You are on page 1of 176

STP 1434

The Use of Glass in Buildings

VaIerie L. Block, editor

ASTM Stock Number: STP1434

ASTM International 100 Barr Harbor Drive PO Box C700 West Conshohocken, PA 19428-2959 Printed in the U.S.A.

Library of Congress Cataloging-in-Publication Data

Symposium on the Use of Glass in Buildings (1st : 2002 : Pittsburgh, Pa.) The use of glass in buildings/[edited by] Valerie L. Block. p. cm.--ASTM special technical publication; 1434 Includes bibliographical references and index. "ASTM stock number: STP1434." ISBN 0-8031-3458-4 1. Glass construction--Congresses. 2. Glazing--Congresses. 3. Safety glass--Congresses. I. Block, Valerie L., 1951- II. Title. TH1560 .S96 2002 691'.6--dc21 2002038238

Copyright 9 2002 ASTM International, West Conshohocken, PA. All rights reserved. This material may not be reproduced or copied, in whole or in part, in any printed, mechanical, electronic, film, or other distribution and storage media, without the written consent of the publisher.

Photocopy Rights Authorization to PhOtocopy items for internal, personal, or educational classroom use, or the internal, personal, or educational classroom use of specific clients, is granted by ASTM International (/L~TM) provided that the appropriate fee is paid to the Copyright Clearance Center, 222 Rosewood Drive, Danvers, MA 01923; Tel: 978-750-8400; online: http://

Peer Review Policy

Each paper published in this volume was evaluated by two peer reviewers and at least one editor. The authors addressed all of the reviewers' comments to the satisfaction of both the technical editor(s) and the ASTM International Committee on Publications. To make technical information available as quickly as possible, the peer-reviewed papers in this publication were prepared "camera-ready" as submitted by the authors. The quality of the papers in this publication reflects not only the obvious efforts of the authors and the technical editor(s), but also the work of the peer reviewers. In keeping with long-standing publication practices, ASTM International maintains the anonymity of the peer reviewers. The ASTM International Committee on Publications acknowledges with appreciation their dedication and contribution of time and effort on behalf of ASTM International.

Printed in Bridgeport, NJ December 2002

The Symposium on The Use of Glass in Buildings was held in Pittsburgh, Pennsylvania on 14 April, 2002. ASTM International Committee E06 on Performance of Buildings served as its sponsor. The symposium chair of this publication was Valerie L. Block.



ASTM C 1036: Does It Work for Field Inspections of Surface Blemishes?-TED W. MAZULA AND IVAR HENNINGS

Codes and Standards Affecting Glass in Buildings: The U.S. and Beyond-VALERIE L. BLOCK The


Impact of Serf.CleaningGlass---CHRISTOPHER J.


PC.Based Stress Measuring System for On-line Quality Control of Tempered and Heat.Strengthened GlasS---ALEXS. REDNER



In-Situ Dew-Point Measurement to Assess Life Span of Insulating Glass



Evaluation of the Condensation Resistance Rating as Determined Using the NFRC 500 Progedure----DAN1EL J. WISE AND BIPIN V. SHAH



Structural Performance of Laminated Glass Made with a "Stiff" Interlayer-STEPHEN J. BENNISGN, C. ANTHONY SMITH, ALEX VAN DUSER, AND ANAND JAGOTA


Development of Design Methodology for Rectangular Glass Supported on Three Sides to Resist Lateral Uniformity Distributed Loads-MOSTAFA M, EL-SHAM! AND H. SCOTT NORVILLE

66 79

Wind Load Resistance of Large Trapezoidal Glass Lites--H. scoyr NORVILLE,


Window Glass Design Software--STEPHEN M. MORSE A Thermal Stress Evaluation Procedure for Monolithic Annealed Glass-W. LYNN BEASON AND A. WILLIAM LINGNELL





