as well as the equilibrium equations of the pipe material and the wrapping layer. 2. Since numerical methods (e.g. FEM) are usually used for stress analysis of butt joints, the proposed procedure is advantageous because it provides analytic formulae for stress distribution and stress concentration factor calculation. 3. From the derived analytical function for the shear stress distribution it can be concluded that the parameters affecting the stress concentration factor are the fiber orientation angle of the FRP pipe and wrapping material, the shear modulus of the adhesive, the thicknesses of the three layers (pipe, adhesive, wrapping), the principle material properties of the anisotropic materials, the number of laminae composing the laminate, and the length of the joint. 4. The maximum values of the shear stress take place at the ends of the bonded materials. 5. The stress concentration factor increases rapidly for small length of butt joint. 6. The capacity of the joint to carry axial force increases versus the length of the butt joint. However, for high values of length, the axial load capacity tends to be constant. 7. Apart from the length of the butt joint, the axial load capacity is affected by the fiber orientation angle. In

Fig. 8. Graphical representation of the parameter m versus the fiber orientation angle β

With the aid of eq. (33) the following dimensionless index of the stress concentration factor: Kt 

t g t N Ex m Gg L

Kh

Conclusion

(38)

versus ϑ can be demonstrated in Fig. 9 for L=30 cm.

Fig. 9. Normalized stress concentration factor K t versus fiber orientation angle ϑ



112

D. G. Pavlou

the proposed example the maximum axial load capacity corresponds to fiber orientation angle ϑ=π/6. The effective length for fiber orientation angles π/6, π/4, and π/3 has been 0.079 m, 0.088 m, and 0.107 m respectively.

Authors’ information Department of Mechanical and Structural Engineering and Materials Science, University of Stavanger, NO-4036 Stavanger, Norway. E-mail: [email protected] Dimitrios G. Pavlou is a Professor in the Department of Mechanical and Structural Engineering and Materials Science at the University of Stavanger in Norway. In 2014 he was elected a full member of the Norwegian Academy of Technological Sciences. He has had twenty years of teaching and research experience in the fields of finite elements, boundary elements, mechanics of solids, and fracture mechanics. Prof. Pavlou has published many research publications and authored / edited five books and conference proceedings. He is a reviewer in more than 21 international journals and participated as a plenary speaker and session chair in many international conferences. Today is a Research Group Leader (Faggruppeleder) of the Mechanical Engineering and Material Science Group.

References [1]

[2]

[3] [4] [5]

[6]

[7]

[8]

[9]

[10]

[11]

[12]

[13]

[14] [15]

Pavlou, D., Kourousis, K., A Phenomenological Approach for Fatigue Damage Accumulation of CF/PEEK Laminates Under Two-Stage Loading, (2013) International Review of Mechanical Engineering (IREME), 7 (7), pp. 1323-1328. D. G. Pavlou, N. V. Vlachakis, M. G. Pavlou, V. N. Vlachakis, Estimation of fatigue crack growth retardation due to crack branching, Computational Materials Science, Vol. 29, n. 4, pp. 446-452, 2004. L. Tong and C. Soutis, Recent Advances in Structural Joints and Repairs for Composite Materials (Springer, 2011). D. G. Pavlou, Composite materials in piping applications (Destech publications, 2013). Mvola, B., Adaptive Gas Metal Arc Welding Control and Optimization of Welding Parameters Output: Influence on Welded Joints, (2016) International Review of Mechanical Engineering (IREME), 10 (2), pp. 67-72. Yang, X., Kah, P., The Analysis of the Relationship Among the Different Factors Influencing Welding Production by the Gephi Software, (2015) International Review of Mechanical Engineering (IREME), 9 (5), pp. 491-498. K. Rege, Undamped vibration of fibre-reinforced polymer overwrapped pipes under fluid hammer conditions, Oil & Oil Products Pipeline Transportation: Science & Technologies, Vol. 23, n. 3, 2016. D. G. Pavlou, Boundary-integral equation analysis of twisted internally cracked axisymmetric biomaterial elastic solids, Computational Mechanics, Vol. 29, n. 3, pp. 254-264, 2002. Pavlou, D., Transfer Matrices Analysis of FRP Pipelines Stability, (2016) International Review of Mechanical Engineering (IREME), 10 (3), pp. 165-172. D. G. Pavlou, Dynamic response of a multi-layered FRP cylindrical shell under unsteady loading conditions, Engineering Structures, Vol. 112, pp. 256-264, 2016. D. G. Pavlou, Undamped Vibration of Laminated FRP Pipes in Water Hammer Conditions, Journal of Offshore Mechanics and Arctic Engineering-Transactions of The ASME, Vol. 137, n. 6, pp. 061701-1-8, 2015. D. G. Pavlou, Pressure-wave propagation in multi-layered fibrereinforced polymeric pipelines due to hydraulic hammer, Journal of Pipeline Engineering, Vol. 1. pp. 29-35, 2014. D. G. Pavlou, Design challenges of steel pipelines in unsteady flow conditions, Journal of Pipeline Engineering, Vol. 14, n. 3, pp. 157-162, 2015. M. Hyer, Stress analysis of fiber reinforced composite materials (Destech publications, 2009). A. D. Polyanin, A. V. Manzhirov, Handbook of Mathematics for Engineers and Scientists (Chapman and Hall/CRC, 2006).



113


MATLAB Simulation and Validation of Fluid Properties in the Cross Flow Wet Cooling Tower N. A. Rawabawale, S. N. Sapali Abstract – The efficiency of steam-based thermal power plant includes the performance of various components such as condenser, boiler, turbine and cooling tower. The cooling tower is a device which is used to reject the waste heat through the cooling water to the ambient air. The performance of cooling tower is normally obtained by means of energy analysis using properties of the fluid at the inlet and outlet. The performance of cooling tower is gauged with its efficiency and effectiveness. The variation of fluid properties inside the cross flow wet cooling tower along with the direction of fluid flow has not been studied exhaustively. As the phenomenon of heat and mass transfer in cross flow wet cooling tower is bidirectional and complex. This paper consists of extraction of fluid properties based on heat and mass transfer principle across a cross flow wet cooling tower using MATLAB simulation. Hence with the view to carry out analysis of the cross flow wet cooling tower, wet-bulb temperature and humidity of air along with the horizontal plane and temperature of water along with the vertical plane are obtained using MATLAB simulation and the results of the simulation are validated through experimentation. Copyright © 2017 Praise Worthy Prize S.r.l. - All rights reserved.

Keywords: Cross Flow Wet Cooling Tower, Evaporative Cooling

The temperate water is admitted at the top of the tower in the form of fine water droplets which moves downward transferring heat and mass of water to the fresh air admitted in the cooling tower. The waste heat present in the temperate water is rejected to the air through convection and evaporation. Through literature evaluation, it is observed that the energy analysis of cross flow cooling tower is done by the researchers based on properties of water and air at the outlet. They have focused on the variation of parameters with respect to ambient conditions. An energy analysis is usually used to examine the performance characteristics of a cooling tower. The different approaches with reference to the investigation of the wet cooling tower have been developed with the goal of exploring the potentialities of technology and enhancement of thermal performance with a view of energy conservation. Baker and Shryock [3] used NTU method for the analysis of counter flow and cross flow cooling tower. They have found that the fluid conditions remain constant across any horizontal section of a counter flow cooling tower, whereas both conditions vary horizontally and vertically in a cross flow cooling tower. Also, they demonstrated the variation of air enthalpy and water temperature graphically in cross flow the cooling tower. ASHARE [18] has confirmed that the lowest possible temperature attained by hot water in a cooling tower performing efficiently was the wet-bulb temperature of the air at the inlet of a tower. Also, the cooling tower performance depends on the air wet-bulb temperature at the inlet of a cooling tower. Whereas, the relative

Nomenclature a V Cp L G h W K R t ω ωi a v w air in db wb

Wetted surface area per unit volume of tower, [m2 m-3] Active volume of tower, [m3 ] Specific heat at constant pressure, [kJ kg-1 K-1] Mass flow rate of water, [kg s-1] Dry air mass flow rate, [kg s-1] Enthalpy, [kJ kg-1] Tower length, [m] Convective mass transfer coefficient, [kg m-2 s-1] Gas constant, [kJ kg-1 K-1] Temperature, [oC] Humidity ratio, [kgw kga-1] Saturated humidity ratio evaluated at water temperature, [kgw kga-1] Dry air Water vapour Water Moist air Inlet Dry bulb Wet bulb

I.

Introduction

Cooling towers are generally an open system direct or indirect contact heat exchangers. These are used thermally to reprocess circulating water for use in power plant condensers, refrigeration condensers, and the heat exchangers [1]-[30]. Copyright © 2017 Praise Worthy Prize S.r.l. - All rights reserved


114

N. A. Rawabawale, S. N. Sapali

humidity, air dry-bulb temperature have negligible influence on the thermal performance of artificial draft cooling tower. A stepwise integration method was used by Montri Pirunkaset [10] to find temperature and enthalpy of water and air in a cross flow cooling tower. It was observed that the temperature and enthalpy of the water decrease from the inlet to outlet, whereas the air temperature and enthalpy increases from the inlet to outlet. It was concluded that the temperature and enthalpy of water and air vary along both horizontal and vertical planes [25]. The variation was bidirectional in the cross flow cooling tower, whereas the variation in counter flow cooling tower was in the vertical plane only. The authors have not made an attempt to validate results by experimentation. A mathematical model based on the mass and heat transfer theory to find the outlet condition of the fluid for the cross flow cooling tower was developed by Mani Saravanan et al. [15]. The model was solved by using iterative method [26]. The analysis carried out had shown that the wet-bulb temperature of the air was the most influencing parameter than the temperature of water, affecting the performance of a wet cooling tower considerably. S.P. Fisenko et al. [6] proposed a mathematical model which includes the aerodynamics of a cooling tower and evaporative cooling rate for the thermal performance of a cross-flow wet cooling tower. The average water droplet radius plays a vital role in improving the thermal performance of a cooling tower was concluded. The performance of a cooling tower was foreseen by Wanchai Asvapoositkul et al. [1] using a basic model, which was identified by the equation of mass evaporation rate. The model was used to do calculations easily and precisely and was used to evaluate the performance of newly designed towers under a variety of operating circumstances. The influence of air humidity, water temperature and air temperature at the inlet of the air washer and the length of air washer on the exergy efficiency and saturation effectiveness were investigated by J.C. Santos et al. [13]. The condition of the air which yielded the best thermodynamic performance was not able to discharge better thermal comfort was observed. J.C. Santos et al. [14] have studied the influence of different parameters for the cooling process in air washers such as temperature, humidity, air flow rate and temperature of circulating water over cooling rate of air. They found considerable enhancement in the effectiveness of the process for reduced cooling water temperature and flow rate of air. Ebrahim Hajidavalloo et al. [20] used a mathematical model to simulate the influence of operating conditions on the thermal performance of a cross flow cooling tower. They observed that the increase in wet bulb temperature at constant dry-bulb temperature results in a decrease of approach, range and evaporation loss significantly [24].

In this paper, fluid properties along with the direction of fluid flow in a cross flow wet cooling tower are simulated using MATLAB and validated by the experimentation.

II.

Mathematical Model for Cross Flow Wet Cooling Tower and Analysis

The water flows from the top to bottom while the air flows in a horizontal direction. The conditions of the fluid vary along both vertical and horizontal directions in the cross flow wet cooling tower. For the purpose of investigation and experimentation some of the important assumptions made are as given below:  Transfer of heat from the fan of a tower to the fluid is negligible.  The coefficients of heat and mass transfer throughout the tower are constant.  The drift loss is negligible.  The specific heat of both the fluid is constant.  The transfer of heat and mass takes place in the direction normal to the fluid flow.  The influence of pressure variation on properties of the fluid is negligible.  The transfer of heat and mass through the walls of a tower to the surrounding is negligible. In order to carry out the investigation of the cross flow wet cooling tower, the condition of fluid flowing through the tower is required to obtain. Hence the entire cross section of the cooling tower is assumed to be divided into sixteen equal cells in order to maintain the resemblance of each cell with the cooling tower [3]. The Lewis number was assumed to be one for the transfer of sensible heat and mass into an overall coefficient based on enthalpy potential as driving force [23]. The heat lost by water is equated with the heat gained by air which gives: =

ℎ=

ℎ

−ℎ

,

(1)

,

The transfer of heat from the interface to the air stream is given by equation (1). By neglecting the film resistance based on driving potential of enthalpy at the bulk water temperature the equation becomes: ⁄ )=

(

=(

ℎ

,

− ℎ

⁄ ) ℎ =(

,

(2) ,

−ℎ

(3)

,

⁄ )

(4)

Change in water temperature in a cell: =


× (ℎ

,

−ℎ

,

)

(5)


115


(

ℎ

⁄ )=

ℎ

−ℎ

,

shown in Fig. 2. The specifications of experimental setup are given in Table I and the properties of the fluid at the inlet of cooling tower are given in Table II.

(6) ,

Change in enthalpy of air in a cell: × ℎ

−ℎ

,

(7)

,

Hot water inlet Cold air inlet

The rate of water loss by evaporation is equal to the rate of gain in moisture content by air. The decrease in water flow rate along with the height of a tower is related to the mass transfer of water vapour across the liquid air interface as given by equation (8). The humidity potential in turn is the difference between the saturation humidity at the interface based on liquid temperature and the humidity of bulk air stream Change in humidity of air:

Region A

Region B

Region C

Region D

Hot air outlet

ℎ =

Cold water outlet Fig. 2. Block diagram of cross cooling tower Hot water spray

×

(

−

)

(8)

Air inlet

=

III. Experimentation The experimental setup comprises i) Water distribution basin at the top of a tower. The basin consists of 280 holes each of 3mm diameter for uniform distribution of hot water, ii) Perforated PVC fill material to provide wetted surface area for heat and mass transfer shown in Fig. 1, iii) Water collection sump fitted with water heater at the bottom of tower, iv) Water circulation system consists of a pump, control valve and rota-meter, v) Exhaust fans are fixed at the exit to suck air uniformly passing through PVC fills of the tower and vi) A set of temperature sensors located at various nodes across the tower.

Air outlet

Water sump

Water heater Water temperature sensor WBT sensor DBT sensor Fan PVC perforated fill

Water Pump

Fig. 3. Schematic diagram of cross flow cooling tower

IV.

Results and Discussion

The temperature of water decreases from the top of a tower to bottom of the tower, through heat and mass transfer to the air. The outermost layer of water droplet absorbs latent heat of vaporization from the inner core of water droplet thereby self-evaporation and cooling inner core of water droplet. The rate of water cooling along the length of a tower does not remain uniform. The rate of water cooling as observed on air inlet side is higher as the air entering has enthalpy and humidity potential. TABLE I SPECIFICATIONS OF CROSS FLOW WET COOLING TOWER Width 0.3 m Length 2.0 m Height 2.0 m Active volume of tower (V) 1.2 m3 Mass flow rate of water (L) 0.4 kg/s Mass flow rate of air (G) 0.4 kg/s Temperature range 10 OC Expected evaporation losses 1.4 % Convective mass transfer co-efficient(K) 0.029 kg/m2s Wetted surface area (a) 9.9 m2/m3

Fig. 1. Sectional view of fill arrangement

The hot water is sprayed from the top of a tower which flows downwards and reaches the sump thereby cooling. The water in a sump is heated and re-circulated for experimentation. The distance moved by air in a cross flow cooling tower is referred as the length of a tower and the distance covered by hot water is referred as a height of tower as shown in Fig. 3. Whole cross section of the cooling tower is divided into sixteen equal sections for the purpose of experimentation, whereas it is again assumed to be divided into four regions A, B, C, and D for easy demonstration, comparison and explanation purpose as

TABLE II CONDITIONS OF AIR AT THE INLET OF CROSS FLOW WET COOLING TOWER Relative humidity (Ø) 57 Wet bulb temperature (WBT) 24 Dry bulb temperature (DBT) 31.3 Enthalpy of air (ha) 72.36 Humidity ratio (ω) 0.016


% O C O C kJ/kg kgw/kga


116


Height (m) 0 0.5 1.0 1.5 2.0 Length (m)

TABLE III WATER TEMPERATURE IN OC MATLAB Simulation result 40.8 40.8 40.8 40.8 40.8 35.7 36.8 37.6 38.3 33.7 31.7 33.4 34.8 36.0 24.7 28.6 30.6 32.3 33.7 24.5 26.2 28.3 30.1 31.7 24.1 0.25 0.75 1.25 1.75 0.25

The method of cross flow integration [18] by dividing the cross section of cooling tower into sixteen small unit volumes is simulated using MATLAB. The results of simulation and experimentation for variation of water temperature are as tabulated in Table III. The surface plot obtained from the result of a simulation is as shown in Fig. 4(a). It is seen from the surface plot that the water temperature decreases from the top to bottom uniformly along the height of a tower at a specified length of the tower. The rate of water cooling along the height of a tower is non-uniform at various lengths. It is seen that the rate of water cooling decreases with increase in length. It is clear from the surface plot obtained by simulation that the cooling rate of water is higher at region A and C compared to region B and D, as the fresh unsaturated air comes in contact with the hot water. The water at the lowest temperature is obtained in region C. The results obtained experimentally demonstrate a similar trend of water cooling as shown in Fig. 4(b). But it is seen that the rate of water cooling is higher in region A which is due to higher enthalpy and humidity potential. The rate of water cooling is very low in region C. The reason for the lower rate of water cooling in region C is the lower difference between the enthalpy of water and enthalpy of air. The rate of water cooling is also less in region B, as the air gets more humid while passing through region A and has less affinity to hold water by evaporation. The water cooling rate is higher in region D compared with region B, as the air entering in the region B is more humid which cannot cool water in

Experimental result 40.8 40.8 36.1 36.4 30.1 32.0 27.0 31.7 26.7 29.6 0.75 1.25

40.8 37.4 33.6 32.0 30.5 1.75

this region. Hence the water entering in region D from region B is having higher enthalpy. The air entering in region D passing through region C with low enthalpy constitute enthalpy potential which leads to higher heat and mass transfer resulting in higher rate of water cooling. The average temperature of water at the outlet obtained experimentally is less by 4.68% than average water temperature obtained by simulation due to the ambient air entering the cooling tower with lower humidity ratio. The wet bulb temperature of air increases along with the horizontal distance travelled by air in a cross flow cooling tower, as the every advancement of air meets hotter water. It is observed that the increase in air wet bulb temperature is not uniform along the height of a tower. As the temperature of water decreases along the height of a tower, the rise in air wet bulb temperature of the air is higher at the top of a tower as compared at the bottom of a tower. The results of simulation and experimentation for variation of air wet bulb temperature are as tabulated in Table IV. The surface plots based on results of simulation and experimentation are obtained as shown in Figs. 5(a) and (b) respectively. From the surface plot for simulated result is observed that the air wet bulb temperature increases along the length of tower uniformly and the highest air wet bulb temperature is obtained in region B. The rate of increase in air wet bulb temperature at the regions A and B is comparatively higher than at the region C and D as the water entering is at the highest temperature.

(a) Surface plot using MATLAB simulation.

