Engineering_02_08_2010092609263635.pdf

transcript

Engineering

www.scirp.org/journal/eng/

Prof. David L. Carroll Wake Forest University, USA

ISSN: 1947-3931 Volume 2, Number 8, August 2010

9 7 7 1 9 4 7 3 9 3 0 0 5 80

ISSN: 1947-3931

ISSN: 1947-3931 (Print), 1947-394X (Online)

http://www.scirp.org/journal/H eng

Prof. David L. Carroll Wake Forest University, USA

Journal Editorial Board

Editor-in-Chief

Editorial Board

Dr. Rehan Ahmed

Prof. MohD A. M. Al-Nimr

Prof. Jan Awrejcewicz

Prof. Shahnor Basri

Prof. Alain. Bernard

Dr. P. Chandramohan

Dr. Hongyang Chen

Prof. Ming Chen

Dr. Rahat Iqbal

Prof. Shouetsu Itou

Prof. Jae Moung Kim

Prof. Chui-Chi Lee

Prof. Yong-Gang Lv

Prof. John Marsh

Dr. Mehrdad Massoudi

Dr. Koduri Venkata Surya Ramam

Prof. Sergey M. Smolskiy

Prof. Hongbin Sun

Prof. Baolin Wang

Prof. Chengshan Wang

Prof. Gaofeng Wang

Prof. Zhao Xu

Dr. Wei Yan

Dr. V.P. Yashnikov

Prof. Tsutomu Yoshihara

Prof. Xiangjun Zeng

Dr. Hongyu Zhang

Prof. Luowei Zhou

Heriot-Watt University, UK

Jordan University of Science and Technology, Jordan

Technical Univers ity of Lodz, Poland

Putra University, Malaysia

Ecole Centrale de Nantes, France

Anna University, India

The University of Tokyo, Japan

Southeast University, China

Coventry University, UK

Kanagawa University, Japan

INHA University, Korea (South)

SHU-TE University, Taiwan (China)

Chongqing University, China

IEEE Photonics Society, UK

U. S. Department of Energy/NETL-Pittsburgh, USA

Universidad of Concepcion, Chi le

Moscow Power Engineer ing University, Russia

Tsinghua University, China

Harbin Inst i tute of Technology, China

Tianj in University, China

Wuhan University, China

Technical Univers ity of Denmark, Denmark

Trend Micro, USA

Russian Academy of Sciences, Russia

Waseda University, Japan

Changsha University of Science and Technology, China

Ceres Inc., Thousand Oaks, USA

Chongqing University, China

Yanna Li eng@scirp.org Scientific Research Publishing, USA

Editorial Assistant

Engineering, 2010, 2, 559-672 Published Online August 2010 in SciRes (http://www.SciRP.org/journal/eng/)

Copyright 2010 SciRes. ENG

TABLE OF CONTENTS

Volume 2 Number 8 August 2010 Experimental Comparative and Numerical Predictive Studies on Strength Evaluation of Cement Types: Effect of Specimen Shape and Type of Sand

H. Hodhod, M. A. M. Abdeen559

Hydrogen Pick up in Zircaloy-4: Effects in the Dimensional Stability of Structural Components under Nuclear Reactor Operating Conditions

P. Vizcano, C. P. Fagundez, A. D. Banchik573

Electrochemical Generation of Zn-Chitosan Composite Coating on Mild Steel and its Corrosion Studies K. Vathsala, T. V. Venkatesha, B. M. Praveen, K. O. Nayana580

Tunable Erbium-Doped Fiber Lasers Using Various Inline Fiber Filters S.-K. Liaw, K.-C. Hsu, N.-K. Chen585

Behaviour of a Composite Concrete-Trapezoidal Steel Plate Slab in Fire T. Hozjan, M. Saje, I. Planinc, S. Srpi, S. Bratina594

The Effect of Initial Oxidation on Long-Term Oxidation of NiCoCrAlY Alloy C. Zhu, X. Y. Wu, Y. Wu, G. Y. Liang602

Highly Nonlinear Bending-Insensitive Birefringent Photonic Crystal Fibres H. Ademgil, S. Haxha, F. AbdelMalek608

Progress in Antimonide Based III-V Compound Semiconductors and Devices C. Liu, Y. B. Li, Y. P. Zeng617

Lie Group Analysis for the Effects of Variable Fluid Viscosity and Thermal Radiation on Free Convective Heat and Mass Transfer with Variable Stream Condition

P. Loganathan, P. P. Arasu625

Statistical Modeling of Pin Gauge Dimensions of Root of Gas Turbine Blade in Creep Feed Grinding Process

A. R. Fazeli635

Wind Turbine Tower Optimization under Various Requirements by Using Genetic Algorithm S. Yldrm, . zkol641

A Device that can Produce Net Impulse Using Rotating Masses C. G. Provatidis648

Computer-Aided Solution to the Vibrational Effect of Instabilities in Gas Turbine Compressors E. A. Ogbonnaya, H. U. Ugwu, C. A. N. Johnson658

Flipped Voltage Follower Design Technique for Maximised Linear Operation C.-M. Chen, K. Hayatleh, B. L. Hart, F. J. Lidgey665

Wire Bonding Using Offline Programming Method Y. L. Foo, A. H. You, C. W. Chin668

Engineering

Journal Information SUBSCRIPTIONS

Engineering (Online at Scientific Research Publishing, www.SciRP.org) is published monthly by Scientific Research Publishing, Inc., USA.

Subscription rates: Print: $50 per issue. To subscribe, please contact Journals Subscriptions Department, E-mail: sub@scirp.org

SERVICES

Advertisements Advertisement Sales Department, E-mail: service@scirp.org

Reprints (minimum quantity 100 copies) Reprints Co-ordinator, Scientific Research Publishing, Inc., USA. E-mail: sub@scirp.org

COPYRIGHT

Copyright2010 Scientific Research Publishing, Inc.

All Rights Reserved. No part of this publication may be reproduced, stored in a retrieval system, or transmitted, in any form or by any means, electronic, mechanical, photocopying, recording, scanning or otherwise, except as described below, without the permission in writing of the Publisher.

Copying of articles is not permitted except for personal and internal use, to the extent permitted by national copyright law, or under the terms of a license issued by the national Reproduction Rights Organization.

Requests for permission for other kinds of copying, such as copying for general distribution, for advertising or promotional purposes, for creating new collective works or for resale, and other enquiries should be addressed to the Publisher.

Statements and opinions expressed in the articles and communications are those of the individual contributors and not the statements and opinion of Scientific Research Publishing, Inc. We assumes no responsibility or liability for any damage or injury to persons or property arising out of the use of any materials, instructions, methods or ideas contained herein. We expressly disclaim any implied warranties of merchantability or fitness for a particular purpose. If expert assistance is required, the services of a competent professional person should be sought.

PRODUCTION INFORMATION

For manuscripts that have been accepted for publication, please contact: E-mail: eng@scirp.org

Engineering, 2010, 2, 559-572 doi:10.4236/eng.2010.28072 Published Online August 2010 (http://www.SciRP.org/journal/eng).


Experimental Comparative and Numerical Predictive Studies on Strength Evaluation of Cement Types: Effect of

Specimen Shape and Type of Sand

Hossam Hodhod1, Mostafa A. M. Abdeen2 1Department of Structural Engineering, Faculty of Engineering, Cairo University, Giza, Egypt

2Department of Engineering Mathematics & Physics, Faculty of Engineering, Cairo University, Giza, Egypt E-mail: {hossamhodhod, mostafa_a_m_abdeen}@hotmail.com

Received May 6, 2010; revised July 23, 2010; accepted July 25, 2010

Abstract Quality of cement is evaluated via group of tests. The most important, and close to understanding, is the compressive strength test. Recently, Egyptian standards adopted the European standards EN-196 and EN-197 for specifying and evaluating quality of cements. This was motivated by the large European invest-ments in the local production of cement. The current study represents a comparative investigation, experi-mental and numerical, of the effect of different parameters on evaluation of compressive strength. Main pa-rameters are shape of specimens and type of sand used for producing tested mortars. Three sets of specimens were made for ten types of cements. First set were 70.6 mm cubes molded according to old standards using single sized sand. Second group were prisms molded from standard sand (CEN sand) according to the recent standards. Third group were prisms molded from local sand sieved and regenerated to simulate same grading of CEN sand. All specimens were cured according to relevant standards and tested at different ages (2,3,7,10 and 28 days). Results show that CEM-I Type of cement does not fulfill, in all of its grades, the strength re-quirements of Ordinary Portland cement OPC specified in old standards. Also, the use of simulated CEN sand from local source gives strengths lower than those obtained using standard certified CEN sand. A lim-ited number of tests were made on concrete specimens from two most common CEM-I types to investigate effect on concrete strength and results were also reported. Numerical investigation of the effect of specimen shape and type of sand on evaluation of compressive strength of mortar specimens, presented in the current study, applies one of the artificial intelligence techniques to simulate and predict the strength behavior at different ages. The Artificial Neural Network (ANN) technique is introduced in the current study to simulate the strength behavior using the available experimental data and predict the strength value at any age in the range of the experiments or in the future. The results of the numerical study showed that the ANN method with less effort was very efficiently capable of simulating the effect of specimen shape and type of sand on the strength behavior of tested mortar with different cement types. Keywords: Cement Type, Sand Type, Mortar Specimen, Strength, Modeling, Artificial Neural Network

