extending the collapse time of non-engineered masonry buildings under seismic loading

7/29/2019 Extending the collapse time of non-engineered masonry buildings under seismic loading

http://slidepdf.com/reader/full/extending-the-collapse-time-of-non-engineered-masonry-buildings-under-seismic 1/14

EWB-UK Research Conference 2009

Hosted by The Royal Academy of Engineering

February 20

Community of Practice: habitatAuthor: Josh Macabuag

Institution: Edge Structures Ltd, London (previously MEng Student, University of Oxford)

Previously published: 14th World Conference of Earthquake Engineering, Beijing, China, Oct 08

The Structural Engineer: Vol. 86, Issue 07, Apr 08

EXTENDING THE COLLAPSE TIME

OF NON-ENGINEERED MASONRY

BUILDINGS UNDER SEISMIC LOADING

J. Macabuag1

and S. Bhattacharya2

1Graduate Structural Engineer, Edge Structures Ltd, London (previously MEng Student, University of

Oxford),

2University Lecturer, University of Bristol, (previously University of Oxford)

Email: [email protected], [email protected]

ABSTRACT :

THE COLLAPSE OF NON-ENGINEERED MASONRY IS ONE OF THE GREATESTCAUSES OF DEATH IN MAJOR EARTHQUAKE EVENTS AROUND THE WORLD.THIS PAPER INVESTIGATES A RECENTLY DEVELOPED RETROFITTING

TECHNOLOGY SPECIFICALLY AIMED AT PREVENTING OR PROLONGING THECOLLAPSE OF ADOBE (MUD BRICK) BUILDINGS UNDER STRONG EARTHQUAKES.THIS TECHNOLOGY USES COMMON POLYPROPYLENE PACKAGING STRAPS TOFORM A MESH, WHICH IS THEN USED TO PREVENT BRITTLE MASONRYCOLLAPSE. THE RETROFITTING TECHNIQUE IS TESTED USING STATIC,DIAGONAL LOADING OF MODEL WALL PANELS. IT IS SHOWN THAT THEPROPOSED TECHNIQUE EFFECTIVELY PREVENTS BRITTLE COLLAPSE OF THEPANEL AND THE LOSS OF DEBRIS. PARTIAL MESHES OF VARIOUSORIENTATIONS ARE ALSO INVESTIGATED IN ORDER TO BETTER IDENTIFY THEACTION OF THE MESH. FINALLY, AN IMPLEMENTATION PROJECT IS PRESENTED,INVOLVING A TRAINING PROGRAMME FOR RURAL MASONS IN NEPAL AND A

PUBLIC SHAKE-TABLE DEMONSTRATION.





February 20





1. INTRODUCTION

1.1. Motivation For This Study

“The replacement of existing dwellings with ‘earthquake-resistant houses’ isneither feasible nor, perhaps, desirable. It has been found more realistic tothink, rather, in terms of low-cost upgrading of traditional structures, with theaim of limiting damage caused by normal earthquakes and giving their occupants a good chance of escape in the once-in-a-lifetime event of a largeearthquake.” [Coburn and Spence]

The great majority of all earthquake fatalities result from building failures with agrowing disparity between vulnerability of those in developing and developedcountries [GHI ]. The greatest risk is by far presented to inhabitants of non-

engineered masonry structures as demonstrated in the 2003 Bam (Iran)earthquake, where many of the thousands of deaths were attributable tovulnerable adobe (mud brick) structures. Similarly vulnerable, non-engineeredmasonry is widespread throughout the developing world and replacement of allsuch dwellings is both infeasible and undesirable, given that they are often theembodiment of local culture and tradition. Therefore, it is often more feasible toconsider low-cost retrofitting of such buildings.

1.2. Masonry Collapse

Structural collapse under seismic loading displays many possible failuremechanisms often related to interaction between structural components (e.g.separation of walls or floor-wall connections). When considering the failure of individual walls, the inertia forces induced by seismic action can act out-of-plane(e.g. causing toppling) or in-plane (e.g. diagonal shear cracks, often from thecorners and openings).