(b) Surface plot using experimental data.

Figs. 4. Variation of water temperature across a cross flow cooling tower in oC



117


Height (m) 0.25 0.75 1.25 1.75 Length (m)

24.0 24.0 24.0 24.0 0

TABLE IV AIR WET BULB TEMPERATURE IN OC MATLAB Simulation result 28.8 31.8 33.9 35.4 24.0 27.8 30.6 32.5 34.1 24.0 27.0 29.5 31.4 32.9 24.0 26.4 28.5 30.4 31.8 24.0 0.5 1 1.5 2 0

The air entering is at the lowest wet bulb temperature and the lowest humidity which has a potential for evaporation of water and contributes to a higher rate of increase in air wet bulb temperature. The rate of increase in air wet bulb temperature varies with the height of a tower. The experimental results obtained also show the similar trend for variation of air wet bulb temperature. The rate of increase in air wet bulb temperature in region C is the lowest as the water entering from region A is at the lowest temperature. The rate of increase in air wet bulb temperature increases while the air passing through region D as the water entering from region B is at a higher temperature. The experimental results show that the increase in air wet bulb temperature is lower than the results predicted by simulation. The average air wet bulb temperature at the outlet is less by 7% than predicted which is due to less rate of evaporation of water than predicted. The air entering in a cooling tower at lower humidity ratio has an affinity to evaporate water. The evaporation

Experimental result 29.1 32.7 34.0 25.7 27.9 30.0 24.6 26.0 26.9 24.5 25.0 26.2 0.5 1 1.5

35.0 33.2 28.1 27.7 2

of water takes place by absorbing latent heat of vaporization from itself thereby cooling water. The enthalpy difference and humidity potential cause evaporation of water and contribute to increasing in air humidity as air passes along the length of a tower. The rate of increase in air humidity varies with the height of a tower. The results of simulation and experimentation for humidity variation are as tabulated in Table V. The surface plots based on results of simulation and experimentation are obtained as shown in Figs. 6(a) and (b) respectively. From the results of a simulation, it is seen that the increase in air humidity is higher in region A and B compared to region C and D. The highest air humidity is attained by the air in region B as the air comes in contact with water at the highest temperature. The results of simulation show that the rate of increase in air humidity at given height of a tower is uniform. The experimental results follow the similar trend as predicted by simulation.


(b) Surface plot using experimental data.

Figs. 5. Variation of WBT of air across a cross flow cooling tower in oC TABLE V HUMIDITY OF AIR IN kgw/kga Height (m) 0.25 0.75 1.25 1.75 Length (m)

MATLAB Simulation result 0.016 0.016 0.016 0.016 0

0.023657 0.020903 0.019128 0.017961 0.5

0.029665 0.025314 0.022320 0.020192 1

Experimental result

0.034380 0.029216 0.025424 0.022674 1.5


0.038079 0.032626 0.028435 0.025237 2

0.016 0.016 0.016 0.016 0

0.024 0.020 0.019 0.019 0.5

0.031 0.023 0.021 0.020 1

0.034 0.026 0.022 0.021 1.5

0.036 0.032 0.024 0.023 2


118



(b) Surface plot using experimental data

Figs. 6. Variation of humidity ratio of air across a cross flow cooling tower in kgw/kga

The rate of increase in humidity in region A is higher and in region B is lower than predicted by simulation as the air in region A gets early saturated. The rate of increase in air humidity in region C is lower than region D as the water entering in region C is at lower temperature than the water entering in region D. It is concluded from the results of experimentation that the rate of increase in air humidity in region A and region D is comparatively higher and depends on the difference of enthalpy and humidity of air. The average increase in humidity obtained experimentally is less than the average increase in humidity predicted by simulation.

V.

increase in humidity ratio of air is less than predicted. The highest air wet bulb temperature is obtained in region B and the lowest air wet bulb temperature is attained by air in the region C, as the water approaching this region is at the lowest temperature. The humidity of air increases along with the length of a tower. The rate of increase in humidity of air at region A is higher due to higher enthalpy difference and decreases at region B, as the air gets saturated while passing through region A than predicted by simulation. The rate of increase in humidity ratio in region C is less, as the water entering in this region is at a lower temperature. Hence the lower enthalpy difference suppresses evaporation. The rate of evaporation increases when air passes through the region D, as the water entering in this region is at higher temperature causes higher enthalpy difference. Hence the rate of increase in humidity also increases. The air at the highest humidity is obtained in region B and at the lowest humidity at region C.

Conclusion

The MATLAB simulation and experimentation are carried out for a cross flow wet cooling tower to obtain the variation of fluid properties within the core of cooling tower. The results of MATLAB simulation and experimentation are studied and compared. It is observed from the study that the water temperature decreases from top to bottom of a tower, the rate of decrease in water temperature is higher on air inlet side due to larger enthalpy difference and humidity difference. The cooling rate of water decreases on air exit side. The lowest water temperature is obtained at bottom of tower on air inlet side at region C. The average water temperature obtained experimentally at the outlet of cooling a tower is less by 4.68% than predicted average water temperature which is due to lower humidity ratio of air at the inlet. The air with lower humidity ratio causes the higher rate of water evaporation by absorbing latent heat of vaporization, thereby cooling water to lower temperature. The air wet bulb temperature increases along with the length of a tower. The rate of increase in air wet bulb temperature at the top of a tower is higher and decreases at a lower height of the tower. The average increase in air wet bulb temperature is less than the predicted by results of a simulation. The difference in predicted and actual air wet bulb temperature is due to the reason that the

References [1]

[2]

[3]

[4]

[5] [6]


Asvapoositkul W. and Treeutok W., A simplified method on thermal performance capacity evaluation of counter flow cooling tower, J. Applied Thermal Engineering, Vol. 38, pp.160-167, 2012. doi: http://dx.doi.org/10.1016/j.applthermaleng.2012.01.025 Bahadori A., Zahedi G., Zendehboudi S., Hooman K., Simple predictive tool to estimate relative humidity using wet bulb depression and dry bulb temperature, J. Applied Thermal Engineering Vol. 50, pp.511-515, 2013. doi:http://dx.doi.org/10.1016/j.applthermaleng.2012.07.033 Baker D. R., Howard A., Shryoc, A Comprehensive Approach to the Analysis of Cooling Tower Performance, J. Heat Transfer, Paper No.60-WA-85, 1960. Bozorgan N., Exergy Analysis of Counter Flow Wet Cooling Tower in Khuzestan Steel Co., J. of Mechanical Research and Application, Vol. 2, pp. 31-37, 2010. Cengel Y.A., Wood B., Dincer I., Is bigger thermodynamically better? Exergy, an Int. J., Vol. 2, pp. 62-68, 2002. Fisenko S.P. and Brin A.A., Simulation of a cross-flow cooling tower performance, Int. J. of Heat and Mass Transfer, Vol.50, pp. 3216-3223, 2007.


119

N. A. Rawabawale Rawabawale,, S. N. Sapali

[7]

[8]

[9]

[10]

[11]

[12]

[13]

[14]

[15]

[16]

[17]

[18]

[19]

[20]

[21]

[22]

[23]

doi:http://dx.doi.org/10.1016/j.ijheatmasstransfer.2006.05.028 doi:http://dx.doi.org/10.1016/j.ijheatmasstransfer.2006.05.028 Kairouani L., Md. Hassairi, Zermani T., Performance of cooling tower in south of Tunisia, J. Building and Environment, Environment, Vol.39, pp. 351 351-355, 355, 2004. doi:http://dx.doi.org/10.1016/j.buildenv.2003.08.017 Lemouari M. and Bo Boumaza umaza M., Experimental investigation of the performance characteristics of a counter flow wet cooling tower, Int. J. of Thermal Sciences, Vol. 49, pp. 2049 2049-2056, 2056, 2010. doi:http://dx.doi.org/10.1016/j.ijthermalsci.2010.05.012 Manuel Lucas, Javier Ruiz, Pedr Pedro o J. Martínez, Antonio S. Kaiser, Antonio Viedma, Blas Zamora, Experimental study on the performance of a mechanical cooling tower fitted with different types of water distribution systems and drift eliminators eliminators,, J. Applied Thermal Engineering, Vol.50, pp. 282 282-292, 292, 2013. doi:http://dx.doi.org/10.1016/j.applthermaleng.2012.06.030 Montri P., Simulation of a Counter Counter--Flow Flow and Cross Cross--Flow Flow Cooling Tower by the Stepwise Integration Method, Kasetsart J. (Nat. Sci.), Vol. 44, pp. 972 972-981, 981, 2010. Muangnoi T., Asvapoositkul W., Wongwises S., An exergy analysis on the performance of a counter flow wet cooling tower, J. Applied Thermal Engineering, Vol. 27, pp. 910 910-917, 917, 2007. doi:http://dx.doi.org/10.1016/j.applthermaleng.2006.08.012 Qureshi B. A., and Syed M. Z., Application of exergy analysis to various psychometric processes, Int. J. Energy Res. Vol. 27, pp. 1079 1079-1094, 1094, 2003. doi:http://dx.doi.org/10.1002/er.933 Santos J.C., Barros G.D.T., Gurgel J.M., Marcondes F., Energy and exergy analysis applied to the evapo evaporative rative cooling process in air washers, Int. J. of refrigeration, Vol. 1, pp. 11-8, 8, 2013. doi:http://dx.doi.org/10.1016/j.ijrefrig.2012.12.012 Santos J.C., Medeiros J.M., dos Santos J.C., Gurgel J.M., Marcondes F., Analytical solution for the simultaneous he heat at and mass transfer problem in air washers, Int. J. of refrigeration, Vol. 34, pp. 353 353-361, 361, 2011. doi:http://dx.doi.org/10.1016/j.ijrefrig.2010.06.016 Saravanan M., Saravanan R., Renganarayanan S., Energy and Exergy Analysis of Counter Flow Wet Cooling To Towers, wers, J. Thermal Science, Vol. 12, pp. 69-78, 69 78, 2008. doi:http://dx.doi.org/10.2298/TSCI0802069S Wang Li and Nianping Li, Exergy transfer and parametric study of counter flow wet cooling towers, J. Applied Thermal Engineering, Vol.31, pp. 954 954-960, 960, 2011. doi: http://dx.doi.org/10.1016/j.applthermaleng.2010.11.019 doi:http://dx.doi.org/10.1016/j.applthermaleng.2010.11.019 Jing Jing Jiang, Xiao Xiao-Hua Hua Liu, Yi Jiang, Experimental and Jing-Jing numerical analysis of a cross cross-flow flow closed wet cooling tower, Applied Thermal Engineering, Vol. 61, pp. 678 678--689, 689, 2013. doi:http://dx.doi.org/10.1016/j.applthermaleng.2013.08.043 doi:http://dx.doi.org/10.1016/j.applthermaleng.2013.08.043 ASHRAE. Handbook Fundamentals, 2000. American Society of Heating, Refrigeration and Air Conditioning Engineers Engineers,, SI Edition (Chapter 36). Thirapong Muangnoi, Wanchai Asvapoositkul, Somachi Wongwises, Effects of inlet relati relative ve humidity and inlet temperature on the performance of counter counter-flow flow wet cooling tower based on exergy analysis, Energy Conservation and Management, Vol. 49, pp. 2795-2800, 2795 2800, 2008. doi:http://dx.doi.org/10.1016/j.enconman.2008.03.019 Ebrahim Hajidavalloo, Re Reza za Shakeri, Mozaffar A. Mehrabian, Thermal performance of cross flow cooling towers in variable wet bulb temperature, Energy Conservation and Management, Vol.51, pp. 1298 1298-1303, 1303, 2010. doi:http://dx.doi.org/10.1016/j.enconman.2010.01.005 Bilal A. Qureshi, Sy Syed ed M. Zubair, Second Second-law law--based based performance evaluation of cooling towers and evaporative heat exchangers, Int. J. of Thermal Sciences, Vol. 46, pp. 188 188-198, 198, 2007. doi:http://dx.doi.org/10.1016/j.ijthermalsci.2006.04.014 J.M. Wu, X. Hung, H. Zhang, Theoreti Theoretical cal analysis on heat and mass transfer in a direct evaporative cooler, Applied Thermal Engineering, Vol. 29, pp. 980 980-984, 984, 2009. doi:http://dx.doi.org/10.1016/j.applthermaleng.2008.05.016 Abdi, G., Benabdallah, T., New Extended Simulation Method at Out Design Operating Conditions for Cooling Towers, (2015) Out-Design International Review of Mechanical Engineering (IREME) (IREME),, 9 (1), pp. 75 75-80.

doi:https://doi.org/10.15866/ireme.v9i1.4672 doi:https://doi.org/10.15866/ireme.v9i1.4672 [24] Aloui, F., Kourta, A., Ben Nasrallah, S., Experimental Study of Synthetic Jets with Cross Flow in Boundary Layer, (2016) International Review of Aerospace Engineering (IREASE) (IREASE),, 9 (1), pp. 13 13-21. 21. doi:https://doi.org/10.15866/irease.v9i1.9130 doi:https://doi.org/10.15866/irease.v9i1.9130 [25] Abdulwahid, A., Lazim, T., Saat, A., Jaafar, M., Kareem, Z., Experimental Thermal Field Measurements of Film Cooling with Twisted Holes, (2015) International Review of Aerospace Engineering (IREASE), (IREASE), 8 (3), pp. 86 86--94. doi:https://doi.org/10.15866/irease.v8i3.6124 doi:https://doi.org/10.15866/irease.v8i3.6124 [26] Abdulwahid, A., Lazim, T., S Saat, aat, A., Kareem, Z., Investigation of Effect Holes Twisted Angle and Compound Angle of Holes on Film Cooling Effectiveness, (2015) International Review of Automatic Control (IREACO) (IREACO),, 8 (3), pp. 244 244--250. 250. doi:https://doi.org/10.15866/ doi:https://doi.org/10.15866/ https://doi.org/10.15866/ireaco.v8i3.6237 ireaco.v8i3.6237 [27] Abdulwa Abdulwahid, hid, A., Lazim, T., Saat, A., Kareem, Z., Thermal Investigations on the Mixing Flow of Film Cooling by Twisted Holes, (2015) International Journal on Energy Conversion (IRECON) (IRECON),, 3 (3), pp. 88 88--94. [28] Thiao, S., Mar, A., Mbow, C., Youm, I., Solar Cooling System: Theoretical Study of Coefficient of Performance (COP) of a Solar Chiller Adsorption in the Site of CERER, (2014) International Journal on Energy Conversion (IRECON) (IRECON),, 2 (5), pp. 147 147-150. 150. [29] Bayrak, G., Cebeci, M., Uslu, A., Karakaya, G., Ornekci, N., A Smart S Solar olar Energy Energy--Based Based Cooling System Design and Application for Sustainable Trout Farming in Keban Dam Lake, (2015) International Journal on Energy Conversion (IRECON) (IRECON),, 3 (4), pp. 120 120--126. 126. [30] Srividhya, P., Muraleedharan, C., Jayaraj, S., Fuel Characteristics and Gasification of Woody Biomasses in Down Draft Gasifiers, (2013) International Journal on Energy Conversion (IRECON) (IRECON),, 1 (6), pp. 288 288--294. 294.

Authors’ information N. A. Rawabawale Born at Udgir Maharashtra (India) on 10th July 1975. Graduated in Mechanical Engineering from R Rural ural Engineering College Bhalki affiliated to Gulbarga University, Gulbarga Karnataka State( India) in the year 1997 and completed M M-Tech Tech in Heat Power Engineering from College of Engineering Pune Maharashtra State (India) in the year 2005. His area of research is thermal engineering, refrigeration refrigeration, and air conditioning. He has published three papers in International Conferences and three papers in national conferences. Mr. Nandkumar is a life member of ISTE. S. N. Sapali Graduated in Mechanical Engineering in the year 1985 and completed ME in the the year 1992 in Mechanical Engineering from College of Engineering Karad Shivaji University Kolhapur, Maharashtra State (India). Awarded Ph.D. Ph . from IIT Khar Kharaagpur gpur in the year 2002. His area of research is cryogenics, refrigeration refrigeration, and air conditioning. He has published twenty-one twenty one papers in international journals, five papers in national journals, twenty five papers in international conferences, eighteen papers in national conferences and authored eight books. He has guided eight Ph Ph.D D. thesis and has published nine patents. He has received the award of “Best Best Teacher Teacher”” from Govt. of Maharashtra. Dr. Sapali is a life member of ISTE, Member of ISHRAE Pune chapter and Fellow member of Institution of Engineers (India). He has worked as BOS member of Savitribai Phule Pune University, Pune and Shivaji University, Kolhapur and as executive committee member of ISHRAE Pune chapter.



120


Attachment Probability of Particle on Bubble Surface and the Stability of Its Aggregate Warjito, Harinaldi, Manus Setyantono, Sahala D. Siregar Abstract – Three processes are involved in the bubble-particle interaction during a flotation process. These processes can be analyzed in terms of collision, attachment, and aggregate stability. Each process plays an important role in determining flotation efficiency. Particle size and geometry also affect the particle-bubble interaction. Information about the irregular shape of particles is important for achieving a full understanding of the flotation process. This research focuses on understanding the effect of particle geometry on the particle-bubble interaction during a flotation process. The experimental setup was made up of a 9 × 9 × 26 cm glass flotation column, a bubble generator system, a particle feeding system, and a high speed video camera. The bubble generator system was made of a 0.3 mm diameter nozzle attached to a programmable syringe pump. The particle feeding system was made of pipettes. This research used real particles collected from an open pit in the Grasberg mine in Timika, Papua. The particle size variation ranges from 38 to 300 µm. Collector type reagent was added to the flotation column. Bubbleparticle interaction is recorded using a high-speed video camera. The data recorded by the highspeed video camera were analyzed using image processing software. The results of the experiment show that particle geometry is dominated by sub-angular geometry. Attachment, aggregate stability, and interaction time are found to depend on particle size. Small particles ranging from 38 and 106 μm have a long interaction time and are able to adhere to bubbles easily. Big particles ranging from 150 to 300 μm have a short interaction time and are unable to adhere to bubbles easily. All particles from 38 to 300 μm size generated the same value of aggregate stability. The aggregate stability of all particles is 1 or stable. Copyright © 2017 Praise Worthy Prize S.r.l. All rights reserved.

Keywords: Aggregate Stability, Attachment, Bubble, Interaction, Particle Size, Geometry

I.