1. Introduction

For decades, engineers used to apply cement based on certain classification [1-3]. This classification refers to its composition and consequently relevant properties. Among these properties, strength was the main target of using a specific type of cement. Ordinary Portland ce-ment (OPC), sulphate resisting cement (SRC) and white cement share almost same values for compressive strength at different ages. One type: namely rapid hard-

ening cement had the higher early strength than others. Recently, end of the year 2006, the Egyptian standards [4] decided to adopt the European standards EN196 & EN 197 [5] for producing, specifying and testing almost all types of cements. The new standard took the designation ES4756 and included all types of cements but SRC. The new standard included a drastic change in specifying cement types, and appeared ambiguous in many aspects since it added ranks and rate of hardening for the same composition of cement. This raised many questions

H. HODHOD ET AL.


560

about the actual composition of cement and its properties. Also, questions were raised about the properties of cem-titious mixes for which cement is used and whether the correlations between properties and type of cement will remain the same or not. Besides, methods of testing ce-ment to evaluate its compressive strength were changed from using cubic specimens (70.6 mm side length) [6] to part of prism (with 40 mm square cross section). More-over, the standard dictated the use of specific type of sand, which is not available locally, for making mortar specimens. This sand must be procured from certified suppliers and is called CEN sand. Such a condition raised a question about the role of this sand in hydration process and strength development too. All these ques-tions motivate the need for research to clarify nature of new cement types and declare their properties and effect on properties of cementitious mixes. One local attempt [7] was made and yielded that new standards are efficient. However, the contradictions between test results of the study and the known size effect rule urge the need for more investigation.

Since the experimental work needs a lot of effort, time and money, which is quite clear from the literature men-tioned previously, the need for utilizing new methodolo-gies and techniques to reduce this effort, save time and money (and at the same time preserving high accuracy) is urged. Artificial intelligence has proven its capability in simulating and predicting the behavior of the different physical phenomena in most of the engineering fields. Artificial Neural Network (ANN) is one of the artificial intelligence techniques that have been incorporated in various scientific disciplines. Solomatine and Toorres [8] presented a study of using ANN in the optimization loop for the hydrodynamic modeling of reservoir operation in Venezuela. Kheireldin [9] presented a study to model the hydraulic characteristics of severe contractions in open channels using ANN technique. The successful results of his study showed the applicability of using the ANN ap-proach in determining relationship between different parameters with multiple input/output problems. Abdeen [10] developed neural network model for predicting flow characteristics in irregular open channels. The developed model proved that ANN technique was capable with small computational effort and high accuracy of predict-ing flow depths and average flow velocities along the channel reach when the geometrical properties of the channel cross sections were measured or vice versa. Al-lam [11] used the artificial intelligence technique to pre-dict the effect of tunnel construction on nearby buildings which is the main factor in choosing the tunnel route. Allam, in her thesis, predicted the maximum and mini-mum differential settlement necessary precautionary measures. Park and Azmathullah et al. [12] presented a study for estimating the scour characteristics downstream of a ski-jump bucket using Neural Networks (NN). Ab-deen [13] presented a study for the development of ANN

models to simulate flow behavior in open channel in-fested by submerged aquatic weeds. Mohamed [14] pro-posed an artificial neural network for the selection of optimal lateral load-resisting system for multi-story steel frames. Mohamed, in her master thesis, proposed the neural network to reduce the computing time consumed in the design iterations. Abdeen [15] utilized ANN tech-nique for the development of various models to simulate the impacts of different submerged weeds' densities, dif-ferent flow discharges, and different distributaries opera-tion scheduling on the water surface profile in an ex-perimental main open channel that supplies water to dif-ferent distributaries. 2. Problem Description To study the effect of specimens shape and types of sand used for producing tested mortars on evaluation of com-pressive as well as flexural tensile strengths, experimen-tal and numerical techniques will be presented in this study. The experimental program and its results will be described in detail in the following sections. After that, numerical approach will be discussed to show the effi-ciency of numerical techniques. The numerical models presented in this study utilized Artificial Neural Network technique (ANN) using the data of experiments and then can predict the strength value in the range of the experi-ment or in the future. 3. Experimental Program The experimental program focuses on evaluating com-pressive strength of mortar made from new cement types. Ten types of cements with different grades and rate of hardening were procured from local market in Egypt. Compressive strength was evaluated for each type using the cubic specimens (70.6 mm side length) and using the testing of part of prism (40*40*160 mm). The last me-thod was employed twice. First with local sand following the same grade specified in ES4756 (and EN 196), and second with certified CEN sand according to same stan-dards. Specimens were tested at ages of 2, 3, 7, 10 and 28 days.

Concrete mixes with same proportions were cast from different types of cements. Slump and compressive strength were measured for each mix to investigate the effect of type of cement on concrete properties. Com-pressive strength was measured at 3, 7 and 28 days.

4. Materials and Specimens

Constituents for mortar and concrete mixes were:

4.1. Water

Tap water was used for mixing and curing of all speci-

H. HODHOD ET AL.


561

mens. 4.2. Cement Ten types of cement were used. All were supplied in bags carrying the symbols of both ES4756 and EN-197. They were all produced locally in Egypt by different Cement Companies. The ten types covered CEM I (ordi-nary Portland cement) with different grades and rates of hardening. The types also included white cement, sul-phate resisting cement (SRC) and CEM II type cements. Table 1 shows the investigated types of cement. 4.3. Sand Two types of sand were used for the current study: CEN sand that was imported from France in sealed transparent bags (Figure 1), and local siliceous sand. Local siliceous sand was used in its natural grading (Figure 2) for cast-ing concrete. This sand was sieved to get the single size sand required for testing mortar cubes according to old ES (still effective as part of local code of practice ECP 203/2001 app.3. The local sand was also used to regener-ate the CEN sand by collecting different sizes in the per-centages specified in EN-196. The grading of this regen-erated sand, and limits of CEN sand, are shown in Fig-ure 3. 4.4. Gravel Local siliceous gravel was used for casting concrete spe-cimens. Gravel has a maximum nominal size of 20 mm. 4.5. Specimens (Cubes and Prisms) Standard cubes with 70.6 mm side were used for evaluat-

ing compressive strength of mortar with mix proportions of water: cement: sand = 0.4:1:3 by weight. Sand was 0.6 0.85 mm local sand. Constituents were mixed manually. Steel molds were used for casting. The other two sets of specimens were prisms (40*40*160 mm) cast from mix-tures with proportions of water: cement: sand = 0.5:1:3 by weight.

For one set, standard CEN sand was used for casting. For the other set, regenerated local sand with grading

Figure 1. Bags of CEN sand.

0

20

40

60

80

100

0.01 0.1 1 10Perc

enta

ge P

assi

ng

Sieve Opening Size (mm)

Figure 2. Grading of local sand used in concrete mixes.

Table 1. Investigated types of cement.

Strength Evaluation (on Mortar) Flexure Compn Flexure CompnNo Type of Cement Manufacturer Cube Compn. Local sand(*) CEN Sand

1 CEM I-52.5N SINAI

2 CEM I-42.5N (SRC)(**) El-MASRIYA 3 CEM I-42.5N (SRC) HELWAN 4 CEM I-42.5N El-MASRIYA

5 CEM I-42.5R HELWAN

6 CEM I-42.5N (White) HELWAN 7 CEM I-32.5R ELKAUMIYA (NCC) 8 CEM I-32.5R (SRC) ASSIUT (CEMEX) 9 CEM II-B-S 32.5N HELWAN 10 CEM II-B-L -32.5N ELKAUMIYA (NCC)

(*) Sand having a grading similar to CEN sand. (**) This cement will be denoted in figures as SRC-1.

H. HODHOD ET AL.


562

0

20

40

60

80

100

0.01 0.1 1 10

Sieve Opening Size - mm

Perc

enta

ge P

assi

ng

CEN Upper LimitCEN Lower LimitRegenerated Sand

Figure 3. Grading limits of CEN sand and grading of locally regenerated sand. similar to CEN was used. Constituents were mixed me-chanically using 5 liter mixer. Steel molds were used for casting. For all sets, specimens were compacted using vibrator and left covered with impervious sheet for 24 hours. Then specimens were demolded and immersed in water till day of testing. Concrete cubes (with 150 mm side length) were cast to evaluate concrete strength. Mix proportions were water: cement: sand: gravel = 0.6:1: 1.5:3.0. Constituents were mixed mechanically using 140 liter tilting type mixer. Dry materials were mixed first for about one minute. Then, water was added gradually and mixing continued till uniform mix was obtained. Con-crete was cast in steel molds and compacted using a vi-brating table. Specimens were covered with plastic sheets for 24 hours. Then molds were removed and spe-cimens were wet cured till age of testing. 5. Test Results The test results are explained in the following sections. 5.1. Cement Setting Time Initial and final setting times measured for different types of cement are shown in Figure 4. One can see that the initial setting time ranges from 70 to 120 min. Final setting time ranges from 140 to 240 min. Generally, final setting time is almost double the initial setting time. The least setting time was recorded for CEM I 52.5 N and the longest setting time was recorded for CEM II B-S-32.5 N. Setting time increases as cement grade decreases, and SRC shows less setting time for same grade. Rate of set-ting (expressed by N or R after grade) does not seem to affect setting time results. Recorded values of initial and final setting times comply with limits of ES 4756 and EN-197.

