Journal of the International Masonry Society · The International Masonry Society Registered Office: British Ceramic Research Ltd, Queens Road, Penkhull, Stoke-on-Trent, ST4 7LQ,

Journal of the International Masonry Societywww.masonry.org.uk 2017/18

ISSN 2398-757X Sfb (2) Fd,e,f UDC 62 + 69 Vol 30, No 1. 1-30 2017/18

FEATURED IN THIS ISSUE:• Construction Practice and Structural Performance of Yemeni Traditional Minarets• Simple Homogenisation Model for the Non-Linear Analyses of 3D Masonry Structures• A Comparative Investigation into Lime Activated Ground Granulated Blast Furnace Slag as a

Sustainable Alternative to Portland Cement in Masonry Mortars

The International Masonry SocietyRegistered Office: British Ceramic Research Ltd, Queens Road, Penkhull, Stoke-on-Trent, ST4 7LQ, UK

Society website www.masonry.org.ukSecretary: Dr K Fisher

Contact details: Postal: Shermanbury, 6 Church Road, Whyteleafe, Surrey, CR3 0AR, UKTel: +44 (0) 20 8660 3633

email: [email protected]

President: Mr M LeonardPresident-Elect: Professor JJ RobertsSecretary: Dr K FisherTreasurer: Mr PE WoodImmediate past president: Professor SW GarrityHonorary Editor: Dr K FisherExecutive Editor: Dr AN Fried

UK Council Members Mr I Harrison Dr L MacoriniMr G Sargeant Dr A Tomor Dr RC de Vekey Mr S HayDr A Smith Professor E Laycock

IMS is an International Society with Members throughout the world. It provides a focus for all those involved in or interested in the manufacture of masonry materials, the design of masonry structures and their economical construction. All masonry units are included: calcium silicate, clay, concrete and stone together with mortar and the ancillary components, dpc’s and straps, ties and fixings.

The Society’s journal, Masonry International, is published three times each year. Each issue consists of original papers with the latest research findings, both experimental and analytical, practical papers, together with Society and International news. Each Member receives Masonry International free of charge. Corporate Members receive two copies. Members may purchase other publications, including the Proceedings, and attend meetings and conferences at preferential Members’ rates. Masonry International is available to libraries and others on subscription.

Technical meetings are held from time to time at which original papers are presented, discussed and which subsequently may be published after peer review. Some of these meetings are major conferences, including the quadrennial International Masonry Conference.

Everyone with any interest in masonry should keep themselves up to date as a Member with the opportunity for discussion with one’s peers. Membership is of special value to: architects, engineers, surveyors,

builders, developers, manufacturers, teachers and research workers, users of masonry products and all those interested in the appearance and performance of the built environment.

Enquiries Enquiries concerning membership, subscriptions to the journal and information concerning meetings should be addressed to the Secretary: Dr K Fisher at the postal or email address at the head of this page.

PaymentCheques in sterling or bankers’ drafts on a UK bank should be made payable to THE INTERNATIONAL MASONRY SOCIETY.The Society also has the facility to accept the following credit cards: EUROCARD/MASTERCARD/VISA/AMERICAN EXPRESS.Persons wishing to pay in this way should send their card number, date of expiry, the name as it appears on the card and the address together with a note authorising the sum to be debited and signed with the recognised signature. Payment can be made online at www.masonry.org.uk

NOTES FOR AUTHORSPapers for publication in Masonry International should be sent to:

The Secretary, Shermanbury, 6 Church Road, Whyteleafe, Surrey, CR3 0AR, UK OR by email to [email protected]

Subject matterPapers on any aspect of design, construction, maintenance, or research relating to masonry and masonry materials are invited for publication in Masonry International.

FormatContributions should be formatted in accordance with the specified template prepared by the Society. This can be found on the Society website, www.masonry.org.uk as listed under Masonry International- Notes for Authors MI Paper – Template v4.DOC.

CopyrightIf authors include in papers material which is not their copyright, for example figures and tables, they MUST not only give appropriate acknowledgement by way of references but also obtain the approval of the copyright holder for the Society to reproduce it.Any queries relating to submissions should be addressed to the Secretary, contact details as above.

CITATIONMasonry International is an archival publication for original papers on research and practice in masonry materials, design and construction. All submissions to the Journal are refereed and edited. The Bulletin contains more practical papers and review articles. Material appearing in this publication, and in the separate Proceedings of the Society, has been subjected to peer review and, as such, may be cited and accepted as significant in relation to professional advancement.

Overseas Council MembersProfessor AW Page (Australia) Dr JM Nichols (USA)Professor G Milani (Italy) Mrs R Popescu (Romania)

Student Council MembersMiss A Isfeld (Canada) Mr M Asenov

Past Presidents - Council Members: Dr AJ Bell Mr CA FudgeDr NE Beningfield Mr BA Haseltine MBEMr WA Ferguson Mr JC HaynesDr DFR Spencer Professor ME PhippsProfessor GJ Edgell Mr P Rogatzki

Masonry International Editorial BoardProfessor John J Roberts (Chairman) (UK)

Dr Rodrigo Andolfato(Brazil)

Dr Michael Griffith (Australia)

Professor Wolfram Jäger (Germany)

Professor Paulo Lourenço (Portugal) Professor EmeritusRG Drysdale (Canada)

Vol 30, No 1 2017/18

Contents Society News

Secretary Contact details ii

AGM 2017 ii

10 IMC ii

Student Awards 2017 ii

Corporate Member News

Lucideon – Guide for On-site Inspecting and Testing of Existing Buildings. ii

– Digital Image Correlation in Masonry Research. iii

Brick Development Association – The UK Clay Brickmaking Process iii

Catnic – A Fabric First Approach (with reference to steel lintels) iii

International ReviewMasonry Calendar ivPublications – BRE – Understanding the factors affecting flashover of a fire in modern buildings iv

British and European Standards v

JournalConstruction Practice and Structural Performance of Yemeni Traditional Minarets A. H. AL-JOLAHY, A. H. ALWATHAF and A. A. AL-MANSOUR 1

Simple Homogenisation Model for the Non-Linear Analyses of 3D Masonry StructuresE. BERTOLESI, G. MILANI 13

A Comparative Investigation into Lime Activated Ground Granulated Blast Furnace Slag as a Sustainable Alternative to Portland Cement in Masonry MortarsS. HETHERINGTON 23 Cover pictureMasonry building on the Temple Meads Castle Park waterway. North Quay, Bristol

The International Masonry Society, c/o British Ceramic Research Limited, Queens Road, Penkhull, Stoke on Trent ST4 7LQ United Kingdom

POSTAL ADDRESSES OF THE SOCIETYThe main postal address of the Society is that of the Secretary.

Secretary: Dr K Fisher, Shermanbury, 6 Church Road, WHYTELEAFE, Surrey, CR3 0ARAccounts and Administration: Mrs J C Wood MAAT, 26 Widecombe Road, Birches Head, Stoke-on-Trent, ST1 6SL.

Tel: +44 (0) 1782 279051; Email: [email protected] Editor: Dr A N Fried

Comments expressed in this publication are those of the authors and not necessarily those of the International Masonry Society.Data, discussions and conclusion developed by the authors are for information only and are not intended for use without

independent substantiating investigation on the part of potential users.

i

Society News

ii

Secretary Contact details – members wishing to contact the Secretary should please note a change of email address.Emails to Dr Ken Fisher, Secretary, should now be sent to either of the following:[email protected] or [email protected] Student Postgraduate Award Entriesshould be sent to either of the above email addresses by midnight 1st December 2017

Society AGMThe Society Annual General Meeting will take place on Thursday 21st September 2017 at the new conference and meeting space of the Institution of Structural Engineers (ISE).Situated at 47-58 Bastwick Street, in the heart of London, it offers state of the art facilities, in a modern event facility.The AGM will take place in the Auditorium, commencing at 11am, followed by a visit and description of the ISE facility.Following a buffet lunch (member expense) it is planned to hold a series of presentations covering student awards and presentations of current work, when corporate and personal members are invited to provide input.Meeting details have been sent to all members and are available on the website.

10th International Masonry Conference (10 IMC)This series of International Masonry meetings was established in 1986 as part of the International Calendar of Masonry Conferences.Co-organised by the Politecnico 0f Milano and The International Masonry society, the event will take place in Milan, Italy, on 9-11 July 2018.The conference website is available on www.10imc.com and provides up-to-date information on the conference, including details regarding submission of papers, and registration fees, with a reduced fee available to IMS members.To date over 300 abstract submissions have been received for consideration at the Conference.Reduced IMS member and Non-member rates are available on early payment until April 16th 2018.Conference organisers are moving to obtain Scopus indexation for 10 IMC. A book of Conference papers and USB is included in the delegate fee, with an electronic version of the conference proceedings being transferred to the Society website archive for future access and interrogation by members.

Student Project Awards 2017The entry forms for the fourth International Masonry Society Postgraduate Student Awards are available on the society website, www.masonry.org.uk

Entries should be submitted to the Secretary, email to [email protected] may be interested to note that the paper ‘Simple homogenisation model for the non-linear analyses of 3D masonry structures’ by Dr Bertolesi and Professor Milani published in this issue was awarded the 2016 Postgraduate Prize. This continues the policy of publishing Award winning entries in the journal.

Corporate Member NewsPublications: - LucideonA Guide for On-site Inspecting and Testing of Existing BuildingsLucideon has released a new guidance document ‘Top Ten Tips for On-site Inspection and Testing of Existing Brickwork Buildings’. The guide covers some of the most important points to consider when planning on-site inspection and testing programmes.Redveloping existing buildings is an important and valuable operation to bring old buildings, often with historical significance, back into use. Inspection and testing is crucial to ensure the safety and suitability for a change of use. The data provided helps to inform the project as a whole and confirm that the building can be redeveloped to the highest standard.Author of the guide, Dr Geoff Edgell, director and principal consultant for Lucideon, notes:‘On-site inspection and testing prior to the redevelopment of buildings can help to make certain the project is a success. The planning that goes into the initial stages is vital, and the more information one has, the better one can design and manage a project.‘The guide provides a step-by-step insight into the best approaches to planning on-site investigations. There are several key elements that need to be considered when developing these tests, not only to ensure all relevant information is gathered, but also to reduce the time and cost of the project. Effective and efficient on-site testing should be at the heart of every good project’.Lucideon provides on-site testing services to a wide range of construction sector clients, for redevelopment of historic buildings through to construction disputes on new builds. Lucideon has dedicated, large-scale structural testing facilities in the UK and USA, and also offers its expertise and cutting edge analytical technologies for on-site testing. Its expert scientists and engineers work to ensure minimal disruption to on-site activities while providing detailed, meaningful data and interpretation that provides real insight.The guide can be downloaded via the Resources contact page on the websitewww.lucideon.com/construction, using the call out box to the right of the page.

New White Paper for Masonry ResearchLucideon has released a new white paper for the construction sector. The paper entitled ‘The Use of Digital Image Correlation in Masonry Research’ is co-written by Dr Geoff Edgell, director and principal consultant for construction at Lucideon, and Dr Cliff Fudge, technical director at H+H UK limited. The paper discusses the use of Digital Image Correlation (DIC) in recent projects on concrete masonry and was initially presented at the 13th Canadian Masonry Symposium, Nova Scotia, Canada on 6 June 2017.DIC is an advanced visual analysis technique which provides insights and understanding into the performance, durability and failure of materials and products when interacting with external forces, weights, strains and stresses. Lucideon has recently used the technique on concrete masonry for the first time with some extraordinary results.Dr Edgell indicated:‘We are always looking to push the limits of construction materials, processes and structures. In order to make this happen, we need reliable data that can be analysed so that we can understand how test specimens are performing and being influenced by applied forces or conditions. DIC allows us to capture real time measurements of strains on individual components or complete structures, and to visualise where these are critical. Additionally, this technology allows us to perform non-destructive testing. As the technique is also non-contacting and non-destructive, it enables testing to be carried out without damaging the materials, products or structures’.Dr Fudge added:‘This is an exciting new development in the field of masonry and other related materials. We have commissioned the technique with Lucideon on a number of occasions and it has enabled us to successfully understand the stresses and strains that take place within materials under load, which hitherto has not been possible. This has led to improvements in design guidance and analysis of structures, giving a greater understanding of how loads are transferred and providing us with information to enable a balance of safety and construction costs’.Lucideon’s investment in DIC technology allows the company to offer advanced analysis, through non-destructive testing, to identify how materials and structures perform under applied stresses and strains. The mobile technology can be used on-site or in Lucideon’s structural testing laboratories in the UK and USA.The White Paper can be downloaded on the the website www.lucideon.com/construction via the Resources panel on the right hand side of the page.

Brick Development AssociationPublication: The UK Clay Brickmaking Process: March 2017

This document has been produced to provide an insight into the various processes and methods employed by the UK Brick Industry in the manufacture of clay bricks.

The contents provide a Process Overview, and detail the key stages:Raw materialsClay preparationForming methodsColours and texturesDryingKilns and firingPacking and distributionProduct quality and compliance

Clay bricks have featured as a construction product for thousands of years with evidence of their use pre-dating the Roman Empire. This product is prevalent across the UK’s built environment today and continues to be a fundamental ingredient in modern architecture.Bricks are a versatile and sustainable building material, which when combined with good building-design, provide the following benefits: 1. Highly durable 2. Offer long-term performance 3. Low maintenance 4. High thermal mass 5. Reusable and recyclable 6. Provide healthy and comfortable environs

The publication can be downloaded under Technical Guides at www.brick.org.uk

Catnic – A Fabric First ApproachFabric First – Steel Lintels -Wall Detailing – Thermal Bridging

Anyone with an eye on Fabric first, or indeed steel lintels specifically, will not have failed to notice the new generation of lintel designs to hit the market. In this news item, Catnic Technical Director, Richard Price, assesses their impact and considers how and why to specify the best lintels for optimal performance.

Interestingly the heat lost from the UK’s average building stock has been reduced by 23% since 1970 as a direct result of the efforts to address and improve building energy efficiency (Palmer & Cooper, UK Housing Energy Fact File, 2013) in part due to changes in lintel design.