There are three aspects which determine the efficiency of bubble-particle flotation. Those aspects are bubbleparticle collision probability, bubble-particle attachment probability, and bubble-particle aggregate stability. Several research projects on particle-bubble interaction have been carried out and have generated various interesting findings with regard to this interaction [1], [2], [3], [4]. Those previous studies utilized spherical glass particles. However, the geometry of real mineral particles is actually very different from the spherical geometry. Consequently, their practical contribution to flotation research is limited. Based on this consideration, we believe that a new experiment using actual geometry particles is necessary and will generate a deeper understanding of the various aspects involved in particlebubble interaction. The aim of this research is to understand the interaction between bubbles and nonspherical geometry particles in various sizes, especially in terms of particle attachment efficiency. The particles used in this study were extracted from the ore with nonspherical geometry and size range, which were collected from the Grasberg mine. Attachment efficiency (probability) is defined as the ratio of real to ideal attachment rates. Attachment efficiency is influenced by

Introduction

Indonesia is well known globally because of its natural resources. One type of natural resources that can be found in Indonesia is mineral. The third largest open pit for mineral exploration in the world can be found in Grasberg, Timika, and Papua. The ore mineral collected from the site contains valuable materials and nonvaluable materials. Valuable minerals are obtained through the processes of grinding and separation. One important particle separation process is flotation. Flotation is the process of separating particles using air bubble in a fluid medium. This process separates hydrophilic and hydrophobic particles, in which the hydrophobic particles will stick to the surface of bubbles. In several cases, chemical reagent is added to modify the surface properties of the valuable materials to make them hydrophobic. Subsequently, when the bubbles and the particles interact during the flotation process, particles with hydrophobic surface will adhere to the air bubbles. Valuable particles which have adhered to those bubbles will rise to the surface. Non-valuable materials will not adhere to bubbles and precipitate at the bottom of the flotation column.



121

Warjito, Harinaldi, Manus Setyantono, Sahala D. Siregar

several variables such as particle and bubble size, particle geometry, and reagent type. Bubble-particle aggregate stability depends on forces balance which is at work in the bubble-particle interaction [5]. The impacts of those forces on bubble-particle stability can be observed when those particles reach the bottom part of the bubbles. The probability of aggregate stability can be estimated by applying the following equation: =1−

1−

1

(1) Fig. 1. Experimental Setup

=

(2)

SEM technique is used to magnify particles so that they can be seen in greater detail. Particle picture magnification varies from 200 to 5000 times of the original size. This technique is used to facilitate the measurement of particle geometry. This technique can also measure particle density. The SEM results of some particles can be seen in Figures 2 below. Particle geometry is defined using Wadell’s equation. The results of our particle geometry measurement are presented in Table I below. This table clearly shows that the particle geometry is dominated by sub-angular geometry consisting of 45 μm to 300 μm particles and angular geometry consisting of 38 μm particles.

where is the Bond number and is the sum of forces that cause particles to detach from bubbles. These forces consist of the weight force ( ), the drag force ( ), and the force generated from pressure due to capillary force ( ). is the sum of forces that keep particles adhering to bubbles. These forces consist of the capillary liquid force ( ) and the hydrostatic force ( ) [6].

II.

Methodology

The experimental setup consists of six main parts, which are a flotation column, a bubble generator, a particle feeding system, a lighting system, an image capturing device, and image processing software. This setup is similar to the one used by Warjito, Harinaldi, and Setyantono in their study [7]. A flotation column is made in the form of a 9×9×26cm glass container. A bubble generator consists of a single nozzle with a diameter of 0.3 mm connected with a programmable syringe pump. As an image capturing device, we used a high-speed video camera with a macro lens and an additional back lighting system. The high-speed video camera recording speed was set at the rate of 500 frames per second. The recorded data was then processed and analyzed using the image processing software called ImageJ. The particle size varies from 38 μm to 300 μm. The programmable syringe pump was set at 15 mL/h in order to get small steady bubbles on the tip of the nozzle. The complete experimental setup in presented in Figure 1. This experiment used particle size variations as its main variable. The particle sizes under investigation are 38, 45, 75, 106, 150, 212, and 300 μm with as many as 30 particles belonging to each size.

(a)

(b)

(c)

(d)

Figs. 2. SEM Image of Particles at 200-time magnification in Various Sizes: (a) 38 μm, (b) 45 μm, (c) 75 μm, (d) 106 μm TABLE I MEASUREMENT OF PARTICLE GEOMETRY USING WADELL’S EQUATION Zingg Roundness Particle Sphericity Roundness Form Degree Size (μm) Index 38 0.6621 0.1245 Oblate Angular 45 0.6695 0.2663 Oblate Sub-angular 75 0.5642 0.2523 Oblate Sub-angular 106 0.5750 0.2503 Bladed Sub-angular 150 0.6232 0.1645 Bladed Sub-angular 212 0.5492 0.2553 Bladed Sub-angular 300 0.6230 0.2058 Prolate Sub-angular

III. Results and Discussion III.1. Particle Geometry and Size Particle size and geometry analysis was performed by using SEM technique. The results of the application of SEM technique have been reported in several previous studies [8], [9]. However, for the purpose of our analysis, those results are explained again in this section.

III.2. Bubble-Particle Interaction Previous research on bubble-particle interaction has also been carried out and reported. For the purpose of this study, an analysis of bubble-particle interaction



122


needs to be included [8]. Bubble-particle interaction on bubble surface may trigger several interesting phenomena such as particle sliding, rolling, attaching, bouncing, and detaching. These types of particle-bubble interaction can be observed in Figures 3.

need to be included here to provide a theoretical background on the subject matter [8], [9], [10]. There are three zones of angular speed which are created when particles are sliding on bubble surface. Those zones consist of the low speed of angular speed zone, the high speed of angular speed zone, and the low speed of angular speed zone. The first zone is characterized by low particle angular speed. This low speed takes place at the initial moment of a particle-bubble collision. When particles collide with bubbles, those particles slowly move from the upper side of the bubbles to the lower side of the bubbles. The transition from the first zone to the second zone cannot be determined in a precise way because of the variation in bubble-particle collision angles. The second zone is the one in which the angular speed of particles gradually increases as the particles move and pass the bubble surface equator. Lastly, the third zone is the one in which the angular speed of particles decreases. The transitional position from the second zone to the third zone varies from one particle size to another. Particles with the sizes of 45, 75, 106 μm have a transition position of 132o to 140o from the bubble center. Particles with the sizes of 150 and 200 μm have a transition position of 150o. Particles with the size of 300 μm will detach at 125o from the bubble center. These zones are presented in Figure 4(a). Figure 4(b) demonstrates the time and position of particles as they move from the point of collision with the bubble surface to the point at which the position of bubble-particle aggregates is established. This figure also indicates that different particle sizes have different interaction times. The collision position is relatively similar at an angle of 20 degrees, and this phase is followed by four types of particle-bubble interaction: rolling, sliding, sticking, and bouncing. However, it is also found that big particles have short interaction time, while small particles have long interaction time. As can be clearly observed in Figure 4(b), the interaction time is approximately 0.04 seconds for large particles with the size of 300 μm and 0.15 seconds for small particles with the size of 38 μm. This figure also shows the progression of three different zones, as evident from the particle velocity gradient curve. Various aspects of a particle-bubble interaction which consist of collision, sticking, stable aggregate, and interaction time have also been pointed out and investigated by A. V. Nguyen and G. M. Evans [11] A numerical study performed by Gregory Lecrivain, et al. demonstrates the zones of particle movement velocity on bubble surface [12]. This current research has generated the same results with those generated by Lecrivain’s. The results of Lecrivain’s study are presented in Figure 5. Long interaction time gives particles a higher chance of adhering to bubbles. This phenomenon is largely caused by the weight force at work in the bubble-particle interaction. Figure 6 shows that weight force varies according to particle size. Particles with bigger sizes

(a)

(b)

(c) Figs. 3. Bubble-particle Interaction with the Particle Sizes of (a) 38 μm, (b) 75μm, and (c) 300 μm

Small particles with the size of 38 μm produce the sliding interaction. At the initial stage of a particlebubble collision, the particles start to slide along the bubble surface. 38 μm particles also have small gravity force. Consequently, sliding interaction can happen. Particles with the size of 45–105 μm produce two types of interaction consecutively: rolling and sliding. This phenomenon can occur because of sub-angular particle geometry. These rolling interactions deserve further study since the resulting image data are not very smooth and sharp. Similar to what happens to small size particles, these particles stop and adhere to bubble surface at quadrant 2. The type of bubble-particle interaction which is produced by relatively big particles with the size of 300 μm is the bouncing interaction. This phenomenon can occur due to the big gravity force at work in these particles. Because of this bouncing interaction, big particles cannot adhere to bubbles. III.3. Particle-Bubble Attachment Angle Previous analyses of bubble-particle attachment angle Copyright © 2017 Praise Worthy Prize S.r.l. - All rights reserved


123


have bigger weight force. This big force shortens particle interaction time and causes particles to detach from bubbles. In order to adhere to bubbles, particles need a certain amount of induction time during which those particles seek to rupture the bubble surface and thus reach their aggregate stability.

The maximum value of weight force is achieved when the particles reach the bubble equator. Therefore, it can be surmised that a maximum value of weight force also indicates a maximum value of particle angular speed. After those particles passed the bubble equator, the weight force and angular speed of the particles gradually decrease. Weight Force 5e-7

4e-7

38 micrometer 45 micrometer 75 micrometer 106 micrometer 150 micrometer 212 micrometer 300 micrometer

Force (N)

3e-7

2e-7

1e-7

0

0

20

40

60

80

100

120

140

160

180

200

Angle (o)

Fig. 6. Weight Force at Work When Particles Slide on Bubble Surface

(a) Attachment Angle Vs Interaction Time 180

III.4. Bubble-Particle Attachment and Aggregate Stability

160 140 38 micrometer 45 micrometer 75 micrometer 106 micrometer 150 micrometer 212 micrometer 300 micrometer

o

Angle ( )

120 100 80

Particle-bubble attachment probability is one of many important factors which determine flotation efficiency. When particles interact with bubbles, those particles will break the bubble surface to create a tiny hole called nucleus. Our experimental results demonstrate that small particles are able to adhere to bubbles easily, but big particles are unable to do that and may easily detach from bubble surface. Figure 7 presents the bubble-particle attachment probability for all particle sizes. Particlebubble attachment probability was measured by performing 30 attempts of bubble-particle interaction. The experimental results as presented in Figure 7 demonstrate that small particles with the size of 38 to 106 µm have high attachment probability at approximately 60% to 90%. Meanwhile, big particles with the size of 150 to 300 µm have small attachment probability at approximately 10% to 40%. This phenomenon is also caused by the weight force at work in the bubble-particle interaction. On the one hand, big particles have big weight force, and big weight force causes short interaction time. Short interaction time decreases the abilities of particles to adhere to bubbles and to reach aggregate stability. On the other hand, small particles have small weight force. Because of this small weight force, those particles take a longer time to interact with bubbles. This long interaction time in turn makes it possible for particles to adhere to bubbles and helps those particles and bubbles to achieve stability. Bubble-particle aggregate stability can only be achieved if particle-bubble collision and attachment processes have previously taken place. Aggregate stability depends on forces balance at work in the particle-bubble aggregate.

60 40 20 0 0,00

0,02

0,04

0,06

0,08

0,10

0,12

0,14

0,16

Time (second)

(b) Figs. 4. (a) Angular Speed Zones of Particles on Bubble Surface and (b) Attachment Angles of Bubbles and Particles in Various Sizes

Fig. 5. Velocity Zones of Particles on Bubble Surface [12]

Figure 6 shows that the weight force of the particles gradually increases as they move from its initial position to the bubble equator, that is, the 90o position.



124


consist of the attachment forces and the detachment forces. Types of attachment forces which enable particles to adhere to bubbles are the hydrostatic pressure force and the capillary force. Meanwhile, the weight force and the bubble capillary force are types of detachment forces that cause particles to detach from bubbles. Particle size plays a significant role in the determination of particle-bubble attachment values. High particle-bubble attachment probability can be reached by small particles, while low particle-bubble attachment probability can be reached by big particles. Table II presents the results of the calculation of various forces in order to determine the values of particle-bubble aggregate stability. Our experimental results show that the particle-bubble stability for all particle sizes is 1 or stable. These results also suggest that all particles with the size of 38 to 300 μm can reach aggregate stability. However, such aggregate stability will only be possible if those particles have collided and stuck to the bubble in the first place. Aggregate stability is not the only factor which enables particles to adhere to bubbles. Other important factors which include interaction time, collision probability, and attachment probability must also be taken into account in order to achieve maximum flotation efficiency.

Particle-Bubble Attachment Probability 100

Percentage (%)

80

60

40

20

0

0

50

100

150

200

250

300

350

Particle Size (micrometer)

Fig. 7. Particle-bubble Attachment Probability

Aggregate stability is expressed by using a nondimensional number called the Bond number. The Bond number (Eq. (2)) with a value lower than 1 indicates weak aggregate stability. The Bond number with a value below 0 indicates that no aggregate stability is achieved or that the particles have detached from the bubbles. The Bond number with a value equal to 1 indicates that the particles and bubbles have reached their aggregate stability. Forces balance at work in the aggregate stability is classified into two types, which

Particle Size (μm)

Weight Force (N)

38 45 75 106 150 212 300

1.06 E-10 1.90E-10 5.71E-10 1.73E-09 6.07E-09 9.93E-09 2.93E-08

IV.

TABLE II RESULTS OF THE CALCULATION OF BUBBLE-PARTICLE STABILITY Force generated by pressure due Hydrostatic Capillary Force Bond Number to capillary force Pressure Force (N) (N) (N) 4.04E-09 3.52E-12 1.07E-06 0.0038 5.96E-09 6.15E-12 1.30E-06 0.0047 1.24E-08 2.13E-11 1.88E-06 0.0069 2.60E-08 6.32E-11 2.72E-06 0.0102 6.00E-08 2.06E-10 4.13E-06 0.0160 8.32E-08 4.04E-10 4.86E-06 0.0191 1.71E-07 1.18E-09 6.97E-06 0.0287

Stability Probability 1 1 1 1 1 1 1

caused by a variation in weight force at work in the bubble-particle interaction. This force affects particlebubble attachment probability. Small particles have high attachment probability, while big particles have low attachment probability. All particles with the size of 38 μm to 300 μm generate the same value of stability probability.

Conclusion

The application of SEM technique on real particles collected from an open pit in the Grasberg mine demonstrates that the particle geometry is dominated by sub-angular and oblate forms. This sub-angular particle geometry does not affect the particle-bubble aggregate stability and attachment probability in any significant way. However, this sub-angular particle geometry affects particle interaction which takes place when those particles are sliding on bubble surface. Small particles with the size of 38 μm produce sliding interaction. Particles with the size of 45 μm to 105 μm produce rolling and sliding interactions consecutively. Big particles with the size of 300 μm produce bouncing interaction. There are three zones of angular speed which are created when particles are sliding on bubble surface. These three zones consist of low angular speed, high angular speed, and low angular speed. This movement is

References [1]

[2] [3]

[4]


Nguyen, A. V., Schulze, H. J., & Ralston, J. (1997). Elementary steps in particle-bubble attachment. International Journal of Mineral Processing, 51, 183–195. Nguyen, A. V., & Schulze, H. J. (2004). Colloidal science of flotation. Marcel Dekker: New York. Ralston, J., Dukhin, S. S., & Mischchuk, N. A. (2002). Wetting film stability and flotation kinetics. Advance Colloid Interface Science, 95, 145 –236. Ralston, J., Fornasiero, D., & Hayes, R. (1999). Bubble-particle attachment and detachment in flotation. International Journal of Mineral Processing, 56(1–4), 133–164.


125


Bloom, F., & Heindel, T. J. (1997) (1997). Mathematical modelling of the flotation deinking process process. Mathematical Computer Modelling, 25 Modelling, 25,, 13– –58. [6] Drzymala rzymala, rzymala, J. (1994). (1994) Characterization of materials by Hallimon tube flotation part 2: Maximum aximum size of floating particles and contact angle angle. International Journal of Mineral Processing, 42, 42 6234 6234– –6237. 6237. [7] Warjito, Warjito, Harinaldi, & Setyantono, M.. (2016) (2016).. Visualization of angular particle particle-bubble bubble surface interaction using a high speed video camera. camera. International Journal of Technol Technology, ogy, 6, 6 1045– 1045 1053 1053. [8] Warjito, Harinaldi, Setyantono, M., & Siregar, S. D. (2015). (2015) Particle bubble aggregate stability on static bubble generated by Particle-bubble single nozzle on flotation process process. Paper presented at the 3rd Regional Conference on Energy Engineering and 7th International Conference of Thermofluids. Thermofluids Yogyakarta, Indonesia. [9] Warjito, Harinaldi, Setyantono, M., & Siregar, S. D. (2015). (2015) Characteristics of small bubble generated by single nozzle on th flotation process process.. Paper presented at the 14 International Conference on QiR (Quality in Research). Lombok, Indonesia. [10] Warjito, W., Harinaldi, H., Setyantono, M., Angular Particle Bubble Attachment Mechanism in Flotation, (2016) International Review of Mechanical Engineering (IREME) (IREME),, 10 (2), pp. 99 99-106. 106. [11] Nguyen, A. V., & Evans, G. M. (2004) (2004). Movement of fine particles on an air bubble surface studied using high speed video microscopy microscopy. Journal of Colloid and Interface Scienc Science, 273, 273, 271– 271 277. [12] Lecrivain, G., Petrucci, G., Rudolph, M., Hampel, U., & Yamamoto, R. (2014) (2014). Attachment of solid elongated particles on the surface of stationary gas bubble bubble. International Journal of Multiphase Flow Flow, 71, 71, 83–93. 83 93.

actice member of Indonesia consortium of Mechanical Engineering Education and asia Fluid Machinery Commite.

[5]

Manus Setyantono, ST.MT, ST.MT, was born in Semarang, Indonesia at March 13th 1965. Degree: ST (1993) in Mechanical Engineering from Universitas Hasanuddin, Ujung Pandang (Makassar) Indonesia. MT (2006) from Mechanical Engineering, Universitas Indonesia, Jakarta, Indonesia. Major field ooff study is Fluid Mechanics. He has student at Mechanical Engineering Departement of Universitas Indonesia until now. Some of publications had written to Internationl Journal of fluid mechanics Research, International Journal of Technology, International Jo Journal urnal Review of Mechanical Engineering and some seminar proceedings. Sahala David, David, ST, MT, was born in Pematangsiantar, Indonesia at 27 January 1993. Degree: ST (2013) Bachelor in Mechanical Engineering from Universitas Indonesia Jakarta, Indonesia. MT (2 (2015) 015) Master in Mechanical Engineering from Universitas Indonesia, Jakarta Indonesia.Major field of study is fluid Jakarta-Indonesia.Major mechanics. After his graduation, he works at mining industry as mechanical engineer until now. Some of his publications had written to Interna International tional Journal of Fluid Mechanics and some seminar proceedings to International Journal of Fluid Mechanics and some seminar proceedings.