In-plane failure is a fundamental failure mode for unreinforced masonry anddetermination of wall shear strength is necessary for defining its resistance tothis mode of failure. For shear failures, the shear strength [τ f ] of plain masonrywalls is given by the frictional relationship presented in equation 1.1 [EC6,

BS1052-3].

f oτ τ µ = + (1.1)

τ f = failure shear strength,τ o = shear strength under zero compressive stress, µ = coefficient of friction.





February 20





1.3. Currently Available Retrofitting Techniques for Non-Engineered Masonry

Methods required to meet the needs of the large populations in danger of non-

engineered masonry collapse must be simple and inexpensive to match theavailable resources and skills [Mayorca]. Notable low-cost retrofitting techniquessuitable for non-engineered, unreinforced masonry dwellings are given in Table1.

Method Developing Institute Description

Polypropylene(PP) Meshing

Institute of IndustrialScience (IIS),Tokyo University, Japan

Encasing masonry walls with amesh constructed of polypropylenestrapping used for packagingworldwide [Mayorca]

Wire Meshing Pontificia UniversidadCatólica del Peru, Peru Similar to pp-meshing, but using asteel wire mesh [Blondet ]External VerticalBambooReinforcement

Sydney University, Australia External vertical bambooreinforcement [Dowling]

TABLE 4: EXISTING RETROFITTING TECHNIQUES FOR UNREINFORCEDMASONRY IN THE DEVELOPING WORLD

This paper focuses on the technique of polypropylene (pp) meshing. PP-meshingwas first formally proposed in 2003 [Mayorca], is still under active research and

currently has application in Nepal, Pakistan and Kathmandu. Figure 1 shows aretrofitted house in Pakistan.

FIGURE 1: RETROFITTED HOUSE IN PAKISTAN BEFORE AND AFTER APPLICATION OF COVERING MORTAR LAYER 6

6Photos have been provided by Meguro Lab, Institute of Industrial Science, Tokyo University, Japan





February 20





1.4. Objectives

This paper is concerned with investigating a method for prolonging the collapselife of non-engineered masonry under seismic loading, using polypropylene

meshing. Validation of a proposed method requires extensive testing undervarious conditions. As masonry is one of the world’s oldest building materials, avast number of masonry types have evolved. Therefore, full-scale testing of allavailable masonry types is infeasible and so efficient modelling techniques arerequired.

Therefore, this paper aims to:

• investigate the effectiveness of pp-meshing in preventing brittle failure of unreinforced masonry specimens, under in-plane loading,

• investigate the scaling issues associated with small-scale modelling of the pp-mesh,

• identify the precise action of the mesh in more detail,• present the implementation of pp-meshing in a seismically active region of

Nepal.

2. DIAGONAL COMPRESSION OF FULL AND SMALL-SCALE MASONRY

Small-scale modelling was conducted at a linear scale of 1:4 as shown in figure4. Determination of masonry shear resistance [τ f ] to in-plane lateral load wasachieved by testing both retrofitted and non-retrofitted square prisms incompression along one diagonal (figure 2). There is as yet no British or

European standard for determining panel shear-resistance, and so the AmericanStandard has been followed [ ASTM E 519-02].

Two full-scale walls and three small-scale models were constructed for non-retrofitted testing under diagonal compression. For retrofitted testing two full-scale walls were constructed (parallel band spacing: 60mm) and nine small-scalemodels (band spacing: 15mm). Three of the small-scale retrofitted modelsshared the same mortar with the non-retrofitted specimens.

In addition to fully retrofitted masonry panels, meshes of various types werealso tested to further isolate and understand the action of the mesh (figures 6e& f). To achieve this, the remaining small-scale specimens were tested in two

batches of three with both batches using a different mortar and consisting of onefully retrofitted specimen, one with only horizontal reinforcement (parallel to themortar bed joint) and one with only vertical (perpendicular).

Standard packaging strapping was used, and fastened with clips provided by theband manufacturers. Note that the clips used do not represent the method usedin practice for applying the mesh. The aim of this test was to examine the effectof the pp-mesh on masonry failure and so recreating the installation method wasnot a requirement.