Retrofitting Commercial Structures with Laminated Glass to Withstand Hurricane E f f e c t s - - P A U L E. BEERS, MARK A. PILCHER, AND


Testing of Annealed Glass With Anchored-Film Glass Retention Systems for Fallout Protection after Thermal Stress Cracking--BRUCE S. KASKEL,



The Advantages of Glazing in Overall Security Strategy--MiCHAELBETTEN


147 153

The Relationship Between Sprinkler Systems and GlasS--JERRY RAZWICK Design Procedure for Blast.Resistant Laminated Glass--H. SCOTTNORVILLE

159 171


This book represents the work of numerous authors at the first Symposium on the Use of Glass in Buildings, April 14, 2002, Pittsburgh, PA. Architectural glass was the broad focus for this symposium. Papers and presentations were targeted to deliver information the user may find useful related to the quality, design, use, and performance of architectural glass. The symposium had a broad focus that incorporated a variety of glass-related topics. Emphasis on glass design was also a key feature to the symposium. The papers contained in this publication represent the commitment of the ASTM E-06.51 subcommittee to providing timely and comprehensive information on glass used in buildings. Common themes throughout the tenure of this symposium can be found in this issue. Papers discussing quality issues, performance assessments, glass design glass in hurricane-prone areas, and glass for fire safety and security were presented.

Quality Issues
Quality issues were addressed from several points of view. One paper focused on the problems associated with the use of ASTM C1036 for field inspections of glass. Another paper examined the interrelationship between building codes and glass standards. A third paper discussed an on-line quality control measuring system for tempered and heatstrengthened glass. A fourth paper assessed the impact of self-cleaning glass.

Performance Assessments
The intent of this section was to present developments around the performance of insulating glass and glass facades. One paper discussed in-situ dew point testing to assess life span of insulating glass units. A second presented an assessment of annual energy consumption of ventilated double glass facades using computer simulation. A third paper focused on the evaluation of a condensation resistance rating as determined using the National Fenestration Rating Council (NFRC) 500 procedure.

Glass Design
A series of papers were presented on glass design. One paper examined the structural performance of laminated 'glass made with stiff interlayers. Several papers dealt with design methodologies for glass, including rectangular window glass supported on three sides, large trapezoidal window glass lites, and window glass design software based on ASTM El300. Another paper introduced a new procedure for thermal stress evaluation of monolithic glass.

Glass in Hurricanes
Glass used in hurricane-prone areas requires special design consideration. In this session, one speaker addressed retrofitting commercial structures with laminated glass to withstand hurricane effects. A second paper discussed testing of annealed glass with anchored-film glass retention systems.



Glass for Fire Safety and Security

This section was developed to cover a broad spectrum of topics, including security glazing, fire rated glass and sprinklers, and a design procedure for blast resistant laminated glass.

Ms. Valerie Block

Narberth, PA


Ted W. Mazula I and Ivar Hennings 2

ASTM C 1036: Does It Work for Field Inspections of Surface Blemishes?

References: Mazula, T.W. and Hennings, I., "ASTM C 1036: Does It Work for Field Inspections of Surface Blemishes?" The Use of Glass in Buildings, ASTM STP 1434, V.

Block, Ed., ASTM International, West Conshohocken, PA, 2002.

Abstract: Glass can be damaged after installation, and often the home or building

owner is left trying to determine if the resulting surface damage is acceptable. Glass quality is addressed in ASTM C1036, Standard Specification for Flat Glass. However, this standard is not intended for use in the field. It is useful for the proper specification of glass quality, and in lieu of any other field inspection standards, parts of ASTM C1036 are helpful in defining acceptable scratch criteria.