According to the Energy Saving Trust, about a third of all heat lost in an uninsulated home, escapes through the walls. Because heat always flows from a warm area to a cold one, surfaces that facilitate this represent weaknesses in the building’s design. Known as thermal bridges, they not only result in colder indoor environments, interstitial condensation and issues with damp, but also increased heating and cooling demands, energy costs and CO2 emissions. They are the building specifier’s nemesis when it comes to the increasingly complicated demands of the Fabric Energy Efficiency Standard, Part L and SAP Compliance.

iii

iv

This means wall detailing and the elements of the wall that represent the potential for thermal bridging must be carefully considered in order to achieve the optimum energy performance for the building. The thermal performance of a building is calculated by taking into account heat loss through the fabric of the building itself, such as walls, roof, floor, doors and windows, and through linear thermal bridges found at junctions between different elements of the building including window heads, jambs and cills.

With more than 10,000 lintel variations to select from, the correct choice of lintel depends on a variety of factors including building location, construction type, loadings and the required thermal performance. Unquestionably there will always be more than one right answer which leaves the building designer with the conundrum of balancing psi values against structural performance criteria, cost and practicalities of use.

Design performance can only be achieved in practice if the products are installed correctly even with the new generation of lintels. Considering all steel lintels have to be CE marked, in accordance with BS EN 845-2: 2013+A1: 2016, it is easy to see how the specifier may simply opt for a generic ‘equal or equivalent to’ specification when it comes to the new generation of ‘thermally broken lintels’ on the market.

However, there is a solution that offers a simple specification choice, marrying structural excellence with the performance criteria requirements of Appendix Q in SAP 2012, providing easier compliance with Part L and reducing heat lost through window head details by 96%. This is achieved through using a thermally broken steel lintel that delivers an independently verified energy transmittance psi value of 0.02 to 0.05 W/m.K available in a composite, site-practical design.

Safe working loads are vital to building integrity. For lintels EN 845-2 and EN 846-9 are the critical compliance standards. Reassessing loads over lintels according to the building type and materials is time-consuming. Therefore a lintel solution that comprises safe working loads within these parameters offers a simple conversion for specifiers and design engineers. Site readiness and ease of use and handling play a part too. With no need for propping onsite thanks to their composite design, such lintels can be installed in the same way as standard cavity wall lintels.

Each of these features represents time, cost and importantly, energy savings for specifiers. By ensuring thermally broken lintel selection encompasses structural performance, thermal performance and ease of use, the most robust specification for cavity wall design and performance will be achieved.

International Review

Masonry Calendar2017Sept 21. IMS AGMOct 10-12. UK Construction Week, NEC, Birmingham, UKOct 23-26. London Build 2017, LondonNov 19. BDA Brick Awards, London

2018Feb 11-14. 10th Australian Masonry Conference, Sydney, AustraliaApr 10-23. Ceramitec, Munich, GermanyJul 9-11. 10 IMC, Milan, ItalySept 13-15. 11th International Conference in Structural Analysis of Historical Constructions, Casco, Peru

2019Jun 16-18. 13th North American Masonry Conference, Salt Llake City, Utah, USA

BSI – Updated StandardsThe following standards of interest to readers are available: BS 8414-1: 2015 + A1: 2017 Fire performance of external cladding systems. Test method for non-loadbearing external cladding systems applied to the masonry face of a building

Amendment 1

Corrigenda to British Standards

BS EN 772-5: 2016 Methods of test for masonry units. Determination of the active soluble salts content of clay masonry units

Corrigendum 1

BS EN 1052-2: 2016 Methods of test for masonry. Determination of flexural strength

Corrigendum 1

British and European Standards

BRE have recently issued the following publication which may be of interest to readers:

Information Paper IP2/17 Understanding the factors affecting flashover of a fire in modern buildings

Author: Richard Chitty

Published April 2017. Price £18.00

ISBN 978-1-84806-468-3

Theoretical and experimental investigations involving flashover have been ongoing since the 1950s, initially using small enclosures and scale models to identify the basic mechanisms in the context of domestic buildings, cellular offices and similar sized enclosures using wall linings and building contents typical of the day. While this work is still relevant today for similar enclosures, different mechanisms may be involved, or become dominant, especially in very large and/or highly insulated spaces encountered in modern buildings.

This Information Paper provides an outline of the process of flashover and the factors that influence its occurrence and development. It will be of value to fire risk assessors, fire safety managers, and other safety managers and building managers with some responsibility for safety, as well as community fire safety officers, fire fighters and fire investigators. Data in this document will also provide a useful summary for those who are new to fire safety engineering.

Ordering details for this publication are available at www.brebookshop.com

Publications

v

vi

Construction Practice and Structural Performance of Yemeni Traditional Minarets

A. H. AL-JOLAHY (1), A. H. ALWATHAF (1) and A. A. AL-MANSOUR (2)

Department of Civil Engineering, Sana’a University, Sana’a, Yemen (1) Associate Professors, (2) Structural Engineer

[email protected], [email protected], [email protected]

ABSTRACT Up-to-date, masonry minarets with traditional architectural style are still, the most popular type of mosque minarets in Sana’a, the capital city of Yemen, and surrounding areas. Their design and construction principles are based on conventional practice. Unfortunately, this practice lacks formal documentation, i.e. technical guidelines, and structural assessment of the minaret performance against various loading conditions. This paper introduces the construction practice of the Yemeni traditional minarets and investigates their structural performance analytically under gravity and seismic loadings using a three-dimensional (3D) Finite Element (FE) model. Based on a field survey, a typical minaret, 41 meters high built with a mix of local stone and kiln-baked brick masonry, was selected to serve the purpose of this study and to demonstrate the features of the construction materials. A linear-elastic dynamic analysis is conducted for the minaret model utilizing spectra characteristics of the Uniform Building Code (UBC 97 [25]) for the specific seismic zone. The analysis results indicate the dynamic response of the minaret in terms of deformations, shape modes, and stresses. The most vulnerable regions in the minaret and susceptibility to damage are also predicted. KEYWORDS: masonry minaret, FE modelling, dynamic analysis, seismic response.

1. INTRODUCTION

Traditional mosque minarets, both historic and contemporary, form a significant character of the urban landscape of Sana’a, the capital city of Yemen. They are masonry structures of either cut stone, kiln-baked brick, or a combination of both. Cement-sand mortar is used in the construction of the contemporary minarets instead of old clay based mortars. Like tower structures, these minarets are supported on a rigid foundation and composed of two main components, the base and the shaft. The base is square and built on the foundation at the ground level while the shaft is mainly constituted of two or more vertical segments, which are polygonal and/or cylindrical, separated by one or two cantilevered circular balconies. The upper segment generally has a smaller diameter and thinner wall than the lower one and is crowned by a hemispherical dome. An interior spiral staircase links the central core/column to the minaret exterior walls and serves as a continuous brace from the bottom to the top of the minaret. At the present time, traditional minarets are the most common type of mosque minarets in the city of Sana’a. They are preferred to reinforced concrete due to their unique architectural style, historical values and adequate performance. Locally, it is believed that clay-brick and stone masonry offer better adaptability features and more stable structures. For example, historical minarets aged 400-500 years are still functioning and serving their purposes [1]. In masonry structures, damaged units or parts

could be removed and easily replaced keeping the original structure intact while extending the operational longevity of the minarets. Traditional minarets are designed and built in Yemen based on local traditional practice rather than an application of engineering principles, the strength and stability of these minarets resulting from using local construction materials and skilled builders. Introducing cement-sand mortar, contemporary construction tools, and new ways for improving the production and preparation of construction materials to the local market allowed builders to construct higher masonry minarets, up to 60 meters, with thinner walls and more accurate vertical alignment than before [2]. It is worth mentioning that Yemen has a history of destructive earthquakes, the most recent one occurred in December 1982. It hit near the city of Dhamar, about 70km south of Sana’a, with a magnitude of 6 on the Richter scale, and a maximum perceived intensity of Vlll (severe) on the Mercalli intensity scale. It killed and injured more than 15,000 people and destroyed about 1,500 settlements, being one of the deadliest earthquakes to hit Yemen [3]. No damaged or collapsed, traditional or new masonry minarets, were reported or seen in the city of Dhamar [4]. Given the “moderate” seismicity of Yemen [5], the significant damage (excluding minarets) that occurred during the 1982 earthquake may reflect to the poor construction of affected masonry residential buildings [6]. Traditional Yemeni minarets are a part of Yemeni architecture and cultural heritage, being important historically [1,7]. A recent study on such minarets, “Minaret Building and Apprenticeship in Yemen” [2] added a significant dimension to previous studies. It provided a technical description of their construction together with a wide view on the expertise of traditional builders who specialized in their construction. Nevertheless, the structural performance of these tower minarets has not yet been assessed. It is important to conduct such an investigation, taking into account the speciality of the unique construction style of Yemeni minarets, the nature of local construction materials, the expertise of Yemeni traditional builders, and the seismicity of Yemen. Thus, the main aim of this study is to provide a clear view on the structural performance of the Yemeni traditional minarets, against gravity and seismic loadings and to link the results to the current construction practice. The outcome of this study will generally assist in identifying any structural deficiencies and weaknesses in this type of minaret, providing an opportunity for proper correction or modification. It will also contribute to the production of technical guidelines for the design and construction of traditional Yemeni masonry minarets. Based on a field survey undertaken to explore their construction practice, a representative minaret of 41 meters height, built from a combination of natural stone for the base and kiln-baked brick masonry for the shaft with cement-sand mortar, was selected for the analytical investigation. SAP 2000 software is used for the 3D FE modelling and analysis for the typical minaret under seismic loading.

Journal

Journal of the International Masonry Society Masonry International Vol 30. No 1. 2017 1

This paper is organized into six main sections: introduction, construction practice of Yemeni traditional minarets - an overview, analytical investigation, analysis results and discussion, assessment of local construction practice in the light of the analytical analysis results, and conclusion and recommendations. Figures 1(a) and 1(b) show images of a historical mosque minaret in the Old City of Sana'a and a contemporary mosque minaret in Sana'a and neighbouring cities respectively.

2. CONSTRUCTION PRACTICE OF YEMENI TRADITIONAL MASONRY MINARETS, AN OVERVIEW

Historically, traditional masonry minarets are built in Yemen based on local requirement and experience. The construction is guided by the master builder through instructions given on site to fulfil architectural and structural requirements. Prior to construction and based on the minaret’s intended height, the geometry and dimensions of the minaret components are identified together with available construction materials suitable for each component of the minaret. However, there is no local guideline, either formal or informal, for the design and construction of traditional masonry minarets. Taking into account construction information provided by the recent study “Minaret Building and Apprenticeship in Yemen” [2], it was deemed necessary to conduct the field survey of several minarets in Sana'a and to meet the master builder, Mr. Almaswary, to obtain a broader overview on the minarets construction practice. The survey included identification of geometrical properties, types and nature of construction materials and structural systems. During the survey, dimensions of various parts of minarets of different typical heights (30, 40/41 and 60 meters) were measured using a measuring tape. Also, checks for out of plumbness (deviation from verticality) were made for the minaret of 1 meters height using surveying equipment, Total

Station-Lica type. In addition, physical and mechanical properties of material used in the construction of the minarets were identified, based on test results obtained from works carried out by the authors and others [8-11]. 2.1 Geometrical properties Generally, traditional masonry minarets are built up to 60 meters in height, most commonly being 30 to 40 meters high. Based on the field survey undertaken by the authors Table 1 shows the geometrical properties for minarets of height 30, 40 and 60 meters, whilst Figure 2 illustrates the geometrical details for the minaret of 41 meters height. 2.2 Construction method and details Initially, a suitable location for the minaret is selected within the mosque surroundings to allow maximum appearance and visibility of the minaret and to ensure good soil condition for establishing the foundation. The ground is excavated to a depth of about four to six meters depth until bed-rock or a solid stratum is reached. The excavated hole is then filled in with compact basalt rubble stones and concrete forming a durable large stone footing, impervious to damp and harmful substances. A square base large enough to accommodate an interior circular space, a central core and a surrounding staircase is built directly on the rigid foundation at ground level from natural stone blocks and cement-sand mortar. The first three courses of the base are damp-proof, built from compact basalt stone units, while the remaining courses are built from vesicular basalt (Habash) and/or volcanic tuff (Abasri), these being lighter and softer stones. The exterior faces of the base wall are of cut stones, ashlar type, while the inner faces are of coursed rubble stones. The base eventually forms about 25% of the total height of the minaret and its weight equals to about 50% of the total weight. This massive part of the minaret contributes to the minaret stability and durability.