5.2. Mortar Compressive Strength Compressive strength measured for all specimens and

Figure 4. Setting time of different types of cement.

types of cements are plotted versus time in Figure 5. One can see that cube specimens specified in old stan-dards give strength lower than part of prisms specified in the current standards. Large size of cubes helps reducing its strength as the grading of the single sized sand does. However, the low w/c ratio is supposed to help increas-ing the strength of cubes. This indicates that the effect of size and confinement of prism specimens and the grading and type of CEN sand could compensate for the increase of w/c ratio of the specimens.

There is a difference between results obtained for CEN sand and regenerated sand composed by adding the proper percentage of each size from local sand. CEN sand always gives higher strength. This implies that it is not only sand grading that contributes to the strength. Shape of particles and probably some chemical charac-teristics of sand may also contribute to this increase of strength. These last two points need more research for clarification.

It must be said that the strength of prisms does not ful-fill the requirements of cement grade for all types. The strength of prisms at 28 days reaches a percentage from 43 to 70% of corresponding grade of cement. Although the compaction was not made using a jolting table (as specified by current standard test method), this is not expected to yield such big difference for all types of ce-ment.

It is note-worthy that similar strength values were ob-tained for all rapid setting cements regardless of their grade. However, the normal setting CEM I 52.5 N gave the highest strength among all other cement mortars.

Strength factor, which is the ratio of 28 day strength to the strength at a specific age, is plotted in Figures 6-8 for different specimens and different types of cement. The small strength values for CEN sand prisms denote the rapid strength development of strength with this sand. However, for same sand, it seems that strength develop-ment does not follow the indication R & N for Type of cement, where smaller factors are observed for normal setting cements. Similar trend was found for other types of sand.

H. HODHOD ET AL.


563

Figure 5. Compressive strength of mortar specimens produced under different conditions, for different types of cement.

H. HODHOD ET AL.


564

Figure 6. Strength factor for mortar prisms produced from CEN sand (left: rapid setting cements, right: normal setting ce-ments).

Figure 7. Strength factor for mortar prisms produced from regenerated sand (left: rapid setting cements, right: normal set-ting cements).

Figure 8. Strength factor for mortar cubes (left: rapid setting cements, right: normal setting cements).

5.3. Mortar Tensile Strength Flexural tensile strength was measured for prism speci-mens since it is the first step in producing compression specimens. Measured flexural strength for all types of cements and for different sands are plotted in Figure 9. One can observe the effect of CEN sand in increasing strength of mortar. This effect confirms the above men-tioned need for investigation of particle shape and chem-ical reactivity of CEN sand.

One more finding can be found when tensile flexural strength is plotted versus compressive strength at differ-ent ages, as in Figure 10. It can be seen that there exists a

significant increase of tensile strength between 7 and 28 days. This could be observed for both types of sand. This implies that the correlation between flexural strength and compressive strength is significantly different at early and later ages.

5.4. Concrete Slump

Concrete slump measuring results are shown in Figure 11 for all cement types. One can identify 3 main ranges of slump: 0-40 mm, 40-80 mm, and 80-120 mm. First low range of slump was recorded for rapid setting ce-ments and 52.5 grade cement. Highest slump range was

H. HODHOD ET AL.


565

Figure 9. Flexural strength of mortar prisms produced different sand types, for different types of cement.

H. HODHOD ET AL.


566

Figure 10. Flexural strength vs. compressive strength (left: cen sand, right: regenerated sand).

Figure 11. Slump values for different types of cement.

observed for CEM II cements. The medium grade was observed for the rest normal setting CEM I type of ce-ment. Since the water consumption is related to rates of hydration and heat evolution. One can conclude that grade 52.5 has high rate of hydration. It is noteworthy that CEM I 52.5 R does not exist in local Egyptian mar-ket. The high slump of CEM II cement mixes can be correlated to low clinker content. 5.5. Concrete Compressive Strength Measured values of concrete compressive strength are plotted versus age, for all types of cement, in Figure 12. Generally, one can see in Figure 12 that the effect of cement grade can be distinguished in the limits where top curve belongs to grade 52. N and bottom curve belongs to grade 32.5 N. Curves for higher grades of cement are shown in Figure 13 Left. One can see that, up to 7 days all 42.5 grade cements show almost same strength. However, at later ages (28 days) the rapid setting type shows higher strength than the normal setting ones. One can also see that SRC cement show slightly higher strength than similar 42.5 N cements. Curves for low grade cements are shown in Figure 13 Right. The effect of setting rate can be identi-fied between 32.5 N and 32.5 R cements.

Still SRC cement shows higher strength at 28 days. Figure 14 shows strength development for rapid setting and normal setting cements, respectively. For rapid set-ting cements, there is no difference between early age

0

5

10

15

20

25

30

0 10 20 30

Age - days

Con

cret

e C

ompr

essi

ve S

tren

gth

(MPa

)

CEM I 52.5NCEM I 42.5N-SRCCEM I 42.5NCEM I 42.5RCEM I 42.5N whiteCEM I 32.5RCEM I 32.5R (SRC)CEM II-B-S32.5NCEM II B-L32.5N

Figure 12. Measured concrete compressive strength for different types of cement.

strength of different grades of cement. At 28 days, SRC of 32.5 grade yields same strength as 42.5 grade. For normal setting cements, there is a clear distinction be-tween strength of different grades at all ages. Strength ratio at 28 days is almost proportional to grade of cement. One can still observe that SRC show higher values of strength than other CEM I cements of same grade.

6. Numerical Model Structure

Neural networks are models of biological neural struc-tures. Briefly, the starting point for most networks is a model neuron as shown in Figure 15. This neuron is con-nected to multiple inputs and produces a single output. Each input is modified by a weighting value (w). The neuron will combine these weighted inputs with reference to a threshold value and an activation function, will de-termine its output. This behavior follows closely the real neurons work of the humans brain. In the network struc-ture, the input layer is considered a distributor of the sig-nals from the external world while hidden layers are con-sidered to be feature detectors of such signals. On the other hand, the output layer is considered as a collector of the features detected and the producer of the response.

6.1. Neural Network Operation

It is quite important for the reader to understand how the

H. HODHOD ET AL.


567

0

5

10

15

20

25

30

0 10 20 30

Age - days

Con

cret

e C

ompr

essi

ve S

tren

gth

(MPa

)

CEM I 52.5NCEM I 42.5N-SRCCEM I 42.5NCEM I 42.5RCEM I 42.5N white

0

5

10

15

20

25

30

0 10 20 30

Age - days

Con

cret

e C

ompr

essi

ve S

treng

th (M

Pa)

CEM I 32.5RCEM I 32.5R-SRCCEM II-B-S32.5NCEM II B-L32.5N

Figure 13. Measured concrete compressive strength (left: high grades of cement, right: low grades of cement).

0

5

10

15

20

25

30

0 10 20 30

Age - days

Con

cret

e C

ompr

essi

ve S

tren

gth

(MPa

)

CEM I 42.5RCEM I 32.5RCEM I 32.5R-SRC

0

5

10

15

20

25

30

0 10 20 30

Age - days

Con

cret

e C

ompr

essi

ve S

tren

gth

(MPa

)

CEM I 52.5NCEM I 42.5N-SRCCEM I 42.5NCEM I 42.5N whiteCEM II-B-S32.5NCEM II B-L32.5N

Figure 14. Measured concrete compressive strength (left: rapid setting cements, right: normal setting cements).

Figure 15. Typical picture of a model neuron that exists in every neural network. neural network operates to simulate different physical problems. The output of ach neuron is a function of its inputs (Xi). In more details, the output (Yj) of the jth neu-ron in any layer is described by two sets of equations as follows:

j i ijU X w (1) and

j th j jY F U t (2) For every neuron, j, in a layer, each of the i inputs, Xi,

to that layer is multiplied by a previously established weight, wij. These are all summed together, resulting in

the internal value of this operation, Uj. This value is then biased by a previously established threshold value, tj, and sent through an activation function, Fth. This activation function can take several forms such as Step, Linear, Sigmoid, Hyperbolic, and Gaussian functions. The Hy-perbolic function, used in this study, is shaped exactly as the Sigmoid one with the same mathematical representa-tion, as in Equation (3), but it ranges from 1 to +1 rather than from 0 to 1 as in the Sigmoid one.

11 x

f xe

(3) The resulting output, Yj, is an input to the next layer or

it is a response of the neural network if it is the last layer. In applying the Neural Network technique, in this study, Neuralyst Software, Shin [16], was used.

6.2. Neural Network Training

The next step in neural network procedure is the training operation. The main purpose of this operation is to tune up the network to what it should produce as a response. From the difference between the desired response and the actual response, the error is determined and a portion of it is back propagated through the network. At each neuron in the network, the error is used to adjust the weights and the threshold value of this neuron. Conse-quently, the error in the network will be less for the same

H. HODHOD ET AL.


568

inputs at the next iteration. This corrective procedure is applied continuously and repetitively for each set of in-puts and corresponding set of outputs. This procedure will decrease the individual or total error in the responses to reach a desired tolerance.

Once the network reduces the total error to the satis-factory limit, the training process may stop. The error propagation in the network starts at the output layer with the following equations:

'ij ij j iw w LR e X (4) and,

1j j j j je Y Y d Y (5) where, wij is the corrected weight, wij is the previous weight value, LR is the learning rate, ej is the error term, Xi is the ith input value, Yj is the ouput, and dj is the de-sired output. 7. Simulation Models To fully investigate numerically the effect of specimen shape and type of sand on the strength behavior of tested mortar with different cement types, seven models are

considered in this study. Two models for sitting time (initial and final), model for cube compression strength, two models for prism compression strength (CEN and regenerated sand) and two models for flexural strength (CEN and regenerated sand). 7.1. Neural network Design To develop a neural network models to simulate the ef-fect of specimen shape and type of sand on the strength behavior of tested mortar, first input and output variables have to be determined. What we have in the current study, to be considered as an input variable, is the types of cement used in the mortar specimen. So from the name of cement type we have to create a certain numeric characteristic values could be used as input variables in the present models as shown in Table 2.