Keywords: damaged glass, scratched glass, glass inspection, glass specification, glass


Inspecting scratched glass in the field is far from an exact science. It is quite common for the project specifications to overlook the type of scratches that are acceptable. The owner and contractor are both exposed to risk in this situation. When a project has damaged glass, the parties look for an industry quality standard, and often turn to ASTM C1036, Standard Specification for Flat Glass to inspect the glass. Under ASTM C1036, medium-intensity scratches are allowed for glass quality Glazing Select (Q3). This level of quality is recommended for architectural applications including reflective and low emissivity coated glass products, and other select glazing applications. It is the most commonly specified quality of glass in the industry I and refers to Table 4 criteria (Figure 1) for the maximum allowable blemishes for 6.0 mm (1/4 in.) or less glass thickness. 1Associate Consultant, Glazing Consultants, Inc., 1325 Rotonda Point, Ste. 329, Lake Mary, FL, 32746. 2Vice President, Glazing Consultants, Inc., 11910 Cypress Links Drive, Fort Myers, FL, 33913.

Copyright9 by ASTM International


ASTM C 1036-01 Table 4 Allowable Linear Blemish Size and Distribution for Cut Size and Stock Sheet Qualities Thicknesses 6.0 mm (l/4in.) or LessA Linear Blemish SizeB Intensity Length Faint < 75ram (3in.) Faint > 75 mm Oin.) Light < 75 mm (3in.) Light > 75 mm (3in.) Medium _<75 mm (3in.) Medium > 75 mm (3in.) Heavy < 150 ram (6in.) Q3 Quality 3 Di~tribr Allowed A/lowed Allowed Allowed Allowed with a minimum separation of 600mm (24in.) None Allowed None Allowed

Heaw > 150 mm f~iin.) None Allowed Glass thicker than 6.0 mm (1/4 in.) and less than or equal to 12.0 mm (1/2 in.) may contain proportionally more and longer blemishes. Table 4 does not apply to glass thicker than 12.0 ram (t/2 in.). Allowable blemishes for glass thicker than 12.0 mm (1/2 in,) shall be determined by agreement between the buyer and the seller. n See 6.1.5 for detection of linear blemishes. Table 4 Blemisll lnt~p~ty Chart (continued) Deteetigl~Dist~ce Blemish Intensity Over 3.3 meters ( 132 in.) 3.3 meters (132 in.) to 1.01 meters (40 in.) 1 meter (39 in.) to 0.2 meters (8 in.) Less than 0.2 meters (8 in.) Heavy Medium Light Faint

Figure 1 - Example of Table 4 Table 4 defaults to allow m e d i u m intensity scratches that are 75 m m (3 in.) long providing any two scratches are not less than 609 m m (24 in.) apart. The inspection is to be conducted per item 6.1.5 Detection for Linear Blemishes (Scratches, Rubs, Digs, and Other Similar Blemishes) as follows: Place samples in a vertical position to the viewer. The viewer shall stand approximately 4 m (160 in.) from specimen and look through the sample at an angle o f 90 ~ (perpendicular) to the surface using daylight (without direct sunlight), or other uniform diffused background lighting that simulates daylight, with a


minimum illuminate of 160 foot-candles. The viewer shall move towards the specimen until a blemish is detected (if any). The distance from the viewer to glass surface when the blemish is In'st detectable is defined as the Detection Distance. Blemish intensity is determined by comparing the Detection Distance to the Blemish Intensity Chart at the bottom of Table 4. Blemish Length is determined by measuring the perpendicular distance between the ends of the blemish. Homeowners do not want to look through scratched glass, especially if they have paid top dollar for a condominium overlooking the ocean. When the sun is setting, even a small scratch in a patio door or window can be disturbing. The owner's first thought is to complain to the developer, who then calls the contractor for warranty service. If there are a large number of windows and/or doors with reported scratches, the cost of replacement may be substantial. In extreme cases, the homeowner may even contact a glass expert to inspect the glass and help solve the problem. The parties review the contract documents to see if the subject of glass quality has been addressed. These documents typically establish glass quality as Q3 from ASTM C 1036 or do not address the issue at all. In any event, the homeowner does not want to inspect the glass at 3.3 m (132 in.) with uniform light as required by the standard. He or she will probably inspect the glass from a much closer distance and in direct sunlight (Figure 2). The end result is that ASTM C1036 is found to be unsuitable and all parties may be forced to expend considerable time, effort, and expense to resolve the situation.~