Figure 1(a) Historical mosque minaret Figure 1(b) Typical contemporary mosque minaret in the Old City of Sana’a – Yemen in Sana’a and surrounding cities – Yemen

2 Journal of the International Masonry Society Masonry International Vol 30. No 1. 2017

A. H. AL-JOLAHY, A. H. ALWATHAF and A. A. AL-MANSOUR

Figure 2 Geometrical properties of a typical 41 meters height minaret (dimensions in meters)

Table 1 Minaret components: geometrical properties

Components Material type

Dimensions (m) Wall thickness

(m)

Minaret total Height (m) Remarks Length Width

Height Diameter

Underground foundation

compact basalt

stone + concrete

5.5 5.5 4-6 - 60 5.0 5.0 4-6 - 40 4.0 4.0 4-6 - 30

Square base above ground

vesicular basalt and tuff stones

4.5 4.5 15 1.0 60 4.0 4.0 10 0.85 40 3.0 3.0 7.5 0.70 30

Cyl

indr

ical

sha

ft lower part Kiln-baked

clay brick masonry

Dext = 4.30 30 0.90 60 Dext = 3.80 20 0.75 40 Dext = 3.00 15 0.60 30

upper part

Dext = 3.90 15 0.70 60 Dext = 3.50 10 0.60 40 Dext = 2.70 7.5 0.45 30

Inte

rior c

ircul

ar s

pace

space Dint = 2.50 60 - 60 Dint = 2.30 40 - 40 Dint = 1.80 30 - 30

central core tuff stone

D = 0.65 57 - 60 D = 0.55 37 - 40 D = 0.50 27 - 30

stair timber + stone

0.92 0.25 0.25 - 60 0.85 0.25 0.25 - 40 0.65 0.25 0.25 - 30

Dext = Exterior diameter, Dint = Interior diameter



Table 2 Material properties: masonry units

Stone type Density (Kg/m3)

Absorption %

Compressive strength (N/mm2) Element

Compact basalt (Aswad) 2900 0.6 155.0 Foundation

Vesicular basalt stone (Habash) 2120 3.2 37.0 Base, exterior wall

Tuff stone (Abasri) 1550 17.0 17.0 Base, interior wall

Kiln-baked clay brick 1200 31.0 3.5 Shaft, balcony, and dome

The shaft consists of two main vertical parts, of either cylindrical or polygonal geometry, separated by a cantilevered circular balcony, and capped with a hemispherical dome at the top of the upper part. It is constructed from kiln-baked clay brick masonry which is much lighter in weight and more adaptable than stone. The shaft height forms the remaining 75% of the total height of the minaret and its weight equals to about 50% of the total weight. A circular balcony constructed above the lower part forms a circular platform with about 0.35 meter projection and a meter height that allows the upper part of the shaft to be constructed with a wall thickness different from the lower part without interruption to the architectural view and structural system. Stairs are built, within the interior circular space, from natural stones and supported by timber beams spanning between the central circular column and minaret exterior walls. Thus, the staircase provides lateral bracing and integral vertical connection for the consecutive parts of the minaret throughout the entire height of the minaret fulfilling structural integrity. As stated by the master builder and noticed on the site [2], all practical and technical measures were taken during the construction process to ensure good execution for each component as well as the whole structure. However, the out of plumbness was checked on site by the authors and found to be 38mm for the 41 meters height minaret (one mm for a meter height), which is within permissible tolerance [12]. 2.3 Material properties Material properties for the various elements of the prescribed minarets are shown in Table 2 for masonry units and in Table 3 for masonry specimens. Compressive strengths and elastic moduli for both natural stone and kiln-backed clay brick masonry used for the construction of the minaret base and shaft were taken from available values determined from compression tests on stone [8] and brick wallette specimens [9]. The masonry wallette specimens were prepared and tested in accordance with relative EN Standards (EN 1052-1:1999) [13]. Tests were also made on stone [10] and brick units [9] and cement sand mortar specimens [11,14]. It is worth mentioning that several tests on natural stone, kiln-baked clay brick and concrete block masonry works [14] were carried out at the Faculty of Engineering, University of Sana’a, Republic of Yemen for postgraduate study and research purposes.

3. ANALYTICAL INVESTIGATION 3.1 Analytical studies review In spite of the distinctive architecture and the noticeable conventional construction practices that are still widely used in the Yemeni minarets, the minarets structural performance

due to the seismic response has not been explored yet. A number of historical minarets in Yemen have survived for hundreds of years, even though frequently exposed to environmental loads, such as winds and earthquakes [1]. This factor has further prompted the Authors investigate the structural performance and assess construction practice of the Yemeni traditional minarets. A Literature review indicated that most previous studies on minarets were conducted in Turkey, which is rich with Islamic heritage [15-17]. These studies addressed typically the dynamic response of minarets against earthquake loading in terms of deformations, shape modes, and stresses indicating the most vulnerable locations to damage in minarets. The FE technique was used in the modelling and analysis of the minarets with different heights and construction materials. The modal analyses of the models showed that the structural periods and the overall structural responses are influenced by the minaret height and spectral characteristics of the input motion. The maximum dynamic internal force demands were compared with calculated capacities. They found that most minaret failures occurred above the base and the largest stresses were calculated at the same location, which is consistent with the earthquake damage observed at those critical locations. Additional field investigations and seismic analyses of historical masonry minarets have been conducted [18-21]. FE modelling was employed to determine the seismic behaviour of the masonry minarets and to illustrate the damage sustained during earthquakes which hit Turkey recently. The damage, displacements, maximum and minimum principal stresses and strains are obtained from the analyses and compared with field observations. 3.2 Finite Element modelling In this study a 3D FE model has been developed to study the dynamic response of traditional minarets. Such minarets in Yemen are commonly built to heights of about 30 to 40 meters, and very rarely outside that range. In this research, a typical minaret of 41 meters height with a single balcony (Figure 2) was selected for FE modelling and analysis. The computer program, SAP2000 was used to generate and analyse the minaret model with SOLID and SHELL elements. The solid element is an eight-node element for modelling 3-D structures and solids (Figure 3(a)). It is based on an isoparametric formulation that includes nine optional incompatible bending modes. The shell element is a three- or four- node formulation that combines membrane and plate- bending behaviour (Figure 3(b)). The membrane behaviour uses an isoparametric formulation that includes translational in-plane stiffness components and a rotational stiffness component in the direction normal to the plane of the element [22,23].



(a) SOLID element (b) SHELL element

Figure 3 SOLID and SHELL elements used in the model [18,19]

Table 3 Material properties: masonry specimens and timber beam

Element Material Density kg/m3

Poison’s ratio

Compressive strength (N/mm2)

Elastic Modulus (N/mm2)

Base Vesicular basalt

+ Tuff-stone masonry

1800 0.17 4.3 3300

Shaft Kiln-baked clay brick masonry

1200 0.15 2.2

1000

Stair support Timber 600 0.15 - 10000

As shown in Figure 4, the 3D FE model includes all components of the minaret. The base/shaft walls, core (central column), and balcony were modelled with SOLID elements whilst the interior spiral stair and top dome were modelled with SHELL elements. The openings for the minaret entrance door at the minaret base and for the windows at the top of the minaret shaft (Figure 2) were also incorporated in the model to get accurate simulation and results. Due to the continuity and good bond between the minaret base and the footing, the footing - base connection at the ground level was defined fully restrained and no soil-structure interaction was considered. The FE model, was based on linear elastic material behaviour and the stiffness degradation, softening, and hardening of materials were neglected. A material model approach was assumed with a homogeneous isotropic continuum neglecting the difference between mortar joints and brick units, as per the macro-modeling technique [24]. The material properties of all components are shown in Tables 2 and 3. As described earlier in section 2.3, the defined properties (Table 3) for the stone base and clay-brick shaft were derived from tests on masonry wallettes of similar materials and construction forms [8,9]. For the stone base, the walls consist of an outer leaf (ashlar stone) and an inner leaf (rubble stone) with a core of crushed stones and mortar [8]. The masonry wallette properties are the average of masonry units and mortars in combination. In addition, the timber properties were

assigned to the stair elements that linked the walls of the base and the shaft to the central core. 3.3 Dynamic analysis attributes Design response spectrum for purposes of seismic analysis and design of structures has not been developed yet for Yemen. However, the nature of earthquakes that occurred in the past and may take place in the future in the country can be understood from limited existing seismology studies [3,5]. A recent probabilistic seismic hazard analysis study for Yemen shows that the Peak Ground Acceleration, PGA, for a 10% probability of exceedance in 50 years (return period 475 years) ranges from 0.2g to 0.3g in the western part of Yemen and generally is less than 0.05g across central and eastern parts [5]. Accordingly, response spectrum dynamic analysis on the minaret model was carried out using the available seismic information for the studied area, Sana’a and surroundings, and corresponding seismic characteristics related to the site and structure as set forth in the UBC 97 [25]. Based on the above-mentioned seismic hazard study [5] and UBC provisions, the studied area is located in a moderate seismic region (2B), and type SC soil class is considered to simulate the site condition. The damping ratio of 5% is assumed and the over-strength and global ductility capacity of the resisting system, R, is taken as 2.9. The design response spectrum determined for the specified zone and soil type is presented in Figure 5.



Figure 4 The developed 3D FE model

0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0 1 2 3 4 5 6 7 8 9 10 11

Accleration, g's

Period, s

Figure 5 Design response spectrum



4. ANALYSIS RESULTS AND DISCUSSION The results obtained from the analysis of the FE model are presented for mode shapes, lateral displacements and principal compressive and tensile stresses. 4.1 Mode shapes Figure 6 shows the mode shapes obtained; in X-direction (modes 1 and 3) and Y direction (modes 2 and 4), as well as torsional mode (mode 5). As seen from the figure, flexure is the dominant behaviour of the minaret structure. The calculated rst or fundamental period of the model, mode 1 in the X-direction and mode 2 in the Y-direction, were 1.579s and 1.573s respectively. Given the symmetry of the minaret, the very small difference between the fundamental periods, T1 and T2, is due to the presence of the minaret entrance opening. It can be noted that the contribution of mode 1 in the earthquake direction X (or mode 2 in the earthquake direction Y) to the dynamic response was significant comparing to other modes. On the other hand, the torsional or the fth mode with a period of 0.192s had nearly no effect on the total response of the minaret structure, due to the symmetry of the minaret geometry. 4.2 Lateral displacement The lateral displacements along with the height of the minaret are shown in Figure 7 for both sides of the

minaret in the excited X-direction, as positive and negative values. The deected shape of the minaret confirms the exure-dominated response with the largest displacement calculated at the top. Although the minaret acts as a cantilever, the deformations are much smaller over the height of the relatively stiff 10 meters high base. The displacements start to increase significantly above the base segment at the bottom of the shaft from 9mm up to a maximum of about 266mm at the top of the minaret. However, UBC 97 [25] gives no limit for lateral displacement in the case of these type of structures; it states that drift limitations given by the code shall be established for structural and non-structural elements whose failure will cause life hazards. Based on the UBC provisions for building structures, the elastic lateral displacement limit (∆max) for structures having a fundamental period (T ≥ 0.7s) can be obtained from the following relation: (∆max ≤ 0.02h/R), where h and R are the height and the ductility factor of the structure respectively. Assuming this relation is applied to the studied minaret, with h = 41 meters and R = 2.9, as an indication of the extent of lateral displacement for a cantilever structure; the corresponding elastic displacement limit is 282mm. Hence, the elastic displacement response of 266mm is below the code limit.

Mode1 Mode 3 Mode 2 Mode 4 Mode 5 T = 1.579 s T = 0.326 s T = 1.573 s T = 0.323 s T = 0.192 s

X-direction Modes Y-direction Modes Torsional Mode

Figure 6 Mode shapes

X

Z

Y

Z



Figure 7 displacement 4.3 Stresses The most critical normal stress contours predicted by the dynamic analysis of the FE model of the minaret are shown in Figures 8 and 9 for compression and tension stresses respectively. It can be seen from the gures that maximum compressive and tensile stresses occur at a height of 10 meters, immediately above the base in the bottom of the shaft. This result is consistent with several studies carried out on masonry minarets in Turkey, where the base-shaft junction is the most common location for minaret failures observed during the past earthquakes [15-18,21]. The maximum obtained axial compressive and tensile stresses were 2.5N/mm2 and 1.4N/mm2 respectively. Such stresses are relatively low compared to those obtained from similar studies on minarets in Turkey. These differences are understandable due to the dissimilar earthquake natures and construction practices of Yemen and Turkey. The structural system of the traditional Yemeni minarets is composed of two vertical elements, the base/shaft walls and the central core. The stairs link the two elements, and as a result enhance their interaction and resistance to the imposed loadings. The cross-sections of Figures 8 and 9 show the stress distribution over the shaft walls and the central core at the most critical section. However, the maximum compressive stress of 2.5N/mm2, at the extreme fibre of the outer bricks of the shaft (Figure 8 – the cross section at the base-shaft junction), is higher than the brick masonry strength of 2.2N/mm2 (Table 3) by about 14%. On one hand, it might be said that the exceeded compressive strength would not cause damage at that part of the shaft section, given that the actual material strength is almost certainly higher than the laboratory value (2.2N/mm2), due to a sort of material confinement within the structure (caused by lateral deformation restraint of the material within the shaft circular cross section) as well as material response in dynamic rather than static or quasi-static conditions [26]. On the other hand, the maximum predicted tensile stress of 1.4N/mm2, at the outer edge of the shaft (Figure 9 – the cross section at the base-shaft junction), is much higher than the tensile strength of masonry material which could be assumed between 10% to 20% of the masonry compressive strength [27], say 0.4N/mm2. Actually, the tensile stress appears to be more critical than the compressive stress and may be capable of initiating damage at the outer edge of the shaft

section. The damage could have the form of crack openings in the outer mortar-brick joints/interfaces or cracks in the brick and mortar materials. As noticed, however, the FE model is efficient in predicting the sections of the minaret most susceptible to damage. This provides invaluable data on how to treat and protect weak sections.

5. ASSESSMENT OF TRADITIONAL CONSTRUCTION PRACTICE IN VIEW

OF ANALYTICAL ANALYSIS RESULTS The local traditional construction practice, as previously described in section 2.2, requires a soil with good properties, i.e. a bed-rock or solid-stratum, for establishing the rigid stone foundation below the ground to support the structure and to eliminate differential settlement. In this study, type SC soil was defined for the analytical analysis to simulate the practice requirement as categorized in the UBC 97 [25]. The type SC soil represents very dense soils and soft rock. In addition, the Syrian Arabic Code [28] indicates that, for simplicity, Sc soil type could be defined for analysis when the allowable bearing capacity of the soil is not less than 250kN/m2. In order to account for a construction case with soil of lower properties than of the considered type, SC, further analysis using the FE model was carried out using SD soil type (stiff soil). The design response spectrum based on the two types of soil SC and SD are shown in Figure 10 and the dynamic response is given in Table 4, as lateral displacement, at the top of the minaret, and maximum compressive and tensile stresses. As indicated in Table 4, SD soil shows higher dynamic response than SC. The displacement is increased by 15% and the compressive and tensile stresses are also increased by 20% and 32% respectively. As per UBC 97 [25] provisions for earthquake analysis, the design spectrum for a specified earthquake zone is given as a function of the coefficients of acceleration (Ca) and velocity (Cv) which vary with soil categorizations. The UBC 97 [25], categorises soil profiles as six types: 1) SA : Hard Rock, 2) SB : Rock, 3) SC : Very Dense Soil and Soft Rock, 4) SD : Stiff Soil Profile, 5) SE : Soft Soil Profile, and 6) SF : Soil needs site evaluation. Therefore, as the soil type gets lower, so the effects of earthquake loading become higher and imposes considerable dynamic effects on the structure.