February 20





A: FULL SIZE BRICK WITH ¼SIZE BRICK USED FOR SMALL-SCALE TESTING

B: FULL-SCALEWALL PANEL

C: SMALL-SCALEWALL PANEL

d: Full-scale retrofitted wallpanel

e: Small-scale fully retrofitted specimen7

Figure 2: Full and small-scale models ready for testing

2.1. Similitude of Model Mortar & Mesh

In order for a small-scale model to accurately represent full-scale behaviour, themodel and prototype should present the same stress-strain profiles whensubjected to equivalent loading types [Harris and Sabnis]. Thus for the full andsmall-scale models to be similar the failure stresses [σ f ] and failure strains [ε f ]must be equal.

7The ruler shown in figure 4e is a 30cm metal rule.

650mm

150mm





February 20





In the diagonal compression test the mortar bed is to be orientated at 45˚ to thehorizontal, and so the stresses in the mortar may be approximated (in theproximity of the applied force [P]) as in figure 5:

2

P N

A

τ ∴ = = (2.1)

Figure 3: Diagonal compression test variables

where,P = diagonally applied load as shown in figure 2N = normal stress component of applied load (P)A = contact area between brick courses

Therefore, substituting these stresses (equation 2.1) into the definition of shearfailure (equation 1.1) and making the initial assumption that brick surface

friction and mortar aggregate size will be controlled such that the coefficient of friction will be comparable for large and small-scale models (µ = const), givesthe condition:

τ mo = τ po (2.2)

Equation 2.2 shows that in order to obtain identical stress-strain curves fordiagonal shear test specimens of different scales, identical mortar strength mustbe ensured. Note that for dynamic testing this analysis no longer applies.

It must be noted that both mesh and masonry must be scaled equally to satisfy

similitude. As the small-scale models used in these tests are at a linear scale of ¼, then the cross-sectional area of the pp-bands must be reduced by a factor of 16. Much of the previous small-scale testing of pp-meshing has used meshes of different scales to the masonry due to difficulties in producing bands of therequired cross-section. To recreate this discrepancy ¼-width bands were usedfor the small-scale testing discussed in this paper to investigate how wellmeshes of this scale describe full-size behaviour.

Δ

P

A = contact area

F

N





February 20





3. RESULTS AND DISCUSSION

All failures of full and small-scale non-retrofitted walls were brittle with nofurther load being maintained whereas retrofitted models continued to carry load

after initial failure (figure 3). Example specimen failure loads are given in figure4.

a: Full-scale specimenat brittle failure

b: Small-scalespecimen at brittlefailure

c: Full-scale specimen aftertesting. Note displacement androtation of corner section

d: Continuing to maintainload after second bandfailure. Further crackingsuggests redistribution of load

e: Intact sectionssuggest littleredistribution of load. Total collapseobserved afterfailure of thesupporting band.

f: Load redistributingthrough specimen(shown by continuedcracking) but littlesupport offered byvertical bands. Noteloss of debris.

Figure 4: Full and small-scale model failures

3.1. Comparing Retrofitted and Non-Retrofitted Failure

Non-retrofitted specimens displayed brittle failure and collapse whereasretrofitted specimens continued to maintain load after initial failure of themasonry (figure 4). There was also no significant loss of debris until several





February 20





bands were broken. During loading, the mesh acted to maintain panel integrityallowing the load to be redistributed throughout the mesh and masonry (shownby the formation of further cracks upon continued loading). Individual bandfailures showed that significant load was also carried by the mesh and it should

be noted that all band failures occurred in horizontal bands at brick vertices.Figure 4 plots the comparative performances.

