vol: xlvii, no. 03 july 2014 - the institution of engineers sri...

Printed by Karunaratne & Sons (Pvt) Ltd.

VOL: XLVII, No. 03VOL: XLVII, No. 03VOL: XLVII, No. 03VOL: XLVII, No. 03VOL: XLVII, No. 03VOL: XLVII, No. 03VOL: XLVII, No. 03 July 2014July 2014July 2014July 2014July 2014

I

ENGINEER JOURNAL OF THE INSTITUTION OF ENGINEERS, SRI LANKA * 42nd Year of Publication *

EDITORIAL BOARD Eng. W J L S Fernando - President (Chairman) Eng. Tilak De Silva - Past President Eng. W. Gamage - Chairman, Library &

Publications Committee

Eng. (Prof.) K. P. P. Pathirana - Editor Transactions Eng. (Prof.) T. M. Pallewatta - Editor „ENGINEER‟ Eng. (Dr.) U. P. Nawagamuwa - Editor „SLEN‟ Eng. (Prof.) (Mrs.) N. Rathnayaka Eng. (Dr.) D. A. R. Dolage Eng. (Miss.) Arundathi Wimalasuriya Eng. (Dr.) K. S. Wanniarachchi The Institution of Engineers, Sri Lanka 120/15, Wijerama Mawatha, Colombo - 00700 Sri Lanka. Telephone: 94-11-2698426, 2685490, 2699210 Fax: 94-11-2699202 E-mail: [email protected] E-mail (Publications): [email protected] Website: http://www.iesl.lk

COVER PAGE

Broadlands Hydropower project The Broadlands hydropower project initiated in 2013 July on a Design-Build basis is expected to be commissioned in early 2017. With two vertical shaft Francis turbines of 17.5 MW rating the power plant is expected to add 35 MW of power at 126 GWh of annual energy, to the national grid. Fed by Broadlands Pond and Kehelgamu River Weir pond concrete gravity type dam contained reservoir is designed to store 200,000 m3 of water to be released to the twin turbines at 70x2 m3/s under a head of 60 m. This project qualified to receive CDM benefits and registered in UNFCCC, is expected to help in cutting down national CO2 emissions by about 83,000 tons annually. Contributed by Eng. Dr. Kamal Laksiri Project Director, Broadlands Hydropower Project Ceylon Electricity Board

CONTENTS

Vol.: XLVII, No. 03, July 2014 ISSN 1800-1122

From the Editor ... III SECTION I

Optimization of the Accuracy of Semi-Indirect Energy Metering System by: G M N K Kumara Galathara,

Eng. Lalith A Samaliarachchi and Eng. S M D P K Sooriyadasa

Wind Loads on High-Rise Buildings by using Five Major International Wind Codes and Standards by: Eng. A U Weerasuriya and

Eng. (Prof.) M T R Jayasinghe Predicting Thermal Performance of Different Roof Systems by Using Decision Tree Method by: Eng. A U Weerasuriya Design and Development of Rasp Bar Mill for Size Reduction of Maize by: Eng. D P Senanayaka,

Eng. H M A P Rathnayake, B M K S Thilakarathne, Eng. B D M P Bandara and Eng. T M R Dissanayake

Criteria to Assess Rock Quarry Slope Stability and Design in Landslide Vulnerable Areas of Sri Lanka: A Case Study at Thalathu Oya Rock Quarry by : Eng. M N C Samarawickrama,

U B Amarasinghe and K N Bandara

SECTION II

Strategic Decision Support for resolving Conflict Existing in Per Aru Basin Located in Vavuniya District by : Eng. G Abira and

Eng. (Prof.) K D W Nandalal

The statements made or opinions expressed in the “Engineer” do not necessarily reflect the views of the Council or a Committee of the Institution of Engineers Sri Lanka, unless expressly stated.

Notes: ENGINEER, established in 1973, is a Quarterly

Journal, published in the months of January, April, July & October of the year.

All published articles have been refereed in anonymity by at least two subject specialists.

Section I contains articles based on Engineering Research while Section II contains articles of Professional Interest.

1

13

27

39

49

61

III

II

Performance and Retrofitting of Unreinforced Masonry Buildings against Natural Disasters – A Review Study by: Miss. W S W Mendis,

Eng. (Dr.) G S Y De Silva and Eng. (Dr.) (Mrs.) G H M J Subashi De Silva

71

II



71

III II



III

FROM THE EDITOR………….. Hunger for energy is now almost a universal factor throughout the world. This trend has been created by a human kind, which has become increasingly dependent on energy consuming appliances for practically all its real needs as well as the dispensable whims and fancies too. With fossil fuels, the main source of energy today, being depleted at an ever increasing rate, the pampered and spoilt human generation of tomorrow is going to end up in a catastrophe, when energy becomes scarce for even the fundamental functions for the existence of a civilization, as we know today. On the other hand, this could even be a „blessing in disguise‟ from the point of view of the environment, which has already become polluted to unacceptable levels, due to fossil fuel emissions in mammoth scales. Whatever the scenario is, the deliverance from impending doom lies in harnessing sustainable and renewable energy sources for prudent usage. By „prudent use‟ what is meant is the use for human necessities in a global scale rather than the personal comforts of a privileged few. In the context of sustainable and renewable energy sources, the most well known and tested is the hydropower, which also has the longest history. In Sri Lanka, hydropower was once synonymous with electricity supply and at some era we even thought that our hydro potential is adequate to export this precious commodity. Within the harsh realities of the present, where we have become near totally dependant on fossil fuel based electricity apart from nuclear, even the last dregs of the hydropower potential need to be harnessed. The Broadlands hydropower project initiated in the last year could be the last of such national scale endeavours to grace this country. What we have embarked upon here is exactly in line with the motto of our illustrious king Parakramabahu the Great, “Not a drop of water from the sky should reach the sea without gainfully serving the people of the country”. Eng. (Prof.) T. M. Pallewatta, Int. PEng (SL), C. Eng, FIE(SL), FIAE(SL) Editor, „ENGINEER‟, Journal of The Institution of Engineers.

V

SECTION I

ENGINEER1

ENGINEER - Vol. XLVII, No. 03, pp. [page range], 2014 © The Institution of Engineers, Sri Lanka

1 ENGINEER

Optimization of the Accuracy of Semi-Indirect Energy Metering System

G. M. N. K. Kumara Galathara, Lalith A. Samaliarachchi and S. M. D. P. K.

Sooriyadasa Abstract: Difference between a measurement reading and the true value of that measurement is known as accuracy. The measuring device or system should have the capability to measure and match the actual value to the quantity that is being measured to an acceptable accuracy declared by respective standards. Measurements of electrical energy in various demand categories of different utilities have their own limits. The employed measuring system is designed according to the demand category. Semi indirect metering system is being used to measure energy consumed between 42 kVA to 1 MVA by the governing Sri Lankan utility, Ceylon Electricity Board (CEB). It requires a current transformer (CT), an energy meter and lead wires only. In general, such systems are installed after certain standard checkups carried out for individual accuracy of CTs and energy meters in a standard testing laboratory. However, it has been observed that there is a considerable deviation of accuracy than expected when considering the accuracy of the whole system setup. Replacing of analogue meters by digital meters in semi indirect measuring environment to establish an automated meter reading system by a Utility has exclusive advantages such as immunity tamper up to certain extent and features supported by enhance data communication facilities etc. However, careful study is required to omit the potential of getting deviated the total system error beyond the desired limits due to the burden mismatch owing to CTs, energy meters and associated lead wires. In this research study, it was observed that considerable percentage of semi indirect metering systems installed are operating at under burden condition. The study focused to minimize the error deviation of semi indirect metering system owing to the above mentioned factors with a new concept of appropriate burden matching. The study also concentrated on to finding the factors affecting the accuracy and how to eliminate the error through a simple error compensation method. In here all the experiments were done in the laboratory for 100/5 semi indirect metering model and it describes the methodology for improving the accuracy clearly by appropriate component selection using developed charts for different contract demands operating at different power factors depending on the application. Keywords: Semi-indirect metering system, burden matching, error compensation 1. Introduction

1.1 Objective Electricity metering system can be classified in to: Direct metering (using energy meter

only) Semi-indirect metering (using energy

meter with a CT) Indirect metering (using energy meter,

CT and a voltage transformer) Categorization of the metering system in case of CEB, depending on the consumer contract demand (D) is as follows: Direct metering (D<42 kVA at 400 V) Semi-indirect metering (42 kVA<D<1000

kVA at 400 V) Indirect metering (D>1000 kVA at 11 kV

or 33 kV)

This paper discusses the methodology for minimizing the error band of the existing semi-indirect energy measuring system by:

Shifting the operating current of CTs in to the best performance region.

Matching the secondary burden of the CTs by appropriate lead wire lengths.

Development of chart based tables for the selection of lead wire lengths for predetermined accuracy.

G.M. N. K. Kumara Galathara, NDT (Moratuwa), B. Tech. Eng (Under Graduate), Open University of Sri Lanka (OUSL), Electrical Superintendent, Meter Testing Laboratory DD-1, Ceylon Electricity Board(CEB) Eng. Lalith A. Samaliarachchi, B.Sc. Eng. (Hons) (Moratuwa), M. Eng.(AIT), MIEE(UK), MIE(Sri Lanka), C.Eng Senior Lecturer, Department of Electrical & Computer Engineering, OUSL Eng. S.M.D.P.K. Sooriyadasa, B.Sc. Eng. (Peradeniya), AMIE (Sri Lanka), Electrical Engineer, Meter Testing Laboratory DD-1, (CEB)

ENGINEER - Vol. XLVII, No. 03, pp. [1-11], 2014© The Institution of Engineers, Sri Lanka

ENGINEER 2ENGINEER 2

1.2 Background Electro-mechanical meters were used for semi-indirect metering system during past few decades in Sri Lanka. Two types of analogue meters are widely being used in this regard to measure electrical energy (kWh) and Maximum electricity demand (kVA). Table 1 shows the meter specifications.

Table 1 – Analogue meter specifications Unit Type

Meter Type

Accuracy Class

Burden (VA) Remarks

kWh GEC 2 2.5 N/A

kVA Landis & Gyr 2 2.5

Accuracy valid between

pf = 0.67-0.98 Replacement of analogue bulk energy meter set with solid state meters result distinguished improvements, in accuracy perspective. However, it has shown certain inaccuracies owing to the associated component mismatch burdens such as CTs, meters and lead wires. It is well known to the public and to the utility that the revenue and loss calculations, demand forecasting and end user satisfaction depend upon the accurate metering system. Hence any fine improvement to an existing metering system will be an effective contribution to the development of any utility. 2.0 Existing System Figure-1 shows the schematic diagram of semi-indirect metering system currently being practised by Sri Lankan institutions that have the distribution licensees. Current transformers, energy meters and lead wires are the main components of this system.

M.C.C.B

P.P.M

T/F Side1 2 3 N

Load Side

CT1

CT2

CT3

Lead wires

Figure 1-Schematic diagram of the

Semi-indirect metering system

2.1 Energy Meters The energy meters installed are of Programmable Polyphase Meters (PPM) of accuracy class 1 with remote meter reading facility [1]. EDMI and PREMIER are the main brand types available in Sri Lanka having burden of 0.5 VA under extreme operating conditions. 2.2 Current Transformers Accuracy improvement of the semi indirect metering system is heavily depends upon the CT specification as per the IEC regulations [6]. Table 2 shows the types of CTs that are widely being used in Sri Lanka for Semi–indirect metering system. It should be noted that the very first metering point is where the CT is installed. Hence the performance characteristics of CTs are identified as the critical and most important area in this research study.

Table 2 – Types of Current Transformers

Type Accuracy Class Burden (VA)

ITL 1/0.5/0.2 15/5/2.5 Horbut 1/0.5/0.2 15/5/2.5 ARW 1/0.5/0.2 15/5/2.5

Wong Young 1/0.5 10/5 Burden of a CT is defined as the apparent power in volt-amperes absorbed at specified power factor at its rated secondary current or in other words it is the maximum load that the CT secondary can handle (Meter and lead wire) [6]. Further the ratio error of a CT is defined as:

100%

primaryI

primaryIsecondaryIratioCTErrorRatio

Phase displacement error of a CT is the difference in phase displacement between the primary and secondary current vectors [6]. Figure 2 and 3 shows the variation of ratio error (%) and displacement error (min) against the nominal current for a CT-ARW Class-1. It also shows that the ratio error increases negatively and displacement error increases positively while varying the nominal current from 2.5% In to 120% In respectively. Total error comes out to be the contribution of both ratio error and phase displacement error.

3 ENGINEER

Rat

io E

rror

(%)

-0.045 -0.075-0.16

-0.22

-0.325

-0.66

-0.955-1

-0.8

-0.6

-0.4

-0.2

0

120 100 60 40 20 5 2.5

In%

Figure 2 – Ratio Error variation

-1.51 -0.71

1.923.77

6.25

12.95

18.66

-5

0

5

10

15

20

120 100 60 40 20 5 2.5Dis

pla

cem

ent

Erro

r(m

in)

In%

Figure 3 – Displacement Error variation

2.3 Lead Wires Even though the lead wire length selection process is not well defined in the respective specifications used by the utilities, for example CEB 57:1997 states that the cross sectional area of the connecting cables should not be less than 2 mm2 [3]. Under the circumstances, an established method has not been practised so far and moreover, there is not much of an attention is given technically when selecting the lead wire lengths in burden matching perspective. Most of the installations are observed to be carried out using 2.5 mm2 or 1 mm2 double insulated multi strand wire lengths. Also, it is worthwhile to note at this point that the Lead wires not only affect the CT burden but also exposed for tampering. Therefore the lead wire selection for semi indirect metering system is an important exercise to be observed very carefully. 2.4 Semi-indirect metering system data A comprehensive survey was carried out at randomly selected factories in Kurunegala area to analyse the present situation of semi indirect metering system. Data collection was done by visiting each location. Appendix-A, Table-A1 shows the technical details gathered. Onsite testing was carried out by using high precision portable test equipment (Zera MT 310, class-1) to measure the energy meter accuracy, overall error and instant parameters. Average power factor and average load were obtained by the load survey of PPM. Here it was observed that the burden of all CTs is 15VA and burden of meters is 0.5 VA. Wire lengths are observed to

be almost around 1m and average power factors were more than 0.89. Operating current (%In) was observed to be very much lower with respect to the rated current of CTs. Although the overall error of semi-indirect metering system at the surveyed location were approximately around 3%, the individual error of the energy meters were within the class of 0.5, thereby the additional error was introduced into the system by CTs and lead wires. 2.5 Ferraris Vs solid state meters Modernization replaced the classical induction meters by solid-state meters. Besides ensuring precise metering, latter enables remote meter reading and various metering systems such as Automatic Metering Management (AMM). In addition to the expanded functions, they differ from the Ferraris in terms of electrical properties, and particularly their self-consumption (meter burden). Since in the course of years electrical properties such as transformation ratio, instrument security factor etc. of measuring VTs and CTs remain more or less unchanged for the majority of end users, their measuring properties are mostly effected by under-burdening of the secondary circuit caused by replacement of the Ferraris meters by the electronic ones. Normally the power consumption of electronic meters is from five to ten times smaller compared to the Ferraris meters [7],[8]. Specification of VA burdens of CTs often come across with specified values of 2.5, 5, 10, 15 VA. These burden values arrived at the specifications are decided or documented by end users, project engineers, process consultants and engineering consultants by making hypothetical burden calculations & inflating the values. These methods invariably lead to select inappropriate burden values of CTs thus leading to inaccurate reading. Old bulk supply metering system (>42kVA) consisted of two separate meters for active and apparent energy measurement where the burden of metering path is much higher with respect to that of static meters. Thus, when replacing two analogue meters with a single digital meter, the operating burden of secondary circuits of measuring CTs get further lowered. 3.0 Literature Survey

Neither any technical literature found nor any research work has been carried out so far as to analyse the performance of the present




Meter Type

Accuracy Class

Burden (VA) Remarks

kWh GEC 2 2.5 N/A




M.C.C.B

P.P.M

T/F Side1 2 3 N

Load Side

CT1

CT2

CT3

Lead wires






ITL 1/0.5/0.2 15/5/2.5 Horbut 1/0.5/0.2 15/5/2.5 ARW 1/0.5/0.2 15/5/2.5


100%

primaryI



3 ENGINEER

Rat

io E

rror

(%)

-0.045 -0.075-0.16

-0.22

-0.325

-0.66

-0.955-1

-0.8

-0.6

-0.4

-0.2

0

120 100 60 40 20 5 2.5

In%


-1.51 -0.71

1.923.77

6.25

12.95

18.66

-5

0

5

10

15

20

120 100 60 40 20 5 2.5Dis

pla

cem

ent

Erro

r(m

in)

In%





ENGINEER3ENGINEER 2



Meter Type

Accuracy Class

Burden (VA) Remarks

kWh GEC 2 2.5 N/A




M.C.C.B

P.P.M

T/F Side1 2 3 N

Load Side

CT1

CT2

CT3

Lead wires






ITL 1/0.5/0.2 15/5/2.5 Horbut 1/0.5/0.2 15/5/2.5 ARW 1/0.5/0.2 15/5/2.5


100%

primaryI



3 ENGINEER

Rat

io E

rror

(%)

-0.045 -0.075-0.16

-0.22

-0.325

-0.66

-0.955-1

-0.8

-0.6

-0.4

-0.2

0

120 100 60 40 20 5 2.5

In%


-1.51 -0.71

1.923.77

6.25

12.95

18.66

-5

0

5

10

15

20

120 100 60 40 20 5 2.5Dis

pla

cem

ent

Erro

r(m

in)

In%






situation. Only routine tests had been carried out to verify the energy meter accuracy only. However, it was observed while surfing the internet that Slovenia and India has faced similar under burden miss-matching metering condition due to replacement of Ferraris meters by solid state meters as in Sri Lanka [4]. 4. Theoretical Analysis 4.1 Effect of the secondary burden Practically, the primary and secondary ampere-turns are not exactly equal in CTs. Depending on the core material used; some of the primary ampere-turns will be used to excite the core. Only the remaining primary ampere-turns will be available for transformation. Thus, the required primary ampere-turns are invariably become less and it creates an error in the operation of CTs. Figure 4 shows an equivalent circuit of an ideal CT and it explains how it would create an error owing to burden mismatch.

Zb

S2P2

Ideal CT

I1

N2N1

P1

jX1 IbR2jX2R1

I2

Im

S1

Zm

U2 Z a

Figure 4 - Equivalent circuit of a CT Symbols on the figure carry their usual meanings and Zb represent the secondary load. Simplified secondary side of the equivalent circuit can be represented as shown in Figure 4. Zm is the magnetization impedance and it is assumed to be a constant. Note that the secondary burden is represented by Za+Zb i.e. secondary impedance + burdens of the meter + lead wire. When the secondary burden of the CT varies, it directly affects the secondary current and phase displacement. Accordingly, the accuracy of the CT operated metering system varies.

2b ImZxZ

mZI

........ (1)

Where Zx = Za+Zb

Zx = Za+Zb can be varied when vary the lead wire length/cross section. When the wire length Zb increases, Ib get decreases. With this low burden condition, positive deviation of accuracy can be observed. According to the above analysis over burden condition may lead to make negative error of accuracy. In this project, lead wire length was varied and selected the best cross sectional area to give the sufficient burden for secondary.

Lead wire burden in VA =

CS

2D2I …… (2)

Where:

I = Secondary current D = Lead wire distance in meter. CS = Cross section area of lead wire. = Conductivity of copper

4.2 Effect of the primary power factor This is a theoretical innovative explanation carried out in this section of the research to illustrate the effect of primary power factor on the accuracy of CTs. Practically the transformation ratio of a CT is 1: N. Figure 5 shows the single line diagram of a CT referred to primary side. Practically current transformer does not reflect the primary phase angle to the secondary side as it is. It allways gives a leading power factor error. Figure 6 vector diagram represents the magnitude and phase displacement error of two states of primary power factors bold and normal. The reflected magnitude component is smaller when primary power factor decreases but power factor improves due to adding of phase displacement error.

I p Ip1 Is

Iw NsIs

Ep

Rp

Im NP

Ie

jXp

Ns

(NP =1)

Zx

Figure 5 - Equivalent circuit referred to primary side Figure 6 illustrate that the phase displacement error (δ) is the different between primary power angle (θ) and phase angle error of the CT (β). Power measurement error due to primary power angle is represented by equation (3) and graphical presentation is given by Figure 7. CT



Zb

S2P2

Ideal CT

I1

N2N1

P1

jX1 IbR2jX2R1

I2

Im

S1

Zm

U2 Z a


2b ImZxZ

mZI

........ (1)

Where Zx = Za+Zb



CS

2D2I …… (2)

Where:



I p Ip1 Is

Iw NsIs

Ep

Rp

Im NP

Ie

jXp

Ns

(NP =1)

Zx


ENGINEER5ENGINEER 4


Zb

S2P2

Ideal CT

I1

N2N1

P1

jX1 IbR2jX2R1

I2

Im

S1

Zm

U2 Z a


2b ImZxZ

mZI

........ (1)

Where Zx = Za+Zb



CS

2D2I …… (2)

Where:



I p Ip1 Is

Iw NsIs

Ep

Rp

Im NP

Ie

jXp

Ns

(NP =1)

Zx


5 ENGINEER

error is increased when the primary power factor decreases.

θ2 NpIp1

β2θ1 β1

Ie

Im

NsIs δ1

NsIs

NpIp

NpIpNpIp1NpIe

NpIe

Ep

δ2

δ = (θ-β)

Iw

Figure 6 – Vector diagram Primary load power factor = θ Phase angle error = β (leading) Actual power = VI Cosθ Measured Power = VI Cos (θ-β)

CT Error =

CosθVI

CosθVIβθCosVI

= Cosβ + Tanθ.Snβ-1 .… (3)

0

2

4

6

8

10

12

14

16

18

20

1 0.9 0.8 0.7 0.6 0.5 0.4 0.3 0.2 0.1

Power Factor

C T

E

r r o

r (%

) 0.0 Deg

0.5 Deg

1.0 Deg

Figure 7 - Power measurement error vs. power

factor and phase angle error. 5. Methodology 5.1 Proposed solution for improvement Laboratory experiments were performed to analyse the existing metering system from which the chart based solution is proposed for optimum accuracy. Figure 8 shows the 100/5

semi indirect metering model prepared for this experiment.

Figure 8 - 100/5 Semi indirect metering model

5.2 Shifting load to the best performance range of the CT It was observed that the best performance range of the CT is 80% In to 120% In. But most of the installations are operating well below this range. IEC regulations states that the performance range of CTs should be maintained up to 120% In [6]. In this research work all the laboratory experiment were carried out for the operating current range from 10% In to 120% In to give a better feeling to the reader. 5.3 Appropriate burden matching According to the surveyed data (2.4), it can be observed that the opposite burdens of the CTs are very much lower than the rated burden. With this lower burden condition, positive deviation of accuracy is observed. The theoretical analysis shows that the over burden condition may lead to make negative error for metering accuracy. Also, in this research study, it is proposed to vary the lead wire length and the cross sectional area to give the sufficient burden for the CT secondary. The experiments were carried out for different wire lengths, loads and power factor conditions to get an insight feeling to see the metering accuracy variation. Length of the lead wires were increased gradually from 1-8 meters and power factor varied from 1.0-0.8. Results were observed using high accuracy working standard of class 0.02. A CT of 100/5 having 5VA and a class 0.5 meter of having 0.5 VA burdens were used to execute the test experiment. It is noted that the burden value variation for different wire lengths, cross


sectional area and secondary CT current values are linearly distributed as shown in Figure 9.

0

1.01

2.02

3.03

4.04

0

0.7021.4

2.1

2.81

0

2.53

5.05

7.5

10.01

0

1.75

3.5

5.26

7.01

0

2

4

6

8

10

12

0 2 4 6 8Wire Length (m)

Wire

Bur

den

(VA

)

2.5 mm2 & 5A

2.5 mm2 & 6A

1.0 mm2 & 5A

1.0 mm2 & 6A

Figure 9-Burdens of lead wires

6. Test Results 6.1 Variation of accuracy against wire lengths under different load conditions The least burden condition of 100/5 semi indirect metering model is observed at the point of 1m wire length and 1mm2 cross sectional area. This is shown in Figure 10. The least burden condition made a positive error which is more than 2% at the optimum range of operating current of the CT and at power factor 1.0. When the wire length is increased gradually, the overall error has come down closer to zero and shifted to the negative region. Even though it is not practical in semi indirect scenario, this can be noted when the wire is at 8m length as shown in Appendix-B, Figure B4. Also, It can be noticed from Figure 10 that the trend lines of overall accuracy between 80% and 120% of nominal current In behaves in more linear manner. Appendix-B, Figure B1 to B4 shows overall accuracy variation of the semi indirect metering system for different wire lengths.

2

2.5

3

3.5

4

4.5

5

5.5

10 20 30 40 50 60 70 80 90 100 120

Nominal current (%In)

Ove

rall

accu

racy

(%)

pf=1

pf=0.98pf=0.95

pf=0.9pf=0.85

pf=0.5

pf=0.8

Accuracy variation for 1 m wire length

Figure 10 - Accuracy variation for different loads and pf for 1m lead wire length

6.2 Variation of accuracy against power factor under different load conditions Maximum expected overall accuracy of this 100/5 test model would be 1% due to 0.5 accuracy class of the energy meter and current transformer. But all the results in every state of the experiment were more than 2%. As per the results obtained by the experiments, wire length (burden) and power factor gives significant deviation for the accuracy. At unity power factor condition, its nominal current variation not much effect for the accuracy (around 0.2%). Appendix-C, Figure C1 to C5 illustrates above variation for power factor 1, 0.98, 0.95, 0.9 and 0.85 respectively.

6.3 Variation of accuracy against voltage Experiments were carried out to find out the effect of voltage variation for accuracy of analogue and digital meters depicted in Table 3 at base current and pf=0.8. Figure 11 shows that both analogue and digital meter are capable of staying at its class even the voltage varied from 150V to 270V. Static meter remained at a significant value of 0.03% all over the range. So the effectiveness of voltage variation for the accuracy of semi indirect metering system is negligible.

Table 3 – Meters used for voltage variation

Error (%) Vs Voltage

0.537

0.957

0.7850.682 0.643

0.583

-0.2

0

0.2

0.4

0.6

0.8

1

1.2

150 200 240 260 265 270

Erro

r (%

)

Voltage Analogue meterVoltage Digital meter

Voltage (V)

Figure 11 - Accuracy variation for different voltages at base current & pf = 0.8

6.4 Research outcome Several experiments were carried out to analyze the effectiveness of the identified parameters and the behavioural patterns are as shown in Figure 12. It could be noted here that the power factor, nominal current, secondary burden

Meter type Model IBase Class Analogue Iskra 20 A 2

Digital Edmi 5 A 0.5



0

1.01

2.02

3.03

4.04

0

0.7021.4

2.1

2.81

0

2.53

5.05

7.5

10.01

0

1.75

3.5

5.26

7.01

0

2

4

6

8

10

12


Wire

Bur

den

(VA

)

2.5 mm2 & 5A

2.5 mm2 & 6A

1.0 mm2 & 5A

1.0 mm2 & 6A



2

2.5

3

3.5

4

4.5

5

5.5

10 20 30 40 50 60 70 80 90 100 120


Ove

rall

accu

racy

(%)

pf=1

pf=0.98pf=0.95

pf=0.9pf=0.85

pf=0.5

pf=0.8







0.537

0.957

0.7850.682 0.643

0.583

-0.2

0

0.2

0.4

0.6

0.8

1

1.2

150 200 240 260 265 270

Erro

r (%

)


Voltage (V)





ENGINEER7ENGINEER 6


0

1.01

2.02

3.03

4.04

0

0.7021.4

2.1

2.81

0

2.53

5.05

7.5

10.01

0

1.75

3.5

5.26

7.01

0

2

4

6

8

10

12


Wire

Bur

den

(VA

)

2.5 mm2 & 5A

2.5 mm2 & 6A

1.0 mm2 & 5A

1.0 mm2 & 6A



2

2.5

3

3.5

4

4.5

5

5.5

10 20 30 40 50 60 70 80 90 100 120


Ove

rall

accu

racy

(%)

pf=1

pf=0.98pf=0.95

pf=0.9pf=0.85

pf=0.5

pf=0.8







0.537

0.957

0.7850.682 0.643

0.583

-0.2

0

0.2

0.4

0.6

0.8

1

1.2

150 200 240 260 265 270

Erro

r (%

)


Voltage (V)





7 ENGINEER

(Wire length) were effectively changed the overall accuracy of the semi indirect metering system. Effectiveness of voltage variation is negligible for solid state meter for wide range of voltage variation.

Error (%)

Power factor

(a)

Error (%)

Nominal currentIn %

(b)

Wire Length (m)

Error (%)

(c)

Error (%)

Voltage (V)

(d)

Figure 12 - Behavioural pattern of the

parameters

a) Variation against power factor. b) Variation against nominal current c) Variation against wire length. d) Variation against voltage

6.5 Component selection criteria Before selecting an appropriate wire length for a particular installation, it is important to select the CT ratio for better accuracy. This describes in the flow chart shown in Figure 13. The wire length selection chart of Figure 14 is developed for 0.5% accuracy only. Here the maximum allowable secondary burden kept at 5VA as mention in CT name plate data. The operating power factor of the installation is highly affected for metering accuracy as discussed earlier. Therefore the wire length selection criterion is restricted on the operating power factor as well as operating current. Hence all the selections are on the safe side and do not fall onto the negative side of the metering error. Flow chart of figure 14 guides the sequence of component selection under two main criterions. Criteria-1, Maximum allowable wire length for opposite burden of a given CT Criteria-2, Maximum permissible wire length for optimum accuracy and given pf. Example: 1 2

Contract demand (kVA) 80 55 Contract demand current (A) 115.9 79.7 CT Secondary current (A) 6.0 4.0 Operating power factor 0.98 1.0 Burden of the CT (VA) 5 5 Wire length (m) 3.2 5.1

Existing system

Calculate the currentfor contract demand

Wire length>Length selected for average pf

(Fig. 14)

Replace the CTWith a closer one

Calculate the

Secondary burdenFor contract demand

Is the secondaryburden = 80% In

of the CT burden

Select the wire lengthFor 80% burden (Fig. 14)

Select the wire length for average pf (Fig. 14))

Improved system

Is the Contract demand ≤ 120 %In

of the CT or closer

No

Yes

No

Yes

Yes

No

Figure 13 - Flow chart

Wire length selection table for 100/5 CT with 5VA burden for + 0.5 % accuracy when the pf=1, 0.98, 0.95, 0.9 and 0.85

0

2

4

6

8

10

12

14

0 2 4 6 8 10Wire Length(m)

Wir

e B

urde

n(V

A)

Burden for 6 A

Burden for 5.5 A

Burden for 5 A

Burden for 4.5 A

Burden for 4 A

Burden for 3.5 A

PF=1

PF=0

.95

PF=0

.98

PF=0

.9

PF=0

.85

5 VA

0.5 VA for m

eter burden +0.5 VA safety

Figure 14


7. Conclusion One of the major findings derived from this research is that the secondary circuits equipped with solid state energy meters installed under semi indirect metering system are operating at under burden conditions depending on the connected CT with high burden. The actual opposite burdens of the measuring CTs are less than 25% of the rated burden. Therefore the system operating at under burden condition does not yield to exact measurement of energy thus; an improvement is required. Method of accuracy improvement through maximization of CT performances and appropriate burden matching is proposed here. It explains the methodology of selecting CTs and lead wire lengths carefully while keeping the utility in the safe side. For given measuring CTs, electricity meters and known cross-section, the desired total burden of the secondary circuit can be achieved by an appropriate selection of the conductor length. Moreover the other charts can be developed for more accuracy such as 0.2% under extreme condition subjected to the p.f. constraints. The attractive feature here is that the low cost of implementation and the least technical expertise requirement. Also reliability is much greater compared to electronic compensation and no routine inspection and periodical calibration are required. Acknowledgement The authors would like to offer their appreciation to Eng. S.R.K. Gamage DGM, Planning & Development Branch, Distribution Division-1 and Eng. Kamal Perera CE, Development Branch, Distribution Division-1 of CEB for their assistance by providing facilities for this research study and many thanks for the helpful comments suggestions and the guidance provided by Dr. Narendra de Silva of LECO. References 1. Ceylon Electricity Board, 2006. Static three

phase energy meters accuracy class 1 CT connected meters for remote reading, CEB Standard 071-3:2006

2. Ceylon Electricity Board, 1998. Low voltage

ring type measuring current transformers, CEB Standard 027-3:1998

3. Ceylon Electricity Board, STEEL BOXES FOR HOUSING BULK SUPPLY METERS, CEB Standard 050-1997.

4. CIRED, 20th International Conference on

electricity distribution, page 0105, Prague, 8-11 June 2009.