Figure 8 Distribution of maximum compressive stresses (N/mm2)



Figure 9 Distribution of maximum tensile stresses (N/mm2)



Figure 10 Design response spectrum for Sc & Sd Soil types

Table 4 The effect of soil type on the dynamic response

Dynamic Response Soil Category (SD-Sc)/SC

% SC SD Displacement (mm) 266 306 15%

Compressive Stress (N/mm2) 2.50 3.00 20% Tensile stress (N/mm2) 1.40 1.85 32%

In local practice and prior to the construction, the geometry and dimensions of the minaret components are identified based on the minaret’s intended height together with available construction materials suitable for each component of the minaret. The selected minaret for this study has a natural stone square base with height and weight equal to about 25% and 50% of the total of the minaret respectively, while the clay brick circular shaft forms the remaining, i.e. 75% and 50% of the height and the weight. This arrangement makes the weight per meter height ratio of the shaft to the base as 1:3 and would be expected to reduce the effects of earthquake forces over the height of the shaft. The analytical results (Figures 8 and 9) show that the largest compressive and tensile stresses occurred in the bottom of the shaft at the junction with the base for which a partial opening in the brick-mortar joints may occur due to the masonry tensile strength being exceeded. The stress concentration phenomenon occurs at the junction where the shaft section reduces and rigidity decreased compared to the base section. Interestingly, similar traditional masonry minarets, in Dhamar city in Yemen, performed well during the 1982 moderate earthquake and no damages were noticed [4]. Such evidence indicates that the analytical solution, using the design response spectrum method, is relatively conservative and showed the most critical situation for minarets structure during the considered earthquake. However, both the results of the analysis and the earthquake incident showed there is no overall collapse in the minaret. Nevertheless, the minaret and loading are typical.

6. CONCLUSIONS AND RECOMMENDATIONS This study reviews and presents construction practice for the traditional Yemeni minarets and investigates their

structural performance analytically under gravity and lateral loadings using a 3D FE method using the SAP2000 software. Based on the results, the conclusions are summarized hereinafter. The traditional Yemeni minaret with a stiff base and slender shaft acts as a cantilever structure under different loadings. The interconnection and interaction between the external walls of the shaft/base and the core enhance the minaret performance against earthquake loading and its subsequent effects. The dynamic response of the minaret to moderate seismic action was flexural-dominant behaviour and with nearly no torsional effect due to the symmetry of the minaret geometry. The lateral displacements are 9mm at the junction of the base and the shaft and a maximum of 266mm at the top end of the minaret. The maximum lateral displacement of 266mm was compared with the corresponding maximum allowable displacement given by the UBC for building structures, as an indication of the extent of lateral displacement for a slender cantilever structure, and found below the code limit of 282mm. The maximum compressive and tensile stresses obtained are 2.5N/mm2 and 1.4N/mm2 and located at the bottom of the shaft at the junction with the base, the most critical section within the structure. The tensile stress is substantial and more critical than the compressive stress, and when compared to the strength capacity of the minaret materials will result in damage/cracks at such a critical section. As well as identifying possible failure of joints, the FE model identifies the most vulnerable sections in the minarets structure and where it is most likely to be damaged. Such knowledge is of great benefit, enabling in practice for possible rectifying or strengthening solutions for the weakest zones which would undergo the greatest damage or complete failure, in the case of stronger earthquakes.



However, the output from the analytical investigation/FE model only presented a preliminary overview of the dynamic behaviour of the Yemeni traditional masonry minarets. The analytical results generally acknowledge the existing local construction practice and provide useful information in terms of mode shapes, deformations and stresses which can be used effectively for risk assessment and for proposing protective measures to assure the stability and safety of this type of minaret structure in case of earthquake events. In addition, such information is also of assistance in the selection of the proper sites for minaret construction. Nevertheless, further research is needed to explicity demonstrate the dynamic behaviour of the traditional Yemeni minarets of different heights, geometries and materials. Non-linear analyses will further improve predictions of behavior in particular how the soil below interacts with the structure. Accurate analysis will enable the structural deficiencies and failure patterns at the vulnerable regions to be better identified and enable effective strengthening and prevention measures. It is also important to provide enough information to support the development of guidelines for design, construction and possible repair of masonry minarets. REFERENCES 1. SAIF, A.S. Sana’a City Minarets, Ministry of Culture

and Tourism, Sana’a, Yemen, 2004. 2. MARCHAND, T.H.J. Minaret Building and

Apprenticeship in Yemen, Routledge Taylor & Francis Group, 2001.

3. ARYA, A.S., SRIVASTAVE, L.S. and GUPTA, S.P. (1985), “Survey of Damages during the Dhamar Earthquake of 13 December 1982 in the Yemen Arab Republic”, Bulletin of the Seismological Society of America, 75, (2), 597–610, 1985.

4. LOCAL AUTHORITY OF DHAMAR GOVERNORATE, Dhamar Earthquake Damages Report, Dhamar Publishing House, Dhamar, Yemen, 1983.

5. MOHINDRA, R., NAIR, A.K.S., GUPTA, S., SUR, U. and SOKOLOV, V. “Probabilistic Seismic Hazard Analysis for Yemen”, International Journal of Geophysics, 2012, 1-14, 2012.

6. AL-MADHAGI, Y.S. “Structural Requirements for Rehabilitation of Residential Buildings with Load Bearing Stone Walls Damaged by Earthquakes”, Proceeding of Disasters' Management and Safety of Buildings in Arab Countries Symposium, Riyadh, Saudi Arabia, 28 March-1 April, 2008.

7. ABDULLA, Y.M. and TAHER, A. Sana’a – Architecture Design Fundamentals and Urban Planning in Different Islamic Centuries, Organization of Islamic Capitals and Cities, 2005.

8. BELL, A.J., AL-JOLAHY, A.M. and KULAIB, M. “Compressive Strength of Natural Stone Masonry”, Proceeding of the 6th International Masonry Conference, No. 9, London, November, 2002.

9. AL-HABARI, A. The Behaviour of Walls built from Kiln-baked Clay Brick, MSc Dissertation, University of Sana’a, Sana’a, Yemen, 2015.

10. AL-JOLAHY, A. “Normalized Compressive Strength of Natural Stone Masonry Units”, HBRC Journal, Housing & Building National Research Center, Cairo, Egypt, 6 (2), 1-7, 2010.

11. AL-JOLAHY, A. “Strength Characteristic of Natural Stone Masonry (Eurocode 6 and BS 5628 methods)”, HBRC Journal, Housing & Building National of Research Center, Cairo, Egypt, 6 (1), 12-21, 2010.

12. EN 1996-1-1, EUROCODE 6, Design of Masonry Structures, Part 1-1: General Rules for Reinforced and Unreinforced Masonry Structures, European Committee for Standardisation, Brussels, Belgium, 2005.

13. EN 1052-1, Methods of Test for Masonry - Part 1: Determination of Compressive Strength, European Committee for Standardisation, Brussels, Belgium, 1999.

14. AL-AMOUDI, M. and ALWATHAF, A.H. “The Behaviour of Hollow Concrete Block Masonry under Axial Compression”, Journal of Engineering Sciences (JES), 3 (2), 2014.

15. DOGANGUN, A., ACAR, R., SEZEN, H. and LIVAOGLU, R. “Investigation of dynamic response of masonry minaret structures”, Bulletin of Earthquake Engineering, 6 (3), 505–517, 2008.

16. SEZEN, H., ACAR, R., DOGANGUN, A. and LIVAOGLU, R. “Dynamic Analysis and Seismic Performance of Reinforced Concrete Minarets”, Engineering Structures, 30, 2253–2264, 2008.

17. TURK, A.M. “Seismic Response Analysis of Masonry Minaret and Possible Strengthening by Fiber Reinforced Cementitious Matrix (FRCM) Materials”, Advances in Materials Science and Engineering, 2013.

18. KOCATÜRK, T. and ERDOĞAN, Y.S. "Earthquake Behaviour of M1 Minaret of Historical Sultan Ahmed Mosque (Blue Mosque)", Structural Engineering and Mechanics, An Int'l Journal, 59 (3), 2016.

19. BAŞARAN, H., DEMIR, A., ERCAN E., NOHUTÇU, H., HÖKELEKLI, E. and KOZANOĞLU, C. "Investigation of Seismic Safety of a Masonry Minaret using its Dynamic Characteristics", Earthquakes and Structures, An Int'l Journal , 10 (3), 2016.

20. BAYRAKTAR, A., ALTUNIŞIK, A.C. and MUVAFIK, M. "Damages of Minarets during Erciş and Edremit Earthquakes, 2011 in Turkey", Smart Structures and Systems, An Int'l Journal, 14 (3), 2014.

21. MUVAFIK, M. " Field Investigation and Seismic Analysis of a Historical Brick Masonry Minaret Damaged During the Van Earthquakes in 2011”, Earthquakes and Structures, An Int'l Journal, 6 (5), 2014.

22. SAP2000 (a), Integrated Software for Structural Analysis & Design, Computers and Structures Inc., California, 2010.

23. SAP2000 (b), CSI Analysis Reference Manual for SAP2000®, ETABS®, and SAFE®, Computers and Structures Inc., California, 2010.

24. HEATH, D.J., GAD, E.F. and WILSON, J.L. “An Automated Finite Element Macro for the Examination of Cosmetic Racking-Related Damage in URM Walls”, Journal of the International Masonry Society, Masonry International, 29, (1), 1-13, 2016.

25. UBC 97, Uniform Building Code Volume 2, International Code Council, 1997.

26. PARK, R. and PAULAY, T. Reinforced concrete structures, John Wiley & Sons, New York – USA, 1975.

27. GUO, P. Investigation and Modelling of the Mechanical Properties of Masonry, Ph.D. Thesis, McMaster University, Hamilton, Ontario, Canada, 1991.

28. SYRIAN ARABIC CODE for Design & Construction of Reinforced Concrete Structures, Addendum 2: Design & Verification of Earthquake Resistance Buildings & Structures, Engineers Syndicate, Damascus, 1997.



Simple Homogenisation Model for the Non-Linear Analyses of 3D Masonry Structures

E. BERTOLESI (1), G. MILANI (2)

(1) PhD, Department of Architecture, Built Environment and Construction Engineering (ABC), Politecnico di Milano, Piazza Leonardo da Vinci 32, 20133 Milano, Italy

[email protected] (2) Associate professor, Department of Architecture, Built Environment and Construction Engineering (ABC),

Politecnico di Milano, Piazza Leonardo da Vinci 32, 20133 Milano, Italy [email protected]

ABSTRACT In the present paper, a simple homogenisation two-step procedure is proposed for the analyses of both in and out of plane loaded masonry walls and 3D structures. The “unit cell” is discretized with 24 triangular plane constant stress (CST) elements and interfaces. Bricks are specified to behave elastically, whereas mortar joints are set as zero thickness non-linear interfaces. The mechanical response of joints is modelled with different holonomic relationships including two dominant failure modes, namely cracking (mode I) and shear (mode II), or a combination of both (mixed mode). In particular, either a piecewise linear or an improved version of the Xu-Needleman exponential law are used, both exhibiting post peak softening. At the structural level, the homogenisation model is implemented into general purpose commercial FE software, modelling the homogenized orthotropic continuum as a discrete assemblage of Rigid Bodies and Homogenized softening Springs (HRBSM). In such a rigid element model, a variety of springs are introduced, to properly characterize both the in plane homogenized shear and normal behaviour, as well as bending and torque (out of plane behaviour). Moment-curvature relationships are evaluated simply by on thickness integration from the knowledge of in plane homogenized stress-strain relationships. Mechanical properties of homogenized springs are identified via classic energetic identification. Three case studies of technical relevance are finally presented for benchmarking purposes: (i) a windowed shear panel, (ii) a church façade modelled with a portion of the perpendicular walls and (ii) a 3D half scale two story masonry building. KEYWORDS: simplified homogenisation strategy, masonry, discrete model, dynamic analyses. Notation: Vectors and tensors are indicated in bold. E and indicate strain and stress homogenized tensors, ( ,

, ) is the macroscopic horizontal (vertical, shear, on direction n) strain, ( , ) is the

homogenized horizontal (vertical, shear) stress, ( , ) is the local horizontal (vertical, shear) stress on

element k, ( , ) is the local horizontal (vertical, shear) strain on element k, L(H) is the brick semi-length (height), , A is the elementary cell (REV) area, ( , e ) is head (bed, generic) joint thickness, ( ) indicate an imposed boundary horizontal (vertical) displacement in the biaxial strain problem, ( ) is the

i-th node unknown horizontal (vertical) displacement, ( ) is the interface normal (tangential) jump of displacements, ( ) is the joint (I: head, II: bed) normal (shear) stress, , , ( ,

) is the brick Young modulus (Poisson’s ratio, shear modulus), ( ) is mortar Young (shear) modulus, ( ) is the ultimate joint normal (tangential) jump of displacements in the multi-linear model, (c) is joint tensile strength (cohesion), , , and are Xu-Needleman interface parameters, , ,

, ( ) indicate an imposed boundary horizontal (vertical) displacement in the shear problem,

.

1. INTRODUCTION Masonry is a traditional heterogeneous material whose behaviour is characterized by some peculiar aspects, the most important being its orthotropy, brittle behaviour and limited tensile strength. In addition, there is a large variability of the mechanical properties of the constituent materials which are dependent on the local availability of construction materials, especially for historical structures; this makes the simulation of masonry a very challenging task. At present, three numerical techniques are widespread: macro-, micro-modelling and homogenisation. In macro-modelling, bricks and mortar are substituted with a homogeneous orthotropic material exhibiting softening and very low or (vanishing) tensile strength, see e.g. [1-3]. Macro-modelling is frequently adopted to analyse large scale structures even in the non-linear dynamic range, because it does not require the separate discretization of bricks and mortar. When the level of sophistication of the model increases [2-3], to better reproduce masonry’s inelastic features, such as anisotropy, post-peak softening and frictional shear behaviour, the number of inelastic parameters grows and the experimental characterization may become costly and cumbersome. Alternatively, a micro-modelling strategy can be adopted. In such an approach, the distinct modelling of joints and bricks at a structural level is adopted. When joints are meshed with 2D/3D elements, more than one element across the joint thickness is usually needed, with the consequent growth of the number of variables, even for small panels. An efficient alternative is the reduction of joints in interfaces [4-8], but the numerical effort still remains high, especially for non-linear computations.