0

0.5

1

1.5

2

2.5

3

0 10 20 30 40 50 60

[∆] Vertical Deformation (mm)

[ P ] C o m p r e

s s i v e L o a d ( k N )

Fully Retrofitted

Non-Retrofitted

FIGURE 5: LOAD VS DISPLACEMENT FOR RETROFITTED AND NON-RETROFITTED SMALL-SCALE MODEL

3.2. Comparing Mesh OrientationHorizontal and vertical retrofitting was investigated individually in order toisolate the effects of the mesh parallel and perpendicular to the mortar bed joint,respectively. These tests showed that the main effect of the mesh is to restrainseparated sections of masonry allowing for redistribution of the load within themasonry itself. Vertical bands provide little direct resistance to lateral sliding butupon panel deformation, band tension acts perpendicular to the horizontalmortar joints allowing frictional effects between sliding rows to resist furthercollapse. However, the specimen shown in figure 4f highlights that theincomplete, vertical mesh was unable to prevent loss of material, so limiting the

redistribution of load. Horizontal retrofitting is shown to resist the separation of bricks within the same row and so is effective when diagonal or vertical crackinghas taken place. That horizontal bands directly resist the load can also be seenin the fact that all band failures occurred in horizontal bands.

3.3. Comparing Full and Small-Scale FailureInitial failure stress and pre-failure behaviour was unaffected by the presence of the pp-mesh due to the masonry’s relative rigidity (figure 4). Therefore, pre-

Δ

P





February 20





failure data of retrofitted and non-retrofitted specimens may be compared.However, a comparison of full and small-scale retrofitted post-failure profilesreveals that the small-scale specimen is able to maintain a far greater ultimateload relative to its initial failure loads. This is to be expected given that the

small-scale mesh used was not of the same linear scale.

If post-failure behaviour were purely defined by the action of the mesh, thenreducing (by a factor of four. i.e. to the scale of the mesh) the load-displacement curve for the small-scale specimen should lead to correlation withthe full-scale curve. However, performing this adjustment gives non-dimensionalloads below that of the full-scale specimen. This trend was shown by all testedspecimens. This therefore highlights the fact that post-failure behaviour is acomplex interaction of masonry and mesh and not solely due to meshproperties.

Therefore, to simulate accurate quantitative post-failure behaviour, mesh andwall must be of the same linear scale. However, the observed retrofitted failurepatterns are common to both full and small-scale specimens despite the lack of similitude. Therefore for the purposes of investigating qualitative mesh/masonrybehaviour and interactions (e.g. for testing of different mesh types/orientations,pitches etc) it is not necessary that similitude of both the masonry and mesh besatisfied.

4. IMPLEMENTATION OF THE PROPOSED RETROFITTING TECHNIQUE

To investigate the practical issues of implementation a pilot scheme is beingconducted in a seismically active region of the Kathmandu Valley, Nepal.

The Himalayan region is an example of one area of constant seismic activity,high population density, and wide-spread use of unreinforced masonry builtoutside of current building standards. Given the high potential for future loss of life, several pp-band implementation programmes have been run in this region,by the Nepalese NGO, National Society for Earthquake Technology.

Given that the dwellings most at risk are built outside of building regulations it isclear that a sustainable solution can only be achieved by raising local awarenessof available methods and allowing the building owners and tradesman to

themselves become the disseminators of the proposed solution.In 2006 a public, low-tech shake-table demonstration was held in Kashmir(following the 2005 earthquake) followed by the retrofit of a full-scale building.In 2007 an implementation project proposal was awarded the MondialogoEngineering Award and granted funding. The Mondialogo scheme will took placein November 2008 as a partnership between Oxford University; the Institute of Industrial Science, Tokyo University; the Indian Institute of Technology,Bombay; Nepal Engineering College and Khwopa Engineering College, Nepal andthe National Society of Earthquake Technology – Nepal.





February 20





The implementation project involved a six-day training course for local, ruralmasons, focusing on both earthquake construction and the pp-retrofittingtechnique. At the end of the course was a public low-tech shake-tabledemonstration of the pp-band technology, inviting the community, press and key

individuals/institutions.The masons were engaged in several aspects of earthquake construction:appropriate site selection, planning and construction techniques (in masonry,timber and RC), strengthening and repairing of existing structures andretrofitting using the pp-mesh. The demonstration was designed to allow themasons to apply what they have learnt and allow the public to graphicallywitness the necessity to improve upon traditional building techniques and tosafeguard existing buildings against collapse. The event received significantradio and television coverage in Nepal.