5. EDISON ELECTRIC INSTITUTE, 2002,

Hand Book for Electricity metering Tenth Edition, EDITION ELECTRIC INSTITUTE, Washington.

6. International Electro Technical commission,

2003-2, Current Transformer, International Standard, IEC 60044–1.


2003-1, Static meter for active energy, International Standard, IEC 62053–22, First edition.


2003-1, Electromechanical meters for active energy, International Standards, IEC 62053–11, First edition.

9. http://www.arwtransformers.co.uk/CTInfo,

Visited, 14th December 2013. 10. http://www.currenttransformers.co.in, Visited, 14th December 2013. 11. http://powermetrix.com,Visited, 14th

December 2013. 12. http://www.umsmeters.co.uk, Visited,

14th December 2013. 13. http://www.zera.de, Visited, 10th January

2014. 14. http://www.edmi-meters.com, Visited,

12th February 2014.






















12th February 2014.

ENGINEER9ENGINEER 8





















12th February 2014.

9 ENGINEER

Accuracy variation for 6m wire length

-1

0

1

2

3

4

5

6

7

8

9

10 20 30 40 50 60 70 80 90 100 120


Ove

rall

accu

racy

(%)

pf=0.5

pf=0.8

pf=0.85pf=0.9pf=0.95pf=0.98pf=1.0


0

1

2

3

4

5

6

7

10 20 30 40 50 60 70 80 90 100 120


Ove

rall

accu

racy

(%)

pf=0.5

pf=0.8

pf=0.85pf=0.9

pf=0.95pf=0.98pf=1.0


1

2

3

4

5

6

10 20 30 40 50 60 70 80 90 100 120


Ove

rall

accu

racy

(%)

pf=0.5

pf=0.8

pf=0.85pf=0.9

pf=0.95

pf=0.98

pf=1.0


-2

0

2

4

6

8

10

10 20 30 40 50 60 70 80 90 100 120


Ove

rall

accu

racy

(%)

pf=0.5

pf=0.8pf=0.85pf=0.9pf=0.95pf=0.98pf=1.0

Appendix-A

Table - A1

** including measurement error

Appendix-B

Customer Factory-1 Factory-2 Factory-3 Factory-4 Factory-5 Contract demand (kVA) 1000 630 400 630 630

Average load (kVA) 300 201 69 320 196 Avg. line current (A) 421 293 92 408 277 Average power factor 0.99 0.89 0.98 0.97 0.96

CT type ITL ARW ITL ITL ARW CT Accuracy class 01 01 01 01 01

CT ratio 1600/5 1200/5 800/5 1200/5 800/5 CT Burden (VA) 15 15 15 15 15

Meter type Premier Premier Premier Premier Edmi Accuracy class 01 01 01 01 01

Meter Burden (VA) 0.5 0.5 0.5 0.5 0.5

Length of Lead wires (m) L1 - 1.1 L2– 1.0 L3 – 0.9

L1- 1.0 L2 – 1.1 L3 – 1.2

L1 – 0.9 L2 – 1.0 L3 – 1.1

L1 – 1.0 L2 – 1.0 L3 – 1.0

L1 – 1.0 L2 – 1.0 L3 – 1.0

Area of the lead wire (mm2) 2.5 2.5 2.5 2.5 2.5 Meter error 0.34 0.31 -0.36 0.25 0.32

Overall error** 2. 94 3. 3 3.15 3.1 3.2 Average current as % In 26% 24.4% 11.5% 34% 34.6%

Figure B1 Figure B2

Figure B3 Figure B4


For Power Factor 1.0

-1

-0.5

0

0.5

1

1.5

2

2.5

3

1 2 4 6 8

Wire Length (m)

Acc

urac

y (%

)

120% In

100% In

90% In

80% In

70% In

60% In


-0.5

0

0.5

1

1.5

2

2.5

3

1 2 4 6 8

Wire Length (m)

Acc

urac

y (%

)

120% In

100% In

90% In

80% In

70% In

60% In


0

0.5

1

1.5

2

2.5

3


Acc

urac

y (%

)

120% In

100% In

90% In

80% In

70% In

60% In

-1

-0.5

0

0.5

1

1.5

2

2.5

3

3.5

1 2 4 6 8 10

Wire Length(m)

Acc

urac

y (%

)

120% In

100% In 90% In

80% In 70% In

60% In


Appendix-C

Figure C1 Figure C2

Figure C3 Figure C4



-1

-0.5

0

0.5

1

1.5

2

2.5

3

1 2 4 6 8

Wire Length (m)

Acc

urac

y (%

)

120% In

100% In

90% In

80% In

70% In

60% In


-0.5

0

0.5

1

1.5

2

2.5

3

1 2 4 6 8

Wire Length (m)

Acc

urac

y (%

)

120% In

100% In

90% In

80% In

70% In

60% In


0

0.5

1

1.5

2

2.5

3


Acc

urac

y (%

)

120% In

100% In

90% In

80% In

70% In

60% In

-1

-0.5

0

0.5

1

1.5

2

2.5

3

3.5

1 2 4 6 8 10

Wire Length(m)

Acc

urac

y (%

)

120% In

100% In 90% In

80% In 70% In

60% In


Appendix-C

Figure C1 Figure C2

Figure C3 Figure C4

ENGINEER11ENGINEER 10


-1

-0.5

0

0.5

1

1.5

2

2.5

3

1 2 4 6 8

Wire Length (m)

Acc

urac

y (%

)

120% In

100% In

90% In

80% In

70% In

60% In


-0.5

0

0.5

1

1.5

2

2.5

3

1 2 4 6 8

Wire Length (m)

Acc

urac

y (%

)

120% In

100% In

90% In

80% In

70% In

60% In


0

0.5

1

1.5

2

2.5

3


Acc

urac

y (%

)

120% In

100% In

90% In

80% In

70% In

60% In

-1

-0.5

0

0.5

1

1.5

2

2.5

3

3.5

1 2 4 6 8 10

Wire Length(m)

Acc

urac

y (%

)

120% In

100% In 90% In

80% In 70% In

60% In


Appendix-C

Figure C1 Figure C2

Figure C3 Figure C4

11 ENGINEER

-0.5

0

0.5

1

1.5

2

2.5

3

3.5

1 2 4 6 8 10

Wire Length(m)

Acc

urac

y (%

)120% In

100% In

90% In

80% In

70% In

60% In


Figure C5

ENGINEER13


ENGINEER 1

Wind Loads on High-Rise Buildings by Using Five Major International Wind Codes and Standards

A. U. Weerasuriya and M.T.R. Jayasinghe

Abstract: A high-rise building of height – 183 m was employed to evaluate similarities and differences of wind load calculations done by using five major wind codes and standards. Evaluation was done in both ultimate and serviceability limit conditions. Member forces in columns, and beams, compressive stress in shear walls and support reactions obtained from finite element modelling was used to assess building responses in ultimate limit condition. Along and across wind, accelerations and drift indices were engaged to estimate serviceability limit state performances. Available 3 second gust wind speeds are converted into mean hourly and 10 minute average wind speeds to calculate wind loads on building. Wind speeds with 5 years return period was used in building acceleration calculation. The simultaneous use of higher terrain-height multiplier and importance factor may be lead to over design, even in cyclone prone areas. The use of post disaster wind speed does not exceed the drift limit but exceeds threshold acceleration value in across insert wind acceleration. Keywords: Wind loading standards, Design wind speeds, Along-wind acceleration, Across-wind acceleration, Drift index 1. Introduction High-rise buildings are more susceptible to wind loads due to their excessive heights, use of lightweight materials for super structure and inner partition walls, low structural damping, etc. Thus, design of high-rise buildings against wind loads is always a challenging task for designers. However, due to scarcity of land and high land cost, and for the glory of developer, building skyscrapers in urban areas is indispensable. In recent time, there has been a national trend to build tall, slender tower type high rise buildings in Sri Lanka especially in Colombo city limit [1]. A properly designed high-rise building should satisfy both ultimate and serviceability limit state conditions against vertical and lateral loads. To ensure a properly designed building, wind loading standards are serving as a primary design tool at early design stage in the absence of more sophisticated testing methods such as Wind Tunnel tests, Computational Fluid Dynamic (CFD) simulations, A proper wind loading standard would be able to calculate wind load fairly accurately and be able to check serviceability criteria by considering both static and dynamic behaviour of the building. However, early wind loading standards are not covering either dynamic analysis or serviceability limit criteria such as maximum allowable wind acceleration. In Sri Lankan context, neither the most common wind loading standard CP3 Chapter V-Part 2: 1972[2] nor the only available mandatory document on wind loading Design Manual-―Design

buildings for high winds, Sri Lanka‖[3] covers the above two aspects. Therefore, Sri Lankan engineers employ different international wind loading standards to calculate wind loads on high-rise buildings. However, direct adaptation of these international standards leads to some problems such as adaptability of country specific factors in Sri Lankan conditions, compatibility of wind loading standards with other design standards, achieved level of risk level remaining unknown, effects of differences and similarities of wind load calculations of different standards, satisfaction of serviceability conditions of a building, etc. In this study, a 183 m tall office building was selected as a case study to compare wind load calculation done by using five major international wind loading standards. Those standards were selected on a rational basis by considering their popularity and familiarity in Sri Lanka. The differences and similarities of wind load calculations are addressed by using forces on structural members such as columns, beams, and shear walls obtained via finite element modelling (FEM). In this paper section 2 explains the reasons for selected codes and standards; section 3 illustrates the case study. Section 4 and 5 show the results for ultimate and serviceability limit

Eng. A.U.Weerasuriya, B.Sc.Eng(Hons), M.Sc. (Moratuwa), AMIE(Sri Lanka), PhD candidate, Department of Civil and Environmental Engineering, Hong Kong University of Science and Technology, Hong Kong. Eng. (Prof). M.T.R. Jayasinghe, B.Sc.Eng(Moratuwa), PhD (Cambridge), C.Eng, MIE(Sri Lanka), Professor, Department of Civil Engineering, University of Moratuwa.


ENGINEER 14

ENGINEER 2

states. Finally section 6 makes a conclusion about advantages and disadvantages of selected wind loading standards and proposes a suitable wind loading standard to assess wind loads on buildings in Sri Lanka. 2. Selected codes and standards

for the study The selection of wind loading standards is mainly based on its ability to address dynamic response of a high-rise building. This is because of slenderness, low damping ratio and flow mechanisms such as vortex shedding which tend to cause high-rise building to excite dynamically. However, the most common wind loading standard use in Sri Lanka, CP3 Chapter V-Part2:1972[2] could not be used to calculate wind loads on dynamically excited buildings as it employs a quasi-static approach. Therefore, many other international standards such as BS 6399.2:1997 [4], AS 1170.2:1989[5], AS/NZS 1170.2:2002[6] and EN 1991-1-4:2005[7] are used to design high-rise buildings in Sri Lanka. The new British standard BS 6399.2:1997[4] uses equivalent static method with an augmentation factor to evaluate dynamic behaviour of high-rise buildings up to 200 m height. In addition to the augmentation factor, there are provisions given to calculate design wind speed and pressure coefficients for dynamic analysis. The building and terrain height factor (Sb) defined in this standard can be used to convert 3 second gust wind speed to mean hourly wind speed [8]. AS 1170.2:1989[5] standard has a detailed method for dynamic analysis, which can be used for a building with height or length-to-breadth ratio greater than 5 and a first- mode of vibration of less than 1 Hz. By using the gust factor method, this standard encounters most of the dynamic nature of wind such as resonance factor, background factor, damping ratio, peak factor for both approaching wind and the building, size of the building and spectrum of the approaching wind. This standard employs a 3-second gust wind speed as the basic wind speed which allows adopting this code directly in Sri Lankan context. Instead of using gust factor, AS/NZS 1170.2:2002 [6] uses aerodynamic shape factor and dynamic response factor to analysis dynamic behaviour of high-rise buildings. The Aerodynamic shape factor adjusts factors such as correlation of pressure on opposite sides of a building, load shearing effects between adjacent areas, local pressure effect, etc. The dynamic factor

considers encountering different characteristics of dynamic behaviour of the building and dynamic nature of the approaching wind. The basic wind speed used in latest Australian standard is mean hourly wind speed. The Euro standard, EN 1991-1-4:2005[7] is the first real multinational wind loading standard. In this study it is used with the National annex prepared for United Kingdom. The dynamic nature of the wind loading is captured via structural factor, which takes in to account the effect on wind actions from the non-simultaneous occurrence of peak wind pressures on surfaces together with the effect of the vibrations of the structure due to turbulence. Comparison of wind loading standards is a common objective accomplished by researchers before adopting them directly in any country other than its origin. Bashor and Kareem [9] compared six major international wind loading standards for both ultimate and serviceability limit states to identify parameters for further theoretical studies. A series of papers was published in Asia Pacific Conference of Wind Engineering (APCWE) to harmonize different international wind loading standards used in Asia-Pacific countries ([10], [11], [12]). These studies employed base shear, bending moments and acceleration at the top of the building for the comparison purpose. Results of the comparison revealed that wind loads calculation using different codes and standards have significant discrepancies though they are evaluated for the same wind climate conditions. Aforementioned 3. Case studies A 183m high Commonwealth Advisory Aeronautical Research Council (CAARC) standard building model [13] was selected as the case study for wind load calculation. The plan dimension of this building is 46 m x 30 m which gives height-to-breadth ratio 6.1 (>5) anticipating dynamic excitation due to wind loads. The same building model was used by previous researchers [10,11,12] to compare wind loads derived by using wind loading standards. For the load calculation purpose, this building is assumed as an office in an urban area of Sri Lanka. It is also designed as a 51 storey building with floor-to-floor height of 3.6 m. To represent the realistic structural behaviour, structural elements and internal spaces are designed in details. The basic structural form is wall-frame structure. The columns and beams form the frame and shear walls located around the lift and service core.

ENGINEER 14

ENGINEER 2




ENGINEER15

ENGINEER 2




ENGINEER 3

Within the service core a hard zoning lift system, washrooms and ducts are located. The dimensions of structural members designed according to vertical loads are given in Table 1. A general purpose finite element computer program SAP 2000 [14] was used to model the 183 m building in order to obtain member forces for comparison. SAP2000 is object-based modelling software which automatically converts the object-based model into an element-based model that is used for analysis. This element-based model is called the analysis model and consists of traditional finite element objects such as frame elements, shell elements and joints (nodes). A frame element with 6 degree of freedom is modelled as a straight line, connecting two points which has its own local axis to define section properties, loads and displacements. Frame elements can carry axial forces, shear and bending forces. Thus, frame members are used to model columns and beams. Shell elements that are used for shear walls and slabs are 4 node elements with membrane and plate-bending behaviour. For the slab, diaphragm constraint was used to preserve membrane action. Other than dead load of structural members, super imposed and

live loads were applied in the model according to the BS6399: Part 1: 1996 [15]. Wind loads on building are applied with respect to wind flow in two orthogonal directions as joint loads at the column - beam junctions on the wind ward and leeward faces separately (Figure 1 (a) and (b)). The accuracy of the FEM model was checked by using fundamental frequency of the model and provision given in Australian standards. Both Australian standards define fundamental frequency as

1

46.................................. (1)n Hz Eq

h

Where, h is the building height. The calculated fundamental frequency is 0.2513 Hz. The first translation mode of the FEM model is 0.2197 Hz shows good aggreement with the calculated fundamental frequency. The slight descripancy may be due to loading and material properties assumed in FEM modelling. Other than that, it can be persumed that FEM model is accurate enough to obtain results for further analysis.

Table 1- Dimensions of structural elements in 183 m height building (all dimensions are in millimetres)

Figure 1 - Finite element 3 – D models of (a) 3-D view of 183 m height building (b) wind loads applied on windward and leeward sides of the building separately.

Structural member

Ground floor – 10th floor

11th floor – 20th floor

21st floor-30th floor

31st floor – 40th floor

41st floor-51st floor

Column 1050 x 1050 950 x 950 800 x 800 700 x 700 500 x 500 Shear wall 300 thickness 250 thickness 200 thickness Beam 600 x 400 (depth x width) Slab 175 thickness

(a) (b)

ENGINEER 16

ENGINEER 4

4. Basic wind speeds with different averaging time The provisions of wind load calculation given in different wind loading standards are based on basic wind speeds with different averaging times. These averaging times are range from 3 second to 1 hour. However, the only available wind speed data in Sri Lanka are based on 3 second gust wind speed. Therefore, it is necessary to convert 3 second gust wind speeds into different averaging time wind speeds by using proper conversion factors. This conversion should be done cautiously as erroneous conversion might lead to have unrealistic wind loads. For this study, conversion was done as follows. First, 3 second gust wind speed is converted in to mean hourly wind speed as proposed by Cook [8] (Eq 1). This method is based on the terrain and building height factor given in BS

6399 - 2:1997[4]. One of the advantages of this method is that it can accommodate site and building characteristics such as the distance away from sea.

..................................... 1mean G bV V S Eq Where, VG is the available 3 second gust wind speed. Sb is the site and building height factor for 10m height in an open terrain as defined in Table 04 of BS 6399 - 2:1997[4] for standard method. The conversion factor 1.06 is used to convert mean hourly wind speeds to 10 minute mean speed, as it was proposed by the Institute of Civil Engineers in United Kingdom (ICEUK) [16]. The wind speeds in all three zones with different average times are shown in Table 1.

Table 2 - Basic wind speeds with different averaging time

5. Wind induced forces Wind loading standards only facilitate calculation of wind pressures on building faces at different heights. Multiplying calculated pressures with corresponding contributory areas will give the wind load at that height. However, this is not the actual load experienced by structural members as they have load sharing mechanisms. On the other hand structural members are designed by considering the actual loads acting on them rather than considering loads on the structure. The actual structural load acting on any member can be obtained by using the 3-D finite element model by applying external loads. Post disaster wind speeds in all three zones as

shown in Table 1 were used for wind calculation. The use of higher post-disaster wind speeds compared to low wind speeds for normal structures is recommended based on views of previous researchers ([17], [18]) considering; uncertainty of derivation of basic wind speeds, lower wind speeds compared to other countries with similar topographical and wind conditions, and as a conservative approach to achieve higher margin of safety, etc. Wind loads are calculated by using dynamic methods as given in standards to encounter anticipated dynamic response of the high-rise buildings. In addition to that, for wind load calculation by using AS 1170.2:1989[5] and AS/NZS 1170.2:2002[6] used higher terrain height multiplier in zone 1 to encounter higher

Zone 1 (ms-1) Zone 2 (ms-1) Zone 3 (ms-1)

Nor

mal

stru

ctur

e

Post

dis

aste

r

stru

ctur

e

Nor

mal

stru

ctur

e

Post

dis

aste

r

stru

ctur

e

Nor

mal

stru

ctur

e

Post

dis

aste

r

stru

ctur

e

CP 3 : Chapter V : Part 2 : 1972 (3 second gust wind speed) 49 54 43 47 33 38

BS 6399 - 2:1997 (Mean hourly wind speed) 27 30 24 26 18 21

BS EN 1991-1-4:2005 (10 minutes mean wind speed) 28 32 25 28 19 22

AS 1170.2 -1989 (3 second gust wind speed) 49 54 43 47 33 38

AS/NZS 1170.2:2002 (3 second gust wind speed) 49 54 43 47 33 38

ENGINEER 16

ENGINEER 4








Nor

mal

stru

ctur

e

Post

dis

aste

r

stru

ctur

e

Nor

mal

stru

ctur

e

Post

dis

aste

r

stru

ctur

e

Nor

mal

stru

ctur

e

Post

dis

aste

r

stru

ctur

e






ENGINEER17

ENGINEER 4








Nor

mal

stru

ctur

e

Post

dis

aste

r

stru

ctur

e

Nor

mal

stru

ctur

e

Post

dis

aste

r

stru

ctur

e

Nor

mal

stru

ctur

e

Post

dis

aste

r

stru

ctur

e






ENGINEER 5

cyclone risk of that area.The importance factor (Mi=1.1)is also employed for AS 1170.2:1989[5] calculation in zone 1 to evaluate effect of use of Mi together with higher terrain-height multiplier on wind load calculation. Member forces used for the comparison of this study are maximum axial force, shear force and bending moment in columns, shear force and bending moment in beams, base moment and base shear at supports and maximum compressive stress in shear wall. The member forces are calculated for the following load combinations: 1. 1.2(Dead loads [G] ) + 1.2(Live load[Q]) +

1.2(Wind load[W]) 2. 1.0 (Dead loads[G]) + 1.4(Wind load[W]) 3. 1.4 (Dead loads[G]) + 1.4(Wind load[W]) and 4. Wind load only([W]).

For the purpose of comparison, the obtained results are shown as normalised forces with respect to CP 3 Chapter V- Part 2:1972[2], which is the most common practice in Sri Lanka. Wind induced forces such in columns, beams, supports and shells on 183 m high building in all three zones are shown in Figure 2 to 10. 6.Results 6.1. Member forces The maximum loads in columns and beams are observed for the load combination 1.2G+1.2Q+1.2W. Thus it can be considered as the governing load combination for structural design. According to Figures 2(a) and (b), there is a significant difference between loads derived for Australian standards and rest of standards especially in zone 1. The loads derived from Australian wind standards are much larger than loads derived from other standards in zone 1. The maximum differences in bending moments in column and beams are 1.35 and 1.48 for wind flow perpendicular to 46m side and 1.35 and 1.22 when wind flow perpendicular to 30m side of the building.. For AS/NZS 1170.2:2002 maximum loads in column and beams are 1.16 and 1.4 respectively for wind flow perpendicular to 46m side and 1.35 and 1.15 for wind flow perpendicular to 30 m side of the building. This is primarily due to the use of higher terrain-height multiplier in zone 1 contributed to larger wind load derivations of Australian standards. However, the substantial differences of loads between AS 1170.2:1989[5] and AS/NZS 1170.2:2002[6] arose from use of importance factor (Mi=1.1) only for calculation done with AS 1170.2:1989[5]. This concludes that the use of higher terrain-height multiplier and importance

factor together may lead to overestimate of wind loads even in cyclone prone areas. Therefore, designer can decide to employ either importance factor or terrain-height multiplier for a design unless necessary to design highly cyclone resilient structure such as cyclone shelter. The effect of different wind loads calculations are clearly visible of reults obtained for load combination 1.0G + 1.4W, where effect of dead load is minimum. The larger wind loads resulted from Australian standards in zone 1 are again reflected from Figures3 (a) and (b). Column and beam loads obtained from zone 2 and zone 3 are approximtely same for all wind loading standards suggests that without using especial factors such as higher terrain-height multiplier,wind load calculations would be similar for all standards. However, the aforementioned fact should be closley inspected as member forces could be varied greatly not only in the use of especial factors but also the manner wind loads are applied on a building. For an example, the same load combination 1.0G + 1.4W displayed larger variation in zone 2 when wind flow perpendicular to 30m building side (Figure 7). There are rather higher coulum and beam loads in zone 2 and 3 for both Euro and British standars (Figure 7). This might be resulted from wind load application method engaged by those two standards. The division-by-parts rule leads to applying higher wind loads over a height range compared to linear variation of wind load with height as assumed in CP 3 Chapter V: Part 2:1972[2]. Bending moments in coulum and beams reach a maximum for load case 1.4G + 1.4W due to the combine effect of higher dead load and wind loads. When wind flow perpendicular to 46 m side, the bending moments obtained for Australian standards can be as high as 1.85 and 1.4 for columns and beams respectively (Figures 4(a) and (b)). When wind flow perpendicular to 30m side, those corresponding values are 1.75 for coulmns and 1.4 for beams for both Australian standards(Figures 8(a) and (b)). Wind load only case represents the individual differences in wind load calculations and its application on buildings. From Figures 5(a) and (b), it reflects that effect of used of both important factor and higher terrain height multiplier in AS 1170.2:1989[5]. The combined effect of these two factors lead to have wind loads two times greater than wind loads derived from CP 3 Chapter V: Part 2:1972[2]. The importance factor alone may increase the wind laod about 50% in columns. More surprisingly, EN 1991-4:2005[7] yields higher wind loads compared to both BS 6399.2:1997 [4]

ENGINEER 18

ENGINEER 6

and CP 3 Chapter V: Part 2:1972[2] due to the use of difference in wind load application and use of a new set of factor different from previous British standards. Particularly, the effect of the use of divison-by-parts rule is

displayed in results of Figures 9 (a) and (b). Both Euro standard and BS 6399.2:1997[4] yeild higher column and beam loads compared to CP 3 Chapter V: Part 2:1972[2].

Figure 2 - (a) Column loads (b) Beam loads for load combination 1.2G+1.2Q+1.2W (wind flow perpendicular to 46 m long side)

Figure 3 - (a) Column loads (b) Beam loads for load combination 1.0G+1.4W (wind flow perpendicular to 46m long side)


(a) (b)

(a) (b)

(a) (b)

ENGINEER 18

ENGINEER 6






(a) (b)

(a) (b)

(a) (b)

ENGINEER19

ENGINEER 6






(a) (b)

(a) (b)

(a) (b)

ENGINEER 7

Figure 5 - (a) Column loads (b) Beam loads for wind load only (wind flow perpendicular to 46 m long side)


Figure 7 - (a) Column loads (b) Beam loads for load combination 1.0G+1.4W (wind flow perpendicular to 30 m long side)

(a) (b)

(a) (b)

(a) (b)

ENGINEER 20

ENGINEER 8


Figure 9 - (a) Column loads (b) Beam loads for wind load only (wind flow perpendicular to 30m long side) 6.2. Base reactions Base reactions are principal measurements of calculated wind loads and its overall effects on the building. Base shear represents the total wind loads on the building and base moment indicates overturning moment resulted from those wind loads. The latter factor is also an indirect indication of wind load distribution along the building height. Figures 10 (a) and (b) show bending moment and shear force in three zones for wind directions perpendicular to 46m side and 30m side respectively. The larger base moment and base shear yielded from AS 1170.2:1989[5] due to use of importance factor. In the zone 2 and zone 3, Euro codes yielded higher base moment as well as base shears values due to both higher wind load derivation and use of division-by-part rule for wind load disrtribution.

The maximum values 2.4 and 2.16 observed for base moment and base shear for AS 1170.2:1989[5]in zone 1 for wind flow perpendicular to 30 m side respectively. BS 6399.2:1997[4] has almost same values for base moment and base shear as CP3 Chapter V – Part 2: 1972[2] values when wind flow perpendicular to 46 m long side. However, when wind flow perpendicular to 30 m side base moment and shear can be as high as1.6 and 1.8 in zone 1 and zones 2, 3 respectively. This is another evidence for difference arising from use of division-by-parts rule in newer British code compared to the previous British code of practice (Table 2).

(a) (b)

(a) (b)

ENGINEER 20

ENGINEER 8




(a) (b)

(a) (b)

ENGINEER21

ENGINEER 8




(a) (b)

(a) (b)

ENGINEER 9

Figure 10 - Base moment and base shear of the 183m building (a) Wind flow perpendicular to 46 m side (b) Wind flow perpendicular to 30 m side

Table 3 - Base moment and base shear of 183m building for two orthogonal directions Wind loading standard Wind flow perpendicular to 46 m side

Base moment (x 106 Nm) Base shear (x 103 N) Zone 1 Zone 2 Zone 3 Zone 1 Zone 2 Zone 3

CP 3 Chapter V - Part 2:1972 2215.9 1478.8 966.7 21625.9 15502.9 10134.1

BS 6399.2:1997 2152.8 1554.8 1007.9 22329.1 16597. 2 10829.3

AS 1170.2:1989 4296.7 1978.4 1358.6 40556.2 19856.4 13871.4

AS/NZS 1170.2:2002 3243.4 1734.0 1120.0 32218.2 17566.8 11345.2

BS EN 1991-1-4:2005 3220.4 1902.0 1315.0 28887.6 20157.4 14046.0

Wind flow perpendicular to 30 m side

CP 3 Chapter V - Part 2:1972 1103.5 772.5 463.8 11520.9 7934.9 4724.1

BS 6399.2:1997 1387.4 1007.9 657.4 16276.6 10828.5 7064.2

AS 1170.2:1989 2626.0 1159.9 745.6 24511.2 11187.2 7196.8

AS/NZS 1170.2:2002 1938.7 1037.8 669.7 19101.5 11659.1 6731.5

BS EN 1991-1-4:2005 1574.4 1363.6 817.6 18197.9 14512.0 8698.6

6.3. Maximum stress in shear walls Maximum compressive stress in shear walls can be observed for wind load derived by using AS 1170.2:1989[5]in zone 1 due to use of higher terrain height multiplier. The normalized maximum stress is about 1.7 and 1.45 for wind flow perpendicular to 46m side and 30m side of the building respectively. However, when wind flow perpendicular to 30m side, Euro standards caused larger stress in shear walls in zones 2

and 3 because of higher wind loads acting on greater area of building sides due to use of divisio-by-part rule. Division-by-part rule also leads to generate higher wind loads, especailly wind flow perpendicular to 30m side for BS 6399.2:1997[4] compared to CP3 Chapter V – Part 2: 1972[2]. .

(a) (b)

ENGINEER 22

ENGINEER 10

Figure 11 - Maximum shell stress in shear wall of the 183m building (a) Wind flow perpendicular to 46 m wall (b) Wind flow perpendicular to 30 m wall

6.4. Drift limit The drift limit is used to evaluate the degree of wind sensitivity of the building. This is an important measurement to prevent damages on non-structural building components including partition walls.The maximum deflection at top of the building in serviceability limit condition is obtained by applying wind loads to the finite element 3-D model. According to the BS 8110-Part 2: 1985[19] the maximum allowable deflection is calculated as hs/500, where hs is the storey height of a single story building. The correspoding allowable deflection for 183m height building is 366 mm. The average drift index is defined as a ratio between maximum deflections to total height of

the building. The calculated drift index values are shown in Table 3. The generally accepted average drift index limit for high rise buildings is 1/500 [20]. By refering Table 3, only wind loads derived from AS 1170.2:1989 in zone 1 exceeds the generally accepted drift limit. This is because of larger wind loads resulted from use of both importance factor and higher terrain-height factor. However, all other cases satisfy the drift index requirement. In zone 3, all models well below the maximum drift limit, approximately by half of its threshold value. Thus, it can be concluded that even wind load calculation done by using post disaster wind speed will not cause conflict with drift index limit in tall buildings.