In such a framework, homogenisation [9-19] may represent an interesting compromise between micro- and macro-modelling, because it allows one, in principle to perform non-linear structural analyses of engineering elements without a distinct representation of bricks and mortar, but still considering their mechanical properties and the actual behaviour at brick and mortar level. Homogenisation is essentially an averaging procedure performed on a Representative Element of Volume (REV), which generates the masonry pattern under consideration by repetition. On the REV, a Boundary Value Problem BVP is formulated, allowing an estimation of the expected average masonry behaviour to be used at structural level. Importantly, the resultant material obtained from the meso-scale homogenisation turns out to be orthotropic, with softening in both tension and compression. A straightforward approach to solve BVPs at the meso-scale is obviously based on FEs [12,16-19], where bricks and mortar are either elasto-plastic with softening or damaging materials. It is also known as FE2 and essentially is a twofold discretization, the first for the unit cell and the second at structural level. However, FE2 appears still rather demanding, because a new BVP has to be solved numerically for each load step, at each Gauss point. In this paper, a simplified two-step homogenisation model is proposed for the non-linear structural analysis of both in- and out of plane loaded masonry. The first step is applied at the meso-scale, where the assemblage of bricks and mortar in the REV is substituted with a macroscopic equivalent material through a so called compatible identification. The unit cell is meshed by means of 24 triangular constant stress (CST) plane stress elements (bricks) and zero thickness interfaces for mortar joints. Triangular elements are assumed linear elastic, whereas the mechanical response of interface elements includes two dominant deformation modes, namely peel (mode I) and shear (mode II) or a combination of the two (mixed mode). Two cohesive relationships are implemented: (i) a piecewise linear and (ii) an improved version of the Xu-Needleman exponential law [20-22]. In the second step the homogenized orthotropic continuum is replaced by a discrete model, indicated hereafter as HRBSM, an acronym standing for Homogenized Rigid Body and Spring Model. HRBSM allows

adopting (for in plane problems) infinitely resistant rigid elements interconnected by shear and normal non-linear homogenized springs. Flexural and torsional homogenized springs are then added to the discrete model in order to suitably reproduce out of plane mechanisms, which are quite common under dynamic excitations. Through simple on thickness integration moment-curvature relationships are evaluated starting from the knowledge of homogenized stress-strain relationships obtained by solving the homogenisation in plane problem. To validate the model proposed, three structural case studies of technical relevance are discussed, namely a windowed shear panel tested up to failure under static conditions, a church façade and a 3D half scale masonry building subjected to dynamic excitation.

2. A SIMPLIFIED COMPATIBLE HOMOGENISATION STRATEGY

Homogenisation is aimed at studying complex masonries by means of averaged quantities, such as the macroscopic strain and stress tensors (respectively E and ) [12,16-18,23] on a representative element of volume Y (REV

or elementary cell, Figure 1), i.e.

dY

A Y )(1 uεεE

and dYA Y σσΣ 1 , where A is the area of the

elementary cell, and are local quantities (strains and stresses respectively) and <*> is the averaging operator. Anti-periodicity is imposed on and periodicity on the displacement field u:

YY

onperiodic-antionperper

σnuuxEu ~

(1)

Where u is the total displacement field, uper is a periodic displacement field, is the local frame of reference (see Figure 1) and E is the homogenized strain tensor.

x

y2H

(2) (1)(3)

REV

14 REV

H

H

a 5-20 cmb 20-30 cm

2L+2ev

Mortar interface

Brick element

(2)(1)(3)

(7)(8)

(9)

(5) (4)(6)

Figure 1 The compatible homogenisation proposed. REV mesh with 24 CST brick elastic elements and holonomic mortar interfaces. Anti-periodicity of the micro-stress field (bottom left)

E. BERTOLESI, G. MILANI


Here, a semi-analytical simplified homogenisation model is proposed for the analyses of running bond masonry patterns. The REV is discretized using 24 constant plane stress triangular elements, whereas joints are reduced to holonomic zero thickness interfaces. In order to reduce variables, two main simplifications are employed: (i) brick-brick interfaces are assumed not active (i.e. the jump of normal and tangential displacements is assumed to vanish) and (ii) all the non-linearity is lumped into mortar joints with zero-thickness interfaces. Indicating

with )(n a stress component belonging to the n-th

element, the stress tensor inside the n-th element (n) is constituted by the components xx(n) (horizontal stress), yy(n) (vertical stress) and (n) (shear). Equilibrium inside each element is automatically satisfied, div = 0, whereas two equality constraints involving stress tensors of contiguous triangles have to be imposed for each internal interface. In particular, for 1-2 interface, the stress vector (normal and tangential component) must be equal passing from element 1 to element 2, i.e. and

, with defined as the ratio between the semi-length of the bricks and its height, i.e. . Analogous equations can be written for 3-2, 3'-2', 2-2' and 2'-1' interfaces. Assuming that the triangular elements are linear elastic, the following relationship in Voigt notation between strains and stresses can be written:

(2)

Where Eb, νb and Gb are block elastic modulus, Poisson’s ration and shear modulus, respectively. As far as mortar is concerned, two holonomic relationships are adopted, namely a multi-linear relationship and an improved version of the Xu–Needleman exponential law [20-22]. In the first case, a complete decoupling of the normal

and tangential responses is assumed. Although not fully realistic, this approach is very straight-forward and allows for an impressive stability and rapid convergence of the algorithms. In order to take into account possible frictional sliding among bricks, a partial coupling can be obtained by means of a classic Mohr-Coulomb criterion, i.e. it is assumed that the peak tangential stress

is influenced by the current normal stress level σ according to the well-known relationship

, where φ and c denote friction angle and cohesion, respectively. The second choice, hereafter called simply “Xu-Needleman”, is characterize by a stress vector T at the interfaces described by the following closed-form expression:

Symbols and denote the work of separation under pure Mode I (i.e. when ) and Mode II (i.e. when

), respectively, while and indicate the relevant characteristic lengths. It is worth emphasizing that Equation (3) implies a strongly coupled response: softening

occurs for both directions, even if the interface is loaded along the other. In compression the response is assumed linear elastic until an interpenetration constraint is activated through a very high stiffness, acting as a penalty factor. 2.1 Semi-analytical simplified homogenisation model The core of the adopted strategy is a semi-analytical two step procedure, which has been conceived in a more general approach known in the literature as “compatible identification”. In such an approach, the behaviour of the REV is investigated assigning a priori-assumed deformation modes, deduced by applying on the boundary non-null components (one or more) of the homogenized strain tensor. The corresponding homogenized stress tensor is estimated by simply solving the resultant BVP in a semi-analytical form. Typically, the biaxial strain state case can be separate from the shear deformation one, which has been studied apart. Such procedure has been already used in a different context in [24] to solve the homogenisation problem in the case of rigid blocks, but it has been here extended to deformable elastic elements and non-linear holonomic interfaces. Unlike the case with rigid elements, when dealing with deformable bricks, a different FE problem is formulated. In the next sections, the homogenized stress tensor is deduced for selected macroscopic deformation modes. The results as well as the procedure employed to solve the non-linear problem is presented and critically discussed. 2.1.1 Horizontal stretching The first case under consideration is a horizontal stretching applied on the REV boundary (i.e. component of the homogenized strain tensor is increased up to failure of the cell). Equilibrium equations written within the REV lead to the following relationships:

LHUUUUfUUUUf

HLUUUUfUUf

HL

yyxxII

tyyxxII

nyy

yyxxII

txxI

nxxxx

4,,

,

659065901

659090)2(31

6590)2(

65902

312

,,

2

yyxxII

t

yyxxII

nyy

xxxxxx

UUUUfUUUUf

LHUUUUfUUUUf

UUf

yyxxII

tyyxxII

nyy

xxI

nxx

4,, 659065903

903

(4)

where symbols have the following meaning: - is the applied boundary displacement, which depends on the assumed homogenized strain Exx, according to a compatible identification procedure; - is the head joint stress-displacement jump holonomic function along the normal direction; - is the bed joint stress-displacement jump holonomic function along the normal direction; -. is the bed joint stress-displacement jump holonomic function along the tangential direction; - is the horizontal (vertical) displacement of the i-th node. Horizontal compatibility written along nodes 1, 2 and 3, leads to the following equation:

2222

22)3()3()1()1(

090

)3()1(0903

LE

LE

LE

LE

UUU

LLUUUU

b

yyb

b

xx

b

yyb

b

xxxxx

xxxxxxxx

(5)

(3)



Where is brick elastic modulus, the Poisson ratio, is the horizontal strain of the i-th element.

Vertical compatibility written on nodes 3-6 and 1-5, after suitable averaging leads to the following equation:

0

0)3()3()1()1(

65

)3()1(65

)3(661

)1(551

HE

HE

HE

HE

UU

HHUU

HUUU

HUUU

b

xxb

b

yy

b

xxb

b

yyyy

yyyyyy

yyyyy

yyyyy

(6)

Where is the vertical strain of the i-th element. Substituting Equation (4) into compatibility Equations (5) and (6), and indicating with and the following equations are obtained:

,,2

20 II

nb

bIIt

b

In

bx f

ELf

HELf

ELU (7a)

0,2

,2

II

tb

bIn

b

bIIn

b

fE

Lf

EH

fEH

(7b)

It is interesting to notice that when and

, then Equation (7) reduces to the following:

Curve I :

II

tb

bIn

b

bx

b

fEH

LfE

LUL

H 2220 1

212

(8a)

Curve II :

II

nb

b

bx Hf

EHLU

20 12

2 (8b)

(8) is a system of two non-linear equations in unknowns that can be solved graphically as follows: 1) Assign a value for in Equation (8a) and find

immediately the corresponding value of . Curve (8a) can thus be plotted in the - plane selecting a suitable range for . Since is the tangential jump of displacements of the horizontal joint, typically the range to inspect is , where is the ultimate tangential jump of displacement of the interface.

2) Assign a value for in Equation (8b) and find immediately the corresponding value of . Similarly to Curve I, Curve (8b) can thus be plotted in the - plane selecting a suitable range for . Again, since is the normal jump of displacements of the horizontal joint, the range to inspect is , where is the ultimate normal jump of displacement of the interface.

3) The intersection between Curve I and Curve II allows the graphical determination of - values.

When shear and normal behaviors of the interfaces are coupled, analogous relations are formally derived:

,

12

12 2220 II

tb

bIn

b

bx

b

fEH

LfE

LUL

H

(9a)

,

12

2

20 II

nb

b

bx Hf

EHLU (9b)

In this latter case, however, the graphical procedure to determine the solution point requires a recursive approach as follows: 1) Assign a value for in Equation (9a) with = 0 in

and find an updated value for , say . Put into and, through (9a) estimate again . Repeat until . Curve (9a) is again plotted in the - plane within the range

2) Assign a value for in Equation (9b) with = 0 and find an updated value for . Put into and estimate a new by means of (9b). range to inspect is again .

3) - values are estimated at the intersection between Curve I and Curve II.

2.1.2 Biaxial strain state When a biaxial strain state is applied to the unit cell, i.e. with both and , it can be shown that Equation (8) slightly modifies into: Curve I :

II

tb

bIn

b

bx

by f

EHLf

ELU

LHU

22200 1

212 (10a)

Curve II :

II

nb

byo

bx Hf

EU

HLU

20 1

22

(10b)

where is an applied vertical boundary displacement, representing , according to the compatible identification procedure adopted. The solution strategy for the non-linear system of Equation (10) is identical to that adopted for problem (8). 2.1.3 Pure shear deformation state When dealing with the application of a macroscopic tangential deformation , the following equations are obtained:

btII

ntI

tb

tty

b

tx

t

GL

fLHfG

LU

GH

U

2422/

2

(11a)

bb

ttxt

xtII

tb

ty

t GL

GH

UUfGLU

22

22

(11b)

tIt

tx

tIIt

b

v fUfGe

2 (11c)

Where: , , ,

, . It is interesting to notice that

(11) is a system of three non-linear equations including variables , and . Assuming an iterative scheme and starting with , Equation (11) provides a graphical solution very similar to that found for the biaxial stress state.



2.2 Numerical simulations at cell level A running bond masonry pattern was used to perform some numerical analyses at cell level by means of the semi-analytical homogenisation approach previously discussed. To this aim, a single leaf masonry pattern constructed with recycled full size wire cut bricks typical of the early 1930s (dimensions 228 x 109 x 69mm3) and 11mm thick mortar, adopted by BOTHARA et al. [25-27] to construct a half scale two story building, is considered. Mechanical properties assumed for constituent materials are summarized in Table 1. It is worth noting that elastic and inelastic parameters of bricks and mortar joints are assumed in agreement with laboratory tests performed in [25-27]. Starting from such data, both Xu-Needleman and multi-linear laws are defined, as depicted in Figure 2.

As already pointed out, the effect of the application of different macroscopic biaxial strain states can be investigated by means of Equation (8). Typically, the applied strain direction is maintained constant during the deformation process, defining a fixed ratio between biaxial strains as . Only two s are here considered for the sake of conciseness, respectively equal to 0° and 90°. When = 0° the unit cell is subjected to a horizontal macroscopic strain, whereas = 90° corresponds to a vertical stretching. The results, in terms of homogenized vertical and horizontal stresses, are comparatively shown in Figure 3(a) and (b), respectively. Shear behaviour is finally depicted in Figure 3(c), for both the multi-linear and the Xu-Needleman joint models.