Table 3 is designed to summarize all neural network key input variables and output for all the seven models presented in the current study. Some abbreviations used in Table 3 due to space limitation as follows:

Str.: Strength Compn.: Compression Flex.: Flexural

Table 2. Characteristic values for types of cement.

No Type of Cement I or II 32.5 or 42.5 or 52.5 N or R A or B S or LSRC or White Manufacturer

1 CEM I-52.5N 1 52.5 19 100 0 60 0

2 CEM I-42.5N (SRC) (**) 1 42.5 19 100 0 67 1

3 CEM I-42.5N (SRC) 1 42.5 19 100 0 67 2

4 CEM I-42.5N 1 42.5 19 100 0 60 0 5 CEM I-42.5R 1 42.5 23 100 0 60 0

6 CEM I-42.5N (White) 1 42.5 19 100 0 50 0

7 CEM I-32.5R 1 32.5 23 100 0 60 0

8 CEM I-32.5R (SRC) 1 32.5 23 100 0 67 0

9 CEM II-B-S-32.5N 2 32.5 19 60 24 50 0 10 CEM II-B-L-32.5N 2 32.5 19 60 12 50 0

Table 3. Key input variables and output for all ANN models.

Input Variables

Model I or II

32.5 or 42.5 or

52.5

N or R

A or B

S or L

SRC or White Manufacturer Days

Output

Initial Sitting Time Initial TimeFinal Sitting Time Final Time

Cube Str. Compn. Str.Prism Str.

(CEN sand) Compn. Str.Prism Str.

(Regenerated) Compn. Str.Prism Flex. Str.

(CEN) Flex. Ten.

Str. Prism Flex. Str. (Regenerated)

Flex. Ten. Str.

H. HODHOD ET AL.


569

Ten.: Tensile Several neural network architectures are designed and

tested for all simulation models investigated in this study to finally determine the best network models to simulate, very accurately, the effect of specimen shape and type of sand on the strength behavior of tested mortar with different cement types based on minimizing the Root Mean Square Error (RMS-Error). Figure 16 shows a schematic diagram for a generic neural network. The training pro-cedure for the developed ANN models, in the current study, uses the experimental data presented in the previ-ous sections of the current study. After the ANN models are settled for all cases, prediction procedure takes place to predict the compression as well as tensile strengths at different age-days rather than those days measured in the experiment (internal and after 28 days).

Table 4 shows the final neural network models for the seven simulation models and their associate number of neurons. The input and output layers represent the key input and output variables described previously for each simulation model.

The parameters of the various network models devel-oped in the current study for the different simulation models are presented in Table 5. These parameters can be described with their tasks as follows:

Learning Rate (LR): determines the magnitude of the correction term applied to adjust each neurons weights during training process = 1 in the current study.

Momentum (M): determines the life time of a cor-rection term as the training process takes place = 0.9 in the current study.

Training Tolerance (TRT): defines the percentage error allowed in comparing the neural network output to the target value to be scored as Right during the train-ing process = 0.001 in the current study.

Testing Tolerance (TST): it is similar to Training Tolerance, but it is applied to the neural network outputs and the target values only for the test data = 0.003 in the current study.

Input Noise (IN): provides a slight random variation to each input value for every training epoch = 0 in the current study.

Figure 16. General schematic diagram of a simple generic neural network.

Table 4. The developed neural network models.

No. of Neurons in each Layer Simulation Model No. of Layers Input Layer First Hidden Second Hidden Third Hidden Output Layer

Initial Sitting Time 4 7 5 3 - 1

Final Sitting Time 4 7 5 3 - 1

Cube Str. 5 8 6 4 2 1

Prism Str. (CEN sand) 5 4 4 3 2 1

Prism Str. (Regenerated) 5 8 6 4 2 1

Prism Flex. Str. (CEN) 5 4 4 3 2 1

Prism Flex. Str. (Regenerated) 5 8 6 4 2 1

Input # 1

Input # 2

Output # 1

Output # 2

Hidden layer 3 neurons

Hidden layer 3 neurons

H. HODHOD ET AL.


570

Table 5. Parameters used in the developed neural network models.

Simulation Parameter

Initial Sitting Time

Final Sitting Time

Cube Str.

Compn. Str. (CEN)

Compn. Str. (Re-generated)

Flex. Str. (CEN)

Flex. Str. (Re-generated)

Training Epochs 1146 4985 672361 301098 179853 315475 505672

MPRE 0.067 0.034 1.175 0.174 0.281 1.512 0.321 RMS-Error 0.0005 0.0005 0.0008 0.0004 0.0003 0.0016 0.0002

Function Gain (FG): allows a change in the scaling

or width of the selected function = 1 in the current study. Scaling Margin (SM): adds additional headroom, as a

percentage of range, to the rescaling computations used by Neuralyst Software, Shin (1994), in preparing data for the neural network or interpreting data from the neural network = 0.1 in the current study.

Training Epochs: number of trails to achieve the pre-sent accuracy.

Percentage Relative Error (PRR): percentage rela-tive error between the numerical results and actual meas-ured value and is computed according to Equation (6) as follows:

PRE = (Absolute Value (ANN_PR AMV)/AMV)* 100

(6) where:

ANN_PR: Predicted results using the developed ANN model

AMV: Actual Measured Value MPRE: Maximum percentage relative error during the

model results for the training step.

8. Results and Discussions Numerical results using ANN technique will be pre-sented in this section for all the seven models. Due to space limitation in the present paper the numerical re-sults of one type of cement (CEM I 52.5 N) will be pre-sented to show the simulation and prediction powers of ANN technique for compressive as well as tensile strengths. 8.1. Sitting Time For the sitting time models (initial and final), Table 6 presents the ANN results with experimental ones. One can see from this table that ANN technique can simulate very efficient the experiment results for different types of cements for mortar specimens.

8.2. Mortar Compressive Strength Three ANN models are developed to simulate and pre-dict the effect of specimen shape and type of sand on evaluating the compressive strength of mortar specimens for all the types of cement presented in the current study

at different ages. Figures 17 and 18 show the ANN results and experimental ones for compressive strength (cube and prism specimens) for one type of cement at the ages of experiment (2,3,7,10,28 days) and then predict the behavior at 14 days (internally) and after 28 days up to 42 days (externally). From these figures, it is very clear that ANN technique succeeded very well to simu-late and predict the compressive strength behavior at different ages for different specimens. 8.3. Mortar Tensile Strength Another two ANN models are developed to simulate and predict the flexural tensile strength for prism specimen for two types of sand (CEN and Regenerated) at different ages. Figure 19 presents the numerical results and ex-perimental ones at the ages of experiments (2,3,7,10,28 days). One can observe that ANN technique can simulate the tensile behavior and then predict the strength at ages different than the ages of experiment (before and after 28 days) very successfully. 9. Conclusions Based on the experimental investigation conducted in the course of the current research, the following can be con-cluded:

1) There is an inverse proportion between setting time and cement grade, and a direct proportion between grade and water requirement for standard consistency.

2) Applying old cement standards, for testing and evaluating mortar compressive strength of cement mortar, results in rejection of new cement types. Using of

C EMI52.5N

0.0

5.0

10.0

15.0

20.0

25.0

0 10 20 30 40 50

Age - days

Com

pres

sive

Stre

ngth

- Mpa

ExperimentANN TrainingANN Prediction

Figure 17. Cube compressive strength.

H. HODHOD ET AL.


571

Table 6. Sitting time models results.

Simulation Model No. 1 No. 2 No. 3 No. 4 No. 5 No. 6 No. 7 No. 8 No. 9 No. 10Exp. 70.0 85.0 75.0 100.0 90.0 80.0 110.0 80.0 120.0 100.0

Initial ANN 69.9 84.9 75.0 100.0 89.9 79.9 109.9 80.0 119.9 100.0 Exp. 140.0 140.0 180.0 210.0 200.0 170.0 225.0 185.0 240.0 200.0

Final ANN 140.0 140.0 180.0 210.0 200.0 170.0 225.0 185.0 240.0 200.0

C EMI52.5N

0

5

10

15

20

25

30

0 10 20 30 40 50

Age - days

Com

pres

sive

Stre

ngth

- Mpa


C EMI52.5N

0.0

5.0

10.0

15.0

20.0

25.0

30.0

0 10 20 30 40 50

Age - days

Com

pres

sive

Stre

ngth

- Mpa


Figure 18. Prism compressive strength (left: cen sand, right: regenerated sand).

C EMI52.5N

0

4

8

12

16

0 10 20 30 40 50

Age - days

Flex

ural

Str

engt

h - M

P


C EMI52.5N

0

4

8

12

16

0 10 20 30 40 50

Age - days

Flex

ural

Str

engt

h - M

P


Figure 19. Prism flexural strength (left: cen sand, right: regenerated sand).

jolting table for compaction is essential for obtaining successful test results according to new standards (EN 196 and ES 4756).