A suitable adobe dwelling has been identified in the area and will soon be

retrofitted by the masons trained in November, under the supervision of NSET.The technique is currently designed for single-storey adobe but research iscurrently being conducted at Tokyo & Bristol Universities to test its applicationup to 2-storeys and for stone masonry. A team of students at Bristol University,Tokyo University, IIT Bombay and Nepal Engineering College are currentlyworking towards reentering the Mondialogo Engineering Awards for 2009. Thecharity Engineers Without Borders-UK are looking at sending volunteers to NSETfor the second consecutive year and there are plans to visit a very similarprogramme running in Peru, to help adapt the technique further.

In addition to further understanding the issues of implementation, the long-termaim is to extend this original training program and demonstration to other areas

of high-risk throughout the Himalayan region, with the assistance of NSET, byfurther engaging municipalities and universities to support research anddissemination.

b: mock mesh reinforcement8

8NB/ The aim of the pp-retrofit during the course was as a short introduction to forming the mesh and

attaching to the wall. PP-band retrofitting is designed for adobe structures where holes can be moreeasily drilled through bricks, allowing finer spacing of through-wall connectors, giving a tighter mesh.Real retrofit is also continued around corners and connected to an RC band at ground level.





February 20





a: trainee masons buildingreinforced masonry model (aboutto prepare for 1st level of horizontal reinforcement)

c: trainee masons building model

Figure 6: Seismic Construction training course and publicdemonstration, Bhaktapur, Nepal 2008.

5. SUMMARY

This paper has investigated the technique of polypropylene meshing forpreventing or prolonging the collapse of adobe buildings under strongearthquakes. The behaviours of full-scale and small-scale models have beencompared to identify scaling issues of modelling retrofitted specimens. Variousmesh types have also been tested to investigate the action of the mesh. Finally,a pilot scheme for pp-mesh implementation in rural Nepal has been introduced.

The main findings of this paper are summarised as follows:• Non-retrofitted walls showed sudden brittle failure and were unable to

maintain further load. It is this brittle failure that poses significant danger tobuilding occupants during earthquakes.• Retrofitting masonry walls with polypropylene meshing allowed specimens to

maintain load after initial failure of the masonry and prevented the loss of debris, even after the failure of several straps. Given the low cost, highavailability and relative simplicity of the pp-meshing technique, thistechnology may potentially be used to prevent/delay brittle collapse of non-engineered structures under seismic loading.

• Separating the effect of horizontal and vertical reinforcement showed that:o Vertical bands apply normal compression once sliding of rows occurs,

increasing the masonry’s frictional resistance to shear sliding.

o Horizontal bands directly bear load by resisting the separation of brickswithin the same row.• Band failure occurred in horizontal bands at brick vertices. This suggests

that further investigation should be focused on reducing the stressconcentrations experienced at masonry corners.

• Small-scale retrofitted models gave good qualitative indication of full-scalebehaviour even where similitude between the mesh and wall was notmaintained. However, quantitative assumptions of full-scale behaviourcannot be obtained from small-scale testing if mesh and wall are not of thesame scale.





February 20





• An implementation project trained rural masons outside Kathmandu, Nepalin earthquake construction and retrofitting and conducted a public, low-techshake-table.

• A full-scale retrofit will be taking place near Kathmandu shortly and furtherwork is being conducted by students at Bristol University, Tokyo University,IIT Bombay, Nepal Engineering College and the National Society of Earthquake Engineering – Nepal.

REFERENCES

Blondet, M., G. Villa Garcia, and S. Brzev (2003), Earthquake-resistantconstruction of adobe buildings: A tutorial. EERI/IAEE World HousingEncyclopedia.

Coburn A., Spence R. (2002), Earthquake Protection (2nd edition), John Wiley & Sons Inc.

Dowling, D.M., and Samali, B. (2006), Low-cost, low-tech means of improvingthe earthquake resistance of adobe-mudbrick houses, Proceedings of theInternational Conference on Earthquake Engineering (ICEE 2006), Lahore,Pakistan.

GeoHazards International, Global Earthquake Safety Initiative – Pilot Project,United Nations Centre for Regional Development (2001).

Harris, H.G. and G.M. Sabnis (1999), Structural modeling and experimentaltechniques. CRC Press.