Table 1 Mechanical properties adopted for the constituent materials

Brick E = 1000MPa ν = 0.2 Mortar joint E = 500MPa ν = 0.15

ft= 0.40MPa C = 2ft

0 0.1 0.2 0.3

t [mm]

0

0.2

0.4

0.6

0.8 Head joint model (multi-linear)Bed joint model (multi-linear)Head joint model (Xu-Needleman)Bed joint model (Xu-Needleman)

(a) (b)

Figure 2 Xu-Needleman and multi-linear laws adopted for head and bed joints: (a) normal and (b) tangential responses

xxyy

[MPa

]

0 2 4 6

xy [-] 10-3

0

0.2

0.4

0.6

0.8

1 Thom multi-linearThom Xu-Needleman

(a) (b) (c)

Figure 3 Homogenised stress ( and ) -strain ( ) curves obtained imposing: (a) and ;

(b) and and (c) a pure shear deformation state



3. STRUCTURAL IMPLEMENTATION The homogenized mechanical properties deduced using the proposed semi-analytical model are implemented on an existing FE code to perform inelastic structural analyses. The simulations, carried out using ABAQUS [28], deal with a variety of masonries subjected to different load conditions for which experimental data are available in the literature, as well as numerical results provided by other authors. The structural implementation is made with rigid infinitely resistant quadrilateral elements [29] and non-linear interfaces exhibiting an orthotropic behaviour. The homogeneous masonry material is modelled with non-linear shear/normal springs placed between adjoining rigid elements and characterized by the homogenized mechanical properties previously estimated. Flexural and torsional springs are added as well, in order to properly capture the out of plane behaviour. Moment-curvature relationships are deduced by on thickness integration of the in plane homogenized stress-strain curves. Elastic properties of springs are suitably tuned by means of classic elastic energy identification.

4. CASE STUDIES

4.1 In plane loaded windowed panel A windowed shear panel experimentally tested by RAIJMAKERS and VERMELTFOORT [30] is herein analysed. The experimental test (two replicates) was carried out on panels of dimensions 990 x 1000mm having a slight eccentric central window. The wall was initially subjected to a pre-compression load of 0.3MPa applied through a steel beam placed on the top edge. The structure was then subjected to an increasing horizontal controlled displacement up to the formation of a failure mechanism. More information about the laboratory investigation as well as on the experimental set-up is available in [30]. The presence of the central window influences considerably the active failure mechanism, which is characterized by the formation of cracks zigzagging between bed and head joints roughly along one of the main diagonals, with a typical and clearly visible stepped pattern close to the opening corners. The masonry wall consists of bricks with dimensions equal to 210 x 52 x 100mm3, and mortar joints 10mm thick. The

mechanical properties adopted for the bricks and mortar are identical to those assumed by Lourenço and Rots in their heterogeneous model, and are not reported here for the sake of conciseness. The reader is referred to [30] for a comprehensive mechanical characterization. The elastic identification of the interfaces conducted by means of the approach previously presented led to the utilization of the following elastic parameters at a structural level: elastic modulus for vertical interfaces = 860MPa, elastic modulus for horizontal interfaces = 372MPa, and shear modulus = 2051MPa. The panel is discretized using 232 rigid quadrilateral elements with dimensions 52 x 52mm2, which represent a fair compromise between numerical efficiency and reliability of the expected results. A comparison between the results obtained experimentally, those obtained with numerical models by other authors and the present load-displacement curve is provided in Figure 4(b). It can be noted, a satisfactory agreement is found in terms of elastic stiffness, peak load and post-peak behaviour, especially when comparing with [31]. According to experimental evidence, the deformed shape at collapse of the model shows that failure is due to the relative rotation of macro blocks inside the panel, as clearly visible from Figure 4(a). The tensile damage map, Figure 4, shows that inelastic deformation has some peaks near the corners of the central opening and near the point of the application of the load, probably because of the high stresses arising in such areas. 4.2 Out of plane behaviour of a church façade A second series of simulations was performed on a masonry church façade, here modelled with a portion of perpendicular walls to favour the 3D behaviour. Analyses performed are both static (pushover) and non-linear dynamic. The façade belongs to a church called “Transfiguration” located in Moggio Udinese (Italy), which collapsed during the devastating seismic sequence which occurred in Friuli (an Italian region) in 1976. The church suffered severe damage after the first main shock, and later collapsed as a consequence of a second strong shake which occurred in September of the same year. After the first shake, a survey of the geometry and damage to the church was conducted by DOGLIONI et al. [32], providing particularly useful information on the actual seismic behaviour observed.

Tensile behavior

(a)

(b)

Figure 4 Windowed shear panel: (a) damage pattern obtained at failure; (b) load displacement curves



0 50 100 150 200Horizontal displacement [mm]

0

100

200

300

400

Casolo & Uva (2013) Mode IPresent model Mode ICasolo & Uva (2013) Mode 0Present model Mode 0

(a)

(b)

Figure 5 Masonry church façade: (a) pushover curves; and (b) control node time-displacement

diagram from non-linear dynamic analyses

(a) (b) (c)

Figure 6 Masonry church façade: (a) tensile damage maps for horizontal bending; (b) vertical bending; and (c) torsion

Here, a portion (2-3 meters) of the perpendicular walls of the unique nave is modelled with the façade in order (a) to properly account for the actual lateral interlocking of the vertical edges and (b) to accurately reproduce the 3D behaviour, which includes a predominant two-way flexural deformation and a possible detachment leading to overturning in the case of insufficient interlocking between façade and nave walls. The mesh is formed using 354 rigid quadrilateral elements, a fair compromise between numerical efficiency (paramount for non-linear dynamic computations) and refinement to properly reproduce actual crack patterns. The façade has global dimensions equal to 16.30 x 18.05m, with walls 55cm thick. Mechanical properties of the constituent materials are assumed in agreement with those available in the literature [33,34] and are not reported here for the sake of conciseness. Homogenized moment-curvature diagrams used at a structural level are obtained considering three levels of vertical pre-compression roughly representative of what occurs on the tympanum (almost zero vertical in-plane load), middle height and base of the façade. The utilization of different moment-curvature relationships dependent on vertical membrane loads is necessary because it has been proved that vertical compression plays a crucial role in the increase of both ductility and out-of-plane strength. In the hypothesis of a static load application up to failure (pushover), two sets of simulations are performed, with either constant (Mode 0) or reverse-linear along the height of the façade (Mode I) horizontal loads.

The results obtained are depicted in Figure 5(a) and compared with those provided by CASOLO and UVA [34]. A satisfactory agreement can be noted, meaning that the proposed homogenisation approach is able to accurately describe the elastic and inelastic behaviour of real scale structures in two-way bending. A non-linear dynamic analysis is then performed in order to deepen the knowledge on the seismic behaviour of the façade under dynamic loads. The results, obtained applying the Fogaria (Friuli 1976) accelerogram, are concisely reported in Figure 5(b) and Figure 6. In particular, Figure 5(b) shows the displacement time history diagram of the control node, located at the top of the tympanum, whereas Figure 6 indicates the tensile damage pattern obtained at the end of the simulations. Good agreement is found with outcomes reported in [34], in terms of both time-displacement history and crack distribution. The façade exhibits a collapse mechanism that mainly involves the upper part, with the resultant overturning of the tympanum. This finding suggests the activation of a failure mechanism reasonably in agreement with that found in [34], where the reader is referred for further details. High levels of damage are reached in correspondence with the horizontal line where the out-of-plane mechanism of the tympanum takes place. It is worth noting that severe damages are also visible along the vertical edges and near the central opening, again in quite reasonable agreement with a previous investigation [34] and post-earthquake surveys.



4.3 3D dynamic behaviour of a masonry building The model is finally applied to the study of the dynamic behaviour of a half scale two story masonry building experimentally tested on a shaking table by BOTHARA et al. [25-27]. The small masonry house has overall plan dimensions equal to 2.8 x 1.92m2 and a total height of 2.48m. First and second floors are 1.34m and 1.14m high, respectively. The global geometry of the construction is depicted in Figure 7(a), within the discretization adopted (946 rigid elements) for the proposed simulations. A sequence of ground motions with increasing level of severity was applied to the prototype along both longitudinal and transversal directions. For the sake of conciseness only the longitudinal direction case is considered herein. Visual observations were used to represent the cracks arising during the seismic sequence, whereas accelerometers and linear potentiometers, placed along the entire construction, were installed [25-27] to monitor both acceleration and displacements in different positions. Timber floor and roof are modelled using beam elements, with geometric features and elastic parameters in agreement with that reported in [25-27]. Homogenized stress-strain curves shown in Figure 3 are here implemented at a structural level utilizing the same procedure discussed for the previous examples, again distinguishing two levels of pre-compression, to properly account for the presence of additional masses placed on the first floor (2.05 tons) and at roof level (2.09 tons). This model is initially analysed in the elastic range in order to compare the frequencies obtained experimentally with those predicted by the proposed model. The outcomes provided by the discrete strategy are in satisfactory agreement with

the experimental results. Indeed, the proposed HRBSM model provides a frequency of 10.6Hz for the first mode along the longitudinal direction and 8.03Hz for the transverse one, against experimental values of 11.7Hz and 9.8Hz, see [25-27] The building is then studied under non-linear dynamic excitation, imposing along the longitudinal direction the same acceleration history used in the shaking table tests. Four, in series accelerograms are applied, the first two with moderate (0.2 and 0.3g) PGAs, the last two with severe (Umbria-Marche 0.5g) and moderate/severe (El Centro 0.348g) PGAs. The obtained numerical time-displacement history of the control node is compared with experimental one in Figure 7(b). A good fit results. For the sake of completeness, in Figures 8 and 9 the in and out of plane crack patters obtained numerically at the end of the application of the Umbria-Marche accelerogram are represented. As expected, the two transverse walls are subjected to predominant out of plane failure mechanisms, while front and back in plane walls cracked mainly due to in plane damages. Out of plane damages are mainly concentrated on the top of the gable walls and near the corners of the openings. In agreement with intuition, the numerical results seem to indicate the activation of an overturning mechanism in the upper part of the transverse walls, in addition to the splitting of the windowed side wall along the vertical central line. Limited differences with respect to the experimental results are found even for in plane interfaces. Severe damages were observed on shear springs, whereas axial interfaces were subjected to moderate damages even at the end of the seismic sequence.

(a) (b)

Figure 7 Masonry building prototype. (a) discretization adopted; (b) time-displacement diagrams at first floor level at the end of application of Umbria-Marche earthquake (PGA = 0.5g) and position of the control point

Front wall Back wall Front wall Back wall

Shear interfaces

Axial interfaces

Figure 8 Masonry building prototype. In plane damage patterns obtained at the end of the application of Umbria-Marche earthquake (PGA = 0.5g)



Solid wall Windowed wall Solid wall Windowed wall

Torsional Interfaces

Flexural Interfaces

Figure 9 Masonry building prototype. Out of plane damage patterns obtained at the end of the application of Umbria-Marche earthquake (PGA = 0.5g)

5. CONCLUSIONS A simple but efficient two step holonomic homogenisation model for masonry subject to in and out of plane loading has been presented. The formulation is semi-analytical and allows a fast estimation of homogenized quantities under biaxial strain states and pure shear deformations. The results obtained at cell level have been benchmarked on a running bond pattern considered next within non-linear dynamic structural analyses. Two cohesive relationships, used to describe the behaviour of the mortar joints, have been comparatively assessed. A variety of different structural examples have been studied, namely a windowed shear wall subjected to static loads, a masonry church façade and a small building prototype, all subjected to non-linear dynamic loads. Comparing model results with literature data (both experimental and numerical), satisfactory results have been obtained, confirming that the present approach can be adopted by practitioners to analyse large scale structures, for which a classical micromechanical approach is not applicable. The meso and macro scale decoupling represents an important advantage especially when dealing with non-linear structural analyses, since the computational effort significantly decreases. Unlike with the time consuming FE2 homogenisation strategies, the present approach is very straightforward, because the mechanical properties of the equivalent homogenized springs used at a structural level are set only once per simulation. In addition, they can be used within general purpose FE software, thus allowing the utilization of robust and verified advanced numerical facilities, such as arc-length routines and non-linear dynamic solvers. Despite the intrinsic mesh dependence of the HRBSM model, in the case of global softening it is not superseded, the procedure produces results which are efficient and reliable because the holonomic laws assumed for mortar allow for a total displacement formulation of the model and it is not necessary to deal with excessively refined meshes -drastically speeding up computations even with non-linear dynamic analyses. REFERENCES 1. DI PASQUALE, S. New trends in the analysis of

masonry structures, Meccanica 27 173-184, 1992. 2. BERTO, L., SAETTA, A., SCOTTA, R. and

VITALIANI, R. An orthotropic damage model for masonry structures, Int J Numer Methods Engng 55, 127–57, 2002.

3. LOURENÇO, P.B., DE BORST, R. and ROTS, J.G. A plane stress softening plasticity model for orthotropic materials, International Journal for Numerical Methods in Engineering 40, 4033-4057, 1997.

4. LOURENÇO, P.B. and ROTS, J. A multi-surface interface model for the analysis of masonry structures, Journal of Engineering Mechanics ASCE 123 (7), 660-668, 1997.

5. MACORINI, L. and IZZUDDIN, B.A. A non-linear interface element for 3D mesoscale analysis of brick-masonry structures, International Journal for Numerical Methods in Engineering 85 (12), 1584-1608, 2010.

6. LOTFI, H.R. and SHING, P.B. Interface model applied to fracture of masonry structures, Journal of Structural Engineering ASCE 120 (1), 63-80, 1994.

7. SUTCLIFFE, D.J., YU, H.S. and PAGE, A.W. Lower bound limit analysis of unreinforced masonry shear walls, Computers & Structures 79, 1295-1312, 2001.

8. MACORINI, L. and IZZUDDIN, B.A. Nonlinear analysis of masonry structures using mesoscale partitioned modeling, Advances in Engineering Software 60-61, 59-69, 2013.

9. DE BUHAN, P. and DE FELICE, G. A homogenisation approach to the ultimate strength of brick masonry, Journal of the Mechanics and Physics of Solids 45 (7), 1085-1104, 1997.

10. LOPEZ, J., OLLER, S., ONATE, E. and LUBLINER, J. A homogeneous constitutive model for masonry, International Journal for Numerical Methods in Engineering 46 (10), 1651–1671, 1999.

11. LUCIANO, R. and SACCO, E. Homogenisation technique and damage model for old masonry material, International Journal of Solids and Structures 34 (24), 3191-3208, 1997.