Table 4 - Drift index calculation for zone 1,2, and 3

7. Along wind and cross wind

acceleration of the building. Only Australian and Euro code facilitate to calculate acceleration at top of the building. Euro code provides a method to calculate only along wind acceleration, while Australian standards offer methods to calculate both along wind and cross-wind accelerations. The

standard deviation of acceleration at height ‗z‘is defined in EN 1991-4:2005[7]

, ,,

. . .. . . ( ).............. (2)f v

a x x l xl x

c b lR K z Eq

m

Where, cf– force coefficient ρ – air density b – width of the structure

Wind loading standard Drift index

Zone 1 Zone 2 Zone 3 CP 3 Chapter V - Part 2:1972 1/961 1/1250 1/1785 BS 6399.2:1997 1/935 1/1219 1/1754 AS 1170.2:1989 1/425 1/862 1/1471 AS/NZS 1170.2:2002 1/565 1/1020 1/1562 BS EN 1991-1-4:2005 1/561 1/1010 1/1538

(a) (b)

ENGINEER 22

ENGINEER 10








, ,,

. . .. . . ( ).............. (2)f v

a x x l xl x

c b lR K z Eq

m




(a) (b)

ENGINEER23

ENGINEER 10








, ,,

. . .. . . ( ).............. (2)f v

a x x l xl x

c b lR K z Eq

m




(a) (b)

ENGINEER 11

lv(zs) – turbulence intensity at the height z = zs above ground R – square root of resonant response Kx– the non dimension coefficient obtained from chart given in Annex B of the code. m1,x - along wind fundamental equivalent mass n1,x– fundamental frequency of along wind vibration of the structureestimated as

1

46n Hz

h

where h is the height of the structure Australian standards use 5 years return period wind speeds to calculate along wind and cross wind accelerations. The serviceability wind

speeds with 5 year return peridod were derived from mean hourly wind speeds with 50 year return period by using probabilistic factor(Sp) proposed in BS 6399.2:1997 as.

5 ln ln 1................ (3)

5 ln ln 0.98p

QS Eq

Where, Q is the annual risk exceedance equals to 0.2 for 5 year return period. Calculated serviceability wind speeds by using Sp = 0.85 are shown in Table 3.

Table 5 - Wind speeds used for acceleration calculations

Comparing Tables 3 and 1, it is clear that the service wind speedswith 5 years return period are closer to wind speed proposed for normal structures in the design manual ―Design building for high winds-Sri Lanka‖[3]. Not only calculated 5-year return period wind speeds are similar to 3-second gust wind speeds for normal structures but also comparable with values proposed in AS/NZS 1170.2:2002[6]. However, the derived value for zone 1 is higher than the value suggested in Australian standard for areas with same wind climate. The along wind and cross wind can be calculated as Equation (4) and (5) according to the Australian standard..

max 20

2

, ,0

2

, ,0

3

1 2

..... (4)

tair R h

v h

h

fig windward des zz

h

fig leeward des zz

SEg I

xm h g I

C v z b z z

Eq

C v h b z z

&&

,

20

0.5ˆ 1.5 .... (5)1

air des fsRm

v h

V Cbgy K Eq

m g I

&&

where, m0- average mass per unit height, in kg/m ρair- density of air which shall be taken as 1.2 kg/m3

Vdes,θ(z)- building orthogonal design wind speeds as a function of height z Vdes,θ(h)- building orthogonal design wind speeds evaluated at height h bz- average breadth of the structure at the section at height z Δz- height of the section of the structure upon which the wind pressure acts b- breadth of a structure, normal to the wind stream gR- peak factor for resonant response (10 min period) given by:

= ce n600log2 nc = first mode natural frequency of vibration of a structure in the crosswind direction, in Hertz gv- peak factor for the upwind velocity fluctuations, which may be taken as 3.7 Ih- turbulence intensity Km- mode shape correction factor for crosswind acceleration Cfs- crosswind force spectrum coefficient generalized for a linear mode shape ζ - ratio of structural damping to critical damping of a structure Et- spectrum of turbulence in the approaching wind stream S -size reduction factor The acceleration values calculated are shown in Table 4. Euro code yields higher along wind acceleration values than Australian standards. However these along wind acceleration values are much less than across wind acceleration values due to slenderness of 183 m height building. In most of the cases, these across-

Return period Wind speed (ms-1) Zone 1 Zone 2 Zone 3

50 – years 54 47 38 5 - years 46 40 32

=

ENGINEER 24

ENGINEER 12

wind acceleration values exceed the threshold value 0.15 ms-2 with 5 year return period [21]. Even for higher wind speed value in zone 1, along wind acceleration values do not satisfy the threshold value set for human comfort. This means that, across wind acceleration would be a significant factor when designing a high-rise building in an urban area of Sri Lanka. Conclusion Major findings of this study can be concluded ed as 1. Structural loads are varying according to the wind loads derived from the standards. Both Australian standards give higher forces in zone 1 due to use of importance factor (Mi) and/or use of higher terrain multiplier for cyclonic areas. 2. Base shear and base moments are two major measurements on determining the effects of wind loads, which directly combine with the wind load values and their distribution along the height of the building. Higher base moment and base shear are obtained from British and Euro codes because they use division by parts rule for contributing wind loads along the height of the building.

3. Division –by – parts rule trends to apply higher loads even at the lower heights of the buildings and subsequently results in higher base moments for the building. Therefore, it is better to choose small strips at intermediate levels rather than choosing larger strips. 4. Only wind loads calculated according to the AS 1170.2:1989 for 183 m high building in zone 1 exceed the drift limit. This could be a result of higher wind loads derived with the use of both importance factor (Mi) and the higher terrain multiplier for cyclonic areas. 5. Along wind accelerations values calculated from Euro code are higher than the those values calculated from Australian codes except in zone 1. The higher serviceability wind speeds values ued for Astralian codes may be the reason for above difference. 6. Across wind acceleration values are considerably higher than the along wind acceleration values. Thus greater concern should give to control across wind acceleration when designing high-rise buildings in Sri Lanka.

Table 6 - Acceleration values at 183 m height in zone 1, 2 and 3

Zone Wind direction Acceleration

type

Acceleration (ms-2)

AS 1170.2:

1989

AS/NZS 1170 .2:

2002

BS EN 1991-1-4:

2005

Zone 1 Normal to 46 m side Along wind 0.155 0.156 0.134

Cross wind 0.239 0.233 -

Normal to 30 m side Along wind 0.109 0.107 0.094

Cross wind 0.227 0.221 -


Cross wind 0.173 0.166 -


Cross wind 0.168 0.159 -


Cross wind 0.118 0.116 -


Cross wind 0.106 0.093 -

ENGINEER 24

ENGINEER 12





type

Acceleration (ms-2)

AS 1170.2:

1989

AS/NZS 1170 .2:

2002

BS EN 1991-1-4:

2005


Cross wind 0.239 0.233 -


Cross wind 0.227 0.221 -


Cross wind 0.173 0.166 -


Cross wind 0.168 0.159 -


Cross wind 0.118 0.116 -


Cross wind 0.106 0.093 -

ENGINEER25

ENGINEER 12





type

Acceleration (ms-2)

AS 1170.2:

1989

AS/NZS 1170 .2:

2002

BS EN 1991-1-4:

2005


Cross wind 0.239 0.233 -


Cross wind 0.227 0.221 -


Cross wind 0.173 0.166 -


Cross wind 0.168 0.159 -


Cross wind 0.118 0.116 -


Cross wind 0.106 0.093 -

ENGINEER 13

Acknowledgement Authors of this paper like to express their gratitude towards National Disaster Management Center (NDMC), who provided financial support throughout this study. They also wish to thank the Civil Engineering Department of University of Moratuwa for facilitating this research. References 1. Karunaratne, S. A., High Rise buildings in

Colombo (Structural Engineer‘s View), Proceedings of the international conference on ―Advances in Continuum Mechanics, Materials Science, Nano science and Nano technology: Dedicated to Professor Munidasa P. Ranaweera‖, University of Peradeniya, Sri Lanka, 2008. pp 291- 302

2. CP 3 Chapter V: 1972, Code of basic data for the design of buildings chapter V. Loading, Part 2 Wind Loads, British Standard Institution, London

3. Design of Buildings for High Winds, Sri Lanka,

Sri Lankan Ministry of Local Government, Housing and Construction, 1980

4. British standard: Loading for building- Part 2: Code of Practice for wind loads; BS 6399- 2:1997, British Standard Institution, London

5. Australian standard for wind loads ; AS 1170.2:1989, Standards Australia, New South Wales

6. Australian and New Zealand standards:

Structural design actions Part 2: wind actions; AS/NZS 1170.2:2002, Standards Australia, New South Wales

7. British Standard: Eurocode 1: Actions on

Structures – Part1- 4: General actions - wind actions; BS EN 1991-1-4:2005 , British Standard Institution, London

8. Cook, N. J., Wind loading, A practical guide to BS 6399-2 Wind loads for buildings, Thomas Telford, 1999

9. Bashor, R., Kareem, A., Comparative study of major international standards, The seventh Asia-Pacific conference on wind ngineering(APCWE-VII), Taipe, Taiwan, November 08-12,2008

10. Tamura, Y., Holmes, J. D., Krishna, P., Guo, Lu.,

Katsumura, A., Comparison of Wind Loads on Medium Rise Building According to the Asia-Pacific codes/.standards, The seventh Asia-Pacific Conference on Wind Engineering(APCWE-VII), Taipe, Taiwan, November 08-12,2008.

11. Holmes, J. D., Tamura, Y., Krishna, P, Wind

Loads on Low, Medium and High-Rise Buildings by Asia - Pacific codes, The Fourth International Conference on advances in Wind and Structures (AWAS'08), Jeju, Korea, May 29-31-,2008.

12. Holmes, J. D., Tamura, Y., Krishna, P., Comparison of Wind Loads Calculated by fifteen Different Codes and Standards, for Low, Medium and High-Rise Buildings, 11th American Conference on Wind Engineering, San Juan, Puerto Rico, June 22-26, 2009.

13. Wardlaw, R. L., Moss, G. F., A Standard Tall Building Model for the Comparisons of Simulated Natural Winds in the Wind Tunnels, 3rd International Conference on Wind Effects on Buildings and Structures, Tokyo, Japan, 1971

14. CSI Analysis Referenc e Manual, Computers and Structures Inc., Berkeley, California, USA, 2013

15. British Standard: Loading for buildings Part 1.

Code of practice for dead and imposed loads; BS 6399-1:1996, British Standard Institution, London

16. Narayan, R. S, Cook, N. J., Smith, B. W., Rees, E.

J., Blackmore, P., Report on the Calibration of Euro code for wind loading (BS EN 1991 - 4) and its UK National Annex against the current UK wind code (BS 6399: Part 2:1997), Report submitted to Department for Communities and Local Government (DCLG), UK, 2007

17. Wijeratne, M. D., Jayasinghe, M. T. R., ―Wind

Loads for High-Rise Buildings Constructed in Sri Lanka‖, Transactions Part 2- Institution of Engineers, Sri Lanka, 1998, pp 58-69

18. Premachandra, W. R. N. R., ―Study of New Wind Loading Code to be Adopting on Sri Lanka‖, M.Sc Thesis, Graduate school, Kasetsart University, 2008

19. British Standard: Structural use of concrete- Part

2: Code of practice for special circumstances; BS 8110- 2:1985, British Standard Institution, London

20. Mendis, P., Ngo, T., Hariots, N., Hira, A., Samali, B., Cheung, J., Wind Loading on Tall buildings, EJSE special issue; Loading on structures, 2007. pp 41-54

21. International Organization for Standardization,

2007. Bases for design of structures—serviceability of buildings and walkways against vibration. Final DraftInternational Standard ISO/FDIS 10137:2007(E), Geneva, Switzerland

ENGINEER27


1

Predicting Thermal Performance of Different Roof Systems by Using Decision Tree Method

A. U. Weerasuriya

Abstract: This paper describes the use of decision tree method to predict thermal performance of several roof systems under different climate conditions. The decision tree method is a data mining technique which has competitive advantages over other methods such as simple and clear procedure, easy to understand without having rigorous mathematical and computational knowledge, etc. Results of 80 energy simulation cases were used to demonstrate the applicability of this method in building energy simulation. These 80 simulation cases are based on five locations in five different climate zones, eight different roof systems, and two extreme climate conditions; warmest and coldest in a year of a particular location. The modelled decision tree has prediction accuracy of 84% on training data and 100% on test data. Addition to that, decision tree automatically ranked the best selection of roof system under prevailing climate conditions. The predicted values shown in each classified data subsets can be used as a reference with an accuracy of 6%to predict the indoor room temperature with the use of a particular roof system. Finally, derived decision rules and simplified guidelines from constructed decision tree are also provided in a tabular format for non-engineer users. Keywords: Decision tree method, Energy simulation, Roof Systems, Thermal performance 1. Introduction

The significant percentage of total energy consumption of a building is used to restore the acceptable occupant thermal comfort level [1, 2]. The direct and indirect heat transfer of building components are the main factors affecting the occupant thermal comfort by increasing the indoor temperature. Among other common building components, roof itself generates significant heat loading due to its vast surface area and the orientation which is directly facing to the sky. Therefore a designer can maintain acceptable indoor thermal environment for occupants by selecting a suitable roof system and carefully controlling its thermal properties. Energy simulation techniques have been widely used to assess the thermal performance of the whole or part of a building [3, 4]. However, its accuracy on predicting energy demand of an occupied building is lower than that for an on-occupied building due to the uncertainty of latent heat generation from human bodies and electrical appliances. There are several drawbacks of energy simulation methods such as steep learning curve to operate software, the necessity to perform separate simulation for every case-study, more suited to evaluate designed buildings rather than those in early design stage, use of simplified methods and limited numbers of factors considered for analysis. Therefore, other techniques which are capable of overcoming these shortcomings have

been adapted to model building energy demand. The traditional regression analysis method and Artificial Neural Network (ANN) method are two of the most popular techniques successfully used by researchers in the past [3]. The simple and efficient regression analysis method is based on statistical analysis and regression equation which is able to combine effect of various climate variables with building physics in order to predict building energy demand [5, 6, 7]. However, complicated nature of regression equation demands that the user has a good mathematical background. The structure of ANN is similar to biological neural networks. It is able to build complex relationships with different factors in the building energy simulation process and thus, get more accurate output [8,9,10,11,12]. Nevertheless an ANN model cannot be understood and interpreted easily as it operates as a ―black box‖ within the analysis process. Decision tree method is one of the data mining techniques that has been used in scientific and medical fields [13, 14, 15. 16, 17] to make decisions based on the consideration of several inputs simultaneously. The flow chart like structure allows understanding and interpreting the analysis easily even for a user without specific mathematical knowledge.

Eng .A.U. Weerasuriya, B.Sc.Eng(Hons), M.Sc. (Moratuwa), AMIE(Sri Lanka), PhD candidate, Department of Civil and Environmental Engineering, Hong Kong University of Science and Technology, Hong Kong.

ENGINEER 28

2

However, the use of the decision tree method in building energy simulation is very sparse. Yu et al.[18] demonstrated the use of decision tree method in building energy simulation in detail analysis by predicting energy use intensity of houses in Japan.Tso and Yau [19] compared the accuracy of the decision tree method with regression analysis and ANN method and found that its accuracy was almost the same as other two methods. 2. Decision tree method The decision tree model is a logical model which predicts the value of a target variable by using the values of a set of predictor variables. Both categorical and numerical attributes can be used as either target or predictor variables unless the use of categorical variables are preferred to numerical variables in terms of accuracy. However, the selection of attributes for the split tests is more significant in a decision tree model as it follows a greedy algorithm. The decision tree algorithm iteratively divides the domain by selecting split attributes that can best separate the target class values. Therefore, the accuracy of the output is heavily dependent on the quality of selection of split attributes. The concept of entropy is used to measure the quality of a split attribute. The quality is by means of the purity of a partitioning in decision tree nodes. Equation (1) shows the method to calculate entropy value for a domain with two types of variables HIGH and LOW by using binary split test at node Di.

2 2

log log .

...........Eq(1)

n n n n

i

HIGH HIGH LOW LOWEntropy D

TN TN TN TN

Where, nHIGH : the number of ‗HIGH‘ variables in node Di and nLOW : the number of ‗LOW‘ variables in node Di; TN : the total number of ‗HIGH‘ and ‗LOW‘ variables in node Di. The entropy value varies in between 0 and 1 for any split test. The value 0 indicates the pure split while 1 shows the 50/50 division of a binary split test. After the split test at node Di, the domain is divided in to two sub domains, can be referred to as DS1 and DS2 with number of record sm and n in respective sub domains. The efficiency of a split test can be evaluted by entropy difference of the parent domain and children domains as shown in Equation (2). This entropy difference is called ―information gain (InfoGain)‖of the ith node.

1 2Entropy DS and DS is the weighted sum of the entropies of subsets DS1 and DS2 calculated according to Equation (3)

1 2 1

2

...... (3)

mEntropy DS and DS Entropy DS

m nn

Entropy DSm n

Eq

The entropy values of DS1 and DS2 are calculated as Equation (1). However, information gain is a biased parameter for domains with many data sets. To avoid this deficiency, information gain is normalized with size of the domain so called the Gain Ratio. The gain ratio is calculated as shown in Equation (4)

...................... (4)Information Gain

Gain Ratio EqSplit Information

Where,

2

2

log

log ..................... (5)

m mSplit Information

m n m nn n

Eqm n m n

The split attribute with the highest gain ratio would be used for the split test at ith node.

3. Energy simulations of different roof systems

3.1. Climate data and building physical data Both input data sets of climate and building physical properties should include wide range of data in order to construct a successful decision tree. The following methodology was adopted to collect necessary input data for decision tree analysis in the absence of field measurement data. Five locations in different climate zones were selected to represent different prevailing climate condition. These five locations are Colombo (Sri Lanka), Athens (Greece), Hong Kong (China), Riyadh (Saudi Arabia), and Chicago (United States) to represent tropical monsoonal, , Mediterranean climate, humid subtropical climate, equatorial dessert hot arid and hot summer continental climate zones respectively. Both warmest and coldest months of these locations were selected to evaluate thermal performance of the roof systems under extreme climate conditions. Mean monthly temperature and the humidity values together with sunshine hours in both warmest and coldest months of these locations are shown in Table 1.These climate factors are also used as input data for energy simulations.

1 2

............. 2

iInfoGain Entropy D Entropy DS and DS

Eq

ENGINEER 28

2


2 2

log log .

...........Eq(1)

n n n n

i


TN TN TN TN



1 2 1

2

...... (3)


m nn

Entropy DSm n

Eq




Where,

2

2

log

log ..................... (5)


m n m nn n

Eqm n m n




1 2

............. 2


Eq

ENGINEER29

2


2 2

log log .

...........Eq(1)

n n n n

i


TN TN TN TN



1 2 1

2

...... (3)


m nn

Entropy DSm n

Eq




Where,

2

2

log

log ..................... (5)


m n m nn n

Eqm n m n




1 2

............. 2


Eq

3

Table 1- Climate data of 5 locations

Climate zone Location Month Sunshine hours

Mean daily temperature(oC)

Mean relative humidity(%)

Max Mini Max Mini

Mediterranean climate

Athens (Greece)

July 12 31.8 22.8 42 59

January 4 13.6 7 63 75

Hot summer continental climate

Chicago (United States)

July 9 28 17 54 82

January 3 -1 -10 66 77

Tropical monsoonal

Colombo (Sri Lanka)

April 7.9 31.1 24.3 68 95

December 6.9 29.8 22.4 61 91

Humid subtropical climate

Hong Kong (China)

July 8 31 27 71 84

January 5 19 14 64 75

Equatorial dessert hot arid

Riyadh (Saudi Arabia)

July 10 42 28 8 16

January 8 19 8 32 60

A two-storey house with total floor area of 99 m2 was selected for the energy simulation modeling. This builiding is similar to the modelled building that was used by Halwathura and Jayasinghe [20]. Sloped and flat roof shapes were used for three basic roof types; concrete flat roof, calicut tile roof and asbestos sheet roof. Some modifications were introduced for these roofs such as a sloped

ceiling for asesbestos sheet roof and calicut tile roof, insulate concrete roof, green roof and concrete roof with parapet walls for concrete roof slab. The insulated slab system is similar to model proposed by Halwathura and Jayasinghe [21, 22]. The green roof has a 10 cm grass layer on top of the roof as proposed by Dareeju et al [23]. More details about roof systems used for this study are shown in Table 2.

Table 2 - Roof systmes use for energy simulations

Roof system Components (from top to bottom)

Asbestos sheet One layer of 6 mm thick fiber cement sheet Calicult tile One layer of 20 mm thick fiber cement sheet Concrete roof 20 mm cement rendering, 125 R/C slab, 10 mm sofit palster Asbestos sheet with a ceiling Fiber cement sheet, 2 mm aluminium foil, 25 mm

polyurethane, 5mm flat asbestos sheet Calicut tile roof with a ceiling Calicut tile layer, 2 mm aluminium foil, 25 mm polyurethane,

5mm flat asbestos sheet Insulated concrete roof slab 40 mm cement rendering, 25 mm expanded cellular

polyethylene layer, 125 mm R/C slab, 10 mm soffit plaster Green roof 10 mm thick grass layer, 25mm soil layer, 125 mm R/C slab,

10 mm soffit plaster Concrete roof with parapet walls at the perimeter

20 mm cement rendering, 125 R/C slab, 10 mm sofit palster surrounded by 1 m high parapet wall

3.2. DEROB-LTH Modelling

ENGINEER 30

4

DEROB-LTH was used as the energy simualation software for this study. It was used by previous researchers [20, 21, 23] and accuracy was evaluated with field measurements[20]. Two types of data inputs are needed for a DEROB simulation, one data set for climate data and other about the building model. The orientation of building is north-south direction and all windows are also in north and south directions only. The 225 mm thick cement plastered brick walls are at the perimeter and 115 mm plastered brick walls used as internal walls. Floor is made with 75 mm thick concrete with a tile paved surface. First floor slab is 125mm thick and at the bottom side there is 15 mm thick soffit plaster and and upside is paved with ceramic tiles. There is a balcony at first floor level, which is protected by a shading device. All windows are wooden framed single glazed windows and doors are timber panelled type. There are altogether 80 simulation cases ( 8 roof types x 5 locations x 2 months) used to build the decision tree. In every case, the indoor temperature of the upper floor volumes were extracted because those volumes are directly under influence of roof system rather than spaces in ground floor level.

4. Preliminary analysis of data 4.1. Analysis of monthly average temperature Outdoor air temperature is one of the main factors influencing occupant comfort level. The amount of variation of the outdoor temperature from the neural temperature would be a better measurement to determine required level of thermal performance of a roof system. Figure 1 shows the boxplot graph for montly average out door air temperature for the selected locations. According to the Figure 1, Colombo has minimum temperature variation, which is minimum and maximum monthly temperature values are close to the annual average temperature. For other four locations larger deviations can be observed among average and highest and lowest temperatures. Thus the selection of two extreme temperature cases for this study can be justified. The annual average temperature is above 9oC for all five locations and that value is close to 20oC except for Chicago city. Only Chicago has the lowest temperature below the freezing point.

Figure 1 - Distribution of monthly average outdoor temperature

4




4




ENGINEER 30

4




4




4




ENGINEER31

5

4.2. Selection of attributes for the decision tree.

There are several climate and building physical factors affecting thermal performance of a roof. Some of these factors are numerical attributes

such as tempeature, humidity and some of them are categorical attributes such as shape of the roof, roof covering material, etc. it is

necessary to convert numerical attributes to categorical attributes to obtain a more accurate decision tree. For the simplicity, only binary categorical attributes were used for this study for an example temperature is simplified in to two categorical attributes―HIGH‖temperature or ―LOW‖ temperature. The annual average values of numerical attributes were used for the binary separation of those attributes The attributes used for constructing decision tree are listed in Table 3. It is necessary to have even distribution of categorical variables in each location to build an unbiased decision tree model. According to the Figure 2, the categorical distribution at each location has fairly even distribution, that percentage is varying between 25% to 47%. In order to demonstrate the thermal performance of a roof

system, normalised average indoor temperature was used as the prediction attribute in the decision tree. This value is calculated as Equation (5). The ‗high difference‘ is defined as the normalised average indoor temperature value exceeds1.04 or 0.96. The advantage of this parameter is that it is directly combined with the outdoor temperature and thus easy to understand even by non-engineering user. It is also more convenient to use in heating/cooling load calculations as it enables the determination of indoor temperature implicitly.

Table 3 - Attributes used for the decision tree

Figure 2 - Categorical distribution of attributes in 5 locations

Attributes Splitting test Remarks Temperature HIGH/LOW Temperature greater than 20oC is high Humidity HIGH/LOW Humidity greater than 60% is high Shape FLAT/SLOPED Insulation WITH/WITHOUT Only apply to concrete slabs Ceiling WITH/WITHOUT Apply only to asbestos and Calicut tile roof Parapet walls WITH/WITOUT Normalized indoor temperature

HIGH/LOW Indoor temperature normalized with the outdoor temperature. If the difference is more than (±4%) then difference is high

.......... (6)

Normalised Indoor temperature

Average temperature of upper floor volumes

Monthly average temperature

Eq

Athens Chicago Colombo Hong Kong Riyadh

each location has fairly even distribution, that such as tempeature, humidity and some of each location has fairly even distribution, that such as tempeature, humidity and some of each location has fairly even distribution, that percentage is varying between 25% to 47%. In percentage is varying between 25% to 47%. In order to demonstrate the thermal performance of a roof system, normalised average indoor temperature was used as the prediction attribute in the decision tree. This value is calculated as in the decision tree. This value is calculated as Equation (5). The ‘high difference’ is defined in the Equation (5). The ‘high difference’ is defined in the decision tree. This value is calculated as Equation (5). The ‘high difference’ is defined decision tree. This value is calculated as Equation (5). The ‘high difference’ is defined as the normalised average indoor temperature value exceeds1.04 or 0.96.

The advantage of this parameter is that it is directly combined with the outdoor temperature directly combined with the outdoor temperature and thus easy to understand even by non-engineering user. It is also more convenient to ring user. It is also more engineering user. It is also more convenient to ring user. It is also more engineering user. It is also more convenient to use in heating/cooling load calculations as it enables the determination of indoor temperature implicitly.

4.2. Selection of attributes for the decision tree.

There are several climate and building physical factors affecting thermal performance of a roof. Some of these factors are numerical attributes such as tempeature, humidity and some of them are categorical attributes such as shape of the roof, roof covering material, etc. it is necessary to convert numerical attributes to categorical attributes to obtain a more accurate decision tree. For the simplicity, only binary categorical attributes were used for this study for an example temperature is simplified in to two categorical attributes“HIGH”temperature or “LOW” temperature. The annual average values of numerical attributes were used for the binary separation of those attributes The attributes used for constructing decision tree are listed in Table 3. It is necessary to have even distribution of categorical variables in each location to build an unbiased decision tree model. According to the Figure 2, the categorical distribution at

4




4




ENGINEER 32

6

5. The decision tree

5.1. Generation of decision tree The steps of constructing a decision tree can be shown as Figure 3. There are two stages of the procedure named, learning and classification. In learning stage, first divide the whole data set in to two subsets called training and test data sets. In this study total 80 data sets were divided into two subsets such as 75 data sets for

training and 5 data sets for test. The decision tree is generated and its accuracy is calculated by analysing the training data set. In the classification stage, if the accuracy of the decision tree is acceptable it can be used for future projects. If the accuracy is not adequate, then it is necessary to identify the reasons and fix them and regenerate a new decision tree.

Figure 3 - Flow chart of making a decision tree (Yu et al. [18])

At each node it is necessary to calculate entropy of parent and children data sets, information gain, split information and gain ratio for selecting the split attribute. The same calculation prodcedure should be repeated at each node until one of the following criteria is met 1. All records in a partition share the same

target class value. 2. There are no remaining predictor attributes

that can be used to further split a partition. 3. There are no more records for a particular

value of a predictor variable. Thus, it is a time consuming repetitive process. Therefore, an open source data mining software WEKA was used for this study. WEKA was originally developed by University Waikato, New Zealand and previously used by Yu et al[18] for a similar study. There are different decision tree algorithms within the WEKA. J48 algorithm was selected for this study by using trial and error method, which gaves the highest

accuracy for training data set. The generated decision tree is shown in Figure 4. The generated decision tree has four levels and 15 nodes. Each node represents either a split test or a decision rule. The Root node and internal nodes show details of split test such as number of data sets and split attribute. Leaf nodes express the decision rules. However, leaf nodes with entropy value 0 are labled as LEAF and otherwise named as STOP. STOP nodes are resulted when there is no significant effects that can be obeserved on information gain ratio in further candidate spliting tests. In both LEAF and STOP nodes, there are information about number of data, calssification result, predicted normalised indoor temperature (NIT), and the lable LEAF or STOP. More details about nodes are shown in Figure 5. The WEKA analysis report shows some information regarding accuracy of the constructed decision tree. According to the report that accuracy is 84%. Though this

Splitting dataset into training and test data

Analysing training data by a decision tree algorithm and generating decisiontree

Estimating the accuracy of obtained decision tree using test data

Identifying reasons and finding solutions Accuracy is considered

acceptable

Applying decision tree to future data

Classification

Learning

YES

NO

Thus, it is a time consuming repetitive process. Therefore, an open source data mining software WEKA was used for this study. WEKA was originally developed by University Waikato, New Zealand and previously used by Yu et al[18] for a similar study. There are different decision tree algorithms within the WEKA. J48 algorithm was selected for this study by using trial and error method, which gaves the highest

accuracy for training data set. The generated decision tree is shown in Figure 4.The generated decision tree has four levels and 15 nodes. Each node represents either a split test or a decision rule. The Root node and internal nodes show details of split test such as number of data sets and split attribute. Leaf nodes express the decision rules. However, leaf nodes with entropy value 0 are labled as LEAF and otherwise named as STOP. STOP nodes are resulted when there is no significant effects that can be obeserved on information gain ratio in further candidate spliting tests. In both LEAF and STOP nodes, there are information about number of data, calssification result, predicted normalised indoor temperature (NIT), and the lable LEAF or STOP. More details about nodes are shown in Figure 5. The WEKA analysis report shows some information regarding accuracy of the constructed decision tree. According to the report that accuracy is 84%. Though this

5 data sets for test. The decision tree is generated and its accuracy is calculated by analysing the training data set. In the classification stage, if the accuracy of the decision tree is acceptable it can be used for future projects. If the accuracy is not adequate, then it is necessary to identify the reasons and fix them and regenerate a new decision tree. At each node it is necessary to calculate entropy of parent and children data sets, information gain, split information and gain ratio for selecting the split attribute. The same calculation prodcedure should be repeated at each node until one of the following criteria is met1. All records in a partition share the same

target class value.2. There are no remaining predictor attributes

that can be used to further split a partition.3. There are no more records for a particular

value of a predictor variable.