Authors’ information Ir. Warjito, Warjito, M.Eng., PhD was born in Cilacap, Indonesia at August 8th 1963. Degree: Ir (1988) in Mechanical Engineering from Universitas Indonesia, Jakarta Jakarta-Indonesia. Indonesia. M. Eng. (1998) from Mechanical Science Hokkaido University, Sapporo Sapporo-Japan. Japan. PhD (2001) from Mechanical Science Hokkaido University, Sapporo Japan. Major field of study Sapporo-Japan. is fluid dynamics. He has work at oil and gas production facilities fabricator after graduated from Universitas Indonesia for 3 years, after which he joined the laboratory of fluid mechanics Mechanical Engineering Universitas Indonesia until now. Some of the publication had written to the Journal of Experimental in Fluid, Applied Mechanic and Material, International Journal of Technology, International Journ Journal al of Fluid Mechanic Research, International Review of Mechanical Engineering and several seminar proceedings. Correspondding author. E-mail mail:: [email protected] Prof. Dr. Ir. Harinaldi Harinaldi,, M. Eng. was born in Jakarta, Indonesia at October 30th 1968. He received his Bachelor degree in Mechanical Engineering from Universitas Indonesia in 1992 and obtained his Master of Engineering degree from Keio University (Japan) in 1997. Furthermore, he received his Doctor of engineering degree also from Keio University in 2001. Currently, he serves as Professor in Department of Mechanical Engineering, Universitas Indonesia. Some of his recent publications include the topic on flow structure and mixing be behind hind a backward facing step with existence of gas injection in Int. Journal of Heat and Mass Trans, 2001; J. Of Chem Eng of Japan, 2001 Applied Mechanic and Material, International Journal of Technology, International Journal of Fluid Mechanic Research, In International ternational Review of Mechanical Engineering and some seminar proceedings. Major field of study is Fluid dynamics, Statistic and Probability for engineering, Engineering Mathematics. He has work in Universitas Indonesia, he joined the laborartory of fluid mechanics, Mechanical Engineering Department of Universitas Indonesia uuntil ntil now. Prof. Dr. Ir. Harinaldi, M. Eng. Is an



126


Heat Carrier Vortex Motion Influence on the Hydrodynamics and Heat Exchange in the Pipes with Transverse Collars and Flow Core Energizers D. Zhumadullayev, A. A. Volnenko, O. S. Balabekov, Zh. Serikuly, S. A. Kumisbekov, L. I. Ramatullayeva Abstract – Analysis of literature data shows that among constructive heat exchange enhancement methods, application of tubes with transversely annular turbulators is of great interest. Results of investigations of the tubes with transverse collars and flow core energizers showed that with increase in Reynolds numbers, values of relative heat exchange and resistance coefficients rise. For the heat exchange enhancement, transition interval with the highest values of the heat exchange and resistance coefficients is the promising. However, increase in the heat exchange coefficients leads the resistance coefficients. It was defined that maximal values of parameters under the investigation fall at the step of arrangement of annular apertures equal to 10. Equations, taking into account vortex interaction mechanisms beyond the collars and spherical thickenings, as well as relation between stream-lined barrier arrangement steps, are offered to describe results. Copyright © 2017 Praise Worthy Prize S.r.l. - All rights reserved.

Keywords: Tubes, Collars, Flow Core Energizers, Spherical Thickenings, Heat Exchange, Resistance Coefficient, Heat Exchange Coefficient

Spiral prominences, spiral wire inserts [1]-[3], spherical prominences and hollows [4]-[5], plane thin ribs along the forming surface [6], wavy surfaces of channels [7], tape inserts [8], craters of different geometrical form on the surface of channels, screw shaped tubes, surfaces with reticulate-wire ribbing [9], etc. are used for these purposes. Application of tubes with transversely annular turbulators is of great interest from the considered methods of induced laminarturbulent transition. Identified regularities during heat exchange process enhancement in such tubes were revealed in the paper [10]. Objective of the research is to study convective heat exchange and hydrodynamic parameters of a heatexchange apparatus with knurled tubes and flow core energizers, and estimate a heat carrier vortex motion influence on the studied parameters. The experimental arrangement, described in the paper [11] was created to study enhancement of the convective heat exchange and hydrodynamic parameters of the heatexchange apparatus with knurled tubes and flow core energizers. Methods and conditions for performance of the experiments are also described here.

Nomenclature D d G h Re Reкр Sl T, t t tяд    в ш кр с ‘ “

Pipe diameter, m Diameter of aperture of diaphragms, m Volume flow, m3/s Height of diaphragms, m Reynolds number Critical Reynolds number Strouhal number Temperature, оС Distance between diaphragms and grooves, m Step between spherical thickenings Heat-transfer coefficient, W/(m2оС) Coefficient of hydraulic resistance Step between ring overhangs Water With a spherical thickening Critical Wall Parameter on an input Parameter on an output

I.

Introduction

It is known that constructive heat exchange enhancement methods are used to reduce boundary-layer thickness by local cracks, streamlined surface or to increase intensity of turbulent exchanges in the immediate neighborhood of heat-exchange surface.

II.

The Main Part

Methods for performance of the experiments contained heating of one heat carrier to desired temperature, switching of hot and cold heat carriers’ pumps, heat carriers’ rate control, systematic observation



127

D. Zhumadullayev et al.

practically does not cause any influence. Therefore, for the heat exchange enhancement, the promising is the field of exchange. Effect of increase in the heat exchange in 3,5 times were received exactly there [13]-[14], that significantly exceeds possible for the dropping liquids effects in the turbulent region.

Relative resistance coefficient, ξ/ξгл

of readings of all thermometers and thermal converters, heat carriers’ flow gages. The following values were measured in the course of the experiments: volume flow rate of the 1st (G1) and 2nd (G2) heat carrier, m3/s; temperature of the 1st heat carrier at the entry and output from the experimental stretch (t11’) and (t12”), oС; temperature of the 2nd heat carrier at the entry and output from the experimental stretch (t21’) and (t22”), oС; temperature field of the wall by the length of experimental tube and diameter in 3 sections; hydraulic resistance in the tube and annular channel. The research was carried out at a temperature value of the tube wall tc=29,1...50,8oС; water temperature at the entry tв’=11,4...17,1оC and output tв”=13,8...26,7оC. Water served as the heat carrier in the annular channel. The water temperature was changed at the entry from 11 to 16,5oС, at the output from 14,4 to 25,7oС; the wall temperature was changed from 32,3 to 50,3oС. The tubes with the annular channel knurling had 20×2,5 mm diameter. The relative height of the tube’s raised portion: d/D=0,96 and 0,875. The flow core energizer represented a wire string with spherical thickenings by diameter 0,6 and 2 mm. The interval between collars – 3 and 7 mm, that corresponded to the intervals between the spherical thickenings (4-5) tяд/dш. As a result of the pursuance of the research, dependencies of the hydraulic resistance coefficients and coefficients of heat exchange of the knurled tubes with the flow core energizers, compared with the results of the research of the knurled tubes without the flow core energizers, were received. As is seen from the graph (Figure 1), the resistance coefficient growth intensity at Re>Reкр in the knurled tubes and knurled tubes with the flow core energizers is higher than in the smooth [12], and increases with reduction of d/D. The maximum increase in the resistance coefficient in the field of exchange is observed for the tubes with apertures d/D=0,875 and consists ξ/ξгл=3 [12], at that the aperture interval consists th/h=10, and interval between spherical turbulators tяд/dш=4,7. As is seen from Figure 2, which presents dependencies of the enhancement effect /гл from Reynolds number, for the tubes with aperture s and tubes with knurls and turbulators with increase in the Reynolds number /гл increases. At that, increase in the enhancement effect is obvious with increase in the aperture height and presence of the flow core energizer (Figure 2). This case is indicative of purposefulness of the turbulization in the field of exchange of heavier wall layers of the flow and its core. At high Reynolds numbers of the dropping liquid flow, vice versa, it is reasonable to use turbulators of low height [13]. At low Reynolds numbers, the heat exchange in the plain tube and tubes with different aperture parameters is the same [12]. It is explained by the fact that at low rate the heat carrier, the free convection in horizontal tubes is large and additional induced laminar-turbulent transition

Reynolds number, Re 1 – d/D=0,875; 2 – d/D=0,96; 3 – d/D=0,875; dш/d=0,145; 4 – d/D=0,96; dш/d=0,04.

Relative heat exchange coefficient, α/αгл

Fig. 1. Dependence of ξ/ξгл from Reynolds number

Reynolds number, Re 1 – d/D=0,875; 2 – d/D=0,96; 3 – d/D=0,875; dш/d=0,145; 4 – d/D=0,96;dш/d=0,04. Fig. 2. Dependence of the enhancement effect from Reynolds number

Figure 3 presents data of influence of the relative interval on increase in the heat exchange coefficient for the annular aperture s of rectangular (experiments of R. Koch) and graded (the research [12] and our data) profiles in the air at equal fixed Reynolds number and fixed height of the aperture. As is seen from the Figure, the maximum increase in the heat exchange is reached at t/h10, with h scale. The results shown are reduced to the regularities of the formation of vortices [15]. Figure 4 presents dependencies of the hydraulic resistance relative coefficient from the interval t/h. As well as at the heat exchange research, increase in the resistance coefficient reaches its maximum at t/h10. Explanation to this – the vortex mechanism for flowaround of the barriers [16]. According to the established regulations for formation of vortexes, toroidal vortexes are formed in the tubes



128


with transverse collars [16]. In arrangement of the string with spherical thickenings along the tube axis, the toroidal vortexes are also formed beyond them [17].

Relative Nusselt number, Nu/Nuгл

vortex 2, and the vortex 1 breaks a vortex 3 and so on. The similar picture is also observed in the vortex interaction in the flow core. In arrangement of the spherical thickenings (Figure 5(a)) in the distance, characterizing time Т (Figure 5(b)) the vortex I breaks an unformed vortex II, and the vortex I breaks a vortex III and so on. It is seen from the considered vortex interaction schemes that such arrangement of the vortex formation sources is possible when coincidence of the vortex formation time and vortex passage time from one source to another is provided. Such arrangement provides the vortex interaction in-phase condition. In this condition, the vortexes, formed simultaneously beyond all sources, separating, fly to the following along the flow motion sources at the moment of completion after them the vortex formation cycle. The vortex capacity summing up occurs. For mathematical description of the considered vortex interaction schemes, as well as in the paper [17], we use concept of the vortex interaction degree : Then, for the case of the collars’ flow-around, where the key parameter is d:

Interval of aperture s, t/h - Koch data; o, - Kalinin data (Re=4×104);  - Yeshankulov data (Re=1,2×104) [12]; х – our data with the flow core energizer (Re=1,2×104)

Relative resistance coefficient, ξ/ξгл

Fig. 3. Influence of the relative interval on increase in the heat exchange coefficient for the annular aperture s and flow core energizers

   t  Sl    0,85  0,15Sin   4  1  2  d  m 

(1)

where Sl=0,164 – Strouchal number, value of m parameter is defined by the formula:

Interval of aperture s, t/h 1 – Koch data; 2 – Kalinin data (Re=4104); 3 – Yeshankulov data [12] (Re=1,2×104); 4 – our data (Re=1,2×104)

m=0,738[1-exp(-t/d)] Fig. 4. Dependence of the hydraulic resistance relative coefficient on the interval between apertures

(2)

Here d – diameter, corresponding to the raised portion diameter d (see Figure 5(а)). In the event that the roughness height h is taken as the defining size, and the vortex formation source interval is characterized by t/h, we propose for the calculations the equations:

Let’s consider the mechanism for formation and interaction of vortexes in the tubes with transverse collars and flow core energizers. In arrangement of the vortex formation sources in the tubes in the form of collars (Figure 5(a)) in the distance, characterizing time Т (Figure 5(b)) the vortex 1 breaks an unformed

   t  h  0,85  0 ,15Sin   4 h  1     2  h  mh mh=1,4[1-exp(-th/h)]

(3) (4)

For the case of the vortex formation and interaction beyond the spherical thickenings, located on the string along the tube axis we use the similar approach. Then for the vortex interaction degree we have the following expression:



129


   t  Sl   яд  0,85  0 ,15Sin   4 яд ш  1   2  d ш  mш  

arranging the spherical thickenings in such a way to align the thickenings’ axis of symmetry with the raised portion’s axis of symmetry, in-phase conditions for the collars and spherical thickenings should be reached simultaneously. In this case, the relation between intervals of the spherical thickenings and collars are defined by the formula:

(5)

where tяд – interval between the spherical thickenings; Slш=0,183 – Strouchal number. Value of mш parameter is defined by the formula: mш=0,874[1-exp(-tяд/dш)]

tяд/dш=0.47. th/h

(6)

Analysis of the heat exchange and hydraulic resistance coefficients research results is indicative of the fact that their maximum values fall at the collars’ interval th/h=10. Size of these intervals corresponds to the interval size of the spherical thickenings tяд/dш=4,7 [16], at which achievement of extreme values of the hydrodynamic and heat exchange characteristics is possible. Therefore,

(7)

Then, the total coefficient, considering the interaction mechanisms beyond the collars and flow core energizers can defined by the formula: θΣ = θh. θяд

(8)

(a)

(b)  Т

(c)  Т (а) the tube with the collars; (b) the vortex interaction scheme at Т; (c) the vortex interaction scheme at Т Figs. 5. The vortex interaction scheme in the tubes with the collars



130


heat exchange surfaces, Moscow: Energoatomizdat, 1998, 408 p. [14] О.S. Balabekov, А.А. Yeshankulov, N.S. Bekibayev, А.А. Volnenko, L.I. Ramatullayeva Enhancement of the heat exchange process in the field of transition regime in the tubes with annular apertures, Bulletin of RK NAS, n.5.pp. 26-31, 2009. [15] Zh. Serikuly, A. Volnenko, E. Ya. Kenig, Hydrodynamics of apparatuses with preformed packing bodies, 7th International Conference Interdisciplinary in Engineering (INTER-ENG 2013), Tirgu Mures, Romania, 2014 pp. 375-381. [16] Serikuly, Z., Volnenko, A., Arginbaevich, K., Mass transfer in the apparatuses with preformed packing bodies, (2014) International Review of Mechanical Engineering (IREME), 8 (4), pp. 779-784. [17] Serikuly, Z., Kaldikulova, A., Ospanov, B., Kumisbekov, S., Industrial Testing Method Hydrodynamic Modeling Apparatus with a Regular Movable Packing, (2015) International Review of Mechanical Engineering (IREME), 9 (4), pp. 336-340.

III. Conclusion Thus, the research results showed that with increase in the Reynolds number, values of the relative heat exchange and resistance coefficients rise. For the heat exchange intensification, transition interval with the highest values of the heat exchange and resistance coefficients is the promising. However, increase in the heat exchange coefficients leads the resistance coefficients. It was defined that maximal values of parameters under the investigation fall at the step of arrangement of annular apertures equal to 10. Equations, taking into account the vortex interaction mechanisms beyond the collars and spherical thickenings, as well as relation between the stream-lined barrier arrangement steps, are offered to describe the results.

Authors’ information Daulet Zhumadullayev 05.01.1990. PhD student, Faculty Mechanical and Petroleum Engineering, M. Auezov South Kazakhstan State University, 160012 Shymkent, Kazakhstan. Current and previous research interests Machines and Apparatus for Chemical Industry.

References [1]

[2]

[3] [4]

[5]

[6]

[7]

[8]

[9]

[10]

[11]

[12]

[13]

Yu.G. Nazmeyev, I.A. Konakhina, A heat exchange enhancement in the flow of viscous liquid in the tubes with screw knurl, Thermal energetics, n. 11, pp. 59-62, 1993. N.V. Zozulya, I.Ya Shkuratov, Influence of spiral inserts for the heat exchange in the motion of viscous liquid inside the tube, Thermal physics and thermotechnics, pp. 65-66, 1964. S.F. Bayev, Marine compact heat exchange apparatus. Leningrad: Sudostroyeniye, 1965. А. I. Leontyev, V.V. Olimpiyev, Ye.V. Dilevskaya, S.A. Issayev, Essence of a mechanism for the heat exchange enhancement on the surface with spherical hollows, Bulletin of RAS. Energetics, n. 2, pp. 117-135, 2002. I. A. Popov, Yu.F. Gortyshov, V.V. Olimpiyev, A.V. Schelchkov Thermal hydraulic efficiency of using the spheroidal hollows for the heat exchange enhancement in the channels, Materials of reports and messages of V Minsk international forum on the heat and mass exchange, Minsk 2004, pp. 28-36. E.N. Saburov, Yu.L. Leukhin, S.I. Ostashev, Heat exchange enhancement in the annular channels with swirling flow of the heat carrier, Proceedings of the Second Russian National Conference of the Heat Exchange, Moscow, 1998, V. 6, pp. 196198. М.К. Ovsyannikov, Ye.G. Orlova, N.Ye Sivtsov. Heat exchange and resistance in the channels of plate heat exchangers, Proceedings of the Second Russian National Conference of the Heat Exchange. Moscow, 1998, V. 6, pp. 170-174. L. Zhargalkhuu, А.F. Ryzhkov, V.Ye. Silin Optimization of gaswater heat exchanger at low Reynolds numbers, Proceedings of the Second All-Russia school-seminar of young scientists and specialists “Energy saving – theory and practice”, Moscow, MEI, 2004. pp. 315-318. Ye.N. Pismenny, V.А. Rogachev, N.V Bosaya. Investigation of thermal efficiency of surfaces with net-wire ribbing at the free convection, Proceeding of the Second Russian National Conference of the Heat Exchange, Moscow, 1998, V. 6, pp. 181183. E.К. Kalinin Law of the heat exchange change on the wall of channels with discrete flow turbulization at the forced convection / E.K. Kalinin [et all], Diploma for scientific discovery № 242 USSR, Moscow 1981. А.А. Volnenko, О.S. Balabekov, D.К. Zhumadullayev, А.А. Yeshankulov, Zh.Ye Khussanov. Investigation of heat exchange and hydrodynamics at the flow of heat carrier in the round tube with transverse collars and turbulators, Bulleting of RK NAS. Series of chemistry and technology, n.2, pp. 37-42, 2013. А.А Yeshankulov. Hydrodynamics and heat exchange at the flow of viscous heat carriers in the heat exchange apparatus with knurled tubes, Doctor of Technical Sciences dissertation, Shymkent, 2009. E.К. Kalinin, G.A. Dreitser, I.Z. Kopp, A.S. Myakochin Effective

Volnenko Alexander Anatolievich 21.12.1955. Doctor of Technical Sciences, Professor Mechanics and Engineering Scientific–Research Institutes, M. Auezov South Kazakhstan State University, 160012 Shymkent, Kazakhstan Current and previous research interests Machines and Apparatus for Chemical Industry. Orazaly Satimbekovich Balabekov 14.10.1941. Doctor of Technical Sciences, Professor, Academician of the National Academy of Sciences RK, Mechanics and Engineering Scientific–Research Institutes, South Kazakhstan State Pedagogical Institute, 160012 Shymkent, Kazakhstan. Current and previous research interests -Machines and Apparatus for Chemical Industry. Zhandos Serikuly 29.01.1985. PhD, Faculty Mechanical and Petroleum Engineering, M. Auezov South Kazakhstan State University, 160012 Shymkent, Kazakhstan. Current and previous research interests - Machines and Apparatus for Chemical Industry.

Kumisbekov Serik Arginbaevich 18.09.1951 Candidate of Technical Sciences, Docent Faculty Mechanical and Petroleum Engineering, M. Auezov South Kazakhstan State University, 160012 Shymkent, Kazakhstan. Current and previous research interests Machines and Apparatus for Chemical Industry. Ramatullayeva Lazzat Imamadinovna 12.03.1971. Candidate of Technical Sciences, Docent, Faculty Faculty of building and transport M. Auezov South Kazakhstan State University, 160012 Shymkent, Kazakhstan. Current and previous research interests - Life’s safety and environmental protection.