3) Standard CEN sand cannot be regenerated locally based only on its grading. Further investigation is re-quired to get its other properties like particle shape and chemical reactivity. There is some evidence on having early strength development when CEN sand is used in mortar.

4) Sulphate resisting cements show higher strength than CEM I cements of same grade, for both mortar and concrete mixtures.

5) There is some evidence that locally available ce-ments do not follow the rate of strength development denoted on packs.

6) For normal setting cements (N coded) there is a

clear distinction between concrete strength obtained for specific cement grade. However, this could not be seen for rapid setting types (R coded).

7) Correlation between flexural tensile and compres-sive strength of mortar differs significantly between early and later ages.

Based on the results of implementing the ANN tech-nique in this study, the following can be concluded:

1) The developed ANN models presented in this study are very successful in simulating the effect of specimen shape and type of sand on the behavior of mortar speci-mens (initial and sitting times, compressive strength, flexural tensile strength) for different types of cement.

2) The presented ANN models are very efficiently ca-pable of predicting the strength behavior at different ages rather than the ages of the experimental results (in the

H. HODHOD ET AL.


572

range of the experiment or in the future). 10. Acknowledgements The Authors would like to express their gratitude to-wards Prof. Dr. Farouk El-Hakim of 15th May institute for Civil and Arch. Engineering, and undergraduate stu-dents (4th yearcivil) for the help they provided during the experimental part of this research. 11. References [1] A. M. Neville, Properties of Concrete, John Wiley &

Sons, New York, 1997. [2] S. H. Kosmatka, et al., Design and Control of Concrete

Mixtures, 5th Edition, Cement Association of Canada, Ottawa, 2002.

[3] Egyptian Standards (ES) 372, Standards of Ordinary and Rapid Hardening Portland Cements, Egyptian General Authority for Standards, Cairo, 1991.

[4] Egyptian Standards (ES) 5476, Standards of Cements, Egyptian General Authority for Standards, Cairo, 2007.

[5] EN 197, CEMENT: Part 1: Composition, Specifications and Conformity Criteria for Common Cements, 2004.

[6] Egyptian Code of Practice 203, Basics of Design and Regulations of Construction of Reinforced Concrete Structures: Appendix III, Guide for Testing of Concrete Materials, Egyptian Ministry of Housing, Egypt, 2001.

[7] K. M. Yosri, Properties of New Cements Produced in Egypt as per ES 4756/2005, HBRC Journal, Vol. 3, No. 3, 2007, pp. 23-33.

[8] D. Solomatine and L. Toorres, Neural Network Ap-proximation of a Hydrodynamic Model in Optimizing Reservoir Operation, Proceedings of the 2nd Interna-tional Conference on Hydroinformatics, Zurich, 1996, pp.

201-206. [9] K. A. Kheireldin, Neural Network Application for Mod-

eling Hydraulic Characteristics of Severe Contraction, Proceedings of the 3rd Internetional Conference, Hydro-informatics, Copenhagen, 24-26 August 1998, pp. 41-48.

[10] M. A. M. Abdeen, Neural Network Model for Predicting Flow Characteristics in Irregular Open Channel, Scien-tific Journal, Faculty of Engineering-Alexandria Univer-sity, Vol. 40, No. 4, 2001, pp. 539-546.

[11] B. S. M. Allam, Artificial Intelligence Based Predictions of Precautionary Measures for Building Adjacent to Tun-nel Rout during Tunneling Process, Ph.D. Thesis, Fac-ulty of Engineering, Cairo University, Giza, 2005.

[12] H. Azmathullah, M. C. Deo and P. B. Deolalikar, Neural Networks for Estimation of Scour Downstream of a Ski- Jump Bucket, Journal of Hydrologic Engineering, ASCE, Vol. 131, No. 10, 2005, pp. 898-908.

[13] M. A. M. Abdeen, Development of Artificial Neural Network Model for Simulating the Flow Behavior in Open Channel Infested by Submerged Aquatic Weeds, Journal of Mechanical Science and Technology, KSME International Journal, Vol. 20, No. 10, 2006, pp. 1691- 1879.

[14] M. A. M. Mohamed, Selection of Optimum Lateral Load-Resisting System Using Artificial Neural Net-works, M.Sc. Thesis, Faculty of Engineering, Cairo University, Giza, 2006.

[15] M. A. M. Abdeen, Predicting the Impact of Vegetations in Open Channels with Different Distributaries Opera-tions on Water Surface Profile Using Artificial Neural Networks, Journal of Mechanical Science and Technol-ogy, KSME International Journal, Vol. 22, No. 9, 2008, pp. 1830-1842.

[16] Y. Shin, NeuralystTM Users GuideNeural Network Technology for Microsoft Excel, Cheshire Engineering Corporation Publisher, Pasadena, 1994.



Hydrogen Pick up in Zircaloy-4: Effects in the Dimensional Stability of Structural Components under Nuclear Reactor

Operating Conditions

Pablo Vizcano, Cintia Paola Fagundez, Abraham David Banchik Centro Atmico Ezeiza, Comisin Nacional de Energa Atmica, Presbtero J. Gonzlez y Aragn N 15,

Buenos Aires, Argentina E-mail: vizcaino@cae.cnea.gov.ar

Received December 3, 2009; revised February 11, 2010; accepted February 15, 2010

Abstract In the present work, the expansion coefficient due to hydrogen incorporation was measured for the axial di-rection of a Zircaloy-4 cooling channel, similar to that installed in the Atucha I PHWR, Argentina, trying to simulate the nuclear power reactor operating conditions. As a first step, the solubility curve of hydrogen in Zircloy-4 was determined by two techniques: differential scanning calorimetry and differential dilatometry. The comparison with classical literature curves showed a good agreement with them, although the calorimet-ric technique proved to be more accurate for these determinations. Dilatometry was able to detect the end of hydride dissolution from concentrations around 60 wppm-H up to 650 wppm-H, where the eutectoid reaction: takes place (at 550oC). We assume that this ability is a good indicator of the aptitude of the technique to measure dimensional changes in the given hydrogen concentration range. Then, the expansion of Zircaloy-4 homogeneously hydrided samples was measured at 300oC, the typical operating temperature of a nuclear power reactor, obtaining a relative expansion of 2.21 10-4% per wppm-H. Considering the rela-tive expansion observed for Zircaloy-4 at room temperature due to hydriding, starting from a hydrogen free sample, the total relative expansion rate is calculated to be 5.21 10-4% per wppm-H. Keywords: Thermal Analysis, Dimensional Change, Hydrides, Zircaloy-4 1. Introduction

Most of the core structural components of the nuclear power reactors are made of Zicaloy-4, a reference zirco-nium alloy in many structural nuclear applications. Dur-ing reactor operation, the initial dimensions of the Zr- base components could increase due to three different degradation processes: hydrogen pick up, irradiation growth and creep.

The hydrogen incorporated into the matrix is a fraction of the total amount of hydrogen produced during the corrosion reaction between the zirconium and the coolant, according to the reaction:

Zr + 2H2O ZrO2 + 4H The crystalline defects produced by the fast neutron

irradiation induce changes in the initial dimensions of the components depending on the fabrication texture. On the other hand, the creep contribution to these processes de-pends on the magnitude of the external stress applied to

the component. The pick up of hydrogen atoms by the metal induces

an expansion of its initial length. This expansion contin-ues after crossing the solubility limit at the reactor oper-ating temperature, since the hydrogen in excess to that limit precipitates as ZrH1,5+x after some supersaturation in solid solution. Due to the higher specific volume of the zirconium hydride with respect to the zirconium matrix, the onset of precipitation induces an additional dimen-sional change. This change in length depends on both, the orientation at which the hydrides precipitated in the matrix and the crystalline texture of the component.

The material under study in the present work is Zir-caloy-4 taken from cooling channels similar to those installed in the Atucha I PHWR. These tubes have a fully recrystallized microstructure, which induces hydride precipitation at the grain boundaries. In addition, these components show a strong texture in a quasi-radial direc-tion: the c axis of the -Zr hexagonal cell is oriented in a cone surrounding the radial direction of the tube [1]. The

P. VIZCANO ET AL.


574

aim of the present work is to determine the expansion coefficient of Zircaloy-4 for the axial direction of the channel at the reactor operating temperature (300C) [2]. 2. Experimental Procedure 2.1. Material and Sample Preparation The Zircaloy-4 samples were taken from an off cut of a cooling channel similar to those installed in the Atucha I reactor. Rectangular samples of dimensions 10 5 1.7 mm were cut from the tube, with the length of the sample parallel to the axial direction of the tube, as it is shown in Figure 1.

The tubes are cold-shaped and welded (by the tungsten inert gas method) from fully recrystallized Zircaloy-4 sheets. The typical Kearns texture factors were measured in a previous work for the [0002] pole (c axis of the hexagonal cell). The range of values was: Faxial = 0.05- 0.07, Ftangential = 0.22-0.26, Fradial = 0.67-0.73. Thus, about 6% of the c poles are aligned in the axial direction, 24% in the tangential and 70% in the radial direction [1]. The microstructure was fully recrystallized with a grain size of 15-20 m. It can be observed in Figure 2.

Figure 1. Orientation and dimensions of the dilatometric samples.

20 m

Figure 2. Fully recrystallized microstructure of a Zircaloy-4 cooling channel. The typical size of the equiaxed grains is 20 6 m.