5.1. Generation of decision treeThe steps of constructing a decision tree can be shown as Figure 3. There are two stages of the procedure named, learning and classification. In learning stage, first divide the whole data set in to two subsets called training and test data sets. In this study total 80 data sets were divided into two subsets such as 75 data sets for training and

ENGINEER 32

6














acceptable


Classification

Learning

YES

NO

Thus, it is a time consuming repetitive process. Therefore, an open source data mining software WEKA was used for this study. WEKA was originally developed by University Waikato, New Zealand and previously used by Yu et al[18] for a similar study. There are different decision tree algorithms within the WEKA. J48 algorithm was selected for this study by using trial and error method, which gaves the highest

accuracy for training data set. The generated decision tree is shown in Figure 4.The generated decision tree has four levels and 15 nodes. Each node represents either a split test or a decision rule. The Root node and internal nodes show details of split test such as number of data sets and split attribute. Leaf nodes express the decision rules. However, leaf nodes with entropy value 0 are labled as LEAF and otherwise named as STOP. STOP nodes are resulted when there is no significant effects that can be obeserved on information gain ratio in further candidate spliting tests. In both LEAF and STOP nodes, there are information about number of data, calssification result, predicted normalised indoor temperature (NIT), and the lable LEAF or STOP. More details about nodes are shown in Figure 5. The WEKA analysis report shows some information regarding accuracy of the constructed decision tree. According to the report that accuracy is 84%. Though this

5 data sets for test. The decision tree is generated and its accuracy is calculated by analysing the training data set. In the classification stage, if the accuracy of the decision tree is acceptable it can be used for future projects. If the accuracy is not adequate, then it is necessary to identify the reasons and fix them and regenerate a new decision tree. At each node it is necessary to calculate entropy of parent and children data sets, information gain, split information and gain ratio for selecting the split attribute. The same calculation prodcedure should be repeated at each node until one of the following criteria is met1. All records in a partition share the same

target class value.2. There are no remaining predictor attributes

that can be used to further split a partition.3. There are no more records for a particular

value of a predictor variable.


5.1. Generation of decision treeThe steps of constructing a decision tree can be shown as Figure 3. There are two stages of the procedure named, learning and classification. In learning stage, first divide the whole data set in to two subsets called training and test data sets. In this study total 80 data sets were divided into two subsets such as 75 data sets for training and

ENGINEER33

6














acceptable


Classification

Learning

YES

NO

7

accuracy is not very high it is acceptable compared to the low number of data sets used as the training set. Information regarding misclassification can be found from the confusion matrix as shown below.

The above matrix implies that 40 LOW NIT cases have 34 correct classifications with 6 missclassification instances. There are 6 misclassification cases in HIGH. NIT cases among total 35 number cases

Figure 4 - Decision tree constructed by using WEKA

Figure 5 - Information show in decision tree nodes

𝑎𝑎𝑎𝑎 𝑏𝑏𝑏𝑏 − − − − 𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝑎𝑎𝑎𝑎𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶 𝑎𝑎𝑎𝑎𝐶𝐶𝐶𝐶𝑎𝑎𝑎𝑎 𝐿𝐿𝐿𝐿𝐿𝐿𝐿𝐿𝐿𝐿𝐿𝐿𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝑏𝑏𝑏𝑏 𝐻𝐻𝐻𝐻𝐷𝐷𝐷𝐷𝐻𝐻𝐻𝐻𝐻𝐻𝐻𝐻 𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝐷

Confusion matrix

Node number

Number of data

Classification result

Predicted NIT value

LEAF or STOP

Node number

Number of data

Split Test

Node number

Number of data

Split Test

75

Green

1

66

Humidity

2

INTERNAL NODE

ROOT NODE LEAF NODE

9

LOW NIT

3

1.004

LEAF

34 6 |

6 29 |

a b classified as

a LOW NIT

b HIGH NIT

ENGINEER 34

8

5.2. Evaluation of decision tree Before, using the constructed decision tree to predict thermal performance of roof systems in future projects it is necessary to assess its accuracy by using test data sets. In this study, only five data sets were randomly selected as test data sets due to limited number of available data. The five test data sets selected are shown in Table 4 with their properties. The predictions of the decision tree are listed in Table 5 with classification result and predicted NIT value.

The percentage error in predicted value and the actual NIT value is also shown. All five test cases are correctly predicted by the decision tree. This prediction accuracy 100% is higher than accuracy (84%) of decision tree. It is believed that this occurs due to limited numbers of test data sets used for evaluation. However, the maximum percentage error is much lower as 5.83 for the test data set, which indicates the better prediction ability of the decision tree.

Table 4 - Data used for the evaluation of the accuracy of decision tree

Case Gre

en ro

of

Hum

idity

Cei

ling

Tem

pera

ture

Insu

latio

n

Shap

e

Para

pet w

all

1 Flat roof Nogreen High Without High No Insulation Flat NoPWall 2 Clay Tile roof Nogreen High Without High No Insulation Sloped NoPWall 3 Flat roof Nogreen High Without Low No Insulation Flat NoPWall 4 Green roof Green Low Without Low No Insulation Flat NoPWall 5 Asbestos roof

with ceiling Nogreen High With Low No Insulation Sloped NoPWall

Table 5 - Summary of results of evaluation of the decision tree

Another aspect of the decision tree is that each LEAF or STOP node represents a decision rule. The constructed decision tree has 8 LEAF and STOP nodes which can be used to derive 8 different decision rules. For an example, node 6 expresses that if roof is not a green roof and with high humidity level and roof with a ceiling then the normalised indoor temperature is low. All derived decision rules are listed in Table 6. The priority order of selection of different roof system under high (>20oC) and low (<20oC) temperature is shown in Table 7. The green roof is the first choice under both climate conditions suggests that it out performed other roof systems in any climate zone. Ceiling is also a better remedy to achieve acceptable indoor temperature in both high and low outdoor

temperature conditions. However, insulated roof system only performs well under high out door temperature condition. The shape of the roof and use of parapet walls are only effective under low outdoor temperatures but yet their significance is less compared to use of green roof or installing a ceiling. 6. Conclusion

The decision tree method was used in this study to predict thermal performance of different roof systems in different climate zones. 80 test cases were simulated by using energy simulation software for a two-storey house. Eight different roof systems used in this study covers traditional roof systems such as asbestos roof sheet, Calicut tile roof, and

Case Actual level Predicted level

Correct/ Incorrect

Actual NIT

Predicted NIT

Error

1 HIGH NIT HIGH NIT Correct 1.123 1.062 5.43% 2 HIGH NIT HIGH NIT Correct 1.054 1.062 0.76% 3 HIGH NIT HIGH NIT Correct 0.947 0.953 0.63% 4 LOW NIT LOW NIT Correct 1.033 1.004 2.81% 5 LOW NIT LOW NIT Correct 1.032 1.018 1.36%

ENGINEER 34

8




Case Gre

en ro

of

Hum

idity

Cei

ling

Tem

pera

ture

Insu

latio

n

Shap

e

Para

pet w

all








Correct/ Incorrect

Actual NIT

Predicted NIT

Error


ENGINEER35

8




Case Gre

en ro

of

Hum

idity

Cei

ling

Tem

pera

ture

Insu

latio

n

Shap

e

Para

pet w

all








Correct/ Incorrect

Actual NIT

Predicted NIT

Error


9

concrete roof with some improvements such as sloped ceiling, green roof, insulated roof slab and roof slab protected with parapet walls. The total 80 numbers of data were divided into two and 75 data sets were used as the training data sets and balance five data sets were used as test data sets. Constructed decision tree has 15 nodes in seven levels. The accuracy of the decision tree is 84% for training data set and 100% for the test data set. There are eight decision rules, which can be derived from the decision tree. However, the accuracy of decision tree is limited by number of data set used for the study. It is also necessary to mention that there are some other parameters needed to consider doing an energy simulation such as accurate ventilation rate, internal latent

heat gain from human bodies and electrical appliances which are not considered in this study. More consideration should be paid when interpreting numerical attributes due to results of splitting tests depend on the used threshold values. Thus, threshold values should be selected in a fair and rational manner.

Even the decision tree method leads accurate predictions on energy simulation results in this study it is recommended to use field measurements to verify its validity under different prevailing site conditions in future studies. It is also necessary to test applicability of this method to design more energy efficient buildings in different building categories such as commercial, public, apartment buildings, etc.

Table 6 - Derived decision rules from the decision tree

Table 7 - Selection of suitable roof system

Potential factors High temperature sites ( > 20oC) Low temperature sites (< 20oC)

Significant factor Rank Significant factor Rank

Green roof √ 1 √ 1

Ceiling √ 2 √ 2 Insulated slab √ 3 Shape √ 3 Parapet walls √ 4

Node Decision Rules

1 3 If roof is GREEN then NIT is LOW 2 4 If roof is NOT GREEN and HUMIDITY is HIGH then NIT is HIGH 3 6 If roof is NOT GREEN and HUMIDITY is HIGH and WITH A CEILING then NIT is

LOW 4 10 If roof is NOT GREEN and HUMIDITY is HIGH and WITHOUT A CEILING and

TEMPERATURE IS HIGH and WITHOUT AN INSULATED SLAB then NIT is HIGH

5 11 If roof is NOT GREEN and HUMIDITY is HIGH and WITHOUT A CEILING and TEMPERATURE IS HIGH and WITH AN INSULATED SLAB then NIT is LOW

6 13 If roof is NOT GREEN and HUMIDITY is HIGH and WITHOUT A CEILING and TEMPERATURE IS LOW and WITH A SLOPED ROOF then NIT is LOW

7 14 If roof is NOT GREEN and HUMIDITY is HIGH and WITHOUT A CEILING and TEMPERATURE IS LOW and WITH A FLAT ROOF and WITH PARAPET WALLS then NIT is LOW

8 15 If roof is NOT GREEN and HUMIDITY is HIGH and WITHOUT A CEILING and TEMPERATURE IS LOW and WITH A FLAT ROOF and WITHOUT PARAPET WALLS then NIT is HIGH

ENGINEER 36

10

Acknowledgement The author of this paper would like to express his gratitude towards Dr. Jack Chen from Department of Civil and Environmental Engineering, Hong Kong University of Science and Technology for his guidance given in throughout this study. The valuable advices given on DEROB–LTH modelling by Dr. R.U. Halwathura from Department of Civil Engineering, University of Moratuwa and Mr. B.S.S.S. Dareeju from Sapienza University, Rome are highly appreciated. References [1]. Annual energy reviews 2011, United States

Energy Information Administration, available www.eia.gov/aer

[2]. Statistical year book for Asia and the Pacific 2011, United Nations economic and social commission for Asia and the Pacific, available http://www.unescap.org/stat/data/syb2011/II-Environment/Energy-supply-and-use.asp

[3]. Jebaraj, S., Iniyan, S., A review of energy models, Renewable and Sustainable Energy Reviews, 10, 2006, pp 281 – 311

[4]. Hong, T., Chou, S.k, Bong, T.Y., Building simulation: an overview of developments and information sources, Building and Environment, 35, 2000, pp 347 – 361

[5]. Catalina, T., Virgone, J., Blanco, E., Development and validation of regression models to predict monthly heating demand for residential buildings, Energy and Buildings, 40 (10), 2008, pp 1825–1832

[6]. Bentzen J, Engsted T. A revival of the autoregressive distributed lag model in estimating energy demand relationships. Energy,26, 2001, pp 45–55

[7]. Ghiaus, C., Experimental estimation of building energy performance by robust regression, Energy and Buildings, 38 (6), 2006, pp 582–587.

[8]. Magnier, L., Haghighat, F., Multi objective optimization of building design using genetic algorithm and artificial neural network, Building and Environment, 45, 2010, pp 739–746

[9]. Aydinalp M, Ismet U. V., Fung A.S., Modelling of the appliance, lighting and space-cooling energy consumptions in the residential sector using neural networks. Applied Energy, 71, 2002, pp 87–110

[10]. Kalogirou, S. A., Bojic, M., Artificial neural networks for the prediction of the energy

consumption of a passive solar building, Energy, 25, 2000, pp 479 – 491

[11]. Kokasl, M. A., Ugursal, V. I., Comparison of neural network, conditional demand analysis, and engineering approaches for modelling end-use energy consumption in the residential sector, Applied Energy, 85, 2008, pp 271-296.

[12]. Zhang, J., Haghighat, F., Development of artificial neural network based heat convection for thermal simulation of large rectangular cross-sectional area earth-to-earth heat exchanges, Energy and Buildings, 42 (4),2010, pp 435–440

[13]. El-Bakry, H. M., Hamada, M., New fast decision tree classifier for identifying protein coding regions, 35th International Symposium on Computer Architecture (ISCA 2008), June 21-25, 2008, Beijing, China, 489 -500

[14]. Fan, C.Y., Chang, P.C.,. Lin, J.J., Hsieh, J. C., A hybrid model combining case-based reasoning and fuzzy decision tree for medical data classification. Applied Soft Computing, 2011, pp 632-644

[15]. Eisenberg, J. N. S., Mckone, T. E., Decision tree method for the classification of chemical pollutants: incorporation of across-chemical variability and within chemical uncertainty, Environment Science Technology, 32, 1998, pp 3396 – 3404

[16]. Tung, K.Y., Huang, I.C., Chen, S.L., Shih, C. T., Mining the Generation Xers‘ job attitudes by artificial neural network and decision tree—empirical evidence in Taiwan, Expert Systems with Applications,29 (4), 2005, pp 783–794.

[17]. Wehenkel, L., Pavella, M., Decision tree approach to power systems security assessment, International Journal of Electrical Power & Energy Systems,15, 1993, pp 13–36

[18]. Yu, Z., Haghighata, F., Fung, B.C.M., Yoshino, H., A decision tree method for building energy demand modelling, Energy and Buildings, 42, 2010, pp 1637–1646

[19]. Tso, G.K.F., Yau, K.K.W., Predicting electricity energy consumption: a comparison of regression analysis, decision tree and neural networks, Energy, 32 (9), 2007, pp 1761–1768

[20]. Halwatura R.U., Jayasinghe, M.T.R., Strategies for improved micro-climates in high-density residential developments in tropical climates, Energy for Sustainable Development,11( 4), 2007, pp 54-65

ENGINEER 36

10























ENGINEER37

10























11

[21]. Halwatura R.U., Jayasinghe, M.T.R., Thermal performance of insulated roof slabs in tropical climates, Energy and Buildings, 40, 2008, pp 1153–1160

[22]. Halwatura, R.U., Jayasinghe, M.T.R., Influence of insulated roof slabs on air conditioned spaces in tropical climatic conditions—A life cycle cost approach, Energy and Buildings, 2009, pp 678–686

[23]. Dareeju B.S.S.S, Meegahage J.N., Halwatura R.U., Influence of grass cover on flat reinforced concrete slabs in a tropical climate, International Conference on Sustainable Built Environment (ICSBE – 2010),2010, pp 469-479


Design and Development of Rasp Bar Mill for Size Reduction of Maize

D. P. Senanayaka, H. M. A. P. Rathnayake, B.M.K.S. Thilakarathne, B. D. M. P. Bandara and T. M. R. Dissanayaka

Abstract: Particle size affects many characteristics in the manufacturing process. Controlling the particle size helps to assure that the milled material will be consistent and repeatable with respect to firstly color, that is, uniform particles assure batch-to-batch color consistency, Secondly, flowability that is critical to packaging, tableting, weighing. Thirdly, uniformity, that is , consistent bulk density. Fourthly, density - helps control shipping costs and minimize dust. Fifthly, reconstitution, that assures the desired dissolution rate, Sixthly, chemical reaction that is vital for uniform, controlled chemical change. Finally, taste that allows precise portion control for consistent taste. The aim of this project is to design and develop a rasp bar mill for size reduction of maize seeds for different purposes. The fabricated rasp bar mill is capable of feeding maize around 60-180 kg/h, delivering of grits 60-90%, with sieve effectiveness 90-100% at a power requirement of 800-1400W,for different sieve perforation sizes of 5, 6, 7, 8 mm respectively.

Key words: Size reduction, Rasp bar mill, Maize gritting

1. Introduction Maize (Zea mays) is the most important coarse grain cereal crop in the Lowland cropping systems, such as Badulla, Moneragala, Ampara, Anuradhapura and Batticalo where in dry zone of the country as well as in the Uplands like Matale , which around 30,000 ha of land area devoted annually, the second highest extent of land next to rice in Sri Lanka[15]. Total Maize production of the country was 202.3 million metric tons in 2012[16].For the rural farmers to maximize profit from their maize, appropriate technology that suites their needs must be used. The processing of agricultural products like maize into quality forms not only prolongs the useful life of these products, but increases the net profit farmers make from mechanization technologies of such products. One of the most important processing operations done to bring out the quality of maize is milling. There are two methods of milling, i.e. dry milling and wet milling. Gritting or size reduction is one of the operations in dry milling. There is no machine in Sri Lanka for this purpose. The objective of

this project is to fabricate and evaluate an affordable gritting machine for size reduction of maize.

Eng. D. P. Senanayaka, AMIE(Sri Lanka), M.phil(Eng.) (Peradeniya),B.Sc.(Mech.Eng)(Peradeniya), Principal, Mechanical Engineer, Institute of Post Harvest Technology, Anuradhapura. Eng. H. M. A. P. Rathnayake,. AMIE(Sri Lanka) M.Tech(india), B.Sc.(Production Eng.)(Peradeniya), Senior Mechanical Engineer, Institute of Post Harvest Technology , Anuradhapura. B. M. K. S. Thilakerathne., BSc.Sp.(Agric. Food Sci.& Tech.)Sri Lanka, MSc (Food Sci.& Tech.)Sri LankaPhD.( Postharvest Tech.Horticulture), India Dip. in rice sciences (China), Dip.in Post harvest biology of fruit and veg.(Israel), Director, Institute of Post Harvest Technology Anuradhapura. Eng. B. D. M. P. Bandara, AMIE(Sri Lanka), M.E. (India) B.Sc.(Mech.Eng)(Peradeniya), Mechanical Engineer, Institute of Post Harvest Technology, Anuradhapura. Eng. T. M. R. Dissanayake, AMIE(Sri Lanka), M.E. (India) B.Sc. (Production. Eng.)(Peradeniya), MechanicalEngineer, Institute of Post Harvest Technology, Anuradhapura









ENGINEER39










ENGINEER 40

1.1 Literature review: In general “size reduction” is taken to mean the disintegration of solid substances by mechanical forces without altering their solid state. This also includes the division of liquids into drops or gases into bubbles. However, the physical and chemical condition of the disintegrated material may alter, particularly when inhomogeneous substances are present. The preparation for separation according to material components, e.g. grinding grain, is therefore, one of the classical tasks of size reduction techniques. Size reduction or „comminution‟ is the unit operation in which the average size of solid pieces of food is reduced by application of grinding, compression or reduction of particle size is an important operation in many chemical and other industries. The important reasons for size reduction are: Easy handling Increase in surface area per unit volume Separation of entrapped components

impact forces

In all types of size reduction, there are three types of forces used to reduce the size of foods:

1. compression forces, 2. impact forces, and 3. shearing (or attrition) forces.

Cereal grains generally are the primary source of energy in feedlot diets. Availability of energy from the grain depends largely on the type of grain used as well processing of that grain.[13].

A variety of grain processing techniques are used including grinding, steam flaking, and compiling high moisture corn to ferment. Each processing method differs in its nutritional efficacy [13] and each has a unique associated cost. [14]. Criteria for size reduction An ideal crusher would (1) have a large capacity, (2) require a small power input per unit of product, and 3) yield a product of the single size. 2. Design details The Comminutor operates by feeding material uniformly into a chamber in which a rotating blade assembly reduces the particles of the material by cutting or impacting them. The material discharges through a screen which

regulates final particle size at the outlet of the milling chamber. The blade and screen act in conjunction to determine final product-sizing.

Figure 1- Sketch of rasp bar mill in operation [9] 2.1 Feed Throat: The feed throat introduces material into the milling chamber. There are several designs of feed throats.[9] A gravity feed throat introduces material tangentially to the rotation of the blades. Other throats are available for production machines, such as a metered feed throat, liquid inlet throat, etc.Here, in this project feed throat was located 90o

to the face plate of the mill machine. 2.2 Blade Profile: [9]The type, quantity and shape of blade helps to determine the degree of reduction achieved based on the material being processed. Some blade styles offer flexibility of knife on one side and impact on the other. Knife-edged configuration is for gentle granulation and impact-edged for more aggressive reduction. In this project, stainless steel flat bars were used as it is food grain processing. Since it is low energy size reduction, reduction of the rasp bar speed should be done to decrease the percentage of cracked seeds by maintaining 600-1400rpm speed range. [3] Therefore 3 hp motor is used to supply power to have rotating speed 2880 rpm for machine operation and the rasp bar rotor speed is kept at 1440 rpm.

ENGINEER 40

1.1 Literature review: In general “size reduction” is taken to mean the disintegration of solid substances by mechanical forces without altering their solid state. This also includes the division of liquids into drops or gases into bubbles. However, the physical and chemical condition of the disintegrated material may alter, particularly when inhomogeneous substances are present. The preparation for separation according to material components, e.g. grinding grain, is therefore, one of the classical tasks of size reduction techniques. Size reduction or „comminution‟ is the unit operation in which the average size of solid pieces of food is reduced by application of grinding, compression or reduction of particle size is an important operation in many chemical and other industries. The important reasons for size reduction are: Easy handling Increase in surface area per unit volume Separation of entrapped components

impact forces

In all types of size reduction, there are three types of forces used to reduce the size of foods:

1. compression forces, 2. impact forces, and 3. shearing (or attrition) forces.

Cereal grains generally are the primary source of energy in feedlot diets. Availability of energy from the grain depends largely on the type of grain used as well processing of that grain.[13].

A variety of grain processing techniques are used including grinding, steam flaking, and compiling high moisture corn to ferment. Each processing method differs in its nutritional efficacy [13] and each has a unique associated cost. [14]. Criteria for size reduction An ideal crusher would (1) have a large capacity, (2) require a small power input per unit of product, and 3) yield a product of the single size. 2. Design details The Comminutor operates by feeding material uniformly into a chamber in which a rotating blade assembly reduces the particles of the material by cutting or impacting them. The material discharges through a screen which

regulates final particle size at the outlet of the milling chamber. The blade and screen act in conjunction to determine final product-sizing.

Figure 1- Sketch of rasp bar mill in operation [9] 2.1 Feed Throat: The feed throat introduces material into the milling chamber. There are several designs of feed throats.[9] A gravity feed throat introduces material tangentially to the rotation of the blades. Other throats are available for production machines, such as a metered feed throat, liquid inlet throat, etc.Here, in this project feed throat was located 90o

to the face plate of the mill machine. 2.2 Blade Profile: [9]The type, quantity and shape of blade helps to determine the degree of reduction achieved based on the material being processed. Some blade styles offer flexibility of knife on one side and impact on the other. Knife-edged configuration is for gentle granulation and impact-edged for more aggressive reduction. In this project, stainless steel flat bars were used as it is food grain processing. Since it is low energy size reduction, reduction of the rasp bar speed should be done to decrease the percentage of cracked seeds by maintaining 600-1400rpm speed range. [3] Therefore 3 hp motor is used to supply power to have rotating speed 2880 rpm for machine operation and the rasp bar rotor speed is kept at 1440 rpm.

ENGINEER41

Photograph 1- Interior view of the gritting chamber

2.3. Design calculation of shaft : The main shaft transmits 3kW at 1440 rpm and is carrying a pulley (D) and a rotor (A) with four sets of rasp bars. The torque (T) transmitted by the shaft,

( .89.1914402

603NmT

………..(1)

T1 and T2 are tensions in the tight side and slack side of the belt on the double v grooved pulleyD respectively. Rotor weight =1250N, pulley weight= 1000N

ø 254 ø 150

B C

100 425 100

A D

Figure 2 - Rotor and pulley mounted on the shaft and shaft mounted on the bearing B and C

RB 1000N

A D

1250N RcFigure 3 - Force diagram on the shaft

Taking moments around the support C(RBx425=1250x525-1000x100 ….(2)) RB=1309N (RB + 1000 = RC +1250 …….(3)) 1309+ 1000= RC+1250 1059N=RC

Where RB - Reaction at bearing B And RC- Reaction at bearing C Bending moments at B and C are zero. Bending Moment at A is, 1250x100=125 x103 Nmm Bending moment at D is, 1000x100=100x103 Nmm Maximum bending moment Mmax= 125x103

Nmm

RD

T1 T3

T2 T4

Figure 4-Tensions of the belts acting on the driving pulley D

T1=T3, T2=T4 214231 2289.19 TTRTTTT D

………..(4) Where, T1 & T2-fan belt tight side and slack side tensions T3&T4-motor belt tight side and slack side tensions RD-Diameter of pulley, mm (

2

1log3.2

T

T ………….(5)) [2]

Coefficient of friction between the belt and pulley is 0.3 [ 2] When ,

57.22

1

T

T

NTNTTT 27.16,33.6945.957.2 2111 Equivalent twisting moments is determined by

22 TKMKT tme ……..(6) Where Km –Shock factor M – Maximum bending moment, Nmm Kt - Fatigue factor [2]

ENGINEER 42

Shock and fatigue factors for bending and torsion [2] Km=2, Kt=1.5

223 27.165.1101252 eT

= 3968 Nmm

( 3

16dTe ………..(7)

33 89.191616

3968 dd

d=10mmconsidering the safety factor as 4.6 as there are impact loads [2] d=46mmUsed 50.8 mm stainless steel bar as the main shaft. 2.4 Energy and power requirements in size reduction[2]The cost of power is a major expense in crushing and grinding, so the factors that control this cost are important. 2.5 Open area[11] Open area or perforation area for a sieve having triangular pitch is determined by

T R

(2

2 69.90

T

R ………….(8))

Where R-hole diameter, mm T-Triangular pitch, mmAccording to the equationfor 5mm sieve, perforation area is22.7mm2 for 6mm sieve, perforation area is32.6mm2 for 7mm sieve, perforation area is19.8mm2 for 8mm sieve, perforation area is25.8mm2

2.6 Crushing efficiency

Empirical relationships determined by Rittinger‟s and Kick‟s law: The work required in crushing is proportional to the new surface created. This is equivalent to the statement that the crushing efficiency is constant and, for a given machine and material, is independent of the sizes of feed and product. If the sphericities F a (before size reduction) and F b (after size reduction) are equal and the machine efficiency is constant, the Rittinger’s law can be written as

where P - the power required, W - the feed rate to crusher, g/s

- the average particle diameter before crushing,mm

, - the average particle diameter after crushing, mm Kr - Rittinger‟s coefficient.[12] Rittinger‟s law is applicable for feed size less than 0.05mm. 2.7 Bond crushing law [12] Bond crushing law is applicable for feed size in between 0.05mm to 50mm.The work required to form particles of size Dp from very large feed is proportional to the square root of the surface-to-volume ratio of the product, sp/vp. Since F s = 6/Dp, it follows that (P/ ṁ = Kb/(Dp)^0.5 ……….(10))

where Kb - is a constant that depends on the type of machine and on the material being crushed P- power required, W ṁ- feed Rate, kg/h From experimental results, following parameters were derived onP=2112 W, =55kg/h Dp=6mm, Kb=3.4x105 Jmm/kg

=72 kg/h Dp=7mm, Kb=3.4x105 Jmm/kg P=3.4x105x72/2.65J kg/h = 92.3x105/3600 =2563W

Therefore 3 hp motor is selected.

Overall efficiency= 100tEnergyinpu

utEnergyoutp…(11)

Overall efficiency=1400/3000x100 =47%

3.Materials and method Maize rasp bar milling or the corn gritting is a kind of dry milling, which can be simply defined as process of impacting corn grains, that leads to grain size reduction and passing through a screen. Electrically operated small scale Rasp Bar Mill was developed and tested for maize. The mill consists with 5 major components; feeding funnel, grinding chamber, power supplying unit, blower unit and collecting outlet. Grinding chamber consists of four (04)

………..(9) sets of rectanguler blades attached to a rotor shaft that leads to size reduction of grains and a sieve, that has round perforations, enclosed the blades to prevent leaving grits from the grinding chamber until they are at least as small as the sieve openings. There are four sieves having 5mm, 6mm, 7mm and 8mm diameter perforations that can use separately according to the preference. Three horse power (03hp)with rotating speed 2880 rpm motor supplies power for machine operation. Following parameters were determined

Physical properties of the maize Out put efficiency(%) gritting capacity (kg/h) feeding capacity (kg/h) percentage of grits (%) screen effectiveness (%)

Machine and operating parameters Screen dimensions Sieve size Screen pitch Hopper capacity Rpm of the unit

The motor was started and spined the rasp bars. Dried maize was fed through the hopper. The hopper capacity was around 1000g. Small bits of plant material exit through the punched holes at the bottom.

3.1 Testing methodologyPerformance of the developed gritting unit was evaluated in terms of output capacity, gritting efficiency, power requirement against different moisture levels.(i.e.8%,9%,10%,11%) Different moisture levels were obtained by oven drying. The Moisture content was measured by Grain Moisture meter GMK 303A G-WON manufacturer. Maize grains flowed under gravity to the gritting chamber where impact of revolving gritting occures. Rotor with 4 sets of rasp bars gritted the grain. Pneumatic cleaning is used to remove light, chaffy and dusty materials out of the grain while heavier materials move downward. Air is generated by a mechanical fan. Light material get collected into the cyclone of which inlet is fixed into the grain falling path.

Photograph 2- Fabricated Rasp bar mill

4. Results and discussion Figure 5 shows the change of gritting efficiency with the sieve size. When the perforation size is larger efficiency is higher.

Figure 5 - Gritting efficiency vs perforation size

Figure 6- power requirement vs perforation size

ENGINEER 42

Shock and fatigue factors for bending and torsion [2] Km=2, Kt=1.5

223 27.165.1101252 eT

= 3968 Nmm

( 3

16dTe ………..(7)

33 89.191616

3968 dd

d=10mmconsidering the safety factor as 4.6 as there are impact loads [2] d=46mmUsed 50.8 mm stainless steel bar as the main shaft. 2.4 Energy and power requirements in size reduction[2]The cost of power is a major expense in crushing and grinding, so the factors that control this cost are important. 2.5 Open area[11] Open area or perforation area for a sieve having triangular pitch is determined by

T R

(2

2 69.90

T

R ………….(8))

Where R-hole diameter, mm T-Triangular pitch, mmAccording to the equationfor 5mm sieve, perforation area is22.7mm2 for 6mm sieve, perforation area is32.6mm2 for 7mm sieve, perforation area is19.8mm2 for 8mm sieve, perforation area is25.8mm2

2.6 Crushing efficiency

Empirical relationships determined by Rittinger‟s and Kick‟s law: The work required in crushing is proportional to the new surface created. This is equivalent to the statement that the crushing efficiency is constant and, for a given machine and material, is independent of the sizes of feed and product. If the sphericities F a (before size reduction) and F b (after size reduction) are equal and the machine efficiency is constant, the Rittinger’s law can be written as

where P - the power required, W - the feed rate to crusher, g/s

- the average particle diameter before crushing,mm

, - the average particle diameter after crushing, mm Kr - Rittinger‟s coefficient.[12] Rittinger‟s law is applicable for feed size less than 0.05mm. 2.7 Bond crushing law [12] Bond crushing law is applicable for feed size in between 0.05mm to 50mm.The work required to form particles of size Dp from very large feed is proportional to the square root of the surface-to-volume ratio of the product, sp/vp. Since F s = 6/Dp, it follows that (P/ ṁ = Kb/(Dp)^0.5 ……….(10))

where Kb - is a constant that depends on the type of machine and on the material being crushed P- power required, W ṁ- feed Rate, kg/h From experimental results, following parameters were derived onP=2112 W, =55kg/h Dp=6mm, Kb=3.4x105 Jmm/kg

=72 kg/h Dp=7mm, Kb=3.4x105 Jmm/kg P=3.4x105x72/2.65J kg/h = 92.3x105/3600 =2563W

Therefore 3 hp motor is selected.