Figure 2 - Typical Surface Blemish (Scratch)

Drawbacks to Using ASTM C1036 in the Field

Despite the difficulties of using ASTM C1036 in the field, it is still utilized to inspect installed glass. As its title suggests, the standard provides more of a guideline for "specifying" glass than it does for "field inspecting" glass. It is recognized that the industry has used segments of the procedures outlined in the standard for inspection on glazing systems installed in the field. There are, however, some inherent problems with


these guidelines. First of all, the procedures to inspect the glass allow significant latitude, which ot~en results in contradictory conclusions by separate inspectors, even on the same piece of glass. For example, an inspector that is 6 t~.-4 in. tall will view the glass differently than an inspector at 5 ft.-8 in. tall due to the geometry of the viewing angle. Second, the natural background (trees, weather conditions, adjacent buildings, etc.) at the exterior of the specimen can either draw to or detract attention from the scratch in question. These conditions will undoubtedly vary from building to building. Third, existing interior conditions perpendicular to the specimen may not provide the mandatory 3.3 m (132 in.) distance required for the inspection. Fourth, fixed glass specimens located on shear walls may not be accessible from the exterior, therefore, cleaning prior to inspection may not be possible. Inspection without consideration of cleaning the exterior glass could skew the results. Finally, requiring the inspector to view "through" the glass as defined in the standard and detect a scratch is extremely subjective and creates discord among the concerned parties.

Suggested Procedures
Quality of glass and the manner in which glass is to be inspected should be specified prior to the construction process. Specifiers need to avoid simply referencing the ASTM C 1036 "Standard Specification for Flat Glass" in general terms. They should scrutinize the ASTM C 1036 Standard to indicate the glass classification (i.e. type, class, style, form, quality, and finish). The typical 6 mm (% in.) thick clear glass product can be represented in Specifications as follows: "Type I - (Transparent Glass, Flat), Class 1 Clear, Glazing Select Quality (Q3) - intended for architectural applications including reflective and low emissivity coated glass products, and other select glazing applications. Blemishes for Type I (Transparent Glass, Flat) shall not be greater than those listed in Table 4." To achieve a higher quality on projects, specifiers should consider specifying Select Quality (Q3) adding criteria as follows: Glass surfaces with detectable linear blemishes that exceed Light Intensity will not be accepted (refer to Table 4 in the standard). In addition to tightening the specifications, proactive steps should be taken by inspecting the glass at key points in the construction schedule to identify if glass damage is present. Implementation of a quality control program to inspect the glass during the product's life cycle from manufacturing through installation is beneficial in detecting surface damage. This requires inspection upon receipt of the product from the manufacturer to the project. Implementation of inspection "sign-off sheets" for the glass and glazing system should be completed and dated immediately after installation. This process assists in identifying damage that can occur during delivery, storage, handling, and installation. This process also establishes a post installation time-line, which can help identify the point at which damage occurred to the glass. Evaluation of this data can reduce the number of trades that may have been performing work in the immediate area where damage took place. The cost to include glass quality and field inspection guidelines in the project specification manual is minimal and is recommended for all


projects. The costs to setup and implement a daily glass inspection schedule may be substantial and should be considered on an individual project basis. Consideration of the project size, type of glass, and access for replacing glass are key components in establishing a glass inspection program.


ASTM C1036 is useful in specifying glass, however, it does not meet the needs of the industry for field inspections to evaluate damaged glass. A new document is needed that will specifically address the field inspection of glass for damage. This document should provide a clear outline with fair and consistent inspection procedures and evaluation criteria to represent all parties (developers, manufacturers, contractors and owners). Furthermore, this document should address all relevant field conditions and eliminate as much subjectivity as possible. In the meantime, frequent inspections from receipt of glass to installation are important in monitoring surface damage.