2.2. Hydriding The hydrogen was incorporated by the cathodic charge technique. The process was carried out in an electrolytic cell at 80 2C. A diluted aqueous solution of sulfuric acid was used as electrolyte, circulating a current density of 5 mA/cm2 through the sample from periods of 18 to 96 h. As a result, hydride layers of different thickness (from a few microns up to 50 microns) formed in the samples.

The hydrogen was diffused into the bulk during the dilatometric experiments. After the experiments, the samples were polished to eliminate the oxide and any remaining hydride layer on the surfaces. Finally, the hy-drogen content was measured using a LECO RH-404 hydrogen meter. The error of these determinations is of 2%.

The hydrogen range of the samples hydrided in this way varied from 50 to 650 wppm-H. 2.3. Differential Dilatometry A Shimadzu TMA-60H vertical push rod differential dilatometer, DD, was used to measure the difference in expansion between a reference sample and a similar hy-drided sample.

The experiments were carried out under inert gas at-mosphere (high purity N2, 99.998%). As reference, an uncharged Zircaloy-4 sample was used, containing about 20 wppm-H, which is incorporated during the fabrication process of the channels. The minimum detection capacity of DD is 0.25 m.

During the test, a constant load of 0.1 N was applied to both samples. All the samples were subject to a nominal thermal cycle made of a heating step at a rate of 5C/min. After keeping the samples 30 minutes at the maximum temperature they were cooled down at 5C/min. To avoid the effect of the transformation, the maxi-mum temperature was a few degrees below 550C. 2.4. Differential Scanning Calorimetry The calorimetric experiments were made using a thermal flux differential scanning calorimeter Shimadzu, model DSC-60. The dimensions of the samples were 4 4 1.7 mm, which were cut from the dilatometric samples after the dilatometric experiments finished.

Two runs were performed for each sample at 5C/min, in order to compare the results with the dilatometric data, but the first one was discarded and TTSSd were deter-mined in the second run. Figure 3 shows the calorimetric heating curve of a hydrided sample where the points usually associated with hydride dissolution are indicated: the peak of the curve (pT, peak temperature), the maxi-mum at the derivative of the DSC curve (msT, maximum

5 mm10 mm

P. VIZCANO ET AL.


575

100 200 300 400 500

-5.0

-4.0

-3.0

-2.0

-1.0

0.0

msT

cT

pT

DSC curve Derived curve

Temperature (oC)

dq/d

t (m

J/se

c)

25oC -0.05

0.00

0.05

0.10

0.15

0.20

0.25

Extrapolation lines

d2q/dt 2 (m

J/sec2)

Figure 3. Calorimetric curve of a sample containing 480 wppm-H, in the heating stage. slope temperature) and the point where the baseline is recovered or completion point (cT, completion tempera-ture). 3. Results 3.1. Diffusion in the Bulk and TTSSd

Determinations A typical diffusive dilatometric run is shown in Figure 4. The differential apparatus needs a hydrogen-free refer-ence sample (in fact it contains 20 wppm-H). Since the reference is identical to the sample before hydrogen charging, the expansion of the reference compensates and cancels the thermal expansion of the -Zr phase in the hydrided sample. Thus, the expansion measured with the differential dilatometer only depends on the hydrogen concentration of the hydrided sample. During the heating stage, the hydride layer at the sample surface dissolves and the hydrogen atoms diffuse into the bulk, raising the concentration in solid solution. This process increases continuously the dimensions of the sample as it is obser-

0 5000 10000 15000 20000 250000

100

200

300

400

500

600 Expansion Temperature

Time (sec)

Tem

pera

ture

()

0

2

4

6

8

10

12

14

L (m)

Lengh increase due to the hydrogen incorporation

Figure 4. Dilatometric thermal cycle to diffuse the hydrogen from the layers into the bulk.

ved in Figure 4. Given enough time at the plateau tem-perature (550C), the hydride layer ends its dissolution and hydrogen distributes homogeneously into the sample. Depending on the thickness of the layer, it will dissolve during the run or during the time at the plateau tempera-ture. Yet, it is possible that a fraction remains undis-solved. This will occur if at the plateau temperature the solubility limit is reached without a complete dissolution of the hydride layer. During the cooling stage, the sample reduces its length but the differential expansion does not return to zero: at room temperature, a difference in length between the sample and the reference still subsists since the hydrogen diffused into the bulk is now precipi-tated as hydrides.

From the description given above, it is inferred that at the first dilatometric run the hydrogen distribution is controlled by the diffusion process. During this transient, TTSSd cannot be determined. Thus, after an additional mechanical polishing to eliminate any possible remain-ing hydride layer at the surface, TTSSd was measured in the second run.

Figure 5 shows a dilatometric curve obtained after the diffusive cycle, in the second run. During heating, the sample increases its length again but when the dissolu-tion finishes, the slope of the curve changes; at this point TTSSd is determined. In Figure 5 this change in the slope or knee is observed, for a sample containing 255 wppm-H, at 403C. This point is identified as the knee temperature, keT. Another possible criterion, which is not used in the present work, is to determine TTSSd at the dilatometric derived curve: the change in the slope at the knee produces a discontinuity, a step in the derived curve, as it is shown in Figure 5 too. It is not an ideal step; the discontinuity extends in a temperature range of 40C to 50C. At the middle height of the step, the step temperature, sT, can be determined. The step criterion proves to be more accurate than the knee criterion.

200 300 400 5000.0

0.5

1.0

1.5

2.0

2.5

3.0

50oC

Expansion curve Derived curve

Temperature (oC)

L(m

)

403oC

401oC

keT

sT

0.0000

0.0005

0.0010

0.0015

0.0020

dL/

dt (

m/s

ec)

Figure 5. Dilatometric curve of the dissolution process and derivative. The arrows indicate the change in the slope, where the dissolution ends (keT) and the middle of the step in the derived curve (sT). The sample contains 255 wppm-H.

(oC

)

P. VIZCANO ET AL.


576

for other zirconium alloys [3], but no difference was observed between them for Zircaloy-4. Thus, TTSSd was measured at the knee point (keT). In any case, it was ob-served that within an uncertainty interval of 2-4C, both temperatures are virtually identical, Figure 5.

The dilatometric TTSSd data are plotted in Figure 6 as TTSSd vs the hydrogen concentration, [H]. The solu-bility equation obtained from these data is:

CkeT = 2.86 105 exp (-4730/keT) (1) On the other hand, with the DSC technique TTSSd

was determined following two criteria commonly used in the literature: the peak, pT, and the completion tempera-ture, cT, as it is shown in Figure 7. The fitting curves are also included.

The solubility equations are:

CpT = 1.85 105 exp (-4362/pT) (2)

CcT = 1.78 105 exp (-4546/cT) (3)

0 100 200 300 400 500 600100

150

200

250

300

350

400

450

500

Dissolution data

Tem

pera

ture

()

[H] (wppm) Figure 6. Dissolution dilatometric data and fitting curve.

0 100 200 300 400 500 600100

200

300

400

500

pT data cT data

Tem

pera

ture

()

[H] (wppm)

Figure 7. Dissolution calorimetric data measured at pT and cT and fitting curves.

4. Discussion 4.1. Terminal Solid Solubility The uncertainty of TTSSd determinations with the dila-tometric technique can be estimated from Figure 5.

Where the expansion curve changes its slope (end of dissolution), its derivative shows a step. This step ex-tends over a temperature interval of about 40 to 50C, an interval larger than the 25C between pT and cT in the calorimetric curve (Figure 3). The knee temperature virtually agrees with the temperature at the middle height of the step in Figure 5. However, there is a higher intrin-sic uncertainty in the dilatometric measurements with respect to the calorimetric ones which affects the accu-rateness of TTSSd determinations. This uncertainty in-creases in the low hydrogen range, where the signal of hydride dissolution is weak. It becomes evident in Fig-ure 6, from 60 to 130 wppm-H, where the scatter of the data is large. For these data, TTSSd error varies from 18C to 15C at the upper extreme of the interval. For higher concentrations (in our case, concentrations higher than 187 wppm-H) the error decreases to 10C, becom-ing constant for concentrations higher than 250 wppm-H, where an error of 8C can be assumed.

Concerning DSC determinations, as it can be inferred from the criteria commented in 2.4, there are some dis-crepancies regarding the exact point where TTSSd should be located in the DSC curve [3-7]. As a brief summary we can say that: Z. L. Pan, measuring Youngs modulus as functions of temperature and hold time dur-ing quasistatic thermal cycles to Zr-2.5Nb hydrided sam-ples, concluded that the most reliable point to associate TTSS is msT [5]. D. Khatamian found the best correla-tion for pT contrasting TTSSd determinations at pT, msT and cT for unalloyed zirconium and Zr-20wt%Nb hy-drided/deuterided samples with neutron diffraction measurements [6]. Recently, the authors of the present work determined TTSSd for Zr-2.5Nb with pressure tube microstructure by DSC using DD as a contrasting tech-nique. In this work, the difficulty of determining the best point to measure TTSSd on the dissolution curves has been discussed thoroughly. Yet, since the accurateness of the DSC data is higher than the DD, it was not possible to obtain conclusions about the best point for TTSSd determination on the DSC curve from this comparison [8].