Overall efficiency= 100tEnergyinpu

utEnergyoutp…(11)

Overall efficiency=1400/3000x100 =47%

3.Materials and method Maize rasp bar milling or the corn gritting is a kind of dry milling, which can be simply defined as process of impacting corn grains, that leads to grain size reduction and passing through a screen. Electrically operated small scale Rasp Bar Mill was developed and tested for maize. The mill consists with 5 major components; feeding funnel, grinding chamber, power supplying unit, blower unit and collecting outlet. Grinding chamber consists of four (04)

………..(9) sets of rectanguler blades attached to a rotor shaft that leads to size reduction of grains and a sieve, that has round perforations, enclosed the blades to prevent leaving grits from the grinding chamber until they are at least as small as the sieve openings. There are four sieves having 5mm, 6mm, 7mm and 8mm diameter perforations that can use separately according to the preference. Three horse power (03hp)with rotating speed 2880 rpm motor supplies power for machine operation. Following parameters were determined









ENGINEER43

sets of rectanguler blades attached to a rotor shaft that leads to size reduction of grains and a sieve, that has round perforations, enclosed the blades to prevent leaving grits from the grinding chamber until they are at least as small as the sieve openings. There are four sieves having 5mm, 6mm, 7mm and 8mm diameter perforations that can use separately according to the preference. Three horse power (03hp)with rotating speed 2880 rpm motor supplies power for machine operation. Following parameters were determined









ENGINEER 44

Figure 6 shows the change in power requirement for size reduction of maize, with respect to the perforation size at moisture level 11%. Maximum power requirement was recorded in the perforation size of 6mm. This may be due to higher perforation area of this sieve compared to the other sieves.

Figure 7 - Gritting capacity vs perforation size

Drastic change of gritting capacity at 11% (wet basis) moisture content was observed in the perforation size 8mm. This may be due to higher opening area in the seive.

Figure 8- shows the change of feeding rate with the perforation size.

Figure 8 - Feeding rate vs perforation size. Drastic change of feeding rate was observed in the perforation size 8mm at moisture level 11%(wet Basis. This may be due to the higher

opening area in the seive. In all cases, seive effectivness are greater than 86%. The speed of the rasp bar rotor and speed of the fan were kept at 1440 and2880 rpm respectivly. Table 1- Results of the perfornance testing

hole size

Feeding rate

Gritting capacity

Out put

efficiency

Power requirement

(mm) (kg/h) (kg/h) (%) (W)

5 24 13.58 56.6 560 6 40 25.07 62.6 1280 7 72 50.87 70.6 960 8 180 135.78 75.4 768

Table 1 shows the important results of the performance testing. In early performance testing the fan power was inadequate. Fan was modified by trial and error method. The table give results obtain after fan modification

Figure 9- Gritting Efficiency vs perforation size

The same experiments were carried out to different moisture levels.i.e.8%, 9%, 10%, 11%(wet basis). An increase of gritting efficiency was noticed at perforation size 6mm. This may be due to highest perforation area as the pitch of this sieve was 10mm compared to other sieves.

ENGINEER 44

Figure 6 shows the change in power requirement for size reduction of maize, with respect to the perforation size at moisture level 11%. Maximum power requirement was recorded in the perforation size of 6mm. This may be due to higher perforation area of this sieve compared to the other sieves.

Figure 7 - Gritting capacity vs perforation size

Drastic change of gritting capacity at 11% (wet basis) moisture content was observed in the perforation size 8mm. This may be due to higher opening area in the seive.

Figure 8- shows the change of feeding rate with the perforation size.

Figure 8 - Feeding rate vs perforation size. Drastic change of feeding rate was observed in the perforation size 8mm at moisture level 11%(wet Basis. This may be due to the higher

opening area in the seive. In all cases, seive effectivness are greater than 86%. The speed of the rasp bar rotor and speed of the fan were kept at 1440 and2880 rpm respectivly. Table 1- Results of the perfornance testing

hole size

Feeding rate

Gritting capacity

Out put

efficiency

Power requirement

(mm) (kg/h) (kg/h) (%) (W)

5 24 13.58 56.6 560 6 40 25.07 62.6 1280 7 72 50.87 70.6 960 8 180 135.78 75.4 768

Table 1 shows the important results of the performance testing. In early performance testing the fan power was inadequate. Fan was modified by trial and error method. The table give results obtain after fan modification

Figure 9- Gritting Efficiency vs perforation size

The same experiments were carried out to different moisture levels.i.e.8%, 9%, 10%, 11%(wet basis). An increase of gritting efficiency was noticed at perforation size 6mm. This may be due to highest perforation area as the pitch of this sieve was 10mm compared to other sieves.

ENGINEER45

Figure 10- Feeding capacity vs perforation size Higher feeding capacity was observed at perforation size 6mm. This may be due to higher perforation area of this sieve. Human error may be the reason for different capacities of the sieve with perforation size 5mm.

Figure 11- Seive effectiveness vs perfortion size A low seive effectiveness was observed in perforation size 5mm at all moisture levels. This may be due to lower perforation area of this seive.

Figure 12 - Power requirment vs perforation size

High power requiement was observed at perforation size 5mm. This may be due to the lower perforation area of this seive. Power was supplied by 3hp motor.The speed of the rasp bar rotor and the speed of the fan were kept at 1440 and 2880 rpm respectivly in all experiments

5. Conclusions and recommendations

The study revealed that the fabricated maize gritting machine is capable of gritting maize at different moisture levels of maize and different seive perforation sizes. i.e. 5 , 6 , 7 , 8 mm in diameter.The product comes out is grits without cover and germ.The optimum performance was obtained at the perforation size 6 and the moisture content of 8%(wet basis). The overall efficiency of unit is 47% as calculated. Size reduction creates heat while in operation but it can be neglected. The brokens do not contain much heat. Therefore no color change appears. The manufacturing cost is calculated as Rs. 150,000/= according to the present market prizes.(2013).This can be a self employment for a person. The brokens can be boiled as it is for consumption with coconut scrapes and can be mixed with milk rice or can be used for making corn flakes.

ENGINEER 46

Acknowledgements

This research was carried out with treasury funds. Authors wish to thank for availing funds from treasury. Authors wish to express there gratitude to the staff of the workshop of IPHT for their technical ideas for the success of this project.

References:

1. Watson, A., Stanly, Rama, E. Paul, 1987, Com:Chemistry and Technology, American Association of Cerial Chemistry, Inc., St.Paul, Minnesota, USA.

2. Khurmi, R. S., Gupta, J. K., 2005, A textbook of Machine Design, Eurasia Publishing House, New Delhi, India.

3. Dong, A., and Edberg, R. J., Hammer mill grain thresher. original version (8/1994). Retrieved 24 October 2013 from of feedlot cattle: http://members.efn.org/~itech/HTM/Converted_Grain_Thresher.html

4. http://lorien.ncl.ac.uk/ming/particle/cpe124p3.html visited on 11/08/2013

5. http://www.lavorazionetutolofollador.it/en/Applications/corn-cob-abrasive-granulates-for-the-vibratory-finishing-and-sifting-of-metal-surfaces.html visited on 14/08/2013

6. http://www.fourleafmilling.com.au/products.php?id=14 visited on 15/08/2013

7. https://www.google.lk/?gws_rd=cr&ei=16CyUq6bD6efiAfDwIHQAQ#q=Macken+et+al.%2C+2006&start=20 visited on 15/08/2013

8. http://benefitof.net/benefits-of-maize/

9. http://www.fitzmill.com/chemical/size_reduction/theory/theory_sr.html visited on 17/08/2013

10. https://www.google.lk/?gws_rd=cr&ei=UvizUpmSHsWpiAfj6IHQDA#q=Macken+et+al.%2C+2006&start=20 visited on 01/09/2013

11. http://www.rmig.com/en/technical+info /formulae/calculation+open+area visited on 02/09/2013

12. http://books.google.lk/books?id=QdEDzpQ9TCIC&pg=PA89&lpg=PA89&dq=Rittinger%E2%80%99s+coefficient&source=bl&ots=HGAUuFan6n&sig=AsRnOCsN3v6vYUfb49ygnt1hpfs&hl=en&sa=X&ei=EUi9UuaZJIqhigfvjoCoDg&ved=0CFoQ6AEwCA#v=onepage&q=Rittinger%E2%80%99s%20coefficient&f=false visited on 03/09/2013

13. Owens, F. N., Secrist, D. S., Hill, W. J., Gill, D. R., 1997, The effect of grain source and grain processing on performance a review. J. Anim., Sdi., 75(3) : 868-879

14. Macken, C. N., 2 Pas, Erickson G.E.,3 PAS, and Klopfentein,T.J.2006, The cost of corn processing for finishing cattle.1.J.The professional Animal Scientist: 22:23-32

15. Department of Agriculture, Sri Lanka Bulletin , 2006,

16. Central Bank of Sri Lanka, (2012)

Breakdown of cost estimate Description

Unit am cost

ount

01 HP single phase motor no 1 14950 1" shafting

ft 3 1395

6"B double V pulley

no 1 960

3"B single V pulley

no 2 980 1/8" B.I.Sheet

sheet 0.5 6950

16G B.I.Sheet

Sheet 1 7200 2" stainless steel round bar ft 3 16350 1" stainless steel flat iron ft 6 2310 transport charges

450

59"B V belt

belt 1 300 62"B V belt

belt 1 400

1.5"x1.5"x4mm Angle Iron bar 1 2200 2"pllow block bearing no 2 8000 1"pillow block bearing no 2 2200 1/4" M.s.Sheet

sheet 0.25 6750

Flat iron 1'x4mm

bar 1 1200

Flat iron 1.5"x4mm

bar 2 750

16mmx1.5" nut & bolt bolt 8 70 5/16"nut &bolt

g 500 100

1/4"nut & bolt

g 250 75 cock sheet

200

Spray paint

L 1 1600 Tinner

L o2 650

Screen(0.5)

no 1 475 Arc welding rod packet no 1 650 D.O.L.starter

no 1 5900

2" round iron

ft 6 1000 gas oxygen welding

500

Sub Total

84565

Welder/Mechanic

24083 Labour

21000

sub total

129648

Profit 15%

19447 Grand total

149095

selling price

150000

ENGINEER 46

Acknowledgements

This research was carried out with treasury funds. Authors wish to thank for availing funds from treasury. Authors wish to express there gratitude to the staff of the workshop of IPHT for their technical ideas for the success of this project.

References:

1. Watson, A., Stanly, Rama, E. Paul, 1987, Com:Chemistry and Technology, American Association of Cerial Chemistry, Inc., St.Paul, Minnesota, USA.

2. Khurmi, R. S., Gupta, J. K., 2005, A textbook of Machine Design, Eurasia Publishing House, New Delhi, India.

3. Dong, A., and Edberg, R. J., Hammer mill grain thresher. original version (8/1994). Retrieved 24 October 2013 from of feedlot cattle: http://members.efn.org/~itech/HTM/Converted_Grain_Thresher.html

4. http://lorien.ncl.ac.uk/ming/particle/cpe124p3.html visited on 11/08/2013

5. http://www.lavorazionetutolofollador.it/en/Applications/corn-cob-abrasive-granulates-for-the-vibratory-finishing-and-sifting-of-metal-surfaces.html visited on 14/08/2013

6. http://www.fourleafmilling.com.au/products.php?id=14 visited on 15/08/2013

7. https://www.google.lk/?gws_rd=cr&ei=16CyUq6bD6efiAfDwIHQAQ#q=Macken+et+al.%2C+2006&start=20 visited on 15/08/2013

8. http://benefitof.net/benefits-of-maize/

9. http://www.fitzmill.com/chemical/size_reduction/theory/theory_sr.html visited on 17/08/2013

10. https://www.google.lk/?gws_rd=cr&ei=UvizUpmSHsWpiAfj6IHQDA#q=Macken+et+al.%2C+2006&start=20 visited on 01/09/2013

11. http://www.rmig.com/en/technical+info /formulae/calculation+open+area visited on 02/09/2013

12. http://books.google.lk/books?id=QdEDzpQ9TCIC&pg=PA89&lpg=PA89&dq=Rittinger%E2%80%99s+coefficient&source=bl&ots=HGAUuFan6n&sig=AsRnOCsN3v6vYUfb49ygnt1hpfs&hl=en&sa=X&ei=EUi9UuaZJIqhigfvjoCoDg&ved=0CFoQ6AEwCA#v=onepage&q=Rittinger%E2%80%99s%20coefficient&f=false visited on 03/09/2013

13. Owens, F. N., Secrist, D. S., Hill, W. J., Gill, D. R., 1997, The effect of grain source and grain processing on performance a review. J. Anim., Sdi., 75(3) : 868-879

14. Macken, C. N., 2 Pas, Erickson G.E.,3 PAS, and Klopfentein,T.J.2006, The cost of corn processing for finishing cattle.1.J.The professional Animal Scientist: 22:23-32

15. Department of Agriculture, Sri Lanka Bulletin , 2006,

16. Central Bank of Sri Lanka, (2012)

Breakdown of cost estimate Description

Unit am cost

ount

01 HP single phase motor no 1 14950 1" shafting

ft 3 1395

6"B double V pulley

no 1 960

3"B single V pulley

no 2 980 1/8" B.I.Sheet

sheet 0.5 6950

16G B.I.Sheet

Sheet 1 7200 2" stainless steel round bar ft 3 16350 1" stainless steel flat iron ft 6 2310 transport charges

450

59"B V belt

belt 1 300 62"B V belt

belt 1 400

1.5"x1.5"x4mm Angle Iron bar 1 2200 2"pllow block bearing no 2 8000 1"pillow block bearing no 2 2200 1/4" M.s.Sheet

sheet 0.25 6750

Flat iron 1'x4mm

bar 1 1200

Flat iron 1.5"x4mm

bar 2 750

16mmx1.5" nut & bolt bolt 8 70 5/16"nut &bolt

g 500 100

1/4"nut & bolt

g 250 75 cock sheet

200

Spray paint

L 1 1600 Tinner

L o2 650

Screen(0.5)


no 1 5900

2" round iron


500

Sub Total

84565

Welder/Mechanic

24083 Labour

21000

sub total

129648

Profit 15%

19447 Grand total

149095

selling price

150000

ENGINEER47

Spray paint

L 1 1600 Tinner

L o2 650

Screen(0.5)


no 1 5900

2" round iron


500

Sub Total

84565

Welder/Mechanic

24083 Labour

21000

sub total

129648

Profit 15%

19447 Grand total

149095

selling price

150000

ENGINEER49


1 ENGINEER

Eng. M. N. C. Samarawickrama, C. Eng., MIE(Sri Lanka), MGS (SL), B.Sc. Eng. Hons.(Moratuwa), M.Sc. (Peradeniya), MBA (Moratuwa), Senior Lecturer in Civil Engineering, Department of Civil Engineering, The Open University of Sri Lanka, Sri Lanka. U. B. Amarasinghe, B.Sc.(Sp.) Hons. (Geol.) (Peradeniya), M.Sc. (AIT), MGS (SL), Senior Lecturer in Geology, Department of Geology, University of Peradenya, Sri Lanka. K. N. Bandaraa, B.Sc.(Sp.) Hons. (Geol.) (Peradeniya), M.Sc. (AIT), MGS (SL), Director, Geotechnical Engineering and Testing Division, National Building Research Organisation.

Criteria to Assess Rock Quarry Slope Stability and Design in Landslide Vulnerable Areas of Sri Lanka:

A Case Study at Thalathu Oya Rock Quarry

M. N. C. Samarawickrama , U. B. Amarasinghe and K. N. Bandara Abstract: The ultimate causative factor for the failure is rapid removal of toe support of the slope due to unplanned mining accompanied with uncontrolled blasting. There is also a natural causative factor behind, a naturally formed highly weathered slip surface, where along the slope failure has taken place. Secondary discontinuity created along the well-developed foliation plane due to an earlier disturbance of rock mass along kinematically more unstable joint planes, is the inception. This has turned into a weaker plane by groundwater seepage for a very long period facilitated by drainage pattern of the area. Intense weathering features of failure zone, chert particles found from the slip surface are good indications for this factor. Furthermore, it was identified that, the shear strength of rock joints can conveniently and rapidly be determined using Rock Mass Rating System and Empirical Equations. Even though these methods provide more conservative values, results will be very useful in initial design work. Results show that the back analysis method is more reliable compared to above two but is conditional as a similar type of a failure need to occur in the same rock mass in order to employ this method. Moreover, it was revealed that Barton‟s theorem can be effectively applied for local rock masses in determining the shear strength of discontinuities and is reliable in using at lower stress levels. When considering the stability of remaining slopes of the same site, these are highly venerable for same type of failure at any moment. According to site geometrical parameters and shear strength parameters found out from back analysis reveals that the natural factor of safety is only around 1.0 for slopes that remain hanging at this site. Further, study reveals that, the most economical method of stabilizing these existing unstable areas in the site is by reduction of the slope height with the use of controlled blasting techniques.

Keywords: Back analysis, empirical equations, Rock Mass Rating System

1. Introduction Introduction Landslides hazard is a major problem in the Central and Sabaragamuwa provinces of Sri Lanka, where high rainfalls are experienced throughout the year. Increase of pore water pressure in joints and discontinuities of rock masses and subsequent reduction in effective stress cause to reduce the shear strength of the failure plane. Furthermore, human activities such as removal of vegetation cover on steeply dipping terrains for agricultural purposes cause to trigger slope failures in overburden areas due to removal of roots network of the vegetation. Worst conditions occur when insufficient drainage is provided, where subsequent stagnation of water remaining at the crest of the slope. Bench and associated slope failures in open cast mines (especially in Rock quarries) sometimes can be turned into disasters. The main causative factors behind are the ignorance of geological structural features of rock mass, careless violation of rules and regulating conditions imposed by the mining regulating authorities, non-removal of the overburden soil mass prior to excavation,

excavating and removal of toe support of slopes and improper designing of benches with poor drainage control. These may contribute either as singly or as combinations. 1.1 Scope and Objectives of the Study This study was carried out to identify the factors related to slope failures in open cast rock quarries, which should be considered before planning and design. In this purpose, a detailed study was carried out for a particular industrial level rock quarry site at ThalathuOya, where slope failure has already occurred due to unplanned quarry face development especially with regards to slope stability considerations. Mining engineers can



utilize the results of this study as a „Model Case‟ for their future mines planning activities. The specific objectives of the study are to, Determine the causative factors behind the

slope failure occurred in Operating face of ThalathuOya Rock quarry.

Carryout detail field and laboratory investigations to determine the shear strength parameters of the slip surface.

Identify remedial measures for possible vulnerable sections of natural permanent slopes of the same site for future slope failures.

Provide recommendations to mine planning professionals, which can be used as a base model for them

2. Methodology The following methodology was adopted in order to achieve the above mentioned objectives. 2.1 Step 01 Desk Study An initial study on the history of quarrying and slope failures occurred in the particular area was carried out and the information was obtained from the respective organizations [3]. The Geology and the Geomorphology of the area was studied using the 1:120000 Structural Geology maps [9] and the 1:50000 Topographic maps [10]. This is in order to identify the joints and major discontinuity patterns and drainage pattern of the study area. 2.2 Step 02 Engineering geological

assessment of the failure site In order to identify the possible causative factors behind and the failure type responsible for this failure, a detailed engineering geological assessment accompanied with a rock joint analysis was performed. In rock joint analysis, readings on the slope geometry, strikes and dips of slip surface and other joints and discontinuities, discontinuity spacing, separation and their geomechanical characteristics and ground water flow of discontinuities were obtained. This data are summarised and presented in Figure 6 and Table 1 and were later used to determine the most possible type of failure pattern. 2.3 Step03 Analysis of effectiveness of

the causative factors. Hydrogeological pattern of the study area was studied through data gathered in the initial steps to analyse the influence of pore water

pressure and degree of leaching of slip surface and thus the influence of ground water for the slope failure. Creeping effects of rock mass was studied by performing a rock mass classification, which assess the inherent quality of rock mass. 2.4 Step 04 Determination of Shear

Strength parameters of Slip surface During the analysis in finding the type of failure pattern in Step02, it was found out that it is of planer type failure and the findings are presented under section 3.6.Hence thereon following methods were employed in determining shear strength parameters rock. 2.4.1 Using the Method of back analysis As sample cases described by Bray & Hoek [2] and Sau Mau Ping Road, Kowloon city case in Hong Kong [11], a range of friction values were given to corresponding factor of safety equations and assuming that the Factor of Safety (F) reaches unity at failure. Different cohesion values were obtained for different internal friction angle values at different possible pore water pressure conditions. Two basic models (Model-01 and Model-02) which were earlier employed to similar cases [2] and [11] were employed in this study. These models represent most possible slope geometries where, Model 01, with a tension crack and Model 02 without a tension crack. Model-01

Figure 1 - Model 01: with a tension crack 𝐹𝐹𝐹𝐹 𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐 𝑊𝑊𝑊𝑊 𝜓𝜓𝜓𝜓𝜓𝜓𝜓𝜓𝜓𝜓𝜓𝜓𝜓 𝜓𝜓𝜓𝜓𝜓𝜓𝜓𝜓 𝜓𝑈𝑈𝑈𝑈𝜓𝑈𝑈𝑈𝑈 𝜓𝜓𝜓𝜓𝜓𝜓𝜓𝜓 𝑇𝑇𝑇𝑇 𝜃𝜃𝜃𝜃 ∅

𝑊𝑊𝑊𝑊 𝜓𝜓𝜓𝜓𝜓𝜓𝜓𝜓 ∝ 𝜓𝜓𝜓𝜓𝜓𝜓𝜓𝜓 𝑈𝑈𝑈𝑈 𝜓𝜓𝜓𝜓𝜓𝜓𝜓𝜓𝜓𝑇𝑇𝑇𝑇 𝜃𝜃𝜃𝜃 (1) Where 𝑍𝑍𝑍𝑍 𝐻𝐻𝐻𝐻𝐻 𝜓 𝜓𝜓𝜓𝜓𝜓𝜓𝜓𝜓 𝜓 𝜓𝜓𝜓𝜓𝜓𝜓𝜓𝜓 (2) 𝑐𝑐𝑐𝑐 𝐻𝐻𝐻𝐻 𝜓 𝑍𝑍𝑍𝑍 𝜓𝜓𝜓𝜓𝜓𝜓𝜓𝜓 (3) 𝑊𝑊𝑊𝑊 𝛾𝛾𝛾𝛾𝑟𝑟𝑟𝑟𝐻𝐻𝐻𝐻 𝜓 𝐻𝐻𝐻𝐻 𝑍𝑍𝑍𝑍 𝜓𝜓𝜓𝜓𝜓𝜓𝜓𝜓 𝜓 𝜓𝜓𝜓𝜓𝜓𝜓𝜓𝜓 (4) 𝑈𝑈𝑈𝑈 𝛾𝛾𝛾𝛾𝑤𝑤𝑤𝑤 𝑍𝑍𝑍𝑍𝑤𝑤𝑤𝑤 𝑐𝑐𝑐𝑐 (5) 𝑈𝑈𝑈𝑈 𝛾𝛾𝛾𝛾𝑤𝑤𝑤𝑤 𝑍𝑍𝑍𝑍𝑤𝑤𝑤𝑤 (6)

Ψp

P

V Z

ψf

Water pressure distribution

Anchor

H

T

W

U

αW ZW

θ














Ψp

P

V Z

ψf


Anchor

H

T

W

U

αW ZW

θ














Ψp

P

V Z

ψf


Anchor

H

T

W

U

αW ZW

θ

3 ENGINEER

∅– Angle of internal friction along the discontinuity

Model-02

Figure 2 - Model 02: without a tension crack 𝐹𝐹𝐹𝐹 𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐 𝑊𝑊𝑊𝑊 𝜓𝜓𝜓𝜓𝜓𝜓𝜓𝜓𝜓𝜓𝜓𝜓𝜓 𝜓𝜓𝜓𝜓𝜓𝜓𝜓𝜓 𝜓𝑈𝑈𝑈𝑈 𝑇𝑇𝑇𝑇 𝜃𝜃𝜃𝜃 ∅

𝑊𝑊𝑊𝑊 𝜓𝜓𝜓𝜓𝜓𝜓𝜓𝜓 ∝ 𝜓𝜓𝜓𝜓𝜓𝜓𝜓𝜓 𝜓𝑇𝑇𝑇𝑇 𝜃𝜃𝜃𝜃 (7) Where

𝑐𝑐𝑐𝑐 𝐻𝐻𝐻𝐻 𝜓 𝑍𝑍𝑍𝑍 𝜓𝜓𝜓𝜓𝜓𝜓𝜓𝜓 (8) 𝑊𝑊𝑊𝑊 𝛾𝛾𝛾𝛾𝑟𝑟𝑟𝑟𝐻𝐻𝐻𝐻 𝜓𝜓𝜓𝜓𝜓𝜓𝜓𝜓 𝜓 𝜓𝜓𝜓𝜓𝜓𝜓𝜓𝜓 (9) 𝑈𝑈𝑈𝑈 𝛾𝛾𝛾𝛾𝑤𝑤𝑤𝑤𝐻𝐻𝐻𝐻𝑤𝑤𝑤𝑤 𝜓𝜓𝜓𝜓𝜓𝜓𝜓𝜓 (10) ∅ – Angle of internal friction along the

discontinuity The Figure 1 and Figure 2 and Equations 1 to 10 were used in determining the value range for cohesion.Here, c‟ and Ø are the cohesion and angle of internal friction of rock and U- pore water pressure, V-thrust force generated in the tension crack , W – weight of the sliding wedge, - coefficient seismic acceleration taken as 0.08g = 0.785, H-slope height, ψf– slope face angle, ψf– dip of slip plane, z- tension crack depth, ZW- water head in tension crack in Model 01, HW-water head in slip plane in Model 02 and T-reinforcement applied through tension bolting (if available). 2.4.2 Using Empirical Equations Shear strength of highly jointed rock mass was determined using empirical equation proposed by Barton [1]. According to section 3.4, the stress levels applied on slip surface due to overburden is comparatively low and hence Barton‟s method was employed in this analysis, which is more effective in low stress levels. The Barton‟s equation can be given as, 𝜏𝜏𝜏𝜏 𝜎𝜎𝜎𝜎𝑛𝑛𝑛𝑛 𝜎𝜎𝜎𝜎𝑛𝑛𝑛𝑛 ∅𝑟𝑟𝑟𝑟 (11) Where,

- Joint compressive strength - Joint roughness coefficient

𝜏𝜏𝜏𝜏 - Shear strength of the joint

∅𝑟𝑟𝑟𝑟 - Residual friction angle 𝜎𝜎𝜎𝜎𝑛𝑛𝑛𝑛 - Normal effective stress Residual friction angle represents the theoretical minimum strength value of a planer and slickensided surface obtained when the roughness is completely worn away. The JRC was determined according to the roughness profiles [1]. The residual friction angle (∅𝑟𝑟𝑟𝑟 )was determined from the Equation 12[1]; and it is with the use of Schmidt hammer rebound value of corresponding weathered and fresh rocks. ∅𝑟𝑟𝑟𝑟 ∅𝑏𝑏𝑏𝑏 𝜓 𝑟𝑟𝑟𝑟 𝑅𝑅𝑅𝑅 (12) Where, ∅𝑏𝑏𝑏𝑏 is the basic friction angle of the rock, „R‟ the schmidt hammer rebound value of fresh rock surface and ‟r‟ the schmidt hammer rebound value of corresponding weathered rock surface 2.4.3 Using Rock Mass Rating System The shear strength of the rock mass can also be determined through rock mass classification. This is by using the geomechanics classification system (Rock Mass Rating RMR System) [4] and Slope Mass Rating (SMR) System [6], which the final score of rock class was correlated to the shear strength parameters of most unfavourable joint set of the rock mass. Here RMR is adjusted into SMR by, 𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑅𝑅𝑅𝑅 𝑅𝑅𝑅𝑅𝑆𝑆𝑆𝑆𝑅𝑅𝑅𝑅 𝐹𝐹𝐹𝐹 ∗ 𝐹𝐹𝐹𝐹 ∗ 𝐹𝐹𝐹𝐹 𝐹𝐹𝐹𝐹 (13) Where 𝐹𝐹𝐹𝐹 𝐹𝐹𝐹𝐹 𝐹𝐹𝐹𝐹 and 𝐹𝐹𝐹𝐹 are the adjusting factors for, dip direction of most vulnerable joint set relative to dip direction slope, dipangle of most vulnerable joint set, dip angle of most vulnerable joint set relative to dip angle slope and the blasting/ excavation method employed respectively. 2.5 Step 05 Introduction of remedial

measures Considering all the analysed data and assuming the worst possible combinations of causative factors for the failure, different slope stabilization techniques were introduced to stabilize the unstable portion that remained hanging at other part of the quarry site. Stabilization techniques such as reduction of slope angle, slope height, providing adequate drainage, and other miscellaneous methods such as tension bolting were analyzed against the factor of safety that can be gained and the cost effectiveness of the remedial action.

T

H


Anchor W

U

αW

Hw θ

ψf

Ψp

P

1/2Hw


2.6 Step 06 Design of safe slope for failed slope

In design of safe slope for already failed slope, the basic concept of achieving the required factor of safety 1.50 for permanent slope was the main objective. In determining the safe slope angle, the Equation 14 proposed by Orr, 1992 [5] was used as the base. 𝑆𝑆𝑆𝑆 𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅 (14) Where, S is slope angle in degrees and RMR is the Rock Mass Rating of particular rock mass. When S< 40º; one of the following options can be adopted. Option 01- Give-up slope design and stop the mining activity. Option 02- Take, S= 40º and minimize the pore water pressure development by improving the drainage simultaneously. 3. Engineering Geological

Assessment 3.1 Location

Scale (1: 50000)

Figure 3 - Topography of the site The particular site under study is at ThalathuOya in the Patahewaheta Divisional Secretariat of Kandy District, at a distance of two kilometers from ThalathuOya town to the left side of the Moragolla road, at an altitude of approximately 600MSL. The study area where slope failure has occurred extends to about two hectares. 3.2 Rainfall The rainfall data for the particular area pertaining to the time of failure (i.e. July 2001) is plotted in Figure 4. As the rain fall graph depicts, area has received 95mm rapid precipitation on 26th July 2001. This may have caused the sudden increase of pore water pressure on the slip surface of the site, which

ultimately caused to initiate the failure as a debris flow on 28th July 2001 (according to the information given by the neighbors). 3.3 Geology and geomorphology The main soil types overlain are residual soils, which formed as a result of weathering of underlying parent rock, charnockiticbiotite gneiss. Due to intense dipping nature of the terrain, the natural vegetation cover consists mostly of grass and few isolated tall trees.Lands have mainly used for cultivation of intercrops such as pepper and cloves, where the dipping terrains are favourable for growth of these crops. The lands that were affected due to the particular slope failure have been used previously for pepper cultivation.