[1] Gana - Glass Association of North America, 1997 Edition, GlazingManual, p. 73.

Valerie L. Block 1
Codes and Standards Affecting Glass in Buildings: The U.S. and Beyond

Reference: Block, V. L., "Codes and Standards Affecting Glass in Buildings: The U.S. and Beyond," The Use of Glass in Buildings, ASTM STP 1434, A.B. Smith and C.D. Jones, Eds., ASTM International, West Conshohocken, PA, 2002. Abstract: This paper examines the development and adoption process of building code requirements and standards related to glass in buildings in the United States. Issues covered include safety glazing, skylights, handrails, and glass strength. The relationship between the building codes and consensus-based standard organizations, such as the American Society for Testing and Materials (ASTM), American National Standards Institute (ANSI), American Society for Heating, Refrigerating, and Air-conditioning Engineers (ASHRAE), the National Fenestration Rating Council (NFRC), and the International Organization for Standardization (ISO) will be reviewed. Specific U.S. glass requirements for safety glazing performance and glass quality will be compared to existing Mexican and Canadian requirements. At the international level, this paper will review the work of ISO Technical Committee 160, its working groups, current activities, and the interrelationship of national and international standards in the workplace.

Keywords: Glass,building codes, nationalstandards,internationalstandards,safety

glazing,and glassquality.

Building codes and standardsgo through specific development and adoption processes in the United States. In many instances,standardsarc referencedor included in the buildingcodes. There arc other cases where standardshave lead to federal regulations. The glass industryhas developed testmethods, performance and quality specifications, and practices through ASTM International and the American National Standards Institute (ANSI). Although the development process is different, the adopted standards clarify and enhance the use of glass in building construction. Over the past twenty years, other organizations have developed standards that impact glass. The American Society for Heating, Refrigerating, and Air-conditioning Engineers (ASHRAE), for example, has produced ASHRAE 90.1, an energy standard that includes building envelope requirements for commercial and high-rise residential buildings. The fenestration performance requirements for thermal transmission (Ufactor) and Solar Heat Gain Coefficient in the 90.1 standard guide the designer's selection of windows, doors, and skylights. To verify performance, manufacturers and building
i Technical Director, Primary Glass Manufacturers Council, 2945 SW Wanamaker Drive, Suite A, Topeka, KS 66614-5321

Copyright9 by ASTM International


code officials look to the National Fenestration Rating Council (NFRC), a national organization that has developed measurement standards for fenestration, as well as a certification and labeling program to assure compliance. Standards writing, testing, and certification activities have occurred in other countries and at the international level. This paper will explore the interrelationship between the building codes and industry standards in order to establish their significance in building construction.

U.S. Building Codes

For many years, three regional building code organizations developed and published building codes in the United States. They were the Building Officials and Code Administrators (BOCA), the International Conference of Building Officials (ICBO), and the Southern Building Code Congress International (SBCCI). These regional code organizations developed "model" building codes that could be adopted by a state or used as a model for a state-developed building code. Because the code groups themselves recognized the duplication and, often times, confusion in building code requirements from one model code to another, the three regional code organizations united in 1994 to form the International Code Council (ICC). The ICC is a nonprofit organization dedicated to developing a single set of comprehensive and coordinated national codes that identify minimum health, safety, and general welfare standards. In 1998, the ICC published its first set of construction codes that included: 9 International Building Code 9 International Residential Code 9 International Electrical Code 9 International Mechanical Code 9 International Fire Code 9 International Plumbing Code 9 International Energy Conservation Code While the three model codes still exist, they are no longer being updated and the regional building code organizations actively promote state adoption of the new I-Codes. As part of the code development process, interested individuals may submit new code proposals and code change proposals. Public hearings are held to give individuals and organizations access to the code development process. At the hearings, individuals can speak for or against a proposal. Committee recommendations are sent to the ICC code official members for ratification and a final vote. These members consist of public building and fire officials from local communities across the country. As impartial officials, they have no vested interest in any specific building product.[1]