In any case, it is evident that the selection of one of the three criteria based on the measurements made with a contrasting technique does not provide physical meaning to the choice, turning it into the true dissolution point. In fact, the certainty of this choice will be strongly de-pendant on the capability of the contrasting technique to detect the disappearance of very small hydrides at the final stage of the dissolution process. In the present cir-

(oC

) (o

C)

P. VIZCANO ET AL.


577

cumstances we judged that it would be most advisable to choose the criteria that better agree with the highly ref-erenced curve of Kearns [9] and the equilibrium solvus line by Zuzek et al. [10], as done by other authors [3,4]. This comparison is shown in Figure 8. Regarding the DSC data, the best agreement with Kearns and Zuzek equilibrium curves corresponds to the completion point (cT). Beyond the criterion chosen for TTSSd determina-tion, the error at pT, msT or cT is always smaller than 2C and the reproducibility is excellent.

Although the present comparison has shown that DD is less accurate than DSC for solubility determinations, it must be alleged in its favor that the technique was capa-ble to detect the hydride dissolution for samples with concentrations from 60 wppm-H. This implies that the sensitivity is good enough to detect dimensional changes at very low concentrations. Since the main objective of the present work is to detect dimensional changes for hydrogen contents like 200 or 300 wppm-H, typical of cooling channels that remained for about 10 years in the reactor at full power operation [11], the performance of the technique is suitable for these purposes. This matter will be developed in the following section. 4.2. Axial Elongation of a Cooling Channel In order to simulate and determine the total elongation of the cooling channels due to the hydrogen pick up in ser-vice, a similar but faster process must be developed in the laboratory. Hydrogen should be incorporated into the bulk, starting from a hydrogen-free material. Instead of the slow hydrogen incorporation due to the corrosion in service, the hydrogen in the hydride layer diffuses into the bulk during the heating ramp and the subsequent iso-thermal annealing at 550C. At this stage, the hydrogen in the bulk remains in solid solution and the sample reaches its maximum length. During cooling, the hydrogen precipitates as hydrides reducing the sample length, but as it was shown in Figure 4, the final length is larger than the initial. After cooling, the final dimension of the sample is measured in situ in the dilatometer, obtaining a differential value. The length increase due to the hydro-gen incorporation into the bulk was measured at room temperature and reported in a recent paper [1]. A linear dependence on the hydrogen concentration was found for fully recrystallized Zircaloy-4. For this type of microstructure, hydrides precipitate on the grain boundaries, but some tendency to precipitate in the rolling direction

0 50 100 150 200 250 300 350 400

100

200

300

400

500

DSC (pT)

Kearns Zuzek

Tem

pera

ture

(oC

)

[H] (wppm)

DSC (cT)

DD (knee)

Figure 8. Comparison between the curves obtained in the present work and classical literature curves. recalling the cold rolling process still subsists after the recrystallization treatment. The competition of these two ways of precipitation generates some scatter in the data. However, a linear model seemed to be a good choice to represent the expansion vs. [H]. The linear assumption was made by simplicity, based on the values of the sta-tistical parameters, considering an error of 0.5 m for the micrometer, Table 1.

The following relation was obtained:

5.6 0.054 [ ] mL m Hwppm

(1) Dividing Equation (1) by the initial length of the sam-

ples from which Equation (1) was obtained (L0 = 18600 m), the relative expansion is:

-4 -6 1( ) 3.2 10 3.00 10 [ ] room0

L HL wppm

(2) where (L/Lo)room is the relative increase after the hy-drogen diffusion into the bulk at room temperature.

As the cooling channels operate in the two-phase field, in order to determine the total expansion in service, the contribution of both, the fraction of H atoms in solid solution and that of the zirconium hydrides at the reactor operating temperature (300C) should be added to the growth due to the hydrides already present at room tem-perature. The measurements made at 300C are listed in Table 2.

The relation found between the expansions and the hydrogen concentration is linear too. Both, the data and

Table 1. Interception, slope, standard errors (SE) and lower and upper confidence limit (LCL and UCL). The standard de-viation (SD) and R-value (R) are also given (97% of confidence).

Intercept Slope Statistics

Value SE LCL UCL Value SE LCL UCL R SD

5.6 2.0 0.6 10.7 0.054 0.004 0.044 0.064 0.9 5

P. VIZCANO ET AL.


578

Table 2. Relative expansion at 300C.

[H] (wppm) l (m) lo (m) l/lo 128 0.85 9781 8.69 E-5

227 5.0 10332 4.81E-4

398 5.7 10035 5.64E-4

358 5.65 10035 5.63E-4

446 9.5 9784 9.75E-4

650 12.9 9968 1.29E-3

the regression line are shown in Figure 9 and the statis-tics parameters in Table 3. The linear equation is:

-4 -6

3001) -1.54 10 2.21 10 [ ] C

0

L HL wppm

(3) where (L/Lo)300C is the relative length increase of the hydrided sample at 300C. Then, combining (2) and (3), the total length increase is calculated as follows:

3000 0 0

) ) ) CTOTAL roomL L LL L L

(4)

Table 3. Interception, slope, standard errors (SE) and lower and upper confidence limit (LCL and UCL). The standard de-viation (SD) and R-value (R) are also given (97% of confidence).

Intercept ( 10-4) Slope ( 10-6) Statistics Value SE LCL UCL Value SE LCL UCL R SD 1.54 1.3 5.5 2.5 2.21 0.33 1.22 3.2 0.96 1.3 10-4

The relative expansion at room temperature and the

total relative expansion are both plotted in Figure 9 too. Then, the total relative expansion coefficient along the axial direction at 300C is 5.21 10-4% per wppm-H.

As it was shown in previous works, the hydrogen iso-tope concentration of the cooling channels measured at different positions along its length after 10 years in ser-vice varies between 150 and 400 wppm-H [11]. If we choose a medium concentration of 250 wppm-H for the whole channel, it is possible to estimate the expansion of the tube at the operating temperature for this concentra-tion. Following Equation (4), the relative expansion will be 0.0015 m/m of tube. Then, if we consider the full length of the tube, 5,300 mm, the total expected expan-sion in the axial direction will be 8 mm, with an error estimated in 15%.

100 200 300 400 500 6000.0000

0.0005

0.0010

0.0015

0.0020

0.0025

0.0030

0.0035

L/LTOTALoL/Lroomo

L/L300oCo

[H] (wppm)

Rel. expansions at 300oC Linear regressions Error bands

L/L

o

Figure 9. Relative expansion at 300C, at room tempera-ture [1] and the sum of both effects. 5. Conclusions The present work was focused on two main objectives: hydrogen solubility measurements and the determination

of the expansion coefficient of Zircaloy-4 for the axial direction of a tube. As a brief summary, the following points must be underlined: The hydrogen solid solubility curve for Zirca-

loy-4 was determined by two techniques, differential scanning calorimetry and differential dilatometry. The comparison with classical literature curves showed a good agreement with them. The solubility curves obtained with calorimetry, measuring TTSSd at the peak and completion temperatures are:

CpT = 1.85 105 exp (4362/pT) CcT = 1.78 105 exp (4546/cT)

and the one obtained by dilatometry, measuring TTSSd at the knee temperature is:

CkeT = 2.86 105 exp (4730/keT) Although the coincidence between them is good, the

calorimetric technique is more accurate for these meas-urements. Dilatometry showed good sensitivity to detect the

end of dissolution from concentrations around 60 wppm-H up to the eutectoid temperature (550C) concentration (650 wppm-H), which is a good indicator of the aptitude of the technique to measure dimensional changes in hy-drided samples in this concentration interval.

The expansion of Zircaloy-4 homogeneously hydrided samples was measured at 300C, the typical operating temperature of a nuclear power reactor, obtaining a rela-tive expansion of 2.21 10-4% per wppm-H. Adding to this value the relative expansion coefficient at room tem-perature due to hydriding, the total relative expansion rate is 5.21 10-4% per wppm-H. 6. References [1] J. C. Ovejero, A. D. Banchik and P. Vizcano, Axial/

Tangential Expansion Coefficients of Zircaloy-4 Chan-nels Due to the Hydrogen Pick up, Advanced in Tech-

P. VIZCANO ET AL.


579

nology of Materials and Materials Processing Journal, Vol. 10, No. 1, 2008, pp. 1-8.

[2] C. P. Fagundez, P. Vizcano, D. Bianchi and A. D. Ban- chik, Dilatometra del Sistema Zr-H, Proceedings of the Congress SAM/CONAMET 2005, Mar del Plata, 18-21 October 2005.

[3] J. P. Giroldi, P. Vizcano, A. V. Flores and A. D. Banchik, Hydrogen Terminal Solid Solubility Determinations in Zr-2.5 Nb Pressure Tube Microstructure in an Extended Concentration Range, Journal of Alloys and Compounds, Vol. 474, No. 1-2, 2009, pp. 140-146.

[4] D. Khatamian and V. C. Ling, Hydrogen Solubility Lim-its In - and -Zirconium, Journal of Alloys and Com-pounds, Vol. 253-254, No. 20, 1997, pp. 162-166.

[5] A. McMinn, E. C. Darby and J. S. Schofield, The Ter-minal Solid Solubility of Hydrogen in Zirconium Al-loys, Proceedings of the 12th International Symposium of the Zirconium in the Nuclear Industry, Toronto, 2000, pp. 173-195.

[6] Z. L. Pan and M. P. Puls, Precipitation and Dissolution Peaks of Hydride in Zr-2.5 Nb during Quasistatic Ther-

mal Cycles, Journal of Alloys and Compounds, Vol. 310, No. 1-2, 2000, pp. 214-218.

[7] D. Khatamian and J. H. Root, Comparison of TSSD Results Obtained by Differential Scanning Calorimetry and Neutron Diffraction, Journal of Nuclear Materials, Vol. 372, No. 1, 2008, pp. 106-113.