Figure 4 - Rainfall in July 2001 to

ThalathuOya (Source- Meteorological Department of Sri

Lanka) When considering the site geological setting, the area belongs to the Highland Series of metamorphic rocks in the central part of the Sri Lanka. Furthermore the structural geology map of the area in Figure 5 reveals that, there is a shear zone running at the North-Eastern boundary of the site. This may be the initiation of the disturbance in the rock mass, which later act as the ground water seepage zone through the rock slope, along a weaker plane of foliation planes, which ultimately act as the slip surface. This can further be justified by the degree of weathering of the rock. The upper and lower parts of this seepage zone is weathered to a lesser degree than the seepage zone, whilst the mid slip surface is excessively weathered and the surface minerals leached due to this same ground water movement. Moreover, chert particles detected from mobilized debris as wells as thin crustal formations at the bottom of the right hand relief surface is good indication of groundwater movement for very longer period, where chert is formed by the

Rain fall July-2001

0.0

20.0

40.0

60.0

80.0

100.0

120.0

1 3 5 7 9 11 13 15 17 19 21 23 25 27 29 31Date

Rain

fall

(mm

)





Scale (1: 50000)






Rain fall July-2001

0.0

20.0

40.0

60.0

80.0

100.0

120.0

1 3 5 7 9 11 13 15 17 19 21 23 25 27 29 31Date

Rain

fall

(mm

)





Scale (1: 50000)






Rain fall July-2001

0.0

20.0

40.0

60.0

80.0

100.0

120.0

1 3 5 7 9 11 13 15 17 19 21 23 25 27 29 31Date

Rain

fall

(mm

)

5 ENGINEER

precipitation of silica under hydrostatic pressure for longer periods. The overburden thickness of the soil varies from 1.0m at the lower part of the slope to 0.25m at the upper part of the slope. The weathered thickness of the rock varies from 1.0m at the left failure slope to more than 6.0m at the right failure slope. The geomorphology of the area is with high relief with highly undulated terrain. The morphological depression at the vicinity of the site is a good indication of an existence of a major discontinuity.

. Scale (1: 120000)

Figure 5 - Structural Geology of the area 3.4 Geometry of failed slope

Figure 6 - Approximate Geometry of failed

slope Approximate geometrical parameters of failed slope were obtained from existing relief faces. The overburden thickness (Yx) varies from 4.50m to 6.50m. The location of probable tension crack was detected at the upper crest of the slope, where it can be traced from remaining extensions in the two relief wedges. The depth of the tension crack was around 6.00m.

3.5 Hydrology and hydrogeology The ground water movement in the particular site is shown in Figure 7 and as it depicts, large volume of ground water as well as surface water is passed through the site in the rainy season. There is a catchment area of around fifteen acres which elevates to more than 40m from the crest of the slope. The existence of Bamboo trees in the upper and upper left side parts of the slope crest is a good indication to prove the shallow ground water existence throughout the year. Also existence of surface water draining canals pointing towards the site location is a good indication for water stagnation effect in the rainy season, which ultimately may have caused to increase the pore water pressure conveniently. As mentioned in section 3.3, this process was in existence for very longer period until the ultimate failure.

Figure 7 - Hydrogeological setting of the site 3.6 Rock Joint Analysis and Failure type Dip and strike of all the joint sets were measured from the exposed relief areas. Moreover, conditions of these discontinuities were also obtained to classify the rock mass. 3.6.1 Stereonet Analysis results of Rock Joints The stereonet plot as shown in Figure 8 was developed based on data in Table 1 and the back analysis results of section 4.1, which was used to plot the friction circle. When considering the kinematic conditions of the possible rock slope failures, it is clear that vertical joint sets 1 and 2 may have contributed to form possible tension cracks due to toppling movements. Individual plane failures are possible for joints 2, 3, 4, 5 and slip surface. Out

Yx

Z= 6m

600

108m 54m

300


of them, most vulnerable is joint 5 and least is along slip surface, which has the least dip. Wedge failures are possible along the combination of joint sets 2&4, 2&5, 3&4, 3&5 and 4&5. From kinematic point of view the most vulnerable of them is the wedging along the intersection of joints 2&5. Table 1 - Orientation and Properties of

discontinuities

Figure 8 - Stereonet plot of great circles of rock

discontinuities 3.6.2 Rock Joint condition Analysis According to Table 1 results, the Joint Separation for all the joints, including slip surface is <0.10 mm and Joint Smoothness is

Irregular to Planer for all. It was very difficult to find any Joint Gouge in all discontinuities. The highest joint water condition is observed in slip surface and joints 1 and 2. Apart from these, it was quite evident that the rock mass is extremely weathered along the slip surface compared to other joints and embedded minerals in rock texture is leached out of their parent rock in this section, which extends to about 2.00m in thickness. 3.6.3 Most Possible type of Failure Pattern Stereonet analysis depicts that the failure type is combination of wedge and plane failures. Moreover, slip surface (which is parallel to foliation planes), which has the least possibility of contributing for the failure. However, initial micro level failures may have opened up these rock joints and later may have act as pipes of drains which brought surface runoff into an underlying weaker-well developed-foliation plane. It is evident that the failure has occurred along the presently exposed slip surface and hence the most contributory joint set out of above joint sets is a well-developed foliation plane and thus is a plane type failure. 4. Shear strength parameters of

slip surface 4.1 Shear strength determination from

back analysis Equations 1 and 7 in section 2.4.1 were rearranged by keeping the factor of safety (F) equals to unity (which is at limit equilibrium) and range of values were given to “”in order to obtain the variation of “c” for two basic slope geometries, which depicts in Figure 9. Previous experimental studies have shown that “” for gneissic rocks is ranged between 270and 340 [8].This range is highlighted from the ellipse in Figure 9, from which, the mid value of the range for “”, which is 300 was taken as the angle of internal friction of the slip surface and the corresponding lowest possible value for “c” from the curves become 10.35T/m2 or 0.1035MPa.

Joint No. (Dip direction/ Dip angle)

Joint Spacing (m)

Joint water condition (l/minute)

Slip surface (Foliation Plane) (600/300)

0.20-1.50 0.80

Joint No.01 (730/900)

0.30-3.00 0.80

Joint No.02 (1800/900)

0.30-3.00 0.80

Joint No.03 (300/400)

5.00-6.00 < 0.80

Joint No.04 (840/500)

14.00-15.00 < 0.80

Joint No.05 (700/600)

> 20.00 < 0.80

Slope Face (600/300-400)

According existing relief area geometry



discontinuities







Joint Spacing (m)



0.20-1.50 0.80

Joint No.01 (730/900)

0.30-3.00 0.80

Joint No.02 (1800/900)

0.30-3.00 0.80

Joint No.03 (300/400)

5.00-6.00 < 0.80

Joint No.04 (840/500)

14.00-15.00 < 0.80

Joint No.05 (700/600)

> 20.00 < 0.80

Slope Face (600/300-400)




discontinuities







Joint Spacing (m)



0.20-1.50 0.80

Joint No.01 (730/900)

0.30-3.00 0.80

Joint No.02 (1800/900)

0.30-3.00 0.80

Joint No.03 (300/400)

5.00-6.00 < 0.80

Joint No.04 (840/500)

14.00-15.00 < 0.80

Joint No.05 (700/600)

> 20.00 < 0.80

Slope Face (600/300-400)


7 ENGINEER

Figure 9 - Cohesion Vs Angle of Internal friction for both failure Models

4.2 Shear strength determination from

empirical equations Experimental results of unconfined compressive strength [7] of rock core samples obtained from slip surface are presented in Table 2. The average unconfined compressive strength of the slip surface from Table 2 is 43.50MPa, which is the Joint Compressive Strength (JCS) of the slip surface. The average Schmidt Hammer rebounce value on weathered rock surface (r) of slip surface was 31.875 and average Schmidt Hammer rebounce value on Fresh rock surface (R) of same slip surface was found to be 46.409. Table 2 - UCS results of rock samples

Sample Number

Failure load (kN)

UCS value (MPa)

01 100.00 58.65

02 475.00 41.80

03 290.00 35.71 04 250.00 49.52 05 400.00 41.42 06 200.00 42.79 07 75.00 38.51 08 175.00 39.61

The basic angle of internal friction for coarse granite (b) can be assumed as 260 [1]. Joint Roughness Coefficient (JRC), is 05 for joints which are Irregular –planer type [1]. Substituting these values to the Equation 12, r becomes 19.730.According to the slope geometrical parameters described in section 3.4, the wedge of the slope exerts low normal effective stress (‟n) level ranges from 08 MPa to 15 MPa on the slip surface. Applying this stress range to Equation 11 results the shear strength profile presented in Table 3.

Table 3- Shear strength range for different effective normal stress levels

Normal effective stress level(‟n) MPa

Shear strength (‟) MPa

08 10 12 14 16

3.45 4.23 4.98 5.71 6.43

Table 4- Shear strength deviation from two

different methods Normal effective

stress (‟n) MPa

Empirical equations (‟)EmMPa

Back analysis

(‟)BaMPa

Difference (‟)

(‟)Em- (‟)Ba MPa

08 10 12 14 16

3.45 4.23 4.98 5.71 6.43

4.718 5.873 7.032 8.183 9.337

1.27 1.64 2.05 2.47 2.91

Using the results of back analysis discussed under section 4.1, the shear strength profile can be developed for the same effective normal stress levels, based on the Mohr –Coulomb criterion. As depicts in above Table 4, the results from two methods provide closer results at lower normal effective stress levels which, indicates the validity of the assumptions and evaluation. Theoretically, Barton‟s method generally provides more conservative values compared to other methods and is more valid for low normal stress levels [1]. This has been proved by the results of Table 4 as the Barton‟s method provide lower shear strength values compared to the values provided by the Mohr- Coulomb criterion and the difference diverges as the normal stress level increases. 4.3 Shear strength determination from

Rock Mass Classification Table 5 - Rock Mass Rating results

Parameter Results Rating(%) Average Intact Rock Strength

43.50MPa (from Table-02)

04

RQD % 42.40 % 08 Joint spacing Minimum Joint

Spacing = 0.30m 10

Joint Condition

Irregular & planer joint surfaces; Continuous; Joint Separation<1.0mm; Soft joint wall rock

20

Ground water condition

0.80 (l/minute) (Damp)

10

Total Rating 52


The Rock Mass Rating (RMR) value for particular rock mass is 52% (Table 5). Section 3.6 has shown that the most vulnerable joint set contributed to the failure was the well- developed foliation plane, which acts as the failure plane. When considering the data in Table 1, the adjustment factors (𝐹𝐹𝐹𝐹 𝐹𝐹𝐹𝐹 𝐹𝐹𝐹𝐹 and𝐹𝐹𝐹𝐹 ) of Slope Mass Rating (SMR) value become 01, 0.60, -50 and 0(since blasting has employed in previous mining process)respectively. Substituting these values to the equation 13, the SMR value of the rock mass becomes 22.Reverting back to the adjusted RMR, i.e. SMR with rock mass shear strength properties the Net RMR (or SMR) is within21-40 and thus, Rock Class Number belongs to category IV and can be concluded as a Poor Rock when considering the stability of a rock slope in an open pit mine. For rock class IV, the corresponding cohesion and the angle of internal friction of the rock mass are 0.100MPa-0.200MPa and 150-250 respectively. 5. Stabilizing the vulnerable areas 5.1 Determination of Best Mitigation

Technique To mitigate future slope failures in the post failure vulnerable areas in the same site, attempts were made to achieve a minimum factor of safety of 1.5 by employing different types of improvement techniques such as, 01. Reduction of Slope height (reduction of

overburden stress) 02. Reduction of Slope angle (reduction of

overburden stress)

03. Drainage improvement (reduction of pore water pressure)

04. Reinforcement of slope.(increase of shearing resistance of slip surface)

The required factor of safety was calculated by assuming the worst possible conditions, i.e. when the pore water pressure is at maximum for both of the assumed failure models (Model 01 and Model 02). The shear strength parameters deduced from back analysis were used in conjunction with Equation 1 and Equation 7 in calculating the respective safety factors and plotted the gained factor of safety against percentage improvement by the respective stabilizing technique. It is evident in Figure 10, that the required factor of safety, 1.5 (which is highlighted by the dotted line) can only achieved through reduction of slope height and the introduction of reinforcements. Out of these two, the most rapid (most economical) and practical method could be the reduction of slope height, as tension bolts will need to penetrate to grater depths than in normal circumstances, since the underlying bedrock is weathered down to a substantial depth. The reason behind lack of effectiveness in reduction of slope angle and drainage improvement can be explained through Figure 6, where slope geometry suggest that reduction slope angle will reduce the overburden weight in very low proportions as natural slope surface is running almost parallel to the slip surface after initial slope angle of 600.

Figure 10 - Variation of Factor of Safety with Percentage improvement for both Failure Models





overburden stress)









overburden stress)





9 ENGINEER

Figure 11 - Model Slope Design for failed slope

The same applies for drainage, where improvements in drainage is less effective compared to pore water pressure generated by large extent of slip plane having about 128m in length. 5.2 Sample Safe Slope Design for the

failed slope This will serve as a model for Engineers in better planning of their mining activities. Using the shear strength parameters determined from the back analysis, with a factor of safety of 1.50 for open pit mine slope, one can carry out the design in iterative basis until the required factor of safety is achieved. As a basis, it was tried to minimize disturbing force by decreasing the weight of the wedge (which happens normally as mining progresses) and simultaneously maintaining the resisting force as constant as possible, by maintaining the effective contact area of the wedge as constant as possible. For the failed slope it is reasonable to assume the Model-01, slope with tension crack behaviour. Different slope parameters were taken in design of the slope and using the Equation 1 the factor of safety was tested against these parameters. This procedure was carried out until required factor of safety was achieved. The final design is shown in Figure 11 and it was arrived in following manner. The Equation 14 gives overall slope angle, S = 0.65*22+ 25 as 39.3º; S < 40º thus, taking S = 40º and slip surface angle =300; Overall Bench angle (or Inter Ramp Angle since multiple sets

of benches need to be employed with several ramps due to slope height) (S) = 400; Individual Bench angle= 600; Individual Bench height = 4.50; Number of Individual Benches per inter ramp= 04, Number of Ramps used= 03 and Ramp Length = 9.00m.The initial weight of the wedge = 1323.52 T. Reduction of up thrust force and the horizontal tension crack force can be achieved by employing proper drainage methods on upper crest of the slope. Thus, U= 0.00 T/m2 and V= 0.00 T/m2. Reduction of the weight of the wedge due to introduction of the benches = 695.474T. Substituting the above values to the Equation 01, gives a Factor of Safety, F = 1.57. However it should be emphasized that the initiation of the bench mining should be started from top of the slope and control basting should be carried out until the required geometry of the slope is achieved. 6. Conclusions and

Recommendations 6.1 Causative factors for the slope failure Structural Geology of the area reveals previous disturbances in this same area due to tectonic movements in adjacent shear zone. Hydrogeological pattern of the area has further worsened the situation by stagnation of water on this disturbed rock mass, may have caused to seep large quantities of flow through opened up more unstable joint sets down to more stable well developed weaker foliation plane, which finally became as the slip surface of the case under consideration. This fact has been further


proved by the presence of chert particles on the slip surface and excessive weathering caused to leach the mineral grains of the slip surface and thus the reduction of the shear strength of the slip surface. Uncontrolled blasting, removal of toe support of the slope and non-employment of drainage facilities have created the reactivation of the previously disturbed area and increase of effective upward force, which ultimately may have caused to displace the material above the slip surface as a debris flow. 6.2 Shear strength parameters in slip

surface According to below Table 6 most conservative and economical testing method is the employment of RMR system in initial design stages. The back analysis can be employed only if a similar type failure has occurred in the vicinity of the site, under the same conditions and provides least conservative values. Use of empirical equations proposed in this study yield more conservative values compared to back analysis method, but cheaper and economical in practical situations. The results of the Table 4 and Table 6 prove the applicability of Barton‟s theorem for local rock masses, where the Barton‟s equation is reliable in using at lower stress levels and it deviates from the actual values at higher stress levels. Table 6 - Comparison of different methods

for Shear strength parameters Method

Shear strength parameters

Angle of internal friction ( )

Cohesion MPa

Back analysis

300

0.1035

Empirical equations

b =260 r = 19.730

-

RMR system

150-250 0.100 -0.200

6.3 Remedial measures to stabilize the

vulnerable areas of the site As the most immediate measure, the canals pointed towards the slip surface should be diverted to elsewhere and sufficient underground draining system should be implemented to reduce the excessive pore water pressures that will be developed in rainy seasons. Furthermore, analysis has revealed that the most economical, rapid and practicable method is to reduce the slope height and this should be carried out with the use of controlled blasting.

Acknowledgement Authors wish to acknowledge the assistance given by the officials of National Building Research organisation, Department of Civil Engineering and Department of Earth Resources Engineering of University of Moratuwa and the Post Graduate Institute of Science of University of Peradeniya in carrying out this study. References 1. Barton, N., Chouby, V., “The Shear Strength of

Rock Joints in Theory and Practise”, Rock Mechanics. Vol.10/ 1-2, pp 1-54, 1977.

2. Bray, J. W., Hoek, E., Rock Slope Engineering. 3rdedition, The Institution of Mining & Metallurgy, London, 1981.

3. Investigation Report on Earth Instability at the Rock Quarry Site at ThalathuOya, National Building Research Organisation, Colombo 05, February-2002.

4. Jaeger, C., Rock Mechanics and engineering. 2nd edition, Cambridge University Press, 1979.

5. Orr, C. M., “Use of Rock Mass Rating (RMR) System in Assessing the Stability of Rock Slopes”, Milestones in rock engineering: the Bieniawski Jubilee collection. A. A. Balkema Publishers, Rotterdam, pp 159-172, 1996.

6. Romana, M., “New Adjustment Ratings for Application of Bieniawski Classification to Slopes”, Proceedings of International Symposium on the Role of Rock Mechanics, International Society for Rock Mechanics, Salzburg: 49-53, 1985.

7. Standard Testing Methods for Unconfined Compressive Strength of Intact Rock Core Specimens, American Society for Testing and Materials – Designation: D 2938 – 79.

8. Wyllie, D. C., Mah, W. C., Rock Slope Engineering; Civil and Mining. 4th edition, Taylor & Francis, London and New York, 2005.

9. 1: 120000 Structural Geology map of Kandy area, Geological Survey and Mines Bureau –Sri Lanka, Dehiwala, 2001.

10. 1: 50000 Topography map of Kandy area, Survey Department of Sri Lanka, 2000.

11. http://www.rocscience.com/hoek/corner/7_A_slope_stability_problem_in_Hong_Kong.pdf, Visited, 15th February 2002.







Cohesion MPa

Back analysis

300

0.1035

Empirical equations

b =260 r = 19.730

-

RMR system

150-250 0.100 -0.200















ENGINEER 10






Cohesion MPa

Back analysis

300

0.1035

Empirical equations

b =260 r = 19.730

-

RMR system

150-250 0.100 -0.200















SECTION II

ENGINEER61

Strategic Decision Support for Resolving Conflict Existing in Per Aru basin located in Vavuniya

district

G. Abira and K.D.W. Nandalal

Abstract: The graph model for conflict resolution (GMCR), along with its associated decision support system GMCR II, is employed for systematically studying the strategic aspects of a conflict existing in Per Aru basin. National Water Supply & Drainage Board has planned to build a reservoir on the Per Aru for the sole purpose of supplying drinking water to the urban areas of Vavuniya district. People who will be affected due to the inundation of their lands have shown resistance to the construction of the reservoir. Moreover, government organizations such as Forest Department and Wildlife Department also have expressed their concerns over the construction of a reservoir since that will have diverse impacts on forest lands and wildlife in the area that will be inundated by the proposed reservoir. The above three stakeholders, viz., National Water Supply & Drainage Board, affected people and government organizations suggest a few options to solve the conflict.The options are specific actions that can actually occur in a conflict. The decision support system, GMCR II, was used to investigate and resolve this conflict. The paper presents all the steps involved in the modeling of the conflict using GMCR II and analysis of the results obtained from the model. The National Water Supply & Drainage Board accepting to reduce the inundation area and agreeing to release sufficient environmental flow was resulted as the best possible solution acceptable for all the three parties.The solution includes affected people and Government Organizations agreeing for the construction of the reservoir at the original location.

Keywords: Conflict resolution, decision support system, graph model

1 Introduction

Vavuniya is one of the districts in the Northern Province of Sri Lanka situated approximately 280 km from Colombo. The rainfall of this area is a bimodal dominated by the North-East monsoon (October – March) and South-West monsoon (May-September). The average rainfall of the area is in the range of 800 mm to 1500 mm.

Major river systems are absent within the Vavuniya district. However, many small to medium scale streams drain through the district and all of them are seasonal streams that are active only during North-East monsoonal rainy periods. KanakarayanAru, ChamaliAru, KiulAru, ChamalankulamAru, Per Aru, KalAru and parts of MalwathuOya are the main river systems draining within the district. These river systems show a remarkable variability in discharge during dry and wet seasons.

The Per Aru shown in Figure 1 originates in the Vavuniya District. TurumpamoddaiAru joins the Per Aru on right side. Afterwards this river

is referred to as ParankiAru. Further downstream of the ParankiAru, PeriyakatteAru enters the ParankiAru from the left side. Finally, this river discharges into the Indian Ocean at Mannar. The river basin has an elongated form with a length of about 70 km and the width ranges from 10 m – 20 m. The catchment of the ParankiAru is located in the northern dry zone and its area is about 832 km2. Annual discharge volume to sea is about 312×106 m3.

The present water supply scheme to Vavuniya urban area draws water from groundwater and the water supply is limited to 3-4 h in a day. Besides, after the end of the civil unrest, the area has become popular for living and the population is expected to rise from 36,000 at present to about 93,000 in the year 2030. The

Eng. G. Abira, BSc.Eng (Peradeniya), MEng (Peradeniya), MIE(Sri Lanka), CEng., Senior Project Engineer, National Water Supply & Drainage Board, Vavuniya.

Eng (Prof.) K.D.W. Nandalal, BSc.Eng (Peradeniya), MEng (AIT, Thailand), PhD (Wagenigen, The Netherlands), FIE (Sri Lanka), CEng., Professor in Civil Engineer, University of Peardeniya, Peradeniya


ENGINEER 62

total water demand will be around 21,000 m3/day and a maximum of about 3000 m3/day can be produced by groundwater sources. The rest has to be sought from surface water sources [1].

National Water Supply and Drainage Board (NWSDB) is interested in withdrawing 12,000 m3/d of water from the Per Aru by constructing a reservoir across it to cater for the drinking water demand in urban and suburb areas of the Vavuniya district. The plan is to capture the runoff during the Mahaseason, store it, treat it and distribute to the urban areas. During the Maha season there is plenty of water with a significant release of water to the sea and there is no competition on water. But during the dry period and especially during the Yala season there is a conflicting situation between water users. Due to the construction of the proposed storage reservoir on Per Aru some of the paddy, high and forest lands will be inundated. This has created a conflict between the farmers who own lands in the proposed reservoir area and the NWSDB. Additionally, Forest Department and Wildlife Department have shown their concern on the project due to a few issues that may arise as a result of constructing a reservoir on the Per Aru.

Figure 1 - Per Aru river system

The area that will be inundated by the proposed reservoir belongs to Sastrikoolankulam, Paranattakal and Puthukulam Grame Niladhari Divisions in the Vavuniya District. There are no large settlements in the above area. However, there are lands under cultivation for which there are 103 claimants either singly or jointly. The

population who would be affected is mostly rural, who generally depend on agriculture.

Due to the construction of the proposed reservoir 326 ha land will be inundated, which consists of 65.87 ha paddy land, 14.93 ha upland cultivation land, 16.55 ha of light jungle and 218 ha of forest land. The uplands are cultivated seasonally with other field crops. The forests comprise of dry mixed evergreen forests, riverine forests and shrub forests. There are 3 small tanks in the project area of which the entire command areas plus one tank will be submerged due to the project while portion of the command area of another will be inundated. A Hindu temple and 3 small houses located in the area would be inundated due to the reservoir.

Construction of the proposed reservoir will reduce water to the downstream area, which will cause impacts on aquatic fauna and flora in the tank bed and downstream. The Wildlife Department shows concern on this aspect of the project.

During the past civil unrest period, some of the land which had been under cultivation had been abandoned and subsequently had turned into a light jungle. However, due to its culmination people are returning to their original land and also many are being resettled in new areas which had been under a forest land use hitherto. In order to facilitate sustainable development while conserving the important ecosystems, an integrated strategic environmental assessment has been done for the Northern Province spearheaded by the Central Environment Authority and coordinated and funded by the UNDP. According to a zoning proposed by this study the planned reservoir and the water treatment plant fall within a proposed jungle corridor between Padaviya tank Sanctuary and Vavunikulam Sanctuary. The forest reserves located in close proximity to the proposed reservoir are Melkulam Forest Reserve and Rasenthirakulam Forest Reserve.

Forest Department is very much concerned on trees to be uprooted in the forest land that will be inundated due to the construction of the proposed reservoir since there are valuable trees in that particular forest reserve.

Thus, NWSDB, Forest Department, Wildlife Department, People in the area have conflicting views on the implementation of the proposed

ENGINEER 62











ENGINEER63











project to supply water to the urban area of the Vavuniya district. The objective of the paper is to investigate the conflict among these parties to propose an effective solution strategy to solve the conflict. The decision support system, called GMCR II [2], [3], [4], [5], [6], developed for implementing the graph model for conflict resolution, is used to rigorously analyze the conflict.

1.1 Multiple Participant – Multiple Objective Decision Making in Water Resources

A conflict over water involves more than one decision maker, participant, or stakeholder. For instance, water conflicts occurring in drainage basins located throughout the world concern many different interest groups, such as national and regional governments, nongovernmental organizations, environmentalists, industries, agriculture, and individual citizens [7]. As water can be used for multiple purposes, often a given interest group is directly associated with a specific use of water such as irrigation, recreation, human consumption, navigation, hydro power generation, etc. Whatever the case, there is almost always disagreement among people and organizations over how water should be utilized in a sustainable fashion that is fair to all of the stakeholders [8].

1.2 The Graph Model for Conflict Resolution

The Graph Model for Conflict Resolution (GMCR) is a comprehensive methodology devised to understand conflict decision-making and conflict resolution [9]. The graph model has been designed to be simple and flexible, as well as to have minimal requirements of information.

The original idea of GMCR was introduced by Kilgour et al. [10] while the first complete presentation was furnished by Fang et al. [11]. The GMCR has been applied to a wide range of application areas: from environmental management to labour management; from military and peace-keeping activities to economic issues; from local to international levels [9], [12].

Figure 2 illustrates the general procedure for applying the methodology of GMCR to a real-world conflict. Two main stages, modeling and analysis, are involved in this procedure. In the modeling stage, essential model elements, such

as the decision makers (DMs), their options, and the relative preferences are identified based on the understanding of the actual dispute. The objective is to find some stable states that represent a resolution of the conflict. The essential parts of a graph model in option form are the DMs and the options available to each DM. In general, a DM may exercise any combination of the options he or she controls to create a strategy. When every DM has selected a strategy, a state is defined. States are derived from the given options. Then this information is fed into the next stage; analysis.

In the analysis stage, the stability of every state is first calculated from each DM’s viewpoint. Subsequently, the overall equilibria, which contain the states that are stable for all DMs, can be obtained. By interpretation and sensitivity analyses, DMs or other interested parties can understand the meaning of resolutions in terms of the real-world disputes. Note that feedback is allowed in the procedure. Feedback means that, at every step of the modeling or analysis stage, one may return to any previous point whenever new information is found. This characteristic makes GMCR more flexible and practical.

Figure 2 - General procedures for applying GMCR

The structure of GMCR II, which is the present version of the GMCR, is illustrated in Figure 3. Through the user interface, “Modeling

ENGINEER 64

Subsystem” build up the graph model with all related information, such as the DMs, their corresponding options, infeasibility information, and last but not least, preferences. Then, the modeling information enters the “Analysis Engine” for further investigation. The important analysis procedures, such as stability analysis, coalition analysis, and status quo analysis, are carried out in this subsystem. Finally, the analysis results or resolutions, for instance, individual stabilities, overall equilibria, and coalition stabilities, are constructed in the “Output Interpretation Subsystem” and delivered to the users through the Graphical User Interface. For a special user who has particular requirements, these requests may be directed additionally from the user interface to the “Analysis Engine”.

Within each subsystem, GMCR II has its specific design to complete the functions. For example, preference information is one of the most critical and intractable issues. In GMCR II, this job could be fulfilled by three technologies, option weighting, option prioritizing, and direct ranking. Each technology is employed to deal with a certain situation, and the direct ranking method can be combined with one of other two to fine tune the preferences. Details about GMCR II may be found in Fang et al. (2003a, 2003b).

Figure 3 - GMCR II structure

2 Deveopment of the model

2.1 Modeling: Putting the Problem into Perspective

The paper focus on issues existing in the Per Aru basin due to the proposal for constructing a reservoir to extract water for drinking purpose. The proposed project has created a conflict between three main parties. They are;

People who would be affected National Water Supply & Drainage Board

(NWSDB) Government Organizations (Forest

Department, Wild Life Department and Central Environmental Authority)

GMCR II is utilized to model the conflict existing in the Per Aru basin by following the steps shown in Figures 2 and 3.

2.2 Decision Makers and Options

As the first step in gathering information to understand the conflict and to make it amenable to systematic modeling and analysis, decision makers and their options were identified. The decision makers should have the ability to make decisions that can directly bear upon the eventual resolution of the dispute. Options are specific actions that can actually occur in a conflict. Each decision maker can select which options to implement during the evolution of the conflict.

Table 1 lists the decision makers and their options in the Per Aru basin dispute. As depicted in Figure 3, a conflict model can be updated at any time during the modeling and analysis process to reflect the obtaining of more information about the conflict or a better understanding of the true situation.

Table 1- Decision Makers and their options in the conflict existing in Per Aru basin

AP : Affected People 1 Change Location: Change the Project

Location 2 Reduce paddy inundation: Reduce the

inundation of paddy land under the reservoir

GOV : Government Organizations 3 Sufficient e-flow: Release of sufficient

environmental flow 4 Reduce total inundation: Reduce the

inundation of forest and paddy land under the reservoir

5 Acceptable compensation: Acceptable compensation to affected people

NWSDB : National Water Supply & Drainage Board

6 Reduce e-flow: Reduce environmental flow

7 Increase capacity: Increase the reservoir capacity

ENGINEER 64

























ENGINEER65

























2.3 Feasible States

The model in Table 1 contains 7 options, each of which can be selected or not, mathematically there are 27=128 possible states. However, not all combinations of options are feasible in practice and of those that are feasible, not all are distinguishable. Subsequent to entering the decision makers and options, infeasible states were identified, which GMCR II will remove from the model, and the groups of states that are effectively identical, which GMCR II will combine into a single state. These steps enable GMCR II to list all states that can actually occur in the model. In the model of the Per Aru basin conflict, there are only 16 feasible states (Figure 6), which is much less than the total number of mathematically possible states.

The model has dialogue boxes for selecting “Mutually Exclusive Options”, “At Least One Option”, “Option Dependence” and “Direct Specification”. In this model two categories of infeasible states are removed: (i) “Mutually Exclusive” options, which remove the states that contain mutually exclusive options, and (ii) “At Least One” option, which is used to specify that, for the set of options, at least one option must be selected.

In the Per Aru basin model, the two options under the control of the “Affected People” are mutually exclusive and, hence, the “Affected People” can only select at most one of their two options, which are numbered as Options 1 and 2 in Figure 4. Thus, if the affected people were to stick with “Change the project location” (Option 1), their other option of “Reduce the inundation of paddy land under the reservoir” (Option 2) could not occur. Likewise, Options 4 and 5 are mutually exclusive for “Government Organizations”. If the inundation areas of the paddy and forest lands are reduced (option 4) their other option of “Acceptable compensation to affected people” (option 5) could not occur. However, Option 3 does not have any mutually exclusive options as environmental flow is mandatory for any reservoir project. For the NWSDB, the options 6 and 7 are mutually exclusive. If the NWSDB could increase the reservoir capacity there is no need to reduce the environmental flow and vice versa. The X’s that are entered in a given column in Figure 4 designate options those are mutually exclusive.