131


An Investigation on Suppression of Vortex Instabilities for Flow Past a Bluff Body Using a Passive Device Sunil A. S., Tide P. S.

Abstract – The present paper aims at investigating the effectiveness of using a passive device on the suppression of vortex instabilities during a fluid flow past a bluff body. A study of fluid flow parameters for the flow past a bluff body fitted with straight fins and helical strakes is presented. A three dimensional numerical investigation is performed on the bluff body for different designs of the fins and helical strakes. A circular cylinder of a diameter of 4cm is used as bluff body. The simulation is carried out for a water flow past a stationary cylinder with Reynolds number 28000. A numerical simulation was performed in a three dimensional computational domain using RANS equations and SST k-ω turbulence model. To reduce the frequency of the vortex shedding, the bluff body is attached with a passive device. The investigation observed that the attachments of straight fins and helical strakes on the surface are found to be highly effective in diminishing the vortex instabilities formed around the structure.The changes in the flow field are analyzed by observing the changes in flow parameters like Drag force and Strouhal number. The results were found to be in reasonable agreement with the available data in the literature. The use of helical strakes largely reduces Vortex induced vibrations (VIV) so that they can be used for the protection of underwater cables used in offshore structures. Copyright © 2017 Praise Worthy Prize S.r.l. - All rights reserved.

Keywords: Straight Fins, Strouhal Number, Vortex Instabilities, Drag Force

I.

The smallest cell was located along the edges of the square cylinder with the value of h/D = 0.01. Doolan [2] has conducted the DNS investigating the grid convergence for three different grid resolutions on DNS around a square cylinder and found that the solution converged when the smallest cell size along the square cylinder edge was h/D = 0.0167. T. Zhou et al. [3] experimentally investigated the vortex-induced vibration of a cylinder with helical strakes. A rigid circular cylinder of a diameter of 80 mm attached with three-strand helical strakes measuring 10d in pitch and 0.12d in height was tested in a wind tunnel. It was found that the helical strakes can reduce VIV by about 98%. Unlike the bare cylinder, which experiences lock-in over the reduced velocity in the range of 5 to 8.5m/s, the cylinder with helical strakes does not show any lock-in region. This paper investigates the reduction in vortex shedding by analyzing the flow variables like Strouhal number, drag force and lift force for a flow past a circular cylinder fitted with fins and helical strakes. The numerical investigation is conducted for different designs of fins and helical strakes on the structure. The effectiveness of using a passive device like fins and strakes to suppress vortex-induced vibrations is investigated in this paper.

Introduction

Whenever a fluid flows past a structure the phenomenon of Vortex Shedding occurs in the wake of the body. The cylindrical structure is a bluff body that can significantly disturb the flow around it. A low pressure zone is created in the wake of the structure inducing both in line and cross vibrations. These vortex induced vibrations due to vortex shedding are highly undesirable and fatally affect the structure .Vortex shedding behind bluff bodies is of concern for many structural engineering applications like bridges, tall chimneys and oil rigs. Therefore, it is important for structural engineers to find and analyze different methods to control and suppress these flow-induced problems. Thus the fluid flow around a cylinder, characterised by flow separation and turbulence transition behind, has been studied in details by many researchers and scientists. They applied some methods and devices for their investigation to control the flow and reported that vortex induced vibrations can be reduced with different methods. Different attachments like fins, shrouds, splitter plates and helical strakes are used in the passive control of vortex induced vibrations. Inoue et al. [1] conducted a numerical analysis using a square cylinder and created a non-uniform mesh but divided the computational domain into three regions, each with a different grid ratio.



132

A. S. Sunil, P. S. Tide

II.

discretization has been done in conformity with the second order implicit scheme. Temporal discretization involves the integration of every term in the differential equations over a time step ∆t. The pressure-velocity coupling scheme, (PISO) was adopted in the calculation of the flow field. A flow domain of length 40d and width 20d was used for numerical simulation. Fig. 1(a) and Fig. 1(b) show the two dimensional and three dimensional views of the domain respectively. Boundary conditions were applied and numerical simulation was conducted to ensure the domain and grid independencies. The results of the grid independence study showed a certain dependency on the grid resolution. From the study it is observed that there is a change in the value of lift force with an increase in the number of cells up to 341856 cells. Beyond this grid cell number, the lift force is approximately the same in all runs. Hence the grid of 341856 cells was chosen in this investigation.

Numerical Simulation Procedure II.1.

Flow Field Formulation

The governing equations on the flow field are the continuity and momentum equations (Navier–Stokes equations), which are represented as follows: Continuity Equations:

u v w   0 x y z

(1)

X Momentum equation:

u u u u u v w  t x y z   2u  2u  2u  1 p     2  2  2   x y z   x

(2)

Y Momentum equation:

v v v v u v w  t x y z   2v  2v  2v  1 p     2  2  2   y y z   x

(3) (a)

(b)

Figs. 1. Flow domain of fluid flow with boundary conditions

Z Momentum equation:

III. Results and Discussion

w w w w u v w  t x y z 

 2w 2w 2w  1 p    2  2  2   z y z   x

In the present study, numerical simulation and analysis of 3 different cases, one on a bare circular cylinder, and two cases on a circular cylinder fitted with straight fins and circular cylinder with helical strakes at two Reynolds numbers (100, 28000) are included.

(4)

Here ρ is the fluid density, ν is the kinematic viscosity, V is the velocity vector of the flow field, p is the pressure, and u, v and w are the velocity components in x, y, and z-directions, respectively. The fluid used in this investigation is water, it is incompressible, and its properties are ρ=998.2 kg/m3 and µ=1.003×10-3 kg/m s. A circular cylinder with a diameter (d) of 4 cm and a length (L) of 40cm is used in this study as a bare cylinder or, after the installation of the straight fin and helical strakes, as a cylinder with a passive device. The fins are arranged on the surface of the cylinder with 0.1d height. The strakes having different dimensions are also used for numerical simulation in this investigation II.2.

III.1. Flow Over a Circular Cylinder without Attachments As a base work, a numerical simulation of the flow over a circular cylinder was conducted at low Reynolds numbers (40 and 100).The objective of this analysis is to study the flow structure, pressure and velocity variations near the cylinder and in the downstream wake region. The Strouhal number which is dimensionless frequency was computed to investigate the changes in the frequency at which the vortex shedding occurs. As the fluid flows past the circular cylinder, due to an increase in the pressure of the fluid flow, an adverse pressure gradient is induced. It is followed by the boundary layer separation from the cylinder. As a result, the fluid began to detach alternatively from top and bottom of the cylinder surface continuously. This leads to both drag and lift forces acting on the structure. This alternate shedding of vortices from either sides of the body into the downstream side is as shown in the vorticity magnitude contour in Fig. 2.

Flow Domain and Grid Independence Study

In this investigation the entire flow domain has been discretized using the finite-volume method on a fixed Cartesian-staggered grid with non-uniform grid spacing. The grids in the region of the embedded boundaries are sufficiently fine, and capturing the vortex phenomenon with reasonable accuracy is ensured. The temporal Copyright © 2017 Praise Worthy Prize S.r.l. - All rights reserved


133


The number of fins provided in the circular cylinder include 3, 4, 6, 8 and 12 which are equally spaced on the periphery of the cylinder as shown in Fig. 4.

Fig. 4. Geometry of cylinders with fins

Fig. 2. Contours of vorticity magnitude

Pressure changes according to the vortices motion in the vicinity of the body. At stagnation point, flow comes to rest and pressure reaches the maximum value as shown in Fig. 3(a). Flow separation occurs when shear stress cannot overcome adverse pressure gradients. Flow separation occurs where wall shear stress is zero, as shown in Fig. 3(b). If the flow takes place at laminar conditions, the variation in wall shear stress along the surface is very less. When the Reynolds number is increased to 100, the flow under investigation becomes more turbulent in nature and displayed a significant variation along the surface of the cylinder. The graph clearly shows that when the wall shear stress approaches the zero value, flow separation occurs and vortex shedding initiates.

The addition of fins is found to enhance the process of flow turbulence. It is observed that fins increase the correlation length and the amplitude of velocity fluctuation at the vortex shedding frequency. The effect of fins on the vortex shedding characteristics around the circular cylinder is investigated by changing the number of fins on the periphery of the cylinder. The contours of vorticity magnitude of different fin arrangements (as in Fig. 5) show that vortex shedding occurs in the wake of cylinder fitted with fins. By increasing the number of fins, a significant change in the flow separation is not observed.

(a)

Fig. 5. Contours of vorticity magnitude for different fin configurations

III.3. Effect of the Number of Fins

Parameters such as the drag coefficient and Strouhal number with different numbers of fins are compared with those of a bare cylinder, as shown in Fig. 6, Table I and Fig. 5. It is found that the drag coefficient marginally decreases as the number of fins increases. But a significant increase in Strouhal frequency is observed. TABLE I STROUHAL NUMBER VARIATION ACCORDING TO THE NUMBER OF FINS No. of fins % reduction in Strouhal number 3 0.29 4 2.05 6 3.28 8 4.1 12 4.6

(b) Figs. 3. (a) Pressure coefficient along the cylinder surface. (b) Wall shear stress along the cylinder surface

III.2. Flow Over Circular Cylinder with Fins of Different Orientation at Reynolds Number 28000

The results show a significant variation in Strouhal number with increase in number of fins. The Srouhal number decreased as the number of fins increased within the investigation range. A decrease in Strouhal number indicates a decrease in vortex shedding frequency. A

Five cases of cylinders with different fin configurations and fin orientations were investigated. The fin height was 0.1d in all cases.



134


maximum reduction of 4.6% was observed by providing 12 fins on the periphery of the cylinder.

The typical vortex structures behind the cylinders with and without fins are shown in Fig. 7(b), and give further evidence of fin effects. In the 4 fin-45° arrangement, the auxiliary fin arrangements decrease the lateral spacing of the first large vortex. The lift force on the cylinder provided with 4fin-00 fin arrangement is higher than a bare cylinder, as shown in Fig. 8(a).

Fig. 6. % Reduction in Strouhal Number according to the number of fins (a)

III.4. Effect of Fin Orientation The fins also increase the nonlinear nature of flow in the wake, as evidenced by a remarkable increase in the amplitude of the higher harmonic components of vortex shedding. However, the correlation length in the wake of the cylinder with fins is found to change with fin orientation around the circular cylinder. This phenomenon seems to be related to an irregular wavy pattern of fin distribution along the cylinder axis. (b) Figs. 8. (a) Coefficient of lift variation with fin orientation (b) Strouhal number variations with fin orientation

However, the 4fin-450 arrangement decreases the lift force on the cylinder. Both fin arrangements were effective in decreasing the Strouhal number. The fins with 00 orientation at incidence of flow are more effective in reducing the Strouhal number and vortex induced vibration. The addition of straight fins on the periphery of the cylinder enhances lift force coefficients and drag force coefficients. When the fin number increases, more and more rotating masses of fluids interact with each other, delaying and reducing the formation of vortex shedding. When the fin number increases the Strouhal number decreases. The same cylinder with a different fin orientation will give different flow characteristics with a change in flow direction. This indicates that the effectiveness of straight fins on vortex shedding suppression depends on the fin orientation in the flow.

(a)

III.5. Flow over Circular Cylinder with Triple Start Helical Strakes at Reynolds Number 28000

(b)

A 3D simulation was performed for a circular cylinder fitted with helical strakes with 5d and 10d pitches and 15d height, as shown in Fig. 9(a), in order to study the changes in the flow characteristics in the wake region of the cylinder. When the fluid flows past the cylinder fitted with helical strakes, these strakes will break the flow and produce vortices at different points along the cylinder

Figs. 7. (a) Geometry of cylinders with fins. (b) Vorticity contours for (a)Bare cylinder (b)4 fin- Oo (c) 4 fin -45o

Two cylinders with different fin orientations (as shown in Fig. 7(a)) were investigated. The fin height is normalized by the cylinder diameter and is taken as 0.1d in all cases. The number of fins is 4 and the angle between two fins is 90°.



135


surface. These vortices, which formed around the structure, are out of phase with one another and lead to partial cancellation of the out-of-phase lift forces at different span wise positions. Hence it is observed that the lift forces on the cylinder with strakes are much smaller than those of a bare cylinder. Fig. 9(b) shows the velocity vector at 3 different Z planes along the length of the circular cylinder with helical strakes.

(a)

The Numerical investigation on fluid flow past a circular cylinder fitted with helical strakes reveals that a decrease in lift forces occurred at the expense of an increase in drag force. Table II shows that there is a significant increase in drag force and a maximum increase of 20% in drag force whenever the height of the strakes increases. The reduction in the lift coefficient and Strouhal number in the investigations with Reynolds numbers100 and 28000 is shown in Fig. 10. At low Reynolds numbers these reductions are relatively low and not appreciable, but the suppression of Strouhal frequency was more predominant in turbulent flows. TABLE II FLOW PARAMETERS FOR CYLINDERS WITH AND WITHOUT STRAKES Cylinder with strakes Reynolds Bare Parameter (pitch 10d, height Number cylinder 0.1d) CL 0.36 0.3

(b)

100 Figs. 9. (a) Circular cylinder fitted with triple start helical strakes (b) Velocity vector at different Z planes for a cylinder with strakes 2800

The helical pattern of the projecting fin induces a span-wise motion in the fluid while it flows around the cylinder fitted with strakes. This will produce a swirling motion in the wake region and disrupt the span-wise vortex formation. It is observed that the strength of this partial cancellation of vortex formation at different planes may be the main reason for reduction of vortex Induced vibration to a low value The flow characteristics of a cylinder fitted with helical strakes shows a significant variation from those of a fluid flow past a bare cylinder. Due to the tremendous interaction between the rotating masses of fluids around the cylinder, the flow separation and vortex shedding frequency were considerably reduced. The pitch and height of the strakes have a major role in determining the formation and strength of vortex shedding. By changing the pitch of strakes, large changes in flow characteristics were obtained .The vortex induced vibrations were reduced along with the increase in pitch of the strakes. Beyond pitch = 10d, a reduction in the suppression of vortex shedding is noticed. A similar effect is observed by changing the height of strakes and a reduction of vortex shedding was obtained with an increase in the height of the strakes. In this arrangement there is a close interaction between the two shear layers, which induces an oscillating wake for the plain cylinder. If the cylinder is provided with helical strakes, the strake height will control the shear layer separation and they will interact at a farther distance from the cylinder body than the flow over a bare cylinder. In the cylinder with strakes wake, the two shear layers do not interact with each other, resulting in the absence of the oscillating wake in the downstream region of the cylinder. This is evident in the decrease of the Strouhal number computed for the cylinder fitted with helical strakes.

CD

1.41

1.61

St.

0.17

0.15

CL

1.01

0.01

CD

1.1

1.3

St.

0.189

0.142

Fig. 10. Strouhal number variation with different fin heights

IV.

Conclusion

A numerical investigation on the flow around a circular cylinder with helical strakes at high Reynolds number was conducted to analyze the flow characteristics around a bluff body. As a passive vortex control method, the straight fins and helical strakes are found to be very effective in suppressing the vortex induced vibrations of the structures. The modification in the design of passive devices by varying pitch and height provides valuable information on vortex shedding phenomenon. The study on helical strakes deduces that the lift force can be reduced at the expense of an increase in drag force, and that the order of its increase has to be further investigated. The numerical simulation is limited to low and medium Reynolds numbers and is confined to incompressible fluid flows. Most of the works in the area are experimental works, only a fewer numerical works are available in the literature.



136


References

Author’s information

Inoue, Iwakam, and Hatakeyama, Aeolian tones radiated from flow past two square cylinders in a side-by-side arrangement, Physics of Fluids, 18(4), 046104,2006 [2] Ali, M.S.M., Doolan, C.J., Wheatley, V. (2009). Grid convergence study for a two dimensional simulation of flow around a square cylinder at a low Reynolds number. In: Seventh International Conference on CFD in The Minerals and Process Industries,2009. [3] T. Zhou, S.F. Mohd. Razali, Z.Hao, L.Cheng., On the study of vortex induced vibration of a cylinder with helical strakes. Journal of Fluids and Structures 27903–917,2011 [4] Shan Huang, Andy Sworn, Hydrodynamic coefficients of two fixed circular cylinders fitted with helical strakes at various staggered and tandem arrangements, Applied Ocean Research 43, 21–26,2013 [5] Lee KeeQuen, Investigation on the effectiveness of helical strakes in suppressing VIV of flexible riser. Applied Ocean Research 44 82–91,2014 [6] Shan Huang, VIV suppression of a two-degree-of-freedom circular cylinder and drag reduction of a fixed circular cylinder by the use of helical grooves, Journal of Fluids and Structures 27,1124–1133,2011 [7] Rathakrishnan, E., Bluff Body Edge Effect on Pressure-Hill and Zone of Influence, (2014) International Review of Mechanical Engineering (IREME), 8 (4), pp. 662-666. [8] Kumaravel, G., Jeyajothiraj, P., Rathakrishnan, E., Formation and Dissipation of Karman Vortex Street in an Accelerating Flow Past a Circular Cylinder, (2015) International Review of Aerospace Engineering (IREASE), 8 (2), pp. 43-55. doi:https://doi.org/10.15866/irease.v8i2.5846 [9] Adebayo, D., Rona, A., PIV Study of the Flow Across the Meridional Plane of Rotating Cylinders with Wide Gap, (2015) International Review of Aerospace Engineering (IREASE), 8 (1), pp. 26-34. doi:https://doi.org/10.15866/irease.v8i1.5290 [10] Beigmoradi, S., Ramezani, A., Effect of the Backlight Angle on the Aero-Acoustics of the C-Pillar, (2013) International Review on Modelling and Simulations (IREMOS), 6 (3), pp. 988-993. [11] Adebayo, D., Rona, A., The Persistence of Vortex Structures Between Rotating Cylinders in the 106 Taylor Number Range, (2015) International Review of Aerospace Engineering (IREASE), 8 (1), pp. 16-25. doi:https://doi.org/10.15866/irease.v8i1.5288 [12] Rostane, B., Aliane, K., Abboudi, S., Three Dimensional Simulation for Turbulent Flow Around Prismatic Obstacle with Rounded Downstream Edge Using the k-ω SST Model, (2015) International Review of Mechanical Engineering (IREME), 9 (3), pp. 266-277. doi:https://doi.org/10.15866/ireme.v9i3.5719 [13] Barata, J., Multiple Jet/Wall/Crossflow Interactions, (2014) International Review of Aerospace Engineering (IREASE), 7 (3), pp. 69-83. [14] Aloui, F., Kourta, A., Ben Nasrallah, S., Experimental Study of Synthetic Jets with Cross Flow in Boundary Layer, (2016) International Review of Aerospace Engineering (IREASE), 9 (1), pp. 13-21. doi:https://doi.org/10.15866/irease.v9i1.9130

Sunil A. S. is an Assistant Professor in Mechanical Engineering at Government Engineering College, Thrissur. He received his Bachelors degree in Mechanical Engineering from the Government Engineering College, Trichur, affiliated to the University of Calicut, M Tech from NIT Calicut. He is a research scholar in Cochin University of Science and Technology. His research works are on fluid flow instabilities and computational fluid dynamics.