[8] P. Vizcano, A. D. Banchik and J. P. Abriata, Calorimet-ric Determination of the -Hydride Dissolution Enthalpy in Zircaloy-4, Metallurgical and Materials Transactions A, Vol. 35A, No. 8, 2004, pp. 2343-2349.

[9] J. Kearns, Terminal Solubility and Partitioning of Hy-drogen in the Alpha Phase of Zirconium, Journal of Nu-clear Materials, Vol. 22, No. 3, 1967, pp. 292-303.

[10] E. Zuzek, J. P. Abriata and A. San Martn, H-Zr (Hy-drogen-Zirconium), Bulletin of Alloy Phase Diagrams, Vol. 11, No. 4, 1990, pp. 385-395.

[11] P. Vizcano, A. D. Banchik and J. P. Abriata, Solubility of Hydrogen in Zircaloy-4: Irradiation Induced Increase and Thermal Recovery, Journal of Nuclear Materials, Vol. 304, No. 2-3, 2002, pp. 96-106.



Electrochemical Generation of Zn-Chitosan Composite Coating on Mild Steel and its Corrosion Studies

Kanagalasara Vathsala, Thimmappa Venkatarangaiah Venkatesha, Beekanahalli Mokshanatha Praveen, Kudlur Onkarappa Nayana

Department of Studies in Chemistry, School of Chemical Sciences, Kuvempu University, Shankaraghatta, India E-mail: drtvvenkatesha@yahoo.co.uk

Received December 16, 2009; revised February 26, 2010; accepted March 6, 2010

Abstract A Zinc-Chitosan composite coating was generated on mild steel from zinc sulphate-sodium chloride electro-lyte by electrodeposition. The electrolyte constituents were optimized for good composite coating. The cor-rosion resistance behavior of the composite was examined by weight loss, polarization and impedance methods using 3.5 wt% NaCl neutral solution as medium. Separate polarization profiles were recorded for composite coating and compared with zinc coated sample. SEM images of coatings were recorded for the pure and composite coating. Keywords: Composite Coating, Chitosan, SEM, Impedance, Electrodeposition 1. Introduction

Zinc electroplating is an industrial process and is widely used to coat on steel for enhancing its service life. As zinc is electrochemically more active than steel and hence it sacrificially protect the steel from corrosion. However zinc itself undergoes corrosion leading to the formation of zinc compounds called white rust on its surface. This tendency of formation of white rust reduces the life of the coating from the expected period. There-fore to enhance the life span of the zinc coating and to avoid the white rust formation the alternative methods like surface modification is adopted. The earlier modifi-cation methods are associated with chromate based for-mulations and the procedure is very simple to generate passive chromate films on corroding zinc coatings. The use of chromate passivation is prohibited because of pollution hazards. An alternate to this chromation is to generate surface films or surface barriers with specific organic molecules or with certain addition agents [1-6]. Also the service life of zinc coating is enhanced by in-cluding the inert materials in its coating. The inclusion is done by codeposition of these materials with zinc and thus generating composite coating. These zinc composite coatings exhibit better corrosion resistance property. Nowadays the nanosized materials are codeposited to get better zinc composite with better corrosion resistance [7-10].

A survey of literature reveals that the conducting

polymers were used for anticorrosive coatings and as inhibitor for steel [11-13]. However limited information connected to zinc - polymer composite coatings on steel is available in the literature and especially with zinc - biopolymer composites.

The chitosan is one such biopolymer used in corrosion inhibition of mild steel without causing environmental problems. Chitosan possess good biocompatibility, chemical resistance, mechanical strength, antimicrobial properties and thermal stability and have been utilized suc-cessfully in biotechnology, for different applications. The hydroxyl apatite chitosan nanocomposite was ob-tained on stainless steel to provide better corrosion pro-tection [14,15]. Chitosan is widely used in industry due to its film forming and gelation characteristics. In dilute solutions it is a linear polycation with high charge den-sity. This electrochemical property was utilized in the present work to get the zinc chitosan composite film on mild steel from electrolysis and its corrosion resistance property was tested. 2. Experimental

2.1. Plating Process Zinc and Zn-chitosan coatings were electrically depos-ited from sulphate-chloride bath. The constituents of the bath were 250 g/L ZnSO47H2O, 40 g/L NaCl, 30 g/L H3BO3 and 0 g/L chitosan (88% deacetylated). In all the

K. VATHSALA ET AL.


581

experiments distilled water and analytical grade reagents were used. The pH of the bath solution was adjusted to 2.5-3 by adding dil.H2SO4 and NaHCO3. The bath was stirred for few hours before subjecting it into plating ex-periments. The cathode was mild steel and anode was zinc (99.99%). The mild steel surface was polished me-chanically, and degreased with trichloroethylene in de-greased plant followed by water wash. Before each ex-periment the zinc surface was activated by dipping in 10% HCl for few seconds and was washed with water. Equal area of anode and cathode was selected for elec-trode position process. The bath temperature was at 300 K. The deposition process was carried at 4 A/dm2 and under mechanical stirring. 2.2. Weight Loss Measurements The coating thickness prepared for corrosion tests was in the range of 1015 m. The corrosion rate by weight loss measurements were performed for mild steel samples coated with pure zinc and Zn-chitosan composite. The electrolyte was 3.5 wt% NaCl solution and the test sam-ples were immersed vertically in the solution which was maintained at room temperature. The difference in weight was measured once in every 24 hours for a period of 15 days. In each weight loss measurement the corroded samples were rinsed in alcohol, dried with hot air, and then the weight was noted. The weight loss evaluated was used for estimating the corrosion rate. 2.3. Salt Spray Test The salt spray test as per (ASTM B 117) was carried out in a closed chamber. The deposited plates were freely hanged inside the chamber and subjected to continuous spray of neutral 5 wt% NaCl vapors. The specimens were observed periodically and the duration of the time for the formation of the white rust was noted. 2.4. Electrochemical Measurements A conventional 3-electrode cell was used for polarization studies. The zinc coated or Zn-chitosan composite coated specimen with surface area of 1 cm2 was used as working electrode. Saturated calomel and platinum foil were em-ployed as reference and counter electrodes respectively. The electrolyte was 3.5 wt% NaCl solution. The corro-sion resistance property of these specimens was evalu-ated from the anodic polarization curves.

The electrochemical impedance measurements were performed using AUTOLAB from Eco-chemie made in Netherlands. The steel specimens and their dimensions were same as that of polarization experiment. The EIS was recorded in the frequency range from 100 kHz to 10 MHz with 5 mV AC amplitude sine wave generated by

a frequency response analyzer. The surface morphology of the coatings was examined

using a JEOL-JEM-1200-EX II scanning electron mi-croscope 3. Results and Discussion

3.1. Corrosion Rate Result The zinc and composite coatings was generated on sepa-rate mild steel plates having the thickness of about 10- 15 m. The steel panels were immersed completely in 3.5 wt% NaCl solution for different time intervals and the weight loss values were used to calculate the corro-sion rate. Figure 1 represents the corrosion rate (wt loss/ hour) profiles with respect to number of hours. The cor-rosion rates of both composite and zinc coatings were very high in the beginning and decrease exponentially in the middle and it becomes constant after 200 and 150 hrs for zinc and composite coatings respectively. At any given time the rate of corrosion for composite was al-ways less than that of zinc coating. This suggests that the composite coating possess higher corrosion resistance property. This property was due to the presence of chito-san in the zinc matrix. 3.2. Salt Spray Test Result

The industrial method of testing the corrosion behavior of zinc-plated objects is salt spray test. The test was conducted by spraying 5 wt% NaCl solution in a chamber. The NaCl drops accumulated on the surface of the coated specimens facilitate the corrosion resulting in zinc salts called white rust. The time taken for the formation of white rust was the indication of the corrosion rate. The higher corrosion resistance delays the production of white rust. In the present case the pure zinc produced the white rust after 19 hrs and the Zn-chitosan composite

0 50 100 150 200 250 300 350 400

2

3

4

5

6

7

8

9

zinc coating composite coating

Cor

rosi

ve v

eloc

ity (1

0-5 k

g/m

2 .h)

Time (h) Figure 1. Variation of the corrosion rate with immersion time for zinc and composite coated samples in 3.5 wt.% NaCl solution.

K. VATHSALA ET AL.


582

produced the white rust after 28 hrs. This test confirms the enhancement of corrosion resistance of zinc in the presence of chitosan in its matrix. 3.3. Electrochemical Result Figure 2 shows anodic polarization profile of zinc and Zn-chitosan coated sample in 3.5 wt% NaCl solution. The linear variation was observed in the beginning up to 1.01 V and afterwards there was gradual increase in current indicating electrochemical oxidation of zinc. However in the case of composite coating, the potential was always more positive for any given current density. This indicates that the composite requires extra potential to bring anodic reaction. Thus the composite possess higher resistance to corrosion process on its surface.

The Nyquist plots for zinc and Zn-chitosan coatings are shown in Figure 3. The larger loop was produced by Zn-chitosan coatings whereas smaller semicircle was obtained for pure zinc. It can be easily observed from the figure that Rp values are higher for composite coating than zinc coating. This indicates that composite coating is more corrosion resistant than zinc coating. 3.4. Surface Morphology The SEM images at lower and higher magnification were represented in Figure 4. Also the SEM images of cor-roded surface of zinc and composite are given in Figure 5 and Figure 6. The SEM images show the practic

Engineering_02_08_2010092609263635.pdf

Documents