Mayorca, P. (2003), Strengthening of Unreinforced Masonry Structures inEarthquake Prone Regions, Ph.D. Dissertation, The University of Tokyo, Japan.





February 20

Community of Practice: Habitat

Author: Andrew Smith & Thomas Redman

Institution: University of Bristol

TABLE 1.1: EARTHQUAKES CAUSING THE GREATEST NUMBER OF CASUALTIES IN THE LAST 100 YEARS (BHATTACHARYA, 2007)

A Critical Review of Retrofitting Methods for Unreinforced Masonry Structures

By Andrew Smith & Thomas Redman

Section 1: Introduction

The consequences of earthquakes can be devastating to human lives, as shownin Table 1.1, which lists the casualties from past earthquakes.

Year Location Casualties

1908 Messina (Italy) 70,000 to 100,000

1920 Gansu (China) 200,0001923 Kanto (Japan) 143,000

1927 Qinghai (China) 200,000

1932 Gansu (China) 70,000

1948 Ashgabat (Turkmenistan) 110,000

1970 Peru 66,000

1976 Tangshan (China) 255,000

2001 Gujarat (India) 20,000

2003 Bam (Iran) 30,000

2004 Sumatra (Indonesia) 220,000

2005 Kashmir (Pakistan) 73,0002008 Sichuan (China) 69,197

Total death toll: 1,556,197

Population of the City of Prague, Czech Republic: 1,212,097 [Czech

This review seeks to collate information on and categorise the various types of retrofitting methods for unreinforced masonry (URM) buildings under research orearly implementation, and critically compare them to help further understand

which methods are most suitable for further research or application indeveloping countries. This work is a continuation of the international, MondialagoAward-winning project, ‘Improving the Structural Strength under SeismicLoading of Non-Engineered Buildings in the Himalayan Region’ and will buildupon the work carried out to date.

Section 2: Types of Buildings Vulnerable to Collapse during Earthquakes

2.1. Introduction

URM buildings can be broadly arranged into three common categories: adobe,

brick and stone masonry. Although cheap and easy to build, all URM buildingshave been observed to be susceptible to earthquakes.





February 20

Community of Practice: Habitat

Author: Andrew Smith & Thomas Redman

Institution: University of Bristol

2.2. Adobe Buildings

The popularity of adobe buildings is illustrated by the high proportion of housingpresent in Latin America, Africa, India, Asia, the Middle East and SouthernEurope. Indeed, 30% of the world’s population lives in adobe dwellings, which

accounts for 20% of the world’s urban/suburban population, Blondet (2003).However, adobe structures are highly prone to collapse during earthquakescausing considerable damage and loss of lives, as shown in table 2.1.

Earthquake Fatalities Adobe Buildings damaged or People effected

El Salvador,

2001

1,100 150,000 1.6 million

Southern Peru,

2001

81 25,000 -

Iran, 2003 26,000 85% of infrastructure destroyed 100,000 lefthomeless

TABLE 5.1: EFFECTS OF EARTHQUAKES ON ADOBE BUILDINGS

[BASED ON INFORMATION FROM BLONDET (2003) & BAM PRELIMINARYEARTHQUAKE REPORT, LINK:HTTP://EARTHQUAKE.USGS.GOV/EQCENTER/EQINTHENEWS/2003/USCVAD/]

2.3. Brick Masonry Buildings

Brick masonry housing is also vulnerable to collapse under seismic loading; the1976 earthquake in China caused the loss of approximately 240,000 lives,mainly owing to the collapse of brick masonry structures.

2.4. Stone Masonry Buildings

Stone masonry buildings tend to perform poorly in earthquakes owing to the lowstrength of the stone and mortar used and the lack of adequate wallconnections. The quality of local construction is often very low due to the lack of

skilled engineers involved. For example, in Nepal, over 98 % of buildings areconstructed by the owners following the advice of local craftsmen, Meguro et al,(2005).

Section 3: Common Failure Mechanisms of URMs

3.1. Introduction

From past earthquakes, the buildings discussed in section 2 have been observedto demonstrate common failure mechanisms.

3.2. Adobe Buildings

extending the collapse time of non-engineered masonry buildings under seismic loading

Documents