[1]

Tide P. S. is a Professor in Mechanical Engineering at Cochin University of Science and Technology, Cochin. He received his Bachelors degree in Mechanical Engineering from the Government Engineering College, Trichur, M Tech from IIT Kharagpur and Doctoral degree from the IIT Madras. His research interests are computational fluid dynamics and compressible fluid flows.



137


Investigation on Failure Strength of Bolted Joints Woven Fabric Reinforced Hybrid Composite D. Sivakumar, L. F. Ng, R. M. Chew, O. Bapokutty Abstract – Understanding of composite bolted joint behaviour is vital since it has been widely used in application involving mechanically fastened joints. The advent of advanced technology had led to the use of hybrid composite to reduce the usage of non-environment friendly material such as synthetic fibre. This current study aims to investigate the effect of geometric parameters on the failure load of bolted joint hybrid composite laminate. Composite laminates were fabricated using hot press moulding compression method and they were cut according to the appropriate dimension with the width/diameter (W/D) and edge-distance/diameter (E/D) ratio of 3, 4, 5 and 6. The hole diameter of each composite laminate was fixed at 6 mm. Bolted joint tests were conducted in accordance to ASTM D5961 using the Universal Testing Machine and several failure modes were identified. The results concluded geometric parameters significantly affect both failure load and failure mode of hybrid composites. The increase in W/D and E/D ratio increase the load carrying capability of the laminate. Copyright © 2017 Praise Worthy Prize S.r.l. - All rights reserved.

Keywords: Bolted Joint, Geometric Parameters, Glass Fibre, Hybrid, Woven Kenaf

I.

The application of lightweight and low-cost natural fibres provides the potential to supersede a significant amount of glass and mineral fillers in automotive industries [13]. It is recently demonstrated by Ishak et al. [14] that natural fibres is an excellent candidate to be used as a car front hood materials. Despite natural fibres possess several desirable properties, the low mechanical properties and high moisture ingestion characteristic have become the retardation of using natural fibre as reinforcement. Due to the adverse effect of natural fibre reinforced composite, hybrid composite which consists of synthetic fibre and natural fibre has been explored to improve the mechanical performance of the composite material. Hybrid composite is formed when more than one type of reinforcement is used within a single matrix. Hence, the advantage of one fibre could compensate the drawback of another fibre. Partial replacement of natural fibre with synthetic fibre improves the mechanical properties [15]. Mechanically fastened joints are common and critical elements in a composite structure such as aircraft structures. It is important to design the joint properly to avoid overweight or defective structures. A bolted joint is engaged to hold two or more parts together to form an assembly in a mechanical structure. Since most of the composite materials display a brittle failure, with little or no margin of safety through ductility, the mechanism of the brittle failure propagation in bolted joint must be fully known [16]. The stress distribution around the hole in the bolted joints is a perplexing phenomenon and it is strongly affected by the geometric parameters, clamping force, stacking sequences, and the clearance between the

Introduction

Composite materials are frequently applied in advanced engineering fields such as in aerospace, leisure, automotive, construction and sporting industries. Composite materials have wide application because of their high specific strength and modulus [1]-[3]. The purpose of using composite structures in the industry is to build a robust and stiff component with low density characteristic. Thermoplastics are preferable than thermosets since they are mouldable after initial process whereas thermoset polymers are permanent and irreversible after solidification. Moreover, thermoplastics are more environmental friendly compared to thermoset matrix since the former matrix is recyclable [4]. Polypropylene is one of the most widely used polymers as it offers some attractive features such as low cost, high toughness, short processing time and low density [5]. Recently, natural fibres are widely used in composite structures due to their low cost, recyclability, comparable specific tensile properties, low density and biodegradable characteristics [6]. The using of natural fibre for composite materials offers a high degree of environmental friendliness which is well known among researchers [7]-[10]. It had been shown natural fibre reinforced composite can be reprocessed with minimal deterioration on its mechanical performance [11]. Furthermore, natural fibres are lightweight with high specific strength, which can reduce approximate 10% to 15% the density of the composite material compared to glass fibre [12].



138

D. Sivakumar, L. F. Ng, R. M. Chew, O. Bapokutty

hole and the pin. Previous research has shown the failure mode is highly influenced by the W/D and E/D ratios [17]. Soykok et al. [18] conducted failure analysis on the effect of temperature on the failure load of glass fibre reinforced epoxy composite. They found out the increase in temperature decreased the failure load and tightening torque increased the load carrying capacity of the joint at room and elevated temperature. Ozen and Sayman [19] numerically and experimentally investigated the bearing strength of epoxy based woven glass fibre reinforced composite with two serial holes. The results showed increase in preload moment, hole to hole diameter (K/D), W/D and E/D ratios improved the failure load of composite. Meanwhile specimens that were immersed in sea water for 24 hours had lower failure load compared to the unimmersed specimen. Net tension and bearing failure modes were observed for specimens without preload moment whereas only net tension failure mode was observed when preload moments were applied. Okutan conducted work [20] to identify the failure strength of mechanically fastened glass fibre reinforced epoxy composite with different geometric parameters. The result showed joint failure with net-tension and shear-out modes is catastrophic and immediate. However, bearing failure is progressive. Hence, bearing failure is more preferred in applications. Besides that, failure mode due to tension and shear is suppressed by providing sufficiently large end distance and width. Higher widths result in bearing failure whereas lower width results in tension failure. Jadee [21] studied the progressive failure analysis and mapped the failure into plain weave glass fibre reinforced polymer bolted joint through experimental and finite element methods. Hasine failure criterion is used to determine the failure load, failure mode and bearing strength. It showed that when W/D and E/D is more than 3, bearing failure mode occurs. However, the net-tension failure mode was shown when W/D is equal to 2, except for the small edge distance when E/D is equal to 2. The shear-out failure mode occurs when short edge laminates (E/D = 1) is used. A recent study investigated the failure behaviour of glass-kenaf hybrid composite under two serial bolted joint holes and the result demonstrated bearing strength increases with increase in K/D and E/D ratio [22]. It was observed that the failure mode at outer hole stabilise when E/D is larger than 2 and bearing failure mode occur when E/D ratio increases. Khashaba et al. [23] presented failure analysis of pinned joint composite with different stacking sequences. The study concluded the [45/-45]2s laminates exhibited the highest failure displacement and bearing strength while laminates with [0/90]2s stacking sequence had the highest ultimate strength and ultimate failure displacement. Bolted joints are generally required in construction of structural components. Nevertheless, joints are always the weakest point of a structural part which results in premature failure. Therefore, the investigation on the bearing strength and failure modes is necessary and vital.

Literatures show limited study on the failure strength of bolted joints hybrid composites. Moreover, the effect of W/D and E/D ratio on the failure strength of bolted joint kenaf/glass hybrid composites is still unexplored. The present study investigates the failure analysis of the bolted joint under geometric parameters with W/D and E/D in the range of 3 to 6 for glass-kenaf hybrid composite laminates to show the potential of using woven kenaf together with thermoplastic polymer in bolted joint applications.

II.

Material and Methods

Al Waha Petrochemical company supplied polypropylene (PP) granules. PP was compressed into a thin PP sheet before the composite fabrication process. Woven E-glass fibre (GF) with the areal weight of 600 g/m2 was obtained from ZKK Sdn. Bhd. Woven kenaf fibre (KF) with the areal weight of 295 g/m2 was obtained from National Kenaf and Tobacco Board. Both woven fibres were cut into the dimension of 170mm x 250mm. II.1.

Composite Panels Fabrication

Composite panels with the stacking sequence of [PP/GF/PP/KF/PP/PP/GF/PP] were produced with the aid of hot press machine. A picture frame mould was used in the composite fabrication process to get a final nominal composite thickness of 3 mm. The temperature for the compression moulding process was 180°C with a pressure of 5 MPa. Table saw was used to cut the panels to the desired dimension as shown in Fig. 1 based on ASTM D5961. Fig. 2 shows the fabricated hybrid composite specimen for bolted joint test. The length of the specimen, L = 135mm. Table I shows the specimen dimension ratio used for the test.

Fig. 1. Specimen geometry

Fig. 2. Hybrid composite specimen

D (mm) 6


TABLE I GEOMETRIC RATIO E/D 3, 4, 5, 6

W/D 3, 4, 5, 6


139


II.2.

Bolted Joint Test

the loading without breaking. This failure mode is called bearing failure mode. The effect of E/D parameter on the load bearing capacity for all specimens shows a similar trend as W/D parameter. As the E/D ratio is getting higher, the composite laminates have higher load bearing capacity.

Universal testing machine INSTRON 5969 was used to perform bolted joint test according to ASTM D5961. Specimens were attached and bolted to the custom made jig. The design of the jig and the fixture of the bolted joint are shown in Fig. 3. The specimen was loaded until force dropped off about 30% from the maximum force to observe the failure mode. The crosshead displacement rate of the experiment was fixed at 2mm/min. The result was then analysed and evaluated to deduce geometric effect on the failure load.

(a)

(b)

Figs. 5. Failure modes of deformed hybrid composite specimens (a) Net-tension failure mode (b) Bearing failure mode

Fig. 6 shows the specimen having a plunging effect on the surface due to the washer plunged into the specimens during the bolted joint test.

Fig. 3. Jig design and fixture of bolted joint

III. Results and Discussion The W/D and E/D ratio were set at 3, 4, 5 and 6. The tests were carried out up to three times at room temperature and the nominal failure strength was calculated during the experiment. Bolted composite specimens commonly consist of four types of failure modes which include cleavage, net-tension, shear-out and bearing are shown in Fig. 4.

Fig. 6. Plunging effect on composite specimen

The plunge effect formed a cut-like opening around the hole that damaged the PP and the fibres. Hence, the full potential of the specimen was not fully established. Throughout the experiment, net-tension failure occurs on the specimen with W/D ratio below 5 whereas specimen with W/D ratio higher than 4 experienced bearing failure. In addition, the specimen became geometric independent when the W/D ratio is 6. Fig. 7 shows bearing strength of composite laminates with E/D variation for W/D ratio equal to 3, 4, 5, and 6. From Fig. 7, it demonstrates that the increase of the edge distance improved the bearing capacity of the laminates until a certain E/D ratio was reached; a further increase of the E/D ratio does not improve the corresponding bearing capacity of the laminates. Aside from that, the increase in the width of the laminates also results in improving joint strength. The main factor that affects the specimen failure mode is the W/D ratio. By increasing the W/D ratio, the specimen changes from net-tension to bearing failure mode. When W/D = 5, the specimen goes through transition failure mode from purely net-tension to bearing with nettension. At W/D ratio = 6, the specimen experiences only bearing failure mode. Thus, the specimen became geometric independent when the ratio is equal to 6 or higher. E/D ratio does not directly affect the specimen failure mode. However, it plays a role in sustaining the load.

Fig. 4. Common failure modes [1]

During the experiment, two types of failure modes were observed which are net-tension and bearing as shown in Figs. 5(a) and (b). The net-tension failure modes can be observed when the W/D ratio is smaller than 5 while bearing failure mode can be observed when the W/D ratio is equal to 6. Mixed bearing and nettension failure mode were observed when W/D ratio is equal to 5. The load-displacement curves are nearly linear for the specimen. It was observed that some specimens fail suddenly due to a small value of W/D and E/D ratio. For numerous specimens, the load increases with growing damage and reaches the ultimate load which is then followed by load decrease with rising damage. However, the specimen continues to withstand



140


Numerous specimens had been observed in which the composite breaks on one side for the net-tension failure mode. In fact, failure due to net-tension should break on both sides. The specimen breaks on one side only is because of the asymmetrical load applied during bolted joint test. Figs. 8 and Figs. 9 show the box plot for the minimum, maximum and median loads for each configuration ratios of W/D and E/D. The upper and

lower edges represent the maximum and minimum values of failure load while the middle line represents the ultimate median load. Some test values obtained from the result are closed to each other while some other data displayed higher standard deviation and spread in test results. This was probably due to the specimen inconsistency and geometrical inequalities.

34 30 26 22 18

19,498

W/D = 4

22,797 23,818

3

Bearing Strength, Mpa

Bearing Strength, MPa

W/D = 3

4

5

25,835 6

E/D

34 30 26 22 18

22,303 26,506 25,277 26,19 3

W/D = 6 Bearing Strength, MPa

Bearing Strength, MPa

3

30,887 32,482 29,714 4

6

(b)

W/D = 5

27,078

5

E/D

(a)

34 30 26 22 18

4

5

6

E/D

34 30 26 22 18

29,604 3

32,92 33,131 31,761

4

5

6

E/D

(c)

(d)

Figs. 7. The effect of E/D variation on the bearing strength (a) W/D = 3 (Net-Tension), (b) W/D = 4 (Net-Tension), (c) W/D = 5 (Bearing/Net-Tension), (d) W/D = 6 (Bearing)

W/D 4 Load (N)

2800 2300 1800 1300 3

4

5

6

E/D ratio (b)

(a)

Figs. 8. Box plots of maximum loads at each configuration of W/D and E/D ratio (a) W/D = 3, (b) W/D = 4



141


(a)

(b)

Figs. 9. Box plots of maximum loads at each configuration of W/D and E/D ratio (a) W/D = 5, (b) W/D = 6

IV.

Conclusion

[3]

The failure analysis on hybrid composite under geometric parameters had been conducted through experimental work. It was concluded geometric parameters have a significant influence on the joint behaviour. Based on the experimental results, following conclusions can be deduced: (a) The maximum load of all composite laminates with different configurations increases as the geometric ratios increase. This implied that the phenomenon of increasing the geometric ratio improves the bearing strength of the hybrid composite joint was established. (b) Bearing and net-tension failure modes were observed in the composite specimens after bolted joint test. The bearing mode is considered as the desired failure mode since it can resist higher load. (c) The specimens with W/D ratio, which is less than 5, are considered to have a weak failure mode as it has a net-tension failure. As the W/D ratio is equal or higher than 6, the failure stabilise and becomes geometric independent.

[4]

[5]

[6]

[7]

[8]

[9]

Acknowledgements [10]

Authors would like to thank Universiti Teknikal Malaysia Melaka for the continuous support on this research project. Authors would also wish to express their gratitude towards National Kenaf and Tobacco Board for the sponsorship of kenaf fibre, Skim Zamalah UTeM provided by Universiti Teknikal Malaysia Melaka and FRGS/1/2015/SG06/FKM/03/F00276 from Ministry of Higher Education Malaysia.

[11]

[12]

References [1]

[2]

[13]

D. Sivakumar, L. F. Ng, J. W. Ng, M. Z. Selamat and Sivaraos, Failure Analysis of Hybrid Fibre Reinforced Plastics for Bolted Joint under Thermal Effect, Journal of Mechanical Engineering, Vol. 1, n. 1, pp. 141-156, 2017. F. Sen, M. Pakdil, O. Sayman, and S. Benli, Experimental Failure Analysis of Mechanically Fastened Joints with Clearance in Composite Laminates under Preload, Materials & Design, Vol.

[14]


29, pp. 1159-1169, 2008. M. A. Che Mahzan, S. D. Malingam, Sivaraos, M. Z. Selamat, and M. R. Said, A Study on the Mechanical and Forming Performance of Oil Palm Fiber Reinforced Polypropylene Composite, American-Eurasian Journal of Sustainable Agriculture, Vol. 8, n. 4, pp. 141-147, 2014. D. Sivakumar, L. F. Ng, M. Z. Selamat and Sivaraos, Investigation on Fatigue Life Behaviour of Sustainable Bio-based Fibre Metal Laminate, Journal of Mechanical Engineering, Vol. 1, n. 1, pp. 123-140, 2017. D. Sivakumar, L. F. Ng, S. M. Lau and K. T. Lim, Fatigue Life Behaviour of Glass/Kenaf Woven-Ply Polymer Hybrid Biocomposites, Journal of Polymer and the Environment. Epub ahead of print 24 February 2017. DOI: 10.1007/s10924-017-0970-0 H. Ku, H. Wang, N. Pattarachaiyakoop, and M. Trada, A Review on the Tensile Properties of Natural Fiber Reinforced Polymer Composites, Composite Part B: Engineering, Vol. 42, n. 4, pp. 856-873, 2011. K. Adekunle, S. W. Cho, C. Patzelt, T. Blomfeldt, and M. Skrifvars, Impact and Flexural Properties of Flax Fabrics and Lyocell Fiber-reinfored bio-based Thermoset, Journal of Reinforced Plastics and Composites, Vol. 30, n. 8, pp. 685-697, 2011. S. Rassmann, R. Paskaramoorthy, and R. Reid, Effect of Resin System on the Mechanical Properties and Water Absorption of Kenaf Fibre Reinforced Laminates, Materials & Design, Vol. 32, n. 3, pp. 1399-1406, 2011. A. Ratna Prasad, and K. Mohana Rao, Mechanical Properties of Natural Fibre Reinforced Polyester Composites: Jowar, Sisal and Bamboo, Materials & Design, Vol. 32, n. 8, pp. 4658-4663, 2011. D. Sivakumar, S. Kathiravan, M. Z. Selamat, M. R. Said and Sivaraos, A Study on Impact Behaviour of a Novel Oil Palm Fibre Reinforced Metal Laminate System, ARPN Journal of Engineering and Applied Sciences, Vol. 11, n. 4, pp. 2483-2488, 2016. S. DharMalingam, M. H. R. Hashim, M. R. Said, A. Rivai, M. A. Daud, Sivaraos, M. A. Che Mahzan, Effect of Reprocessing Palm Fiber Composite on the Mechanical Properties, Applied Mechanics and Materials, Vol. 699, pp. 146-150, 2015. M. J. Sharba, Z. Leman, M. T. H. Sultan, M. R. Ishak, and M. A. Azmah Hanim, Effects of Kenaf Fiber Orientation on Mechanical Properties and Fatigue Life of Glass/Kenaf Hybrid Composites, BioResources, Vol. 11, n. 1, pp. 1448-1465, 2016. J. Holbery and D. Houston, Natural-Fiber-Reinforced Polymer Composites in Automotive Applications, The Journal of The Minerals, Metals & Materials Society, Vol. 58, n. 11, pp. 80-86, 2006. N. M. Ishak, S. D. Malingam, M. R. Mansor, Selection of Natural Fibre Reinforced Composites Using Fuzzy VIKOR for Car Front Hood, International of Materials and Product Technology, Vol. 53, n. ¾, pp. 267-285, 2016.