Figure 5 shows the dialogue box to select the “At Least One Option”. The “Affected People” should select at least one of their two options,

which are numbered as Options 1 and 2 in Figure 5. Thus, the affected people were to stick with “Change the project location” (Option 1) or their other option of “Reduce the inundation of paddy land under the reservoir” (Option 2).

Figure 4- Remove infeasible state using “Mutually Exclusive Options”

Likewise, the government organizations selected “Reduce the inundation of paddy and forest land (Options 4) or “Acceptable compensation to affected people” (Option 5). The NWSDB should select at least one of their options, which are numbered as options 6 and 7 in Figure 5. Thus, the NWSDB selected “Reduce the environmental flow” (Option 6) or “Increase the reservoir capacity” (Option 7).

Figure 5- Remove infeasible state using “At least one option”

After GMCR II removes the infeasible states, the remaining feasible states are listed as columns of Y’s and N’s, where “Y” indicates “yes,” the option opposite the Y is selected by the decision maker controlling it, and “N” means “no,” the option is not taken. The 16 feasible states for the Per Aru basin dispute are listed in Figure 6. For convenience, each state is assigned a number. For example, Figure 6 shows that at State 16, the “Affected People” select its second option, but not its first option, to form its strategy NY. The “Government

ENGINEER 66

Organizations” have taken Option 3 and 5, but not 4, and therefore, have adopted its strategy YNY. Finally, the NWSDB select its second option, but not its first option, so has followed the strategy NY. When strategy selections of all decision makers are combined, State 16, or (NY YNY NY) written horizontally in text, is the result.

2.4 Allowable State Transition

At any state of a conflict model, a particular decision maker may be able to unilaterally cause a transition to another state by changing his or her option selection. The GMCR II automatically calculates all possible state transitions, if any, from each state for each decision maker. However, some transitions may be infeasible.

Figure 6 - Feasible States

This occurs, for instance, when an option is irreversible - after the option is selected, it cannot be undone. In this Per Aru basin conflict model affected people decided to change their option selection “Change the project location” from Y to N. In the same manner NWSDB decided to change their option selection “Reduce the environmental flow” from Y to N. Therefore, as illustrated in Figure 7 all the options are reversible other than options 1 and 6.

Figure 7 - Allowable state transition in the Per Aru basin conflict

2.5 Relative Preferences

Before carrying out a stability analysis, the GMCR II requires that the feasible states be ranked from most to least preferred for each DM, where ties are allowed. The GMCR II possesses two flexible approaches, called “Option Weighting” and “Option Prioritization,” for conveniently specifying preference information in terms of options for each DM. An internal algorithm then automatically orders the states for the DM based upon this preference information.

Option Weighting allows users to assign a number or numerical weight to each of the options from the viewpoint of each DM, where a positive or negative number means the DM likes or does not like the option, and the magnitude of the number reflects the degree of preference. Option prioritization provides an intuitive specification based on preference statements listed from most important at the top to least important at the bottom. In addition to these two means to specify the ranking of feasible states for each DM, the GMCR II also allows users to fine-tune the preference ranking by directly reordering states, joining two or more states into an equally preferred group, and splitting an equally preferred group apart. Option Prioritization along with “Direct Ranking” is employed to come up with the preference ranking for the DMs in the Per Aru basin conflict. Table 2 lists the preference statements using option numbers in order of priority for each DM.

Table 2 - Lexicographic Preference Statements for the DMs expressed in terms of options

AP GOV NWSDB 2 3&-6 -1 1 IF 7 4 & 2 & -7 -2 & -4 5 5 5 3 & -6 -1 6 & -3 4 7 IF 3

Consider the preference statements of the Affected People (AP) given in the left-hand column in Table 2. The Option 2 located at the top of the first column means that AP most prefers to reduce the inundation area of the reservoir. As indicated by “1 if 7” written below the Option 2, the AP’s next preference is to change the reservoir location if the reservoir capacity is to be increased. In the order of

ENGINEER 66

Organizations” have taken Option 3 and 5, but not 4, and therefore, have adopted its strategy YNY. Finally, the NWSDB select its second option, but not its first option, so has followed the strategy NY. When strategy selections of all decision makers are combined, State 16, or (NY YNY NY) written horizontally in text, is the result.

2.4 Allowable State Transition

At any state of a conflict model, a particular decision maker may be able to unilaterally cause a transition to another state by changing his or her option selection. The GMCR II automatically calculates all possible state transitions, if any, from each state for each decision maker. However, some transitions may be infeasible.

Figure 6 - Feasible States

This occurs, for instance, when an option is irreversible - after the option is selected, it cannot be undone. In this Per Aru basin conflict model affected people decided to change their option selection “Change the project location” from Y to N. In the same manner NWSDB decided to change their option selection “Reduce the environmental flow” from Y to N. Therefore, as illustrated in Figure 7 all the options are reversible other than options 1 and 6.

Figure 7 - Allowable state transition in the Per Aru basin conflict

2.5 Relative Preferences

Before carrying out a stability analysis, the GMCR II requires that the feasible states be ranked from most to least preferred for each DM, where ties are allowed. The GMCR II possesses two flexible approaches, called “Option Weighting” and “Option Prioritization,” for conveniently specifying preference information in terms of options for each DM. An internal algorithm then automatically orders the states for the DM based upon this preference information.

Option Weighting allows users to assign a number or numerical weight to each of the options from the viewpoint of each DM, where a positive or negative number means the DM likes or does not like the option, and the magnitude of the number reflects the degree of preference. Option prioritization provides an intuitive specification based on preference statements listed from most important at the top to least important at the bottom. In addition to these two means to specify the ranking of feasible states for each DM, the GMCR II also allows users to fine-tune the preference ranking by directly reordering states, joining two or more states into an equally preferred group, and splitting an equally preferred group apart. Option Prioritization along with “Direct Ranking” is employed to come up with the preference ranking for the DMs in the Per Aru basin conflict. Table 2 lists the preference statements using option numbers in order of priority for each DM.

Table 2 - Lexicographic Preference Statements for the DMs expressed in terms of options

AP GOV NWSDB 2 3&-6 -1 1 IF 7 4 & 2 & -7 -2 & -4 5 5 5 3 & -6 -1 6 & -3 4 7 IF 3

Consider the preference statements of the Affected People (AP) given in the left-hand column in Table 2. The Option 2 located at the top of the first column means that AP most prefers to reduce the inundation area of the reservoir. As indicated by “1 if 7” written below the Option 2, the AP’s next preference is to change the reservoir location if the reservoir capacity is to be increased. In the order of

ENGINEER67

decreasing preference, AP would like to accept the compensation package (Option 5). Next, AP would like sufficient e-flow (Option 3) and AP do not want NWSDB to reduce the e-flow (Option 6). Finally, AP like to reduce the inundation of forest land.

The middle column of Table 2 lists Government Organization (GOV)’s preference statements. As can be seen, it most prefers to release sufficient e-flow and it prefers NWSDB not to reduce the e-flow (Options 3 & 6). The second level preference is for it to reduce the inundation area of the paddy land, reduce the inundation of forest land and NWSDB not to increase the reservoir capacity (Option 2 & 4 & -7). This is followed by acceptable compensation package to affected people (Option 5). At last it prefers NWSDB not to change the location of the reservoir (Option 1).

As indicated in the third column from the left in Table 2, NWSDB most prefers not to change the project location (Option 1). Next, NWSDB prefers not to reduce the inundation of paddy land and forest land (Option -2 & -4). Then it prefers to give the acceptable compensation package to affected people (Option 5). Subsequently, NWSDB prefers to reduce the e-flow (Option 6) and not to release sufficient e-flow (Option 3). Finally, NWSDB prefers to increase the reservoir capacity if the sufficient environmental flow is mandatory (Option 7 IF 3).

A remarkable advantage of the option prioritization approach to preference elicitation is that it closely reflects the way in which a person thinks about preferences in a given dispute. What clients find to be stunning about the graph model methodology is that relative preference information can furnish detailed and accurate analytical results, such as the potential resolutions to the conflict under study.

When employing the GMCR II, a user only needs to enter known preference statements expressed in terms of options for a given decision maker. In fact, these preference statements adhere to all of the rules of first order logic. Assuming transitivity of preferences, the GMCR II possesses an algorithm for taking the prioritized preference statements for each decision maker to produce a ranking of states for each decision maker in which ties are permitted. Often, a given decision maker’s most and least preferred preference statements are initially known.

After the GMCR II ranks the states based on a limited number of preference statements, there may be blocks of equally preferred states contained within the ordering of states. As more preference information becomes available, these equally preferred blocks will become less prevalent and will fully disappear when there are sufficient preference statements to produce complete ranking of states from most to least preferred with no ties. However, even when there are blocks of equally preferred states for one or more decision makers, the GMCR II can still carry out stability analyses. In practice, this means that one can commence with a “quick and dirty” analysis and subsequently refine preference statements as more information becomes available. Finally, it should be stressed that a user only has to supply preference statements in terms of what is usually not a large number of options. The user does not have to order the states, which in some cases can be relatively large,the GMCR II expeditiously determines the ordering using the lexicographic preference statements supplied by the user. In fact, an appealing feature of the graph model methodology is that the user only needs to supply relatively small amounts of information to calibrate a conflict model upon which highly accurate analytical results can be ascertained.

Table 3 displays the preference ranking of states from most preferred at the top to least preferred at the bottom for each of the three decision makers.

Table 3- Ranking of States for Decision Makers

AP GOV NWSDB 8 16 6 6 15 14 2 12 16 4 11 8

16 4 2 14 2 10 12 14 12 10 8 4 15 6 5 7 7 15

13 5 13 5 13 7

11 10 1 3 9 11 9 3 9 1 1 3

ENGINEER 68

Notice that AP’s most preferred State is 8, as indicated by the placement of 8 at the top of the left-hand column in Table 3, and least preferred State 1 is, shown at the bottom. Further, the “power” held by a participant in a dispute is reflected by the options available to the decision maker and the decision maker’s preferences over states.

3 Analysis and results

3.1 Deciding What to Do

After a model has been established, the analysis phase of the GMCR II is invoked, as is indicated in the lower portion of Figure 2. A GMCR II analysis includes a determination of the stability of every state, for every decision maker, under all solution concepts listed in Table 4. These stability definitions describe the patternsof interaction that a decision maker may expect. If it is not advantageous for a given decision maker to depart unilaterally from a particular state according to a given solution concept, then the state is deemed to be stable for that decision maker under that solution concept. If a state is stable according to a given solution concept, for all of the decision makers, it constitutes an equilibrium under that solution concept. It is, therefore, a compromise resolution, since no decision maker has an incentive to unilaterally move away from it. Fang et al. (1993) present mathematical definitions and comparisons as well as original

references for the solution concepts given in Table 4.

Subsequent to executing a stability analysis, the GMCR II can display stability results separately for each decision maker as well as the equilibria. Table 5 lists the individually stable states for the three decision makers and the equilibrium states for the Per Aru basin conflict.

Table 4- Individual Stability States and Equilibrium States

Individual Stability Equilibria AP GOV NWSDB

2 2 1 2 4 4 2 16 6 5 5 8 7 6

10 15 7 12 16 8 14 9 16 10

11 12 13 14 15 16

Figure 8 is the Equilibrium list for the Per Aru basin conflict, which illustrates the two possible resolutions’ strong stabilities.

Table 5- Solution Concepts and Human Behaviour

Solution concept Stability description

Nash stability DM cannot unilaterally move to a more preferred state

General metarationality (GMR)

All DM’s unilateral improvements are sanctioned by subsequent unilateral move by others

Symmetric metarationality (SMR)

All DM’s unilateral improvements are still sanctioned even after possible responses by the original DM

Sequential stability (SEQ) All DM’s unilateral improvements are sanctioned by subsequent unilateral improvements by others

Limited-move stability Lh All DM’s are assumed to act optimally and maximum number of state transitions (h) is specified

Nonmyopic (NM) Limiting case of limited move stability as the maximum number of state transitions increase to infinity

ENGINEER 68









2 2 1 2 4 4 2 16 6 5 5 8 7 6

10 15 7 12 16 8 14 9 16 10

11 12 13 14 15 16












ENGINEER69









2 2 1 2 4 4 2 16 6 5 5 8 7 6

10 15 7 12 16 8 14 9 16 10

11 12 13 14 15 16












R, GMR, SMR, SEQ, NM, and L (2) in the lower rows correspondinglyrepresent Nash Stability, General Metarationality, Symmetric Metarationality, Sequential Stability with a horizon of two, Non-Myopic Stability and Limited Move Stability. The tick in the column of each state indicates that the state is in equilibrium under the corresponding solution concept in that row.Undoubtedly, States 2 and 16 are equilibria under all listed solution concepts.

Figure 8 - Equilibrium list for the Per Aru basin conflict

3.2 State transitions from Status Quo to Final Outcome

Table 6 shows the sequence of state transitions from the status quo to the final equilibrium state 16, where arrows indicate the location and direction of option changes during the evolution of the conflict. At the status quo, “Affected People” enforced to change the reservoir location or forced to reduce the inundation of paddy land. However, attempts by the NWSDB to reduce the inundation of paddy land by extend the bund, would encourage the Affected People to accept the project location without changing as shown in the joint transition from Status quo to State 6. Hence, by exhibiting a spirit of cooperation, Affected People, Government Organizations and NWSDB can jointly move the conflict to State 14, which is more preferred than State6 by all decision makers. Finally, State 16, in which NWSDB decided to release sufficient environmental flow, could be an acceptable solution for all the parties.

Table 6 - State transitions from Status Quo to Final Outcome

4 Conclusions

As demonstrated by the Per Aru basin conflict study, the Graph Model for Conflict Resolution methodology, in conjunction with the decision support system GMCR II, furnishes a valuable decision technology for systematically and rigorously investigating real-world conflict. Moreover, the analytical results provide strategic insights into how the conflict evolved from a status quo situation to its final resolution.

In the Per Aru basin conflict the affected people most preferred to change the project location. As per the preference order Government Organizations, most preferred to release sufficient environmental flow. However, as the NWSDB accepted to reduce the inundation of paddy land and agreed to release sufficient environmental flow, the affected people and Government Organizations agreed to construct the reservoir at the original location.

TransitionStatus Quo

6 14 16

Change location Y N N N

Reduce paddy inundation

N Y Y Y

Sufficient e-flow N N N Y

Reduce total inundation (Forest)

N N N N

Acceptable compensation

Y Y Y Y

Reduce e-flow Y Y N N

Increase reservoir capacity

N N Y Y

State Number

Affected People

Government Organizations

NWSDB

N Y Y Y

Y Y N N

N N Y Y

N N N Y

ENGINEER 70

5 References

1. Environmental Impact Assessment (EIA), 2012. Surface water extraction from a reservoir across Per Aru.

2. Fang, L., Hipel, K.W., Kilgour, M.D., and Peng, X.,“A Decision Support System for Interactive Decision Making, Part 1: Model Formulation”, IEEE Transactions on Systems, Man, and Cybernetics, Part C, Vol. SMC-33, No. 1, 2003.

3. Fang, L., Hipel, K.W., Kilgour, D.M., and Peng, X., “A Decision Support System for Interactive Decision Making, Part 2: Analysis and Output Interpretation”, IEEE Transactions on Systems, Man, and Cybernetics, Part C, Vol. SMC-33, No. 1, 2003.

4. Hipel, K.W., Kilgour, D.M., Fang, L., and Peng, X., “The decision support system GMCR II in environmental conflict management”,Applied Mathematics and Computation 83, No.2 and 3: 1997, pp.117-152.

5. Hipel, K.W., Kilgour, D.M., Fang, L., and Peng, X., “Strategic Decision Support for the Services Industry.” IEEE Transactions on Engineering Management 48, No.3: 2001, pp.358-369.

6. Kilgour, D.M., Hipel, K.W., Fang, L., and Peng,X.,. “Coalition Analysis in Group Decision Support.” Group Decision and Negotiation 10, No.2: 2001, pp.159-175.

7. Wolf, A., ed.,Conflict prevention and resolution in water systems, Elgar, Cheltenham, U.K.2002.

8. Gleick, P. H., “Water and conflict: Fresh water resources and international security”, Int. Secur., 18(1),1993.

9. Kilgour, D.M. and Hipel, K.W. The graph model for conflict resolution: past, present, and future," Group Decision and Negotiation, 14(6),2005.

10. Kilgour, D.M., Hipel, K.W., and Fang, L. “The graph model for conflicts," Auto-matica, 23(1),1987.

11. Fang, L., Hipel, K.W., and Kilgour, D.M. Interactive Decision Making: The Graph Model for Conflict Resolution. Wiley, New York,1993.

12. Nandalal, K.D.W. and Hipel,K.W. “Strategic Decision Support for Resolving Conflict over Water Sharing along the Syr Darya River in the Aral Sea Basin”, Journal of Water Resources Planning and Management, ASCE, Vol.133, Issue 4, 2007, pp.289-299.

ENGINEER 70

5 References













ENGINEER71

5 References














1 ENGINEER

Performance and Retrofitting of Unreinforced Masonry Buildings against Natural Disasters – A

Review Study

W. S. W. Mendis, Sudhira De Silva and G. H. M. J. Subashi De Silva Abstract: Un-Reinforced Masonry (URM) buildings are popular all over the world including Sri Lanka because of their durability, low cost, construction easiness and architectural character, need of less skilled labour, eco-friendliness and use of locally available materials such as ashlar or rubble, adobe and brick. However, these buildings have a higher probability of failing under natural disasters such as earthquakes, tsunamis and storm surges, floods, cyclones and landslides. In Sri Lanka, winds, landslides and floods have frequently occurred. In addition, a massive tsunami adversely affected the people in 2004 and its effects to islands on the Indian Ocean have been continued since December, 2004. Minor earthquakes have come off recently with experiences of wall cracks and no deaths. It is also believed that, there is a defused plate boundary in the making some 500 km south of the southern tip of Sri Lanka which might be the cause of these tremors or minor quakes. Further, an earthquake occurred in Colombo area in 1615, has caused around 2000 of human deaths. Therefore investigation of performance of URM buildings against these natural disasters and possible retrofitting methods are increasingly important. In this review study, an attempt is made to identify the performance of URM buildings against natural disasters and identify retrofitting methods that can be applied to existing building, to enhance the strength properties of structural components. Common failure mechanisms for URM structures consist of separation of walls at corners, diagonal cracking or vertical cracking in walls, separation of roofing from walls, out-of-plane wall failure, in-plane wall failure, shear cracks and de-lamination. These damages on a wall diminish the service life of building. In addition, different kinds of retrofitting methods: ferrocement, poly propylene mesh and bamboo reinforcement, for URM structures to be seismic resistant are presented. Mechanisms of failure of URM walls and effects of retrofitting techniques to reduce the damage are also discussed. Keywords: Un-Reinforced Masonry (URM) buildings, retrofitting, natural disasters, failure mechanisms, control mechanisms 1. Introduction Throughout the centuries, natural disasters have taken a high toll of human lives and caused great property losses all over the world and unfortunately mostly in developing countries. The worst death toll from an earthquake, in the past century, occurred in 1976 in China, where it is estimated that 240,000 people were killed and most of the deaths were due to the collapse of brick masonry buildings (D‟Ayala [11]). Further, Sri Lanka had also experienced a tsunami in 26th December, 2004 which caused large amount of deaths and damages. Most of the damaged structures in Sri Lanka were domestic buildings, which had been constructed using masonry (i.e., bricks or cement sand blocks). A natural hazard is said to be a natural disaster when the hazard affects humans and the built

environment. Natural hazards (i.e. earthquakes, landslides, volcanic eruptions, floods and cyclones), will never result in a natural disaster in areas without vulnerability. Therefore, the financial, environmental or human loss resulted from a disaster depends on capacity of the population to resist the disaster. Besides natural disasters, there are also man-made disasters which are the results of technological or human hazards such as fires, transport accidents,

Ms. W.S.W. Mendis, B.Sc. (Hons) (Ruhuna), Rresearch Student, Department of Civil and Environmental Engineering, Faculty of Engineering, University of Ruhuna, Sri Lanka. Eng. (Dr.) Sudhira De Silva, PhD (Saitama), M.Eng (Saitama) B.Sc. Eng. (Hons) (Moratuwa), C.Eng. MIE (Sri Lanka), Senior Lecturer, Department of Civil and Environmental Engineering, Faculty of Engineering, University of Ruhuna, Sri Lanka. Eng. (Dr.) (Mrs). G. H. M. J. Subashi De Silva, PhD (Saitama), B.Sc. Eng. (Hons) (Moratuwa), C.Eng. MIE (Sri Lanka), Senior Lecturer, Department of Civil and Environmental Engineering, Faculty of Engineering, University of Ruhuna, Sri Lanka.



industrial accidents, oil spills and nuclear explosions. Further, earth cutting, which is mechanically excavated or blasted out with carefully placed explosives, often causes landslides, if appropriate protective measures are not provided. Piling work, which has been provided as a support for a structure, also induces ground vibration. These vibrations disturb residents at near area in a manner similar to an earthquake. The economic cost of natural disasters over the last ten years in Sri Lanka exceeded LKR 257 billion, (i.e., USD 1.95 billion) (Rajasingham [29]). In Sri Lanka, natural disasters such as minor earthquakes, tsunami and storm surges, floods, cyclones and landslides are encountered. Apart from the environmental implications, deforestation in Sri Lanka has caused ill effects such as flooding, landslides and soil erosion from exposure of the deforested areas (Keerthisinghe [21]). Weather changes in Sri Lanka showed that, not only landslides and floods (originated with precipitation), but also extreme wind events have frequently occurred. Minor earthquakes have come off recently with experiences of only wall cracks and no human death. Besides, it is believed that, there is a defused plate boundary in the making some 500 km south of the southern tip of Sri Lanka (Dissanayake [14]). Sri Lankans were adversely affected by the tsunami created by the earthquake that occurred in the coastal zone near Sumatra Island in December, 2004 with a huge catastrophe to human lives. Reasons for such a loss of human lives are that, people have no awareness on behaving in a disaster and the collapse of man-made buildings or structures resulted in most of the deaths. In general, buildings can be divided into two main categories: engineered buildings and non-engineered buildings. Their percentages are quite different in developed, developing, and under-developed countries. Past destructive disasters have showed that non-engineered buildings were the one which have been damaged mostly. Skilled technicians, engineers and architects are generally not involved in this type of construction (Blondet et al. [7]). In Sri Lanka, most of dwellings are non-engineered buildings constructed as URM buildings. URM buildings are popular because of having inherent advantages such as durability, low cost, construction easiness and architectural character, need of less skilled labour, use of locally available materials, eco-friendliness, heat and sound insulation and fire resistance.

The damages of structures due to natural disasters are encountered day to day and these damages cause structural failures, casualties and deaths. Therefore, investigation of performance of URM buildings and introducing required retrofitting methods to improve their resistance against natural disasters are increasingly important. Objectives of the current study are to review published literature and, Identify performance of URM buildings

against natural disasters Identify retrofitting methods that can be

applied to existing building, to enhance the strength properties of structural components, decrease the amount of damage and enhance the time duration for collapse, which helps people to evacuate

However, a suitable retrofitting technique for Sri Lanka should be efficient not only in improvement of seismic resistant characteristics such as strength, ductility, but also in economy and availability of construction materials and required labour skills. Damage patterns and causes for strength losses in URM buildings due to natural disasters (i.e., earthquakes, tsunami and floods) are identified and presented in this paper. In addition, retrofitting methods that can be applied to unreinforced masonry buildings in Sri Lanka, so as to resist dynamic loads induced by earthquakes, tsunamis and floods are also presented. 2. Damages in URM Buildings From the observation of structural performance of buildings during an earthquake, the strong and weak aspects of the design as well as the desirable qualities of materials, construction techniques and site selection can be clearly identified (Boen [8]). This can be applied to other natural disasters as well and the study of damages provides an important step in the evolution of strengthening measures for URM buildings. The most of damages in dwelling houses occurred because of the poor designing and construction. It has been reported that the most important weaknesses of the masonry structures were the lack of interlocking units between external and internal wythes of the wall sections and the lack of connection between crossing walls (Velazquez-Dimas et al.













3 ENGINEER

[36]). Both of them give rise to possibility of out-of-plane behaviour, as their formation increases net length of the walls. Also, roof placed directly on the walls without bond beams does not provide a diaphragm and due to free end at the top of walls, probability of out-of-plane failure mechanisms increases. Placement of openings near the corners of the walls is another common problem where crack propagation is concentrated around these openings. With the mass of evidence from past earthquakes, tsunamis and floods, the typical damages to URM buildings are discussed, below. 2.1 Damages Due to Earthquake During earthquakes, the ground shakes in all directions and generates inertia forces that the structure should be able to withstand. Under seismic loading, URM walls have two main possible failure mechanisms: in-plane and out-of-plane (Saatcioglu et al. [32]). In-plane failures are characterized by a diagonal tensile crack pattern while out-of-plane failures are characterized by cracks that are primarily along the mortar bed joints. The principal in-plane failure mechanisms of URM walls subjected to earthquake actions are shear failure, sliding failure, rocking failure and toe crushing (Figure 1).

Figure 1- In-plane failure modes of a laterally

loaded URM wall (ElGawady et al. [19]): (a) Shear failure, (b) Sliding, (c) Rocking,

(d) Toe crushing The typical out-of-plane failure patterns of URM wall resulted from an earthquake are shown in Figure 2.

Figure 2 - Typical crack patterns of URM buildings due to out-of-plane failure

(Bartolome et al. [4]) The performance of small adobe and low-quality mud-brick constructions varied from no damage to collapse and, within any specific area, the performance of these buildings depended on a number of parameters, including wall thickness, roof mass, size of rooms, and quality of materials (Webster [37]). Earthen structures have less ductility and are very brittle resulting in sudden failures under seismic loading without any warning. The traditional earthen buildings are vulnerable due to a perverse combination of the mechanical properties of their walls where earthen walls are dense and heavy, have extremely low tensile strength resulted from weak material, lack of reinforcement, poor workmanship and null maintenance (Bartolome et al. [5] and Blondet et al. [7]). Common failure modes of adobe structures were reported by Blondet et al. [6] (Figure 3). The same failure modes can be expected for other types of masonry buildings (Bartolome et al. [4] and [5], Blondet et al. [7], D‟Ayala [11] and Kaplan et al. [20]) (Figure 4).

Figure 3 - Seismic deficiencies of adobe masonry (Blondet et al. [6])

Since the tensile strength is very low, significant cracking starts (with the initiation of an earthquake) in the regions subjected to tension. Vertical cracking starts at the lateral corners of

Roof collapse

Beams prone to collapse due to

the loss of support

Vertical cracks in

wall

Out-of-plane collapse of a

long wall

Vertical cracks at wall corners

Diagonal cracks

Diagonal cracks above lintels

Failure of wall corners

Collapse of mud and stone walls

Parapet collapse

Direction of motion induced by earthquake

Bracing wall


the walls, where the tensile stresses are higher due to out-of-plane bending produced by seismic forces perpendicular to the walls (Figures 2 and 3). The continuity of ground movement produces large vertical corner cracks, which tend to separate the walls from one another (Figures 2 and 3). Shear forces generated by lateral seismic forces acting within the plane of the walls, produce diagonal cracks, which usually follow stepped patterns along the mortar joints (Figures 1and 3). Due to the stress concentration at the corners of openings (i.e., doors and windows), diagonal cracks often start at these locations (Figure 3). Front walls are, usually the first to collapse in an earthquake, overturning onto the adjacent street (Blondet et al. [7]) as shown in Figure 4(a) and (b).

Figure 4 - Seismic cracks in URM houses: (a) Collapse due to out-of-plane failure

(severe earthquakes in Peru) (Bartolome et al. [4]),

(b) Typical cracks on adobe houses due to out-of-plane seismic forces (Bartolome et al. [4]),

(c) Shear cracks initiated at the corners of openings in wall of house in La Tinguiña (in-plane failure) (Bartolome et al. [5]), (d) Wall-diagonal crack and vertical corner crack (both in-plane and out-of-plane failure) (Kaplan et al. [20])

According to Blondet et al. [7], earthen houses built without any structural reinforcement, with several stories, thin walls, large window and door openings, and irregular plan and elevation configurations are extremely vulnerable and suffer significant damage or collapse during earthquakes. 2.2 Damages Due to Tsunami Damages of URM buildings by tsunami effects could be due to hydrostatic, hydrodynamic, impulsive, impact and buoyancy forces (Figure

5). Both static and dynamic loads are dependent on wave height as wave velocity is related to its height (Dias et al. [13]). Therefore depending on the wave height, damage level can be differed in various types of structures. It has been found that most single storied masonry structures completely pushed off their foundations in sliding mode, due to no tying down of structure and less self weight caused by thin wall (130 mm) of half brick (Dias et al. [13]). Renuka and Lewangamage [30] found that main failure types in URM structures were bending, diagonal tension and compression, overturning and sliding (Figure 6), by conducting a Finite Element Modeling (FEM) of single storied and two storied houses while Dias et al. [13] have also stated that overturning and sliding are two of main threats from tsunami wave on a structure. Figure 5 - Components of tsunami induced forces, where h is inundation depth, ρ is density of water and g is gravitational acceleration (Renuka and Lewangamage [30]) The URMs performed very poorly in resisting the lateral forces of the tsunami. Bending capacity of unreinforced brick masonry was very low against the hydrostatic forces of the tsunami (Maheshwari et al. [26] and Renuka and Lewangamage [30]). Overturning moment increases with higher pressure while higher building weight and gravel type soil around base will reduce the overturning effect. Figure 6 - Damage to URM buildings by December 26, 2004 Sumatra Earthquake: (a) Brick masonry walls in Talenguda-sliding

failure (Maheshwari et al. [26]), (b) Brick masonry in Trincomalee- overturning due to foundation scouring (Khazai et al. [23]), (c) Buildings in Meelamanakudy- bending failure (Maheshwari et al. [26])

Hydrostatic pressure

hρg

Hydrodynamic pressure

0.625hρg

Impulse pressure

0.5hρg

hρg h

h











hρg


0.625hρg

Impulse pressure

0.5hρg

hρg h

h











hρg


0.625hρg

Impulse pressure

0.5hρg

hρg h

h

5 ENGINEER

2.3 Damages Due to Flooding Damages of URM structures by a flood resulted from storm surge, riverine flooding, or urban flooding mainly occur due to physical forces such as hydrostatic loads, hydrodynamic loads, impact loads and buoyancy (Figure 7) (Caraballo-Nadal et al. [9] and Rogers [31]).

Figure 7 - Typical forces generated by flooding (Caraballo-Nadal et al. [9]) Lateral hydrostatic forces are generally not sufficient to cause deflection or displacement of a building unless there is a significant difference in water elevation on opposite sides of the wall in contact with the flood water. However, if there is significant difference, permanent deflections and damage to structural elements within the building may occur. Hydrodynamic forces are a function of flood water velocity and the building geometry and are capable of collapsing structural walls or floor systems. Figure 8 shows failure of a URM wall due to influence of flood water.