12. PEGON, P. and ANTHOINE, A. Numerical strategies for solving continuum damage problems with softening: application to the homogenisation of masonry, Computers & Structures 64, (1-4), 623-642, 1997.

13. MILANI, G. Simple homogenization model for the non-linear analysis of in-plane loaded masonry walls, Computers & Structures 89, 1586–1601, 2011.

14. MILANI, G., LOURENÇO, P.B. and TRALLI, A. Homogenised limit analysis of masonry walls. Part I: failure surfaces, Computers & Structures 84 (3-4), 166-180, 2006.



15. MILANI, G., LOURENÇO, P.B. and TRALLI, A. Homogenization approach for the limit analysis of out-of-plane loaded masonry walls, Journal of Structural Engineering ASCE 132 (10), 1650-1663, 2006.

16. MASSART, T., PEERLINGS, R.H.J. and GEERS, M.G.D. Mesoscopic modeling of failure and damage-induced anisotropy in brick masonry, Eur J Mech A/Solids 23, 719–35, 2004.

17. MASSART, T., PEERLINGS, R.H.J. and GEERS, M.G.D. An enhanced multi-scale approach for masonry wall computations with localization of damage, International Journal for Numerical Methods in Engineering 69, 1022–1059, 2007.

18. MERCATORIS, B.C.N and MASSART, T. A coupled two-scale computational scheme for the failure of periodic quasi-brittle thin planar shells and its application to masonry, International Journal for Numerical Methods in Engineering 85 (9), 1177–1206, 2011.

19. COLLIAT, J.B., DAVENNE, L. and IBRAHIMBEGOVIC, A. Modélisation jusqu'à rupture de murs en maçonnerie chargés dans leur plan, Revue francaise de génie civil 4, 593-606, 2002.

20. XU, X.P. and NEEDLEMAN, A. Potential-based and non-potential-based cohesive zone formulations under mixed-mode separation and over-closure, Part I: Theoretical Analysis 2, 417-418, 1993.

21. MCGARRY, P., MÁIRTÍN, E.O., PARRY, G. and BELTZ, G.E. Potential-based and non-potential-based cohesive zone formulations under mixed-mode separation and over-closure. Part I: Theoretical analysis, Journal of the Mechanics and Physics of Solids 63, 336–362, 2014.

22. FEDELE, R. Simultaneous Assessment of mechanical properties and boundary conditions based on Digital Image Correlation, Experimental Mechanics 55, (1), 139-153, 2015.

23. SUQUET, P. Analyse limite et et homogeneisation. Comptes Rendus de l'Academie des Sciences - Series IIB – Mechanics 296, 1355-1358, 1983.

24. CECCHI, A., MILANI, G. and TRALLI, A. A Reissner-Mindlin limit analysis model for out-of-plane loaded running bond masonry walls, International Journal of Solids and Structures 44, (5), 1438-1460, 2007.

25. BOTHARA, J.K., DHAKAL, R.P. and MANDER, J.B. Seismic performance of an unreinforced masonry building: An experimental investigation, Earthquake Engineering and Structural Dynamics 39, 45-68, 2010.

26. BOTHARA, J.K., MANDER, J.B., DHAKAL, R.P., KHARE, R.K. and MANIYAR, M.M. Seismic performance and financial risk of masonry house, ISET Journal of Earthquake Technology 44, 421-444, 2007.

27. BOTHARA, J.K. A shaking table investigation on the seismic resistance of a brick masonry house, Master Thesis. University of Canterbury, Christchurch New Zealand, 2004.

28. ABAQUSTM, Theory manual, version 6.6, 2006. 29. KAWAI, T. New discrete models and their application

to seismic response analysis of structures, Nucl. Eng. Des. 48, 207–229, 1978.

30. RAIJMAKERS, T.M.J. and VERMELTFOORT, A. Deformation controlled tests in masonry shear walls (in Dutch), Report B-92-1156, TNO-Bouw, Delft, The Netherlands, 1992.

31. LOURENÇO, P.B. and ROTS, J. A multi-surface interface model for the analysis of masonry structures, Journal of engineering mechanics 123, (7), 660–668, 1997.

32. DOGLIONI, F., MORETTI, A. and PETRINI, V. Le Chiese e il Terremoto, Lint press: Trieste, 1994.

33. CASOLO, S. and MILANI, G. A simplified homogenization-discrete element model for the non-linear static analysis of masonry walls out-of-plane loaded, Engineering Structures 32, 2352-2366, 2010.

34. CASOLO, S. and UVA, G. Nonlinear analysis of out-of-plane masonry façades: full dynamic versus pushover methods by rigid body and spring model, Earthquake Engng Struct. Dyn. 42, 499-521, 2013.



A Comparative Investigation into Lime Activated Ground Granulated Blast Furnace Slag as a Sustainable

Alternative to Portland Cement in Masonry Mortars

S. HETHERINGTON (1)

(1)Department of the Natural and Built Environment, Sheffield Hallam University, Sheffield, South Yorkshire, S11WB England

[email protected]

ABSTRACT Portland cement production is known to have a detrimental impact on the global environment and hence the construction industry has made moves to introduce more sustainable materials and practices. The introduction of hydraulic lime based mortars for masonry being one example. Nevertheless there are other binders that are also deemed sustainable, for example, lime or alkali activated ground granulated blast furnace slag (GGBS). Very little information is available on the performance of lime activated GGBS mortars, especially their tensile bond strength characteristics. This project was designed to compare the performance characteristics and properties of two sets of mortar mixes. One set of samples was based on traditional Portland cement, lime, sand mixes. The other set was based on GGBS, lime, sand mixes, where the lime content served as a workability agent and as an activator for the GGBS. Tensile, flexural and compressive strength samples were fabricated from both mortar sets and tested at 28 days. The results indicated that mortar mixes containing lime activated GGBS could be used as sustainable alternatives to Portland cement based mortars for masonry applications as they showed similar properties to their Portland cement based counterparts. KEYWORDS: masonry, tensile bond strength, lime activated GGBS, Portland cement mortar, comparative study, sustainable mortars.

1. INTRODUCTION In recent years the building industry has become aware of the need to change to more sustainable materials. This is highlighted by a move away from the use of cement when in 2012 the global production generated approximately 2.4 billion tons of carbon emissions, which comprised of between 5%-7% of the total global emissions [1]. The introduction of blended cements containing Portland cement blended with pulverised fuel ash (PFA) or ground granulated blast furnace slag (GGBS) for concrete production has reduced the construction industries reliance on Portland cement. This trend away from the more traditional cement based binders in favour of more sustainable blended cements containing GGBS and pulverised fuel ash (PFA) has resulted in a move towards more sustainable binders. In the UK, the majority of masonry now uses ready mixed mortars, and virtually all of these contain slag, ash or ground limestone. Pure CEM I mortars are virtually non-existent. However the reduction of Portland cement for use in mortars could go further. Mortars for masonry use a considerably smaller quantity of cement compared to concrete but moves have also been made to introduce more sustainable alternatives to cement for masonry mortars, as for example Hydraulic limes or CEM II mortars. It is well documented that Hydraulic lime mortars take longer to set than cement based mortars with the compressive strengths of lime-based

mortars usually quoted at 91 days rather than 28 days [2]. This move away from the use of Portland cement as the principle binder in the production of masonry mortars to the use of Natural Hydraulic limes is seen as being a more sustainable option due in part but not limited to the lower kiln temperatures needed to produce Hydraulic limes and the subsequent reduction in CO2 emissions from the manufacturing process. Hydraulic limes also offer a number of other advantages, including a more open pore structure that creates a pathway for water to pass freely from the building units into the mortar and evaporate. However, Natural Hydraulic limes do not show the same performance characteristics as Portland cement with generally lower compressive and flexural strengths and as stated previously, slower setting times. These factors limit their application within the construction arena to the use in low rise domestic dwellings and generally to areas where they will not be subject to severe exposure or aggressive environments although numerous historic examples exist of applications in those areas still performing satisfactorily. The use of blended cements, as for example those containing GGBS, may provide another sustainable option for masonry mortars especially if the same performance characteristics associated with the use of GGBS in concrete vis a vis non-blended cement are shown in the production of mortars. If this is the case then the use of GGBS, as one example, could be seen as a more sustainable direct replacement for Portland cement without compromising performance. Very little published work has been identified on the properties and performance of mortars containing GGBS as the principle binder used for masonry due to the industries wide use of mortars manufactured from other binders, as for example hydraulic limes and CEM II cements. GGBS is an attractive alternative as it is a by-product of the steel manufacturing process and is subsequently considered to be a more environmentally-friendly binder than Portland cement. GGBS is a "latent" hydraulic binder and offers little or no cementitious reaction on its own as it requires alkali activation (AA) to initiate the setting procedure. In blended cements used for the production of concrete the alkali activation (AA) is achieved by the inclusion of varying percentages of a Portland cement component depending on the binder designation. For example CEM III/C can have as little as 5% Portland cement content [3] with this amount being sufficient to initiate the setting reaction. In this study the alkali activation (AA) of the GGBS is provided by hydrated lime Ca(OH)2. This has been chosen as it is generally known that a Ca(OH)2 based AA GGBS binder has a practical advantage in regard to a relatively moderate workability loss and stable strength development [4]. Both of these qualities are desirable in terms of the mortars ease of use, application and performance. The use of lime as the activator for the GGBS reduces the reliance on Portland cement even further even though the production of lime still liberates CO2, a mortar produced with lime as opposed to Portland cement is a more sustainable product.


2. MATERIALS The workability and rheology characteristics of each mortar type produced can be influenced to a greater or lesser extent by the type of sand being used in its production. The sand used for this study was a general purpose building sand that conformed to BS 1199 and BS 1200; 1976 [5]. Nevertheless, it is widely recognised that the finer the sand the more water is required to produce the same flow, (often referred to as the workability) [6]. However, with regard to this study the same sand has been employed throughout the production of all of the mortars meaning the influence that the sand has on the performance of the different mortars was standardised and predictable. Other factors that affect the performance of the mortar are the particle size of the different constituents. With GGBS being 2-3 times smaller than Portland cement [7], this affects the mortar’s water retentivity and workability and the ability of the freshly mixed mortar to retain water against the dewatering action of an absorbent substrate [8]. This is an important factor in the development of bond strength although the choice of the correct building unit is also of importance; this aspect of the experimental procedure is discussed later in this paper. All the components used in the production of the mortar for this work conformed to their respective standards. Portland cement and Ground granulated blast furnace slag were used as the two principle binders with the Portland cement conforming to BS EN197-1:2011 [3] and GGBS conforming to BS EN 15167-2:2006 [9]. The Hydrated lime used as a workability agent and as an alkali activator for the GGBS conformed to BS EN 459-1:2015 [10].

3. MORTAR PREPARATION Each set of mortars was produced in accordance with the proportions stated in table 7 of BS 4551:2005 [11]. Traditional mortar designations of i, iii and v were used and these equate to volume proportions of binder, lime and sand of 1:1/4:3, 1:1:6 and 1:3:10. These mortars were chosen for three reasons. Firstly to provide mortars that would offer the opportunity to directly compare the two binder systems with the lime content of each mix being used for increased workability and the alkali activation of the GGBS. The second reason being to obtain a good indication of the performance of the mortar systems across the whole spectrum of binder proportions specified in table 7 of BS 4551:2005 [9], again so that a direct comparison of the performance of the GGBS and the Portland cement could

be observed. Thirdly that these are not ‘contrived’ mortar mixes in so far as the mix types containing a binder a workability agent (lime) and aggregate have been used for some time and could be termed as 'traditional' mixes. The only difference between the two mortar mixes is the different binder type being employed. All the mortar mixes were produced using a Hobart mechanical mixer shown in Figure 1 and to the procedure specified in BS 4551:1998 [12]. The mortars were all produced by volume proportioning the ingredients as set out in table 7 of BS 4551:2005 [9]. The water content was determined by measuring the workability of the mortar using two tests, the dropping ball apparatus as stated in BS 4551:2005 [9] and the flow table as in BS EN 1015-3:2003 [13]. Final adjustments to the workability were achieved by adding water on the spot board to give the required workability of between 10mm and 11mm when tested using the dropping ball technique described in BS 4551:2005 [11], in which the "mortar mix prepared in the laboratory should have its consistency adjusted to a penetration of 10 + or - 0.5mm" [9]. The workability of each mix was also tested using the flow table to the method specified in BS EN 1015-3:2003 [13]. The results of the two tests can be seen graphically represented in Figures 2 and 3 where the GGBS lime mortars are labelled GLS and the Portland cement lime mortars are labelled CLS. A common workability reading of between 10mm and 11mm on the dropping ball test was chosen as a standard for all the mortars as opposed to a water to binder ratio as this would provide a consistent set of workable mixes where trying to control the water cement ratios would give a series of mortars with differing workabilities. Also, controlling the workability of all the mortars including the ones that contain large volumes of lime and or GGBS, reduces any variability that these mortars are likely to have on the bond strength due to their high water retentivity. This aspect of finely divided binders having greater water retentively than Portland cement binders is well known with mortar mixes with lime needing more water to reach the same consistency of mortars without lime which is essentially related at least partly to the smaller particle size [14]. Therefore, it was deemed prudent to use this parameter (workability) as a controlling factor for the mortar production. Another factor that influenced the choice of workability over water to binder ratio is the fact that workability is considered one of the most important properties of mortar because it influences directly the bricklayer's work [12] and would most likely be a factor influencing the mortars production on site.

Figure 1 Hobart mechanical mixer

S. HETHERINGTON


Figure 2 Average and Standard deviation of each Figure 3 Average and Standard deviation of each mortars workability tested using the mortars workability tested using the dropping ball technique flow table technique The results of the workability tests carried out indicate that each mix was produced to a constant standard with no significant difference between mortars when referring to the dropping ball test as a gauge of workability, see Figure 2. However, this was not the case for the results from the flow table when used as a test of workability where the workability increased in both the GGBS and Portland cement mortars from the high binder to sand ratios to the low binder sand ratios, see Figure 3, indicating that these two tests do not in this case show a correlation. This illustrates that the choice of test can influence the results but for this experiment the dropping ball test was chosen as the principle test method for workability. Each mortar was used to produce 3 mortar prisms (40mm * 40mm * 160mm) in accordance with BS EN 1015-11 1999 [15] which when tested would give an indication of the flexural and compressive strength of each mortar. In addition to the 3 mortar prisms, 5 brick couplets were fabricated so that the direct tensile bond strength of the brick mortar interface could be determined.