142


[15] N. Venkateswaran, A. Elayaperumal, and G. K. Sathiya, Prediction of Tensile Properties of Hybrid-Natural Fiber Composites, Composite Part B: Engineering, Vol. 43, n. 2, pp.793-796, 2012. [16] K. I. Tserpes, G. Labeas, P. Papanikos, and T. Kermanidis, Strength Prediction of Bolted Joints in Graphite / Epoxy Composite Laminates, Composite Part B: Engineering, Vol. 33, n. 7, pp. 521-529, 2002. [17] B. G. Kiral, Effect of the Clearance and Interference-Fit on Failure of the Pin-Loaded Composites, Materials & Design, Vol. 31, n. 1, pp. 85-93, 2010. [18] I. F. Soykok, O. Sayman, M. Ozen, and B. Korkmaz, Failure Analysis of Mechanically Fastened Glass Fiber/Epoxy Composite Joints under Thermal Effects, Composite Part B: Engineering, Vol. 45, n. 1, pp. 192-199, 2013. [19] M. Ozen, and O. Sayman, Failure Loads of Mechanical Fastened Pinned and Bolted Composite Joints with Two Serial Holes, Composite Part B: Engineering, Vol. 42, n. 2, pp. 264-274, 2011. [20] B. Okutan, The Effects of Geometric Parameters on the Failure Strength for Pin-Loaded Multi-Directional Fibre-Glass Reinforced Epoxy Laminate, Composite Part B: Engineering, Vol. 33, n. 8, pp. 567-578, 2002. [21] K. J. Jadee, Progressive Failure Analysis and Failure Map into Plain Weave Glass Fibre Reinforced Polymer Bolted Joint, American Journal of Materials Science and Engineering, Vol. 3, n. 2, pp. 21-28, 2015. [22] D. Sivakumar, L.F. Ng and N.S. Salmi, Eco-Hybrid Composite Failure Behavior of Two Serial Bolted Joint Holes, Journal of Engineering and Technology, Vol. 7, n. 1, pp. 114-124, 2016. [23] U. A. Khashaba, T. A. Sebaey, and K. A. Alnefaie, Failure and Reliability Analysis of Pinned-Joint Composite Laminates: Effects of Stacking Sequences, Composite Part B: Engineering, Vol. 45, n. 1, pp. 1694-1703, 2013.

Raymond Mitchell Chew received his degree in Mechanical Engineering from Universiti Teknikal Malaysia Melaka in 2016.

Omar Bapokutty is working as a Senior Lecturer in the Faculty of Mechanical Engineering, Universiti Teknikal Malaysia Melaka. He received his PhD in Mechanical Engineering from Universiti Kebangsaan Malaysia in 2015. His research interests include creep, fatigue and advanced material.

Authors’ information High Performance Structure Research Group, Centre for Advanced Research on Energy, Faculty of Mechanical Engineering, Universiti Teknikal Malaysia Melaka, Hang Tuah Jaya, 76100 Durian Tunggal, Melaka, Malaysia. Sivakumar Dhar Malingam is working as a Senior Lecturer in the Faculty of Mechanical Engineering, Universiti Teknikal Malaysia Melaka. He received his PhD in Mechanical Engineering from the Australian National University in 2012. His research interests include metal and composite forming, fibre metal laminate, bio-composite and FEA. Ng Lin Feng received his degree in Mechanical Engineering from Universiti Teknikal Malaysia Melaka in 2015. He is currently an MSc student at Universiti Teknikal Malaysia Melaka, Malaysia. His research interests include fibre metal laminate, natural fibre and fatigue.



143


CMM Path Planning, Position and Orientation Optimization Using a Hybrid Algorithm Gabriel Mansour1, Dimitrios Sagris2, Αpostolos Tsagaris3 Abstract – In the present paper a Hybrid Optimization Method is applied in order to determine the optimal placement of a workpiece on a coordinate measuring machine (CMM) as well as the optimal movements of the CMM probe, aiming to reduce the measurement time. Further to the review of the toolpath computation and optimization methods, the framework of the optimization problem is defined. A modeling methodology for the problem is proposed and a solving method is presented. The proposed methodology is a combination of a genetic algorithm, a quasi-Newton algorithm and a constraints-handling method. The final outcome is the determination of the best positioning location of the object as well as the best probe positioning location for the scanning. The evaluation is done comparing two methods; the first one uses a genetic algorithm while the second combines a genetic algorithm with a hill climbing method. Copyright © 2017 Praise Worthy Prize S.r.l. - All rights reserved.

Keywords: Genetic Algorithms, Optimization, System Integration, CNC Machining

In order to achieve this, two restrictions must be met: a)the minimization of the sum of squared deviations between the predefined and the calculated positions of the measuring probe and b)the minimization of the total motion time along measured surfaces, from the initial to the final point, while selecting simultaneously the optimal sequence of measured surfaces. In order to evaluate the method, an algorithm is developed in MatlabTM environment and tested on a SCIROCCO-RECORD by Brown & Sharpe DEA CMM machine (Figure 1).

Nomenclature CMM QNA GA CNC

Cοordinate Measuring Machine Quasi Newton Algorithm Genetic Algorithm Computerized Numerical Control

I.

Introduction

In order to reduce production time length, a great deal of manufacturing research focuses on methods for optimizing the production processes and the quality control. Existing research work is focused in the optimization process using industrial robots; a fact which is evident from the number of publications on this subject. In the last few years the path planning optimization is also used on CNC machines and on reverse engineering methods. Evolutionary approaches such as Tabu Search (TS), Artificial Neural Networks (ANN), Genetic Algorithm (GA), Particle Swarm Optimization (PSO), Ant Colony Optimization (ACO) and Simulated Annealing (SA) are used to solve the path-planning problem, yielding to efficient solutions, but not always optimal ones. In most cases, CMMs are programmed manually by moving the measurement probe through a sequence of points of interest; often repeated during the measurement process. The repeated path and the non-optimal movement of the CMM lead to a longer inspection path, which in turn increases the process time. The aim of the proposed methodology is to determine an optimum relative position between the base of the coordinate measuring machine (CMM) and the workpiece to be measured.

Fig. 1. Brown & Sharpe DEASCIROCCO-RECORD



144

Gabriel Mansour, Dimitrios Sagris, Αpostolos Tsagaris

II.

This method uses a genetic algorithm with an innovative crossover function (greedy selection crossover GSX) which reduces the overall motion trajectory of the tip, thus reducing the total measurement time [4]. Syed Hammad Mian et al. present a time optimization technique for measuring the part using a CMM. A genetic algorithm, a method of simulated annealing and a brute force method are compared, in respect of the effectiveness and measuring time [5]. Steven N. Spitz et al. propose an optimal path calculation algorithm avoiding obstacles. The basic idea is to develop a map of the free space of the piece, in which the measuring points are represented as nodes of a network. The optimal route is calculated by solving the appropriate problem, when the measuring points appear as a single connected component of the map[6]. Lu et al. use a GA model for the optimization of the inspection path on a CMM. Genetic operators like encoding, heuristic population initialization, fitness evaluation, crossover and mutation are used in order to reach the best solution. The experimental results indicate that the GA path planning system provides reasonable results [7].

State of the Art

The methodology tested on the current paper is an evolution of previous works developed mainly in the area of robotics. In [1] (Sagris et al.) a methodology is presented in which Cartesian spatial robotic arms are designed. This methodology is based ona hybrid algorithm consisting of a genetic algorithm, a quasiNewton algorithm and a constraints handling method. The algorithm is applied to an open loop spatial robot with three revolute joints. The base position, the geometry of the links and the joint angles of the robot were included in the design process. The calculated solution places the end-effector at certain locations, avoiding at the same time singular configurations. The same optimization algorithm is used in [2] for a point cloud alignment, without user involvement. Stojadinovic et al. introduced a set of feature optimization of a CMM planning process. Beginning with a reduction of the initialization time through the automatic generation of a measuring protocol, a reduction of the measuring time through the optimization of the measuring probe path is achieved, followed by an analysis of the placement of the workpiece using the accessibility analysis and automatic configuration of measuring probes [8]. Mansour et al. present an expert CNC system, using genetic algorithms for the optimization of the path keeping time short. The algorithm produces an optimized machining path, using graphic features and geometric parameters from a CAD part drawing. The efficiency of CNC machining is highly improved and the cutting time is reduced. The proposed method is tested on a piece with 8 indentations, reducing the overall processing time by 18.63% [3]. In another research case by Zhang et al. an intelligent CMM system is presented which is controlled by a CCD camera. The position as well as the orientation of the part is identified by the CCD camera (which is mounted on the CMM). Using CAD, the geometric and measuring information of the part is extracted. Following that, a measuring program is created determining the selection of the probe and the accessories, the determination of the features and parameters to be measured, as well as the number and the positions of the measured points [9]. Hwang’s proposes [10] an inspection plan involving the workpiece setup and the configuration of the measuring probes; some elements of this approach are used in the present paper. Other key elements are collision avoidance and accessibility analysis. Collision avoidance is issued by Lin et al. [11],[12] and accessibility analysis by Wu et al. [13], Alvarez et al. [14] and Rico et al. [15]. In the field of reverse engineering several papers are published regarding the usage of intelligent techniques. Qu et al. presents a method of finding the optimal measuring path using a coordinate measuring machine (CMM).

III. Mathematical Formulation The optimal placement of the workpiece could be formulated as an optimization problem, where the objective function (F) takes into account the deviations between the prescribed and the real probe poses (F1), as well as the total motion time among all surfaces of the workpiece (F2): =

+

∙

(1)

where: ( , )−

=

(, )

=

(2)

(3)

n is the number of the prescribed measuring points, ( , ) is the real calculated value of the element (i, j) ( , ) is the prescribed value of the of the matrix, element (i, j), tt is the total required motion time in order to reach all of the surfaces and β is a weighting factor. In Figure 2, the coordinate system of each part according the Denavit-Hartenberg is shown. Using the homogenous transformation matrices (Figure 2), the pose of the C.M.M. probe P6 with respect to the fixed frame PS is given by: =

∙

∙

∙

∙

∙

∙

(4)

where the matrix describes the pose of frame i whit respect to frame i-1, though the corresponding DenavitHartenberg parameters (Table I).



145


IV.

Proposed Algorithm

The described mathematical model is solved by means of a hybrid method that combines a genetic algorithm (GA)and a quasi-Newton algorithm (QNA). The input data for the algorithm consist of the number and the type of joints, the number of independent variables and the upper and lower limits of these variables, default sensor measurement locations and parameters of the algorithm. These parameters include the initial parameters of the genetic algorithm (GA), such as the initial population of individuals, the possibility of crossover, the probability of mutation as well as the number of repeats of the genetic algorithm (GA) and the climbing method (QNA). The objective function F is used in all steps of the algorithm (Figure 3). Input Data  End Effector Machining position  Limits  Algorithm parameters  Fitness Function

Fig. 2. Denavit-Hartenberg parameters of C.M.M. Brown & Sharpe DEA SCIROCCO TABLE I DENAVIT-HARTENBERG PARAMETERS OF C.M.M. BROWN & SHARPE DEA SCIROCCO Link θi(°) αi(°) ai (mm) di (mm) 0 θ0 α0 a0 -799,18 1 90 -90 1174 d1 2 -90 90 223 d2 3 180 180 169 d3 4 θ4 -90 0 0 5 θ5 -90 0 0 6 180 180 0 145

Max of GAGen es

No

QNA Method

Max of QNA Repeats

The total travel time required to reach all the surfaces of the workpiece is given by the second part of the objective function (F2): =

Genetic Algorithm (GA)

No

Optimal Solution

(5) Fig. 3. Proposed algorithm flow chart

where tk is the travel time between the task point k and task point k+1 and n is the number of the prescribed poses of the C.M.M. probe. The total travel time is calculated as the minimum required time. Taking into account all the available routes among the target poses of the C.M.M. probe, the corresponding travel times are calculated and the minimum one is selected as total travel time. The time tk can be written as: =

,

,

̇

, i =(1,2,…,6)

In the first step of the proposed algorithm, the initial population of individuals is randomly generated, to define the values of the variables and they are used to calculate the value of the objective function. The genetic algorithm uses specific procedures for creating new generations, which are the selection, elitism, crossover and mutation. The new generations of successive iterations converge to a minimum, which is not necessarily the in general. The number of repetitions is limited by either a maximum number of iterations, or the achievement of a critical value of the objective function, both predetermined by the user. The minimum calculated value of the objective function is selected as a final value of the genetic algorithm so it determines the optimal vector of the independent variables in this step. These optimal values of the independent variables are inserted in the second step, as input data in the climbing

(6)

where qi,k is the ith joint displacement for kth probe pose and ̇ is the maximum allowed velocity of joint i, imposed by the constructor. The slowest joint determines the motion time between poses k and k+1.



146


V.

Scanning Points

method (QNA). They form an initial estimation variables vector. The Climbing method (QNA) modifies the values of the variables in this vector, using a gradient method of finite differences, so that the objective function moving in the largest slope, which locates in the search area. The changes in the values of the variables inside the climbing process are repeated until either a predetermined maximum number of iterations is obtained, or until the a local minimum is detected. These steps are repeated a few times, including repetitions of the first step, to identify various local minima, through the genetic algorithm and the approach of the general minimum, the use of climbing method (QNA). The number of these repetitions is limited to some maximum number of iterations or to a critical value of the objective function; both specified by the user. The minimum calculated value of the objective function is chosen as the final value of the climbing method (QNA) and the optimal vector of the independent variables is defined in this step. The minimum calculated value of the objective function determines the values of the independent variables selected in the third step, which consist at the same time the optimal values of the proposed algorithm. These variables determine the optimal placement of the arm base, and the formations through the variable of the angles of joints on all the predetermined end processing sites. Although a part of the objective function is discontinuous, the successive application of the genetic algorithm (GA) and the climbing method (QNA), based on the elementary slopes (gradient), minimizes the degradation behavior of the proposed algorithm to the minimum.

Var

Start End

386,876 443,493

Start End

484,878 503,673

Start End

413,896 460,917

Start End

412,574 493,287

Surface 1 -708,755 -620,383 Surface 2 -681,132 -639,649 Surface 3 -609,218 -611,55 Surface 4 -716,543 -706,939

Angles Α Β (°) (°)

-574,161 -574,82

0 0

0 0

-575,707 -602,218

45 45

-90 -90

-573,273 -615,589

90 90

0 0

-590,277 -606,314

90 90

180 180

To make evident of the advantages of the proposed methodology, we compare four alternatives in each of the three numerical examples.

Experimental Results

The proposed methodology, implemented using the programming language MatlabTM, is applied to the CMM SCIROCCO-RECORD company Brown & Sharpe DEA, consisting of three sliding joints and two rotating joints. The input data used in the algorithm are: the algorithm parameters, the predetermined positions of the probe in an absolute Ps system (Table III), and the limits of variables (Table II).

θ0(°) α0(°) a0(mm) d1(mm) d2(mm) d3(mm) θ4(°) θ5(°)

TABLE III INITIAL PROBE POSITION Cοordinates X Y z (mm) (mm) (mm)

Fig. 4. Workpiece placement on C.M.M

The first method uses a genetic algorithm, the second combines a genetic algorithm with a constraints handling method, the third one combines a genetic algorithm with a hill climbing method and the fourth is a combination of all. The parameters used in all the tests, particularly those involving a genetic algorithm, are held constant throughout. They are selected as the best ones, after several trials, using a population of parents equal to 80, a crossover probability of 80%, a mutation probability of 8% and the constant b is 1e-7 . The rest of the parameters of the algorithm used in each test are different, and each example is depicted in the Table IV.

TABLE II VARIABLE LIMITS AND MAXIMUM JOINT SPEED Max speed Max speed min max (°/s) (mm/s) -180 180 -180 180 0 1488 150,7 1400 500 628 1488 500 369 1029 500 -180 180 90 -180 180 90 -



147


TABLE IV PARAMETERS AND RESULTS OF THE IMPLEMENTATION OF THE FOUR METHODS Parameters

Results

GA

QNA

CHM

Variable limits

Repetitions

GA

3500

-

-

-

GA &QNA

50

100

-

-

method

Calc. time

2min 30s 4h 54min

F.

27,7 1,7e-6

As it is can be clearly seen in table IV, the proposed methodology is consistently better than the other three methods. The final value of the objective function on the three numerical applications makes clear the advantage of the proposed methodology (Figure 5).

Fitness Function

1,00E+06

1,00E+03

1,00E+00 Fig. 6. Initial and optimum object position 1,00E-03

1,00E-06 0

1

2

3

4

5

θ0 α0 a0

(°) (°) mm

d1 d2 d3 θ4 θ5

Start point mm mm mm (°) (°)

d1 d2 d3 θ4 θ5


d1 d2 d3 θ4 θ5


d1 d2 d3 θ4 θ5


Time (h) GA

GA & QNA

Fig. 5. The minimum value of fitness function

The values of the variables corresponding to the optimum value of the objective function are reached and depicted in Table V. In Table VI the best probe position for surface scanning is presented. On Fig. 6 the initial position of the objects is shown, together with the best position calculated by the proposed method. Figure 7 provides a measurement set up, taking into account the optimal positioning and the optimal measurement sequence. The measurement starts from the surface on the left side of the object (top left) and after a complete surface scan the measurement continues to the front inclined surface (top right), then to the right side surface (lower right) and finally to the upper end surface (bottom left). Thereby, the overall motion trajectory from one surface to the other is minimized and the total measurement time is significantly reduced. Copyright © 2017 Praise Worthy Prize S.r.l. - All rights reserved

TABLE V BEST PARAMETERS Base Position 90 90 419,883 Surface 1 332,702 d1 1128,635 d2 801,981 d3 -90 θ4 0 θ5 Surface 2 430,704 d1 1101,013 d2 805,527 d3 180 θ4 45 θ5 Surface 3 359,722 d1 1029,101 d2 803,093 d3 -90 θ4 90 θ5 Surface 4 358,399 1136,426 820,097 90 90

d1 d2 d3 θ4 θ5

End Point mm mm mm (°) (°)

389,319 1030,263 804,64 -90 0


449,498 1059,534 832,038 180 45


439,113 1031,433 836,134 -90 90


439,113 1126,819 836,134 90 90


148

Scannin g Points


Start End Start End Start End Start End

TABLE VI BEST PROBE POSITION FOR SURFACE SCANNING Cοordinates Angles X Y z Α Β (mm) (mm) (mm) (°) (°) Surface 1 359,362 -386,702 -572,161 0 -90 457,734 -443,319 -574,82 0 -90 Surface 2 386,985 -382,173 -533,237 45 180 428,468 -400,986 -559,748 45 180 Surface 3 313,899 -413,722 -428,273 90 -90 311,567 -493,113 -461,314 90 -90 Surface 4 496,574 -412,399 -445,277 90 90 506,18 -493,113 -461,314 90 90

we get the matrix: − =

145 − + 1488 − − 145 − 54 (7) 374,82 − 145 − 1

− 0

0 0

0

(where ci = cosθi and si = sinθi). The initial data are: = 145

−

=−

+ 1488

(8)

− 54

(9)

− 145

= 374,82 − 145

−

(10)

=

(11)

=−

(12)

From (11) and (12) we have: =

−

(13)

and: =

+

(14)

− 54 −

(15)

+ 1488 −

(16)

From (9) we have: = −145 From (8) we have: = 145 From (10) we have: = 374,82 − 145

−

(17)

From (13) and (17) we have the position of the CMM for a given input position.

Pin position 1η 2η 3η 4η 5η 6η 7η 8η

Fig. 7. Optimal Measurement sequence

VI.