Figure 8 - Brickwork wall broken by flood water, Malton, U.K., November 2000 (Kelman and Spence [22]) When the buoyant forces, which are also called as vertical hydrostatic force, associated with the flood, exceed the weight of the building components and the connections to the foundation system, the structure may float from its foundation. The force acted vertically through the centre of mass of the displaced volume. Impact loads are the direct forces associated with waves, as typically encountered during coastal flooding, or the impact of floating debris (Figure 9) within the flood

waters. These loads are especially destructive because the forces associated with them may be higher in magnitude than the hydrostatic and hydrodynamic forces.

Figure 9 - Brickwork wall damaged due to impact force, Cambridge, U.K., December 2001 (Kelman and Spence [22]) Kelman and Spence [22] found that the vertical and diagonal yield lines comprise a sequence of horizontal and vertical segments (Figure 10) due to failure along the mortar joints rather than through the masonry units from conducting a yield line analysis for URM failing due to flood water pressures resulted from the combination of hydrostatic and hydrodynamic forces using small (area = 38 m2), medium (area = 55 m2) and large (area = 84 m2) cavity wall dwellings in U.K. This result is also confirmed from practical situations shown in Figures 8 and 9.

Figure 10 –Yield line in brick work (Kelman and Spence [22]): (a) Vertical yield line and (b) Diagonal yield line Khazai et al.[23] found that, URM walls would collapse, if height difference of 1 – 1.5 m in opposite directions of the wall with no flood velocity. With flood velocity, the height difference in opposite directions of the wall for collapse can be below 0.5 m for some masonry walls. 3. Retrofitting Techniques Retrofitting techniques are usually introduced to prevent the sudden collapse of buildings during natural disasters allowing people to evacuate (Bartolome et al. [4]). Retrofitting will upgrade the disaster resistance of an existing unsafe building, or a damaged building while repairing (Arya [2]). Though it may not be


designed to be totally disaster-resistant but to avoid its collapse, adequate reinforcements should be provided. Seismic retrofitting techniques for URM buildings that have been studied in previous studies are ferrocement, PP-band mesh, bamboo reinforcement, old tire strip, FRP, steel strip mesh and diagonal steel bracings (Figure 11).

Figure 11 - Retrofitting techniques: (a) Ferrocement (Shah [34]), (b) PP-band mesh reinforcement in testing stage (Sathiparan et al. [33]), (c) External horizontal bamboo (outside), external vertical bamboo (inside), internal horizontal chicken wire mesh and ring beam (Dowling et al. [15]), (d) Old tire strips (Kaplan et al. [20]), (e) Application of FRP reinforcement (Velazquez-Dimas [36]) 3.1 Retrofitting of URM Buildings against

Earthquakes The range of earthquake magnitude that may affect the country region should also be considered when introducing or selecting a retrofitting technique for existing and damaged URM buildings and, when introducing seismic resistant features for building construction. This is to eliminate unnecessary time and money wasting. In Sri Lanka, earth tremors of magnitude of less than 5 on Richter scale have been reported, in some areas in Nuwara Eliya, Buddulla, Matara, Galle, Ratnapura, Kalutara, Colombo, Hambantota, Ampara and Monaragala districts (De Silva [12]). A technique to be used to strengthen structures must be compatible with economy of the particular country as well as efficient in seismic resistant against particular range of earthquakes that the country could be affected. Therefore, three retrofitting techniques: ferrocement, poly-propylene band mesh and bamboo reinforcement, are explained where they meet with the economic feasibility in rural area, technology skills of labourers in Sri Lanka and level of required resistant level against earthquakes. In addition, there are other retrofitting techniques: Fiber Reinforced Polymer (FRP) laminating (Ehshani and Saadatmanesh [16], Ehshani et al. [17] and ElGawady et al. [18])), steel bracketing and introducing columns and tie beams (frame structure), which give higher

earthquake resistant to the building. These three techniques seemed to be very expensive and needed specialized labour for installation; therefore, suitable for urban areas and where structure is more valuable. Also, making a frame structure with reinforced columns and tie beams and infilling the frame using masonry or concrete blocks can be seen in building construction for residents in urban areas in Sri Lanka. 3.1.1 Ferrocement Ferrocement is a composed of steel wire mesh that is completely penetrated with a mortar which is rich in cement. The mortar is a mixture of sand, water and cement where sand particle size is not bigger than about 5 mm. The wire mesh is very thin and the wires are closely spaced (i.e., commonly named as “chicken mesh”). Several layers of meshes can also be used. Effect of ferrocement has been evaluated by Shah [34] by conducting an experiment under axial compression (Figure 11-(a)) using twenty one masonry columns of 221 mm x 221 mm x 784 mm and they were tested. He found that, encasement of unreinforced masonry brick columns with ferrocement cover of 6.125 mm thickness and 1:2 cement sand mortar with water cement ratio of 0.5 and mesh spacing of 12.25 mm, increased the ultimate failure load 121% or 1.33 times. Importantly, test results have showed that, failure of columns initiate only after failure of the casing. Ferrocement can also be used to repair columns, which have been loaded close to failure. The method discussed in Sha [35] can be applied in masonry walls as columns and walls are nearly similar in behaviour theoretically and practically. Shahzaada et al. [35] found that retrofitting method of ferrocement overlay (i.e., with 12.5 mm pitched steel wire mesh and 5 mm thick 1:2 cement-sand mortar overlay) improved the overall strength of unreinforced brick masonry walls by 48% and also increased their ductility and avoided brittle failure as observed is compressive loading on twenty brick masonry walls size 20 x 16 x 9 inch3 (i.e., 508 mm x 406.4 mm x 228.6 mm). A kind of in-plane cyclic loading test of increasing intensities with constant vertical loading has been carried out on URM walls (i.e., with size of 3087 mm x 3262 mm x 225 mm) before and after retrofitting by

(a) (b) (c) (d) (e)






(a) (b) (c) (d) (e)






(a) (b) (c) (d) (e)

7 ENGINEER

ferrocement overlay by Ashraf et al. [3]. As they have stated, lateral in-plane strength and lateral stiffness of the URM wall could be increased by 110% and 68%, respectively after retrofitting the wall by ferrocement overlay of 12.5 mm mesh pitched steel welded wire mesh (i.e. 1.0 mm diameter wires) and 19 mm thick 1:3 cement –sand mortar. These methods are simple, cost effective, required low technology and adding limited mass to the existing structure. 3.1.2 Poly-Propylene Packaging (PP-band)

Strip Mesh Reinforcement Poly propylene bands have been proposed as a cost-effective retrofitting material in Japan (Meguro et al. [28]). The method is simple, cost effective with no requirement of special technology and knowledge. Polypropylene is durable, inexpensive, harde and available worldwide. This material has no corrosion or insect failure effect and possesses excellent resistance to organic solvents and degreasing agent as well as electrolytic attack. The behavior of walls strengthened with various PP-band mesh arrangements have been studied by Macabuag et al. [25] using diagonal compression test on both full scale and ¼ scale wallets. They found that PP-band mesh helps masonry walls to enhance the diagonal shear capacity and deformation by 2.75 times and 15 times those of URM walls without retrofitting. To determine the resistance to in-plane and out-of-plane loading, diagonal compression (Figure 11) and flexural bending, tests for PP mesh reinforced wallets and unreinforced wallets have been conducted (Sathiparan et al. [33]). The diagonal compression tests showed that PP mesh strengthened walls provide higher residual strength after formation of the first diagonal shear cracks. The out-of-plane tests also indicated the effectiveness of PP mesh after the walls have cracked. The strength and deformation of PP mesh reinforced walls were 2 times and 70 times those of the un-retrofitted wallets, respectively in diagonal compression tests. In out-of-plane bending tests, they were 2 times and 25 times respectively. Mayorca et al. [27] have carried out a series of lateral reversed in-plane shear test on clay brick masonry walls and proved that although the mesh does increase the peak strength before initial failure, it contributes to improve wall‟s

performance including the post peak strength after the first crack occurrence. This statement is as similar as the conclusion regarding masonry wall with PP-band mesh, given by Macabuag et al. [25] and Sathiparan et al. [33]. These tests (Macabuag et al. [25], Sathiparan et al. [33] and Mayorca et al. [27]) proved that initial failure stress is unaffected by the presence of the PP mesh due to the much lower stiffness of PP mesh compared to masonry; but, after the initial masonry failure, retrofitted walls continue to maintain the load and even avoid the sudden collapse and brittle failure. Residual strength after crack initiation and residual stiffness of masonry wall retrofitted with PP-band mesh are directly proportional to PP-band density up to an optimum value and afterwards, they do not increase with the PP-band density (Sathiparan et al. [33]). It can be said that this is due to the fact that the behaviour of masonry walls retrofitted with PP-bands is effected not only by the effect of PP-bands, but also the combination of both masonry and bands as a composite. Looseness of the PP-band attachment with specimen reduces the residual strength after crack initiation of the specimen. However, an application of surface finishing gives a beneficial effect in residual strength (Sathiparan et al. [33]). Further, vertical bands mainly redistribute load through the specimen but offer little support, and horizontal bands offer little redistribution of load but directly bear the load to prevent loss of debris (Macabuag et al. [25]). 3.1.3 Bamboo Reinforcement Dowling et al. [15] suggested that a significant improvement in the earthquake resistance of adobe mud-brick structures can be obtained by using external vertical and horizontal bamboo reinforcement, internal horizontal chicken wire mesh reinforcement and a ring beam (Figure 11-(c)). They tested five 1:2 scale u-shaped adobe mud-brick walls; each 150 mm thick, 1800 mm wide and 1200 mm high. One of them was a control specimen with no retrofitting and others were retrofitted: one with only corner pilasters; one specimen with internal horizontal chicken wire mesh, external vertical bamboo (inside and outside) and timber ring beam; one specimen with internal vertical bamboo reinforcement, internal horizontal chicken wire mesh and a timber ring beam; and another


specimen with external vertical (inside) and horizontal (outside) bamboo reinforcement, internal horizontal chicken wire mesh and a timber ring beam. Each specimen was identified as summarized in Table 1. A downward restraining pressure loading of 125 kPa was applied to the top of the „wing‟ walls (acting as in-plane shear walls) of all specimens by tension bars between timber plates and beam resting on the walls, and the concrete base. Table 1- Specification of u-shaped wall units (Dowling et al. [15])

Specimen Description 3A Unreinforced, traditional 3B Corner pilasters/buttresses only

3E

Internal horizontal chicken wire mesh (every three courses), External vertical bamboo (inside and outside), Timber ring beam

3G

Internal horizontal chicken wire mesh (every three courses), Internal vertical bamboo, Timber ring beam

3I

Internal horizontal chicken wire mesh (every three courses), External vertical bamboo (inside), External horizontal bamboo (outside),Timber ring beam

The specimens were subjected to transient dynamic loading using the uni-axial shaking table to evaluate the response to out-of-plane seismic forces. Test results indicated that significant improvement in the earthquake resistance of adobe mud brick structures can be obtained by using technique specification used in 3I specimen. Although the specimen has showed severe damage at (100%) x 4 intensity time-scaled simulation of the January 13, 2001 El Salvador earthquake (which possessed magnitude of 7.7 on Mw (moment magnitude scale)), the collapse of wall was not imminent. In order to evaluate the effectiveness of the bamboo band mesh retrofitting technique on masonry walls against earthquakes, shake table tests have been conducted using retrofitted and non-retrofitted 1/4 scaled masonry houses with sinusoidal ground motion inputs (Meguro et al. [28]). According to the test, retrofitted masonry building could withstand over twice larger input energy than that of non-retrofitted specimen.

These methods seem to be relatively simple and easy to undertake, and utilize low-cost and readily-available materials, making them appropriate for application in householders with low income in developing countries. However, it is important to consider precautions against insect attack on bamboo. 3.2 Retrofitting of URM Buildings against

Tsunami and Flooding Because there are similar causes of effect by tsunami and flooding as by hydrostatic, hydro dynamic, impulsive, impact and buoyancy forces, same techniques can be used to strengthen the URM buildings against them. Retrofitting techniques should include anchoring the building or ensuring that the building itself is heavy enough against buoyancy forces and pressure. Also mechanical connections between the floor system and foundation must resist vertical and horizontal forces induced by flood and tsunami. Introducing the bracing members to the walls facing sea and minimizing the wall lengths can effectively reduce the flexural stresses developed by lateral pressure. As suggested by Maheshwari et al. [26] bending failure caused by hydrostatic forces of tsunami could only be avoided by using reinforced walls designed for lateral forces of tsunami. To fulfil this purpose, retrofitting techniques of reinforcement such as bamboo, PP-band mesh, FRP which proved as seismic resistant may be effectively used. Increased wall thicknesses will enhance the overall resistance of the structure and 225 mm thick walls are suitable instead of 113 mm thick walls which are not suitable for structures in coastal zone having a threat of tsunami loading. Kreibich et al. [24] have shown that, building a flood adapted house structure, e.g. using an especially stable building foundation or water proof sealing the cellar, is generally quite expensive. These structures can fail especially during extreme floods. However, steel frame and brick buildings tend to be less susceptible to collapse than those built with other materials. Water proof dry wall will hold up for long periods of inundation. To prevent penetration of surface water and ground water, any openings in the building must be raised or sealing measures must be implemented. Buildings can be sealed by using bitumen or strips of plastic or by constructing the base and walls of buildings using concrete that is almost non-permeable. The maximum height of water




3E


3G


3I








3E


3G


3I





9 ENGINEER

proofing should be approximately one meter above the ground. 4. Discussion A structure adequately resists the forces applied to it if there is a suitable path for the force to follow, from the source of the force to the soil below the foundation and each part of the structure, that is, various connections – the walls, the foundation and even the soil, are sufficiently strong to withstand the effect of the inertial force, generally induced by externally applied forces. To enable the house to withstand the inertial forces due to strong ground motion without any damage being done to it, various parts of the house must be larger than they are now. However, it would be very costly to make them large. Instead of that, the magnitude of the inertial force can be very significantly reduced if the parts of the house which resist the inertial force can be made “ductile”. Unfortunately, masonry blocks alone cannot be made ductile as they are made from brittle materials. Ductility in buildings can be achieved by placing ductile materials at certain places so that relevant parts of the building (walls, beams and columns) achieve sufficient level of ductility. However, materials cannot achieve 100% ductility and hence cracks will develop, but the building will not collapse. Steel, PP-bands, bamboo and also rubber are ductile materials which can be used in retrofitting URM walls against natural disasters. The mesh reinforcement of these materials is said to be more effective rather than the use of as steel rods, bamboo poles, PP-strips and rubber tires. The distribution the meshes of ductile materials on the wall imparts beneficial properties of those materials to the masonry. This advantage can also be enhanced by embedding of meshes into brittle mortar as mentioned under the retrofitting technique of PP-band mesh reinforcement by Sathiparan et al. [33] and according to Clarke and Sharma [10]. The main effect of the external vertical and horizontal PP-band mesh is to restrain the separated sections of masonry allowing for redistribution of the load within the masonry itself. Vertical bands apply normal compression once sliding of rows occurs, resulting in increasing the masonry‟s frictional resistance to shear sliding. Through friction, energy from

alien stresses is dissipated. Vertical bands also redistribute the load eliminating the concentration of load at particular regions but allowing deformation of wall panel. Horizontal bands directly bear load by resisting the separation of bricks within the same row allowing vertical bands to keep redistribution of the load over long period preventing/ delaying loss of debris. Both types of bands together give an improving of ductility with rigid box like action of enclosures. Further investigation should be focused on reducing stress which causes masonry corner cracking and separations and effect of real dwellings subjected to dynamic loads or reversed loads. Steel mesh reinforcement in ferrocement casing improves ductility and energy dissipation and redistribute lateral loads applied to wall as discussed in PP-band methods. Compared to PP-band reinforcement, ferrocement improves compression strength of masonry walls leading to enhance their ultimate load bearing ability and tensile strength in higher amount. Ferrocement casing also delays crack originating or appearing while PP-band reinforcement gives their distribution on seismic resistance for masonry walls after first crack created on the walls. Compared with conventional reinforced concrete, ferrocement is reinforced in two directions; therefore, it may have homogenous isotropic properties. Because the specific surface of ferrocement reinforcement is higher than that of reinforced concrete, ferrocement generally has a high tensile strength and high modules of rupture. Therefore larger bond forces develop with matrix resulting in average crack spacing and crack width of smaller magnitude than that of URM walls and even conventional reinforced concrete. However, the premature cracking can occur if the excessive mortar thickness applied to cover wire mesh or the ferrocement is not properly cured (Sha [35]). Preventing the delamination of ferrocement casing, increasing of reinforcement into ferrocement skin, using an alternative material as reinforcements instead of steel and resistance of ferrocement strengthening of wall panel for both in-plane and out-of-plane should be focused in further investigations. The retrofitting method used in 3I masonry specimen presented in Table 1, improves the flexural tensile strength of the wall. External horizontal bamboo poles restrain bending about vertical axis of the wall and tend to reduce flexural tension normal to bed joints and


external vertical bamboo poles restrain bending about horizontal axis of the wall and tend to reduce flexural tension parallel to bed joints. Therefore, this combination restrains both in-plane and out-of-plane bending of the structure. Internal horizontal chicken wire meshes redistribute the in-plane shear load/stress along the mortar bed joint improving diagonal tensile strength and shear strength against bed joint failure. The connection of ring beam on top of the wall resists the uplifting forces and shear forces and it applies normal compression together with vertical bamboos to resist shear sliding along mortar bed joints by increasing frictional resistance. Due to downward restraining force applied to top of the wing walls as mentioned in Section 3.1.3, overturning on walls and axial load resulted from overturing are prevented and wing walls acted as in-plane shear walls which resist lateral loads. This action lead to transfer the bulk of dynamic loads simulated in the test on out-of-plane long wall and corners of the specimens so that effectiveness of bamboo reinforcement as out-of-plane load resisting technique could be effectively tested. The external vertical and horizontal bamboo reinforcement, internal horizontal chicken wire mesh reinforcement and a ring beam altogether create an integrated matrix which restrain movement, and absorb, dissipate and redistribute energy within the structure. The behaviour of URM walls reinforced by external bamboo strip mesh reinforcement with different angles for in-plane and out-of-plane strengthening should be the focus of further investigation. The application of surface finishing after the retrofitting of URM walls gives a beneficial effect in residual strength (as mentioned by Sathiparan et al. [33] in section 3.1.2) as it fills the gap between mesh and wall. This phenomenon is common in any situation of retrofitting of a building by any kind of material due to which proper application of external plaster or surface finishing fill gap between the wall and the retrofitting material and even makes a bond between them. It is important to repair existing or damaged buildings before applying a new retrofitting on them, because repairing retains the original shape of the structure treating cracks and deterioration in them. As mentioned by Agarwal and Shrikhande [1], injection of cement grout containing admixtures or epoxy with and without inserting of reinforcement is

recommended for hairline cracks and 5-20 mm wide cracks, respectively. In the case of 20 mm wide cracks or wall material dislodging, cement grouting (with through stone/bond stone for stone masonries) may be effectively used. Deteriorated or damaged units/bricks should be replaced by new ones of same appearance and material property. Mortar deteriorated due to its poor quality must be strengthened by re-pointing technique. 5. Conclusions In Sri Lanka, natural disasters such as minor earthquakes, tsunami and storm surges, floods, cyclones and landslides are generally encountered. Weather changes showed that not only landslides and floods but also extreme wind events frequently occur. Forces induced by these effects on URM walls were identified as lateral loads including in-plane and out-of-plane induced by earthquakes and hydrostatic, hydrodynamic, impulsive, impact and buoyancy forces induced commonly by tsunamis and floods. Most of domestic buildings in Sri Lanka are constructed by using unreinforced masonry (URM) units. These URM buildings were frequently collapsed due to natural disasters. The most important weaknesses of the damaged masonry structures were the lack of interlocking units between external and internal wythes of the wall sections and the lack of connection between crossing walls. Possible failures of structures were identified as in-plane failure, out-of-plane failure and connection failure. The main in-plane failure mechanisms of URM walls due to earthquake actions are identified as shear failure, sliding failure, rocking failure and toe crushing while out-of plane failures consist of vertical centre breaking on main wall perpendicular to the earthquake and vertical corner cracking on intersection of main wall and bracing walls. Bending, sliding, overturning, cracking by diagonal tension and crushing by diagonal compression are the failures induced by tsunami loading. It can be concluded that, failures due to flooding are also similar to those by tsunami where almost both of them induce similar type of forces in varying magnitudes acting on URM buildings. Formation of openings near the corners of the walls was identified as another common problem where crack propagation is concentrated around these openings.







11 ENGINEER

Retrofitting of existing and damaged buildings to resist the forces induced by natural disasters is simple, cost effective and time saving than reconstruction of the building. The main objective of the retrofitting is to enhance the earthquake resistance of masonry structural elements, in order to avoid failure modes that manifest in brittle and unforeseen manner. There are numbers of retrofitting methods for URM buildings, including ferrocement, bamboo reinforcement, PP-band mesh reinforcement and FRP. Commonly bamboo and ferrocement retrofitting methods explained and discussed in this paper are very significant in using against earthquakes in order to either in-plane or out-of-plane loads. But it is important to study about newly introduced retrofitting techniques such as PP-band reinforcement for strengthening of wall behaviour for both loads at same time with same environment. Use of low quality materials and construction techniques cause extensive damages to unreinforced masonry buildings even when the magnitude of the natural disaster is quite low where no damages or very limited cracks are expected. Because of less awareness of peoples living in rural areas, generally, the domestic/rural area buildings are subjected to above situations. The most deaths occurred because low quality URM buildings have suddenly collapsed without giving sufficient time to people to evacuate for safe areas. Increasing of public awareness about the importance of strength properties of URM buildings are also recommended. Acknowledgement The research is funded by the National Research Council under NRC 11-193. The authors wish to express their special thanks to National Research Council (NRC) for providing necessary funds to carry out this research work.

References 1. Agarwal, P., and Shrikhande, M., “Earthquake resistant

design of structures”, 4th Edition, Prentice-Hall of India Private Limited, 2007.

2. Arya, A. S., “Non-engineered Construction in Developing Countries – An Approach towards Earthquake Risk Prediction”, The 12th World Conference on Earthquake Engineering (WCEE), Auckland, New Zealand, 30 January 2000.

3. Ashraf, M., Khan, A. N., Naseer, A., Ali, Qaisar and Alam basher, “Seismic Behaviour of Unreinforced and Confined Brick Masonry Walls before and after

Ferrocement overlay Retrofitting”, International Journal of Architectural Heritage, 2012, vol. 06, no. 06, pp. 665-688.

4. Bartolome, A. S., Quiun, D., and Zegarra, L., “Effective System for Seismic Reinforcement of Adobe houses", 13th World Conference on Earthquake Engineering (WCEE), Vancouver, Canada, 01-06 August 2004.

5. Bartolome, A. S., Quiun, D., and Zegarra, L., “Performance of Reinforced Adobe houses in Pisco, Peru Earthquake”, 14th World Conference on Earthquake Engineering (WCEE), Beijing, China, 12-17 October 2008.

6. Blonde,t M., Garcia, G. V., and Brzev, S., Earthquake-Resistant Construction of Adobe Buildings: A Tutorial, 2003, (available online http://www.world-housing.net/uploads/WHETutorial_Adobe_English.pdf [accessed on 12/06/2012]).

7. Blondet, M., Vargas, J., and Tarque, N., “Low-Cost Reinforcement of Earthen houses in Seismic Areas”, The 14th World Conference on Earthquake Engineering, Beijing, China, 12-17 October 2008.

8. Boen, T., “Earthquake Resistant Design of Non-engineered Buildings in Indonesia”, 2001, (available online http://www.buildingtrustinternational.org/Indonesian%20Earthquake%20Resistant%20Design.pdf [accessed on 23/05/2012]).

9. Caraballo-Nadal, N. C., Zapata-Lopez, R. E., and Pagan-Trinidad, I., “Building Damage Estimation due to Riverine Floods, Storm Surges and Tsunamis: a Proposed Methodology”, Fourth LACCEI International Latin American and Caribbean Conference for Engineering and Technology (LACCET),Mayaguez, Puerto Rico,21-23 June 2006.

10. Clarke, R., P., and Sharma, A. K., “The Earthquake Strengthening of Single-Storey Unreinforced Block Masonry houses in Trinidad and Tobago using ferrocement”, 1998, (available online http:// richardpclarke.tripod.com/hurri/manual.pdf [accessed on 27/05/2012]).

11. D‟Ayala, D., “Unreinforced Brick Masonry Construction”, 2011, (available online http://www.world-housing.net/wp-content/uploads/2011/08/Type_Brick.pdf [accessed on 27/06/2012]).

12. De Sliva, L., “Shifting of Tectonic Plates Cause for Tremors”, Sunday Observer, 19 April 2012, (available online http://www.sundayobserver.lk/2009/04/19/new13.asp [accessed on 19/04/2012])

13. Dias, P., Dissanayake, R., and Chandrathilake, R., “Lessons learned from Tsunami Damage in Sri Lanka”, Proceeddings of ICE – Civil Engineering, pp. 74-81.

14. Dissanayake, C. B., “A new plate boundary near Sri Lanka: implication for future geohazards”, Journal of Natural Science Foundation, Sri Lanka, Vol.33, No.1, 2005, pp. 5-8.

15. Dowling, D., Samali, B., and Li, J., “An Improved Means of Reinforcing Adobe Walls - External Vertical


Reinforcement”, Sismo Adobe, PUCP, Lima, Peru, 16-19 May 2005.

16. Ehshani, M., and Saadatmanesh, H., “Seismic Retrofit of URM walls with Fiber Composite”, TMS Journal, 1996, pp. 63-72.

17. Ehshani, M. R., Saadatmanesh, H., and Velazquez-Dimas, J. I., “Behaviour of Retrofitted URM Walls under Simulated Earthquake Loading”, Journal of Composite for Construction, 1999, pp. 134-142.

18. ElGawady, M. A., Lestuzzi, P. and Badoux, M., “Aseismic Retrofitting of Unreinforced Masonry Walls using FRP”, ELSEVIER, Composites: Part B 37, 2006, pp. 148-162.

19. ElGawady, M. A., Lestuzzi, P. and Badoux, M., “Retrofitting of Masonry walls using Shotcrete”, New Zealand Society for Earthquake Engineering (NZSEE) Conference, Napier, New Zealand, 10-12 March 2006.

20. Kaplan, H., Yilmaz, S., Nohutcu, H., Cetinkaya, N., and Binici, H., “Experimental Study on the use of Old Tires for Seismic Strengthening of Masonry Structures”, The 14th World Conference on Earthquake Engineering, Beijing, China, 12-17 October 2008.

21. Keerthisinghe, A. I., “Environmental Protection and Sustainable Development in Sri Lanka”, The Sunday Leader, 9 July 2012, (available online http://www.thesundayleader.lk/2012/07/08/environmental-protection-and-sustainable-development-in-sri-lanka/ [accessed on 08/07/2012]).

22. Kelman, I. and Spence, R., “A Limit Analysis of Unreinforced Masonry Failing under Flood Water Pressures”, Masonry International, 2003, Vol. 16, No. 2, pp. 51-61.

23. Khazai, B., Franco, G., Ingram, J. C., Rio, C. R. D., Dias, P., Dissanayake, R., Chandratilake, R., and Kanna, S. J., “Post-December 2004 Tsunami Reconstruction in Sri Lanka and its Potential Impacts on Future Vulnerability”, Earthquake Spectra, 2006, Vol.22, No.S3, pp. S829–S844.

24. Kreibich, H., Thieken, A. H., Petrow, T., Muller, M. and Merzl, B., “Flood Loss Reduction of Private Households due to Building Precautionary Measures – Lessons Learned from the Elbe flood in August 2002”, Natural Hazards and Earth System Sciences, 2005, Vol. 5, pp. 117-126.

25. Macabuag, J., Bhattacharya, S. and Blakeborough, T., “Extending the Collapse Time of Non-Engineered Masonry Buildings under Seismic Loading”, The 14th

World Conference on Earthquake Engineering, Beijing, China, 12-17 October 2008.

26. Maheshwari, B. K., Sharma, M. L. and Narayan, J. P., “Structural Damages on the Coast of Tamil Nadu due to Tsunami Caused by December 26, 2004 Sumatra earthquake”, ISET Journal of Earthquake Technology, June-September 2005, Vol. 42, No. 2-3, pp. 63-78.

27. Mayorca, P. and Meguro, K., “Proposal of an Efficient Technique for Retrofitting Unreinforced Masonry Dweellings”, 13th World Conference on Earthquake

Engineering (WCEE), Vancouver, Canada, 01-06 August 2004, Paper No. 2431.

28. Meguro, K., Soti, R., Navaratnaraj, S. and Numada, M., “ Dynamic Behaviour of Masonry houses Retrofitted by Bamboo Band Mesh”, The 15th World Conference on Earthquake Engineering, Lisbon, Portugal, 24-28 September 2012.

29. Rajasingham, K. T., “Saving Sri Lankan Lives through Open Data”, Asian Tribune, 14 September 2012, Vol. 11, No. 532, (available online http://www.asiantribune.com/news/2012/09/13/saving-sri-lankan-lives-through-open-data [accessed on 11/07/2012]).

30. Renuka, I. H. S., and Lewangamage, C. S., “Structural Aspects of Post Tsunami Domestic Constructions in Sri Lanka”, Annual Research Journal of SLSAJ, Sri Lanka, 2011, Vol. 11, pp. 65–71.

31. Rogers, C.D.P.E., “Structural damage due to floods”, 2012, (available online http://www.rimkus.com/craig_rogers_article_in_claims_magazine [accessed on 11/07/2012]).

32. Saatcioglu, M., Ghobarah, A. and Nistor, I., Reconnaissance Report on the December 26, 2004 Sumatra earthquake and tsunami, 2005, (available online http://www.caee.uottawa.ca/Publications/PDF%20files/CAEE%20Sumatra%20Report.pdf [accessed on 27/062012])

33. Sathiparan, N., Mayorca, P., Nesheli, K. N., Guragain, R. and Meguro, K., “Experimental Study on In-plane and out-of-plane Behaviour of Masonry Wallets Retrofitted by PP-band meshes”, Seisan-Kenkyu, 2005, vol. 57, no. 06, pp. 530-533.

34. Shah, A. A., “Applications of ferrocement in Strengthening of Unreinforced Masonry Columns”, International Journal of Geology, 2011, vol. 5, no. 01, pp. 21-27.

35. Shahzada, K., Javed, M., Alam, B., Khan, M., Ali, Z., Khan, H., and Shah, S. S. A, “Strengthening Brick Masonry against Earthquake Loading”, International Journal of Advanced Structures and Geotechnical Engineering, July 2012, Vol. 01, No. 01, pp. 10-14.

36. Velazquez-Dimas, J. I., Ehsani, M. R. and Saadatmanesh, H., “Out-of-plane Behaviour of Brick Masonry walls Strengthened with Fibre Composites”, ACI Structural Journal, May-June 2000, Vol. 97, No. 3, pp. 377-387.

37. Webster, F.A., “Earthen Structures: Assessing Seismic Damage, Performance and Interventions”, Terra Literature Review: An Overview of Research in Earthen Architecture Conservation, 2008, pp.69-78, (available online http://www.getty.edu/conservation/publications_resources/pdf_publications/terra_lit_review.pdf [accessed on 10/07/2010]).

Printed by Karunaratne & Sons (Pvt) Ltd.

vol: xlvii, no. 03 july 2014 - the institution of engineers sri...

Documents