4. BRICKS All of the tensile test samples were fabricated using multi perforated 'engineering' bricks, that conformed to the specification set out in BS EN 771-1:2003 [16]. This type of brick was chosen as it has low initial rates of suction compared to facing bricks and as being more predictable and consistent in performance in respect to their absorbency. Ten samples were tested in accordance with the procedure set out in BS EN 772-11:2011 [17]. The average initial rate of absorption was found to be 0.881kg/m2/min with a standard deviation of 0.138kg/m2/min and a coefficient of variation of 15.6%. These values indicate that this sample set expressed very little variability in relation to the initial rate of suction. This is an important factor when considering the mortar combinations used as some mortars tend to have greater water retentivity than others and this aspect of the mortars’ characteristics needs

to be acknowledged when choosing the substrate. The choice of a series of bricks with an unpredictable initial rate of suction would offer another unwanted variable. Pre-wetting, "docking", or suction rate adjusting of the bricks can be one way of reducing the water demand and variability of a high suction rate brick or building unit. This could still lead to some variability unless a clear regime of testing and categorising of bricks into 'groups' of similar initial rates of suction is undertaken along with a well-defined structure and procedure for 'docking' or pre-wetting the bricks similar to the regime suggested by BARR et al [18] with a pre wetting procedure consisting of a variety of different time periods depending on the bricks initial rate of suction.

5. TENSILE BOND STRENGTH SAMPLE MANUFACTURE AND TESTING

The tensile samples were fabricated in a standard jig that was designed to produce a consistent tensile strength test sample that would not be susceptible or sensitive to operator error. The jig used to prepare the samples is in two parts. The first part is designed to control the amount of mortar that is applied to the lower brick. The mortar applied in this instance is an excessive amount, some 15mm in depth (see Figure 4). The second jig (Figure 5) provides control over the placing of the upper brick and by the application of a downward force on the upper brick the ultimate mortar depth of 10mm can be achieved this being controlled by the use of gauging bars (see Figures 5 and 6). This manufacturing process was adopted in order to reduce the variability in the sample set manufacture and by controlling the workmanship in this way the test specimen couplets can be fabricated and "at the same time are a reasonable simulation of bricklaying" [19]. The samples (Figure 7) were then cured in a curing chamber with the temperature and humidity controlled at 200C and 65% relative humidity for 28 days in keeping with the recommended curing time and conditions for the mortar test prisms.


A Comparative Investigation into Lime Activated Ground Granulated Blast Furnace Slag as a Sustainable Alternative to Portland Cement in Masonry Mortars

Figure 4 Initial 15mm of mortar Figure 5 Second jig with 10mm gauging bars ready to receive the upper brick

Figure 6 Placing of the upper brick Figure 7 Completed couplet sample

6. PROPERTIES OF THE HARDENED MORTAR The mean mortar flexural and compressive strengths and corresponding standard deviations were obtained by initially testing a series of three 40mm * 40mm * 160mm prisms that had been manufactured in accordance with BS EN 1015-11 1999 [15]. The results from these tests can be seen in Figure 8. After performing the flexural tests on the prisms the remaining halves were used to obtain the compressive strengths as can be seen in Figure 9. The water absorption and apparent porosity were also obtained. The results from these tests can be seen in Figures 10 and 11, and Table 1. The water absorption and apparent porosity were obtained by vacuum saturation testing on a series of five samples from each mortar type. The samples used in this test were taken from the remaining material after the compressive strength tests had been performed. A variety of different techniques could have been used to obtain information on the properties of the mortars used in this work. However the procedure chosen in this case was based on a vacuum saturation technique that is set out in BS 3921:1974 [20]. This method was chosen as opposed to a simple 24 hour

immersion test as immersion does not result in complete saturation due to air becoming entrapped within the porous solid [21]. Although it is acknowledged that the entrapped air would eventually dissolve into the water the rate of this absorption cannot be guaranteed, hence vacuum saturation was chosen as it has been shown that a 6 minute pumping followed by a 15 minute soak essentially gives complete saturation [21]. The vacuum saturation test will give a more accurate representation of the mortar's properties. The procedure requires the samples to be dried in a drying oven at 110oC until they reach a constant mass usually after 48 hours. The samples are left to cool and the dry mass of each sample obtained and recorded. The samples are then placed into a vacuum chamber and evacuated down to below 20mm Hg. Water is then introduced into the chamber and the samples left to soak for 10 minutes. After this period the samples are weighed in water (suspended weight) and their saturated weights taken. From the dry weights, suspended weights and saturated weights the percentage water absorption and apparent porosity can be calculated. The results of the tests can be seen in Table 1 and Figures 10 and 11.


S. HETHERINGTON

Figure 8 Means and standard deviations Figure 9 Means and standard deviations of the of the flexural strengths of the mortar prisms compressive strengths of the respective mortar mixes

Figure 10 Percentage water absorption

Figure 11 Percentage apparent Porosity



Table 1 Porosity and water absorption of the mortars

Figure 12 Direct tensile test apparatus Figure 13 Direct tensile test apparatus

Figure 14 Means and standard deviations of the tensile bond strength results

Mortar type 1: 1/4:3 CLS 1:1:6 CLS 1:3:10 CLS 1:1/4:3 GLS 1:1:6 GLS 1:3:10 GLS

%porosity 36.35 38.01 39.68 34.69 37.35 35.84 %water absorption 21.88 23.54 26.16 20.95 22.92 22.84

S. HETHERINGTON


7. TEST RESULTS AND DISCUSSION Direct tensile tests were carried out on all the couplet samples prepared after 28 days curing at 200C and 65% relative humidity. The direct tensile test rig used to test the couplets comprises of two cradles that fit into spaces in the mortar joint between the bricks that make up the couplet test sample as illustrated in Figures 12 and 13. The calculation of the tensile strength of the couplets is obtained by dividing the load at failure in Newton's by the gross mortar to bed face contact area of the sample in mm2 and expressing the results in N/mm2 or MPa. It can be seen from the results illustrated in the graph in Figure 14 that the lime activated GGBS mortars perform as well if not better in some instances than their respective cement based counterparts. A statistical analysis of the results comparing like for like mortar mixes using an unpaired ''t' test resulted in no significant difference between the respective 1:1/4:3 and the 1:1:6 mortar sets. However a significant difference was recorded between the 1:3:10 mortars with the bond strength of the cement based mortar being significantly stronger than the GGBS based mortar. This trend was not repeated in the compressive strengths of the respective mortars with all of the GGBS mortars expressing significantly higher compressive strengths than their respective cement based counterparts as can be seen illustrated in Figure 9. The results of the flexural strengths of mortar prisms reflected these results with the exception of the 1:1/4:3 GGBS mortar, which showed a large standard deviation and lower than expected average result. When consideration is given to the overall general performance of the two binders it would appear that the GGBS based mortars are slightly superior for each test being applied. However it would also appear that the GGBS based mortars show a greater degree of variability especially when considering the results of the compressive and flexural strength. This could in part be due to poor distribution of the hydrated lime throughout the mortar which consequently could lead to a slower activation of the GGBS.

8. CONCLUSIONS It would appear form the tensile bond strength results illustrated in Figure 14 that the GGBS based mortars could serve as more sustainable alternatives to their respective Portland cement based counterparts. The readings indicate that the tensile bond strength of the GGBS samples perform as well as their Portland cement based counterparts. Indeed both the compressive strength of the GGBS based mortar cubes, along with the tensile bond strength readings indicate that these mortars can serve as viable replacements for Portland cement mortars as they have similar performance characteristics. Although no comparison has been made in this paper with natural hydraulic lime mortars it is well documented that natural hydraulic lime has a slower strength gain than Portland cement with its ultimate compressive strength being quoted at 91 days as opposed to 28 days. Natural Hydraulic lime mortars are considered a more sustainable alternative to Portland cement mortars especially for new build low rise domestic dwellings due to the lower carbon emissions during their production and its absorption of carbon dioxide from the atmosphere. Nonetheless these mortars do not possess the same performance characteristics as Portland cement.

This then raises the question as to how to maintain the level of performance that is expected form masonry and provide the construction industry with low carbon sustainable materials. From the results given in this paper it can be seen that mortars containing GGBS could serve as a sustainable alternative to Portland cement and with the construction industry looking for sustainable building materials GGBS based mortars could be a viable option. A possible disadvantage of using GGBS mortars is the generally lower open porosity and percentage water absorption compared to Portland cement mortars. Although these are not significantly lower there could be some issues around the effect that a low porosity and low water absorption mortar might have if combined with high porosity and high water absorption building units. REFERENCES 1. CHIA-JUNG TSAI, RAN HUANG, WEI-TING LIN,

HSIANG-WEI CHANG, 2014. Using GGBOS as the alkali activators in GGBS blended cements. Construction and Building Materials available online at http://www.sciencedirect.com/science/article/pii/S0950061814009015.

2. NHBC, 2008. The use of lime based mortars in new build, available online at http://www.nhbcfoundation.org/Publications/Guide/The-use-of-lime-based-mortars-in-new-build-NF12.

3. BS EN 197-1:2011 Cement Part 1: Composition, specifications and conformity criteria for common cements. London BSI.

4. KEUN-HYEOK YANG, AH-RAMCHO, JIN-KYU, SANG-HO NAM, 2012. Hydration products and strength development of Calcium hydroxide-based alkali activated slag mortars. Construction and Building Materials available online at https://www.highbeam.com/doc/1G1-284323194.html

5. BS 1199 and 1200, 1997 Specification for Building sands from natural sources London: BSI.

6. B BV VENKATARAMA REDDY, AJAY GUPT, 2007. Influence of Sand grading on the characteristics of mortars and solid-cement block masonry Construction and Building Materials available online at http://www.sciencedirect.com.lcproxy.shu.ac.uk/science/article/pii/S0950061807001766.

7. CORDEN, E., PAVIA, S., 2007. A Comparative study of the moisture transfer properties and durability of PC and GGBS mortars, available on line @ https://www.tcd.ie/civileng/local/notes/Sara.Pavia/S8/research%20papers/SPCP2007ConcretePBelfast.pdf last accessed 5-5-2016.

8. INCE, C. OZTURK, Y., CARTER, M.A. WILSON, M.A., 2013. The influence of supplementary cementing materials on water retaining characteristics of hydrated lime and cement mortars in masonry construction.

9. BS EN 15167-1:2006 Ground granulated blast furnace slag for use in concrete, mortar and grout. Definitions, specifications and conformity criteria. London BSI.

10. BS EN 459-1:2015 Building lime. Conformity evaluation. London BSI.

11. BS 4551:2005. Methods for testing mortar, chemical analysis and physical testing. London: BSI.

12. BS 4551:1998 Methods for testing mortar, chemical analysis and physical testing. London: BSI.

13. BS EN 1015-3:2003 Methods of test for mortar for masonry. Determination of consistence of fresh mortar (by flow table). London: BSI.



14. HAACH, VLADIMIR G.; VASCONCELOS, GRACA; LOURENCO, PAULO B., 2011. Influence of aggregates grading and water/cement ratio in workability and hardened properties of mortars. Construction and Building Materials, June 2011, Vol.25 (6), pp.2980-2987.

15. BS EN 1015-11 1999. Methods of test for mortar for masonry. Determination of flexural and compressive strength of hardened mortar. London: BSI.

16. BS EN 771-1:2003 Specification for masonry units Part 1: Clay masonry units. London: BSI.

17. BS EN 772-11:2011. Methods of test for masonry units. Determination of water absorption of aggregate concrete, autoclaved aerated concrete, manufactured stone and natural stone masonry units due to capillary action and the initial rate of water absorption of clay masonry units. London: BSI.

18. BARR, S., MCCARTER, W.J., SURJANTO, B., 2015. Bond-strength performance of hydraulic lime and natural cement mortared sandstone masonry. Available online at http://www.limes.us/wpcontent/uploads/2013/04/BondStrengthPerformancePaper.pdf.

19. TAYLOR FIRTH. A. TAYLOR. I., 1990. A bond tensile strength test for use in assessing the compatibility of brick mortar interfaces. Construction and Building materials Journal Volume 4 no 2, Surrey, Butterworth Heinemann.

20. BS 3921:1974. Specification for clay bricks and blocks. London: BSI.

21. WILSON, M.A., CARTER, M.A., HOFF, W.D., 1999. BRITISH STANDARD and RILEM water absorption tests: A critical evaluation. Materials and Structures vol 32 October 1999 available online at http://www.civil.ist.utl.pt/~cristina/RREst/Aulas_Apresentacoes/07_Bibliografia/durabilidade%20betao%20(durability)/Outros/British%20Standard%20and%20RILEM%20water%20absorption%20tests%20A%20critical%20evaluation.pdf.

S. HETHERINGTON


Printed by: Sherwin Rivers Ltd, Waterloo Road, Stoke on Trent ST6 3HRTel: 01782 212024 Fax: 01782 214661 Email: [email protected]

Materials Development • Testing • Assurance

Find out more at www.lucideon.com/mi

SUPPORTING THE CONSTRUCTION INDUSTRY

We are experts in all things construction thanks to our extensive testing facilities, knowledge of current regulations and industry experience.

Our services include: materials classification, testing to international standards, product development, resource efficiency, structural and environmental testing and failure analysis. And this year we have added dynamic wind load testing.

Lucideon - helping clients to improve performance, solve problems, comply with regulations, meet specifications, realise projects, develop novel products/processes, and improve existing ones.

LUCIDEO

N – the new

name for Ceram

Materials • Products • Systems • Structures

Journal of the International Masonry Society · The International Masonry Society Registered Office: British Ceramic Research Ltd, Queens Road, Penkhull, Stoke-on-Trent, ST4 7LQ,

Documents