Mathematical Model

To verify the above data the reverse kinematic model of the system has to be solved. The reverse kinematic model computes the d1, d2 d3 parameters and the rotation C4 and C5 of the CMM for specify position of the end effector. Setting θ0 = 0 °, α0 = 90 ° and a0 = 1488 mm, which are the constant values of the base of the CMM system

TABLE VII OPTIMAL VALUES FOR THE VARIABLES Variables θ4(°) θ5(°) d1(mm) d2(mm) 0 0 332,702 1128,6350 0 0 389,319 1030,2629 0 45 430,704 1101,0129 0 45 449,499 1059,5345 0 90 359,722 1029,1012 0 90 439,113 1031,4332 0 90 358,400 1136,4262 0 90 439,113 1126,8192

d3(mm) 801,981 804,640 805,527 832,038 803,093 836,134 820,097 836,134

Comparing the values in Table VII with the values in Table V, it is clear that there is a deviation of the joint values θ4. This is because the tangent of 0 ° and 180 ° are 0, the tangent of 90° and -90° is + inf and -inf



149


[11] J.Y. Lin, P. Murugappan, A new algorithm for determining a collision-free path for a CMM probe, Int. J. Mach. Tools Manuf 39 (1999) 1397–1408. [12] Z.C. Lin, C.C. Chen, Collision-free path planning for coordinate measurement machine probe, Int. J. Prod. Res. 39 (9) (2001) 1969–1992. [13] Y. Wu, S. Liu, G. Zhang, Improvement of coordinate measuring machine probing accessibility, Precis. Eng. 28 (2004) 89–94. [14] B.J. Alvarez, P. Fernandez, J.C. Rico, S. Mateos, C.M. Suarez, Accessibility analysis for automatic inspection in CMMs by using bounding volume hierarchies, Int. J. Prod. Res. 46 (20) (2008) 5797–5826. [15] J.C. Rico, G. Valino, S. Mateous, E. Cuesta, C.M. Suarez, Accessibility analysis for star probes in automatic inspection of rotational parts, Int. J. Prod. Res. 40 (6)(2002) 1493–1523.

respectively. In other words, we are not able to calculate with a numerical method the value of the joint. But having achieved very good approximation of the remaining variables, undoubtedly the test is successful and that the values of the variables correspond to reality.

VII.

Conclusion

The current paper determines the optimum positioning of an object and the optimum measurement sequence of its surfaces, when a hybrid algorithm is used. The objective function calculates the sum of the deviations between the pre-required placements of the sensor probe of a coordinate measuring machine and the calculated placements, as well as the total time of movement between the surfaces, when the constraints of variables limits are given. The algorithm can be easily modified to fit on any object and type of CMM, which is a great advantage of the proposed methodology in comparison with the analytical methods. Compared with existing methodologies the optimization of the objective function is improved in every step of the hybrid method. The most significant improvement of the proposed algorithm is noted in the reduction in the data processing time and the time to export the results.

Authors’ information 1,2

Laboratory for Machine Tools and Manufacturing Engineering, Mechanical Engineering Department, Aristoteles University of Thessaloniki, Greece. E-mails: [email protected] [email protected] 3

Department of Automation, Technological Educational Institution of Thessaloniki, Greece. E-mail: [email protected] Professor Gabriel Mansour obtained his Dipl. and his Ph.D. in Mechanical Engineering from the Aristoteles University of Thessaloniki. From 2010 he is Associate Professor in Laboratory of Machine Tools and Manufacturing Engineering at Department of Mechanical Engineering, Polytechnical School of the Aristoteles University of Thessaloniki, His research interests include machine tools foundation, CAD/CAM, mechanisms, robotics, vibration measurements, 3D measurements, reverse engineering, Nanocomposite and composite materials

References [1]

D. Sagris, S. Mitsi, K.-D. Bouzakis, G. Mansour, Optimum geometric design for robot arm with geometric restrictions by means of a hybrid algorithm, Journal of the Balkan Tribological Association, Vol. 18, No 3, 325-333 (2012) [2] G. Mansour, S. Mitsi, K.-D.Bouzakis, D. Sagris, E. Varitis, Developed hybrid genetic algorithm for optimizing reverse engineering methods, International Journal of Modern Manufacturing Technology, ISSN 2067 3604, Vol II No. 1/2010. [3] G. Mansour, A. Tsagaris, D. Sagris, CNC machining optimization by genetic algorithms using CAD based system, 3rd International Conference on Diagnosis and Prediction in Mechanical Engineering Systems DIPRE 12. [4] L. Qu, G. Xu, G. Wang, Optimization of the measuring path on a coordinate measuring machine using genetic algorithms, Elsevier Measurement 23 (1998) 159-170. [5] Syed HammadMian, Abdulrahman Al-Ahmari, Enhance performance of inspection process on Coordinate Measuring Machine, Elsevier Measurement 47 (2014) 78-91. [6] Steven N. Spitz, Aristides A. G. Requicha, Multiple-Goals Path Planning for Coordinate Measuring Machines, Programmable Automation Laboratory University of Southern California, National Science Foundation under grant DMI-96-34727. [7] C. G. Lu, D. Morton, M. H. Wu and P. Myler, Genetic Algorithm Modeling and Solution of Inspection Path Planning on a Coordinate Measuring Machine (CMM), Int J AdvManufTechnol (1999) 15:409–416 [8] Slavenko M. Stojadinovic, Vidosav D. Majstorovic, Numan M. Durakbasa, Tatjana V. Sibalija, Towards an intelligent approach for CMM inspection planning of prismatic parts, Measurement, Volume 92, October 2016, Pages 326-339. [9] G.X. Zhang, S.G. Liu, X.H. Ma, J.L. Wang, Y.Q. Wu, Z. Li, Towards the Intelligent CMM, CIRP Annals - Manufacturing Technology, Volume 51, Issue 1, 2002, Pages 437-442 [10] C.Y. Hwang, C.Y. Tsai, C.A. Chang, Efficient inspection planning for coordinate measuring machines, Int. J. Adv. Manuf. Technol. 23 (2004) 732–742.

Dr. Eng. Sagris Dimitrios obtained his PhD from the Aristotle University of Thessaloniki in 2008. During the period 2008-2012 worked as Research Assistant in the Laboratory for Machine Tools and Manufacturing Engineering of Mechanical Engineering Department of Aristotle University of Thessaloniki. Since 2010 he is a Senior Research Assistant in the Mechanical Engineering Department of Technological Education Institute of Central Macedonia in Serres. His PhD Thesis is related to robotics optimization. His research interests include robotics, automations and mechanisms, reverse engineering, optimization processes and algorithms. Assistant Professor Apostolos Tsagaris received the Bachelor degree in Automation Engineer (Alexander Technological Educational Institute of Thessaloniki-ATEI, 1994), MSc in Design of interactive and industrial products and systems (Aegean University, 2005), Msc in Mechatronics (University of polytechnic from Catalonia – Barcelona in corporation with T.E.I of West Macedonia, 2007), Med in Adult Education (Open university of Patras, 2010) and his Ph.D. degree in Human Mechatronic System Interaction in Applied Informatics department from University of Macedonia (Thessaloniki-Greece) in 2014. He is currently a Assistant Professor in Robotics and CAD/CAM/CAE in Automation Department in ATEI(Greece).His research interests include automation system integration, mechatronic and robotic systems, CAD/CAM, reverse engineering technologies.



150


Neural Network Model to Predict Exhaust Emissions on a Stationary Diesel Engine Operating with Castor-Oil-Plant Biodiesel Fuel F. O. Narváez Argoty, A. Lyons Cerón, F. E. Sierra Vargas Abstract – A stationary diesel engine connected to an electricity generator was set up to determine the exhaust emissions of polluting gases and opacity when the engine is operated at different engine loads and with different castor-oil-plant biodiesel mixtures, as an alternative to diesel fuel and palm-oil biodiesel blends. The data was employed to model a feed-forward multilayer neural network, using the software NNModel to train the neural network, and predict the behavior of emissions from the combustion of the fuel based on two input variables: the engine load and the castor-oil plant biodiesel fuel mixture. The results of NNModel were compared against experimental data and analyzed to identify how the engine emissions are affected by the implementation of an alternative biodiesel fuel. Copyright © 2017 Praise Worthy Prize S.r.l. - All rights reserved.

Keywords: Biodiesel, Castor Oil, Diesel Engine, Emissions, Neural Network

This means that almost all petroleum consumed nowadays is derived from old deposits. Considering that the economic growth in developing countries is increasing the petroleum demand, shortages and price increases are leading to an alarming situation all over the world. An alternative fuel to gas-oil petroleum derivate is biodiesel, made by 100% biodegradable and natural resources and proved to have lower polluting emissions. As stated by the International Energy Agency (IEA), the production of biodiesel has increased more than four times since 2000. Nowadays, biodiesel is produced in more than 30 countries all over the world from different sources, such as animal fat wastes, vegetable oils, plants, algae oil and frying oil waste. In Latin America, including Colombia the implementation of biodiesel fuels is crucial for the future years, as it encourages the decrease in fossil fuel consumption and also reduces the amount of imported fossil fuels. The main contributions on research related with production and implementation of biodiesel in Latin America have been made in Brazil, Mexico, Argentina, Uruguay and Peru. In Colombia, the most relevant research on biodiesel fuel have been made by Agudelo [1], who made a research on the advantages in mechanical and environmental performances of vegetable oil biodiesels, Acevedo [2], who analyzed emissions of palm-oil and petrochemical biodiesel mixtures, Agudelo [3], who made research on quality and characterization of biodiesel in Colombia, and Fajardo [4] who studied the performance of diesel engines working with biodiesel. In big cities, such as Bogota, most pollution gases are emitted by diesel engine vehicles that operate for public transportation and cargo. Studies have shown how the air in Bogotá is barely affected by nitrous oxides and carbon

Nomenclature ACPM BH E FFNN HC Ln NN OP R2

Diesel fuel derived from petroleum Biodiesel mixture Error Feed forward neural network Hydrocarbons Neural network layer Neural network Opacity Coefficient of determination

I.

Introduction

Colombia, as many other developing countries, has developed a plan to promote sustainable energy sources and decrease environmental pollution caused by fossil fuel consumption in the transport sector. Nevertheless, due to the elevated costs of small scale biodiesel production, palm oil biodiesel is the only fuel that is commercialized in the country. However, diesel engines working with palm oil biodiesel have presented some disadvantages because this fuel tends to crystalize at ambient temperatures in cities like Bogota, with low pressure and temperature conditions, and altitudes higher than 2000-2500m AMSL. For this reason, alternative biodiesel fuels such as castor-oil-plant biodiesel can offer a solution for biodiesel implementation in cities with this altitude conditions, provided that these fuels meet the standardized requirements and quality standards for biofuels. As stated by the Association for the Study of Peak Oil (ASPO), since the 1980s, petroleum consumption has surpassed petroleum deposits discoveries. Copyright © 2017 Praise Worthy Prize S.r.l. - All rights reserved


151

F. O. Narváez Argoty, A. Lyons Cerón, F. E. Sierra Vargas

monoxide and is highly affected by particulate matter [5]. To mitigate this problem, the Colombian government aims to implement over 26 states diesel fuel derived from petroleum (ACPM) mixed with 5% biodiesel fuel. For this reason, the government promotes research projects related with production and implementation of biodiesel fuels. It has been demonstrated that engines working with palm oil biodiesel can’t operate properly at low pressure and temperature conditions. For this reason, castor-oilplant biodiesel offers a promising alternative for biodiesel mixture fuels. As stated by Franco [6], castoroil-plant has a promising prospect, as its oil is of great value on the international market, and this plant has a great agricultural production potential, where a whole new agro industrial activity could be created with a technicized production of this plant. Some studies were conducted using castor-oil-plant biodiesel as an alternative fuel for diesel engines in Medellin; however there aren’t any known studies for castor-oil-plant biodiesel in Bogota [7]. The representation of a system or physical process such as the emissions made by a diesel engine [16]-[17] can be obtained as a theoretical model, or as a physical experiment.. In order to obtain models, different system identification techniques can be implemented, including differential equations. The main problem is that these techniques are limited, as many systems have complex and non-linear behaviors with several perturbations. As an alternative, artificial neural networks are used to identify different kinds of dynamic systems. In the recent years, neural networks have been implemented as virtual sensors that predict data for diesel engine emissions reducing costs on the engine control and research. Maaß [8] made a study on how a non-linear autoregressive neural network was used to predict the smoke emissions of a diesel engine operated in a nonroad transient cycle. Furthermore, Brace [9] applied neural networks on a diesel engine to predict transient changes in emission levels under different operating conditions, predict emissions during a legislative driving cycle, and to minimize fuel consumption by setting ideal points for engine speed and load. Neural networks can also be implemented to predict torque, power, fuel consumption, and pollutant emissions on turbo charged diesel engines [10], and to predict a diesel engine performance and emissions when it works with waste cooking biodiesel [11]. This paper presents an experimental study and a supervised feed-forward neural network model of polluting emissions of a stationary diesel engine connected to an electricity generator. Tests were performed in Bogota (2600m AMSL) under eight different load conditions, using different blends between castor-oil-plant biodiesel fuel and ACPM. The experiments were made in the Laboratory of Automotive Dynamic Tests of the Transportation Technologies Center at the SENA institute in Bogota. Castor-oil-plant biodiesel fuel was produced and characterized in the

Biofuels laboratory of the Universidad Agraria and the Chemical Engineering Laboratories of the Universidad Nacional de Colombia. Five different biodiesel mixtures (BH %Vol) were used; BH0,0, BH7,5, BH12,5, BH17,5 and BH22,5 at the first stage to train the neural network, and at the second stage 3 different biodiesel mixtures (BH5,0, BH10,0, BH15,0) were used to compare the experimental results with the predictions made by a feedforward multi-layered neural network modeled by NNModel to analyze CO2, CO, NO, NO2, Hydrocarbons (HC) and O2 emissions under different engine loads. The experimental results and the neural model predictions were compared; likewise, the relative errors, standard deviations and inputs and outputs sensibility were analyzed.

II.

Methodology

II.1.

Biodiesel Selection

Creating a Neural Network model which can predict the exhaust emissions of a diesel engine requires a proper identification of the fuel that is going to be used. Identifying the proper method to produce this fuel, the international policies which regulate quality and specifications of this type of fuel, and the characteristics and origin of the plant which is used to extract and produce this fuel is so relevant. Transesterification is an essential process to produce diesel fuels, therefore, it was necessary to make a research on the existing technologies for biodiesel production, this technologies can be classified by the raw material used, the catalytic system employed and the heating method used during the process. After understanding all the production methods, the most suitable method for Castor-oil-plant biodiesel production selected was Basic Homogeneous Transesterification To achieve the quality and specifications of diesel fuels, the most relevant international policies and regulations for diesel and biodiesel were consulted. The main policies are issued and constantly modified by the ASTM International and the CEN (European Committee for Standardization). The ASTM D975 policy on Standard specification for diesel fuel oils characterizes all biodiesel fuels and classifies them in 3 main groups; 1D, 2D and 3D based on their volatility and applications. The quality requirements for diesel fuels are regulated by the ASTM D 975-08. As for biodiesel the main regulation policies are the ASTM D 6751 and the EN 14214. Based on these policies, some developing countries have created their own regulations for Biodiesel, such as the NTC 5444 in Colombia. Biodiesel blends from B6 to B20 are regulated by the ASTM D7467-08. The main characteristics analyzed for Diesel are; viscosity, volatility, low temperature performance, lubricity, purity and self-ignition. Biodiesel policies regulate Flashpoint, cetane number, acidity, distillation, and contents of calcium, magnesium, sulfur, sodium, potassium, phosphorus, sulfated ash,



152

F. O. Narváez Argoty, A. Lyons Cerón, F. E. Sierra Vargas

TABLE II PROPERTIES OF CASTOR OIL FROM CHINA, ANALYSIS MADE BY GLOBE CHEMICALS GMBH Parameters Results Requirements Acid value 0,64 2 Max Pale yellow, Viscous Appearance Passes Liquid Color in 1 inch Lovibond cell 1,1 Y + 0,1 R 2,2 Y + 0,3 R Max Tintometer Iodine value 84,71 82-90 M.I.V 0,11% 0,25% Max Optical rotation 4,21 3,5 min Refractive index at 40°C 1,472 1,470-1,474 Saponification Value 179,81 177-185 Unsaponifiable matter 0,34% 1%max Weight/ml (at 30°C) 0,9572gm/ml 0,954-0,960 Specific gravity @30/30°C 0,954 0,954-0,960 Hydroxyl value 160 160-168 Heavy metals N/a N/a

sediments and water [3]. Castor-oil-plant was selected as the feedstock to produce biodiesel, a high quality oil plant that can be found in intermediate and high temperature climate zones with the following characteristics [12]: • Small size plants up to 1.6m high • Medium size plants 2-2.5m high • Maximum size 12m high • Climate conditions from 20°C to 28°C • Altitude conditions from sea level to 2800m AMSL • Oil content in seeds from 45-55% seed weight II.2.

Biodiesel Preparation

A research was made on the most common methods used to extract oil from castor-oil-plants in order to analyze these methods and select the most suitable method for the preparation of the biodiesel fuel used on the stationary engine. There were 2 main methods found for the oil extraction. The first, a 2-stage mechanical extraction method at low temperatures and then high temperatures, creating light and bright color oil with low acidity. The second method creates dark and high acidity oil. As a way to optimize the biodiesel fuel production, a preliminary production method was made with smallscale laboratory equipment. After this preliminary method 2 different methods were used to obtain castoroil-plant biodiesel fuel. The main properties of castor-oilplant biodiesel can be found on Table I [13]. TABLE I CASTOR-OIL-PLANT BIODIESEL PROPERTIES Properties Units Density kg/m3 Viscosity at 40°C mm2/s Cetane number cetanes Ignition point °C Pour point °C Residual carbon mass % Heat of combustion MJ/kg Distillation °C

TABLE III OPTIMAL CONDITIONS FOR CASTOR-OIL-PLANT BIODIESEL FUEL PRODUCTION Condition Value Temperature 50°C Molar rate (methanol-oil) 6:1 NaOH concentration 0.5% Washing agent Water

For the production of castor-oil-plant- biodiesel fuel the two methods employed used castor oil imported from China and different catalysts and alcohols. The overall process of biodiesel production can be seen on Table IV. TABLE IV CASTOR-OIL-PLANT BIODIESEL PRODUCTION PROCESSES Biodiesel 1 (NaOH Biodiesel 2 (KOH and Methanol) and Ethanol) Catalyst 0,5% weight NaOH 1% weight KOH Alcohol 99,5% Methanol 99,8% Ethanol Washing agent water water Transesterification 60°C 83°C temperature Mechanical agitation, high Mechanical temperature oil, cold agitation, hot water, Temperature Control water, electric semiautomatic resistances and control system semiautomatic control system Phase separation Gravity (12hours Gravity (24hour (biodiesel-glycerol) rest) rest) Transesterification 20L 25L equipment capacity

Value 926 14,9 38 84 -18