Glass requirements are found in Chapter 24 of the International Building Code (IBC), Section R308 of the International Residential Code (IRC), and in the International Energy Conservation Code (IECC). The requirements of the IBC are diverse and focus on wind, snow, and dead loads on glass, sloped glazing and skylights, safety glazing, glass in handrails and guards, glazing in athletic facilities, and glass in floors and sidewalks. (Table 1). The IRC specifically addresses safety glazing and skylights (Table 2), and the IECC includes requirements for thermal transmission (U-factor), Solar Heat Gain Coefficient, Visible Transmittance, and air leakage. The glass and fenestration industries have a voice in the code process via their trade associations or on an individual company basis. The Glazing Industry Code Committee (GICC) has represented the interests of the U.S. glass and fenestration industries for many years and, through its code consultants, has actually participated in writing the chapters on glass in the codes. Federal regulations and other consensus standards are often referenced in building codes. For instance, within Section 2406, Safety Glazing, of the International Building Code, the federal regulation, CPSC 16 CFR 1201, and the American National Standard, ANSI Z97. I, are referenced. Both of these standards contain test methods to evaluate the safe performance of glass. In Section 1609.1.4 of the International Building Code, ASTM E1886 and E1996 are referenced under Protection of Openings. These ASTM standards give testing information and use parameters for fenestration and storm shelters installed in hurricane-prone areas. In some cases, parts of the standards are included verbatim in the code. For example, in Section 2405 of the IBC on Sloped Glazing and Skylights, twelve nonfactored load charts are reprinted from ASTM E1300, Standard Practice for Determining Load Resistance of Glass in Buildings. Each chart covers a specific thickness of monolithic annealed glass. By using the charts and appropriate factors for single and insulating glasses, the building official is able to confirm the load resistance of glass. The final part of the building code process is adoption through the appropriate local or state legislative and administrative processes. It is important to note that until the building codes are adopted, they have no basis in law and are much like any other standards that may or may not be referenced by the designer of a building. Voluntary standards that are referenced or excerpted in the code become mandatory once the code has been adopted.

TABLE 1 - - 2000 International Building Code, Chapter 24 Glass and Glazing. General Definitions General Requirements Wind, snowand dead loads on glass Sloped glazingand skylights Non-factoredload charts SafetyGlazing Coversglass, light-transmittingceramicand light-transmitting plastic panels; glazingreplacement DaUeGlass, DecorativeGlass, etc. Identification(labeling),glass supports, framing,interiorglazed areas, louveredwindowsor jalousies Vertical glass Allowable glazingmaterialsand limitations; screening Vertical and slopedglazing; factors Human impactloads, identificationof safetyglazing,multiqight



Glass in handrails and guards Glazing in athletic facilities Glass in Floors and Sidewalks

assemblies, hazardous locations, fire departmentaccess panels Materials, loads, support, parking garages Testing Design loads, laminated glass, desi~ formula

TABLE 2 - - 2000 International Residential Code, Section R308 Glazing. Identification Permanent label required for safetyglazingunless building code official approvesof certificate; tempered spandrel may have removalpaper label, tempered glass must have permanent label One pane fully labeled, others can have "16 CFR 1201" No thinner than 4.76mm(3/16 inch), no longerthan 1219 mm (48 inches) ; wired glass prohibited with wire exposed on longitudinal edges Same as IBC Must complywith 2404 of IBC Definition, permitted materials, screens, screens not required, glass in greenhouses, screen characteristics, curbs for skylights

Identificationof multipane assemblies Louveredwindows or jalousies Hazardous locations Site built windows Skylights and sloped glazings

U.S. Standards There are thousands of voluntary standards in the United States that benefit both the manufacturer and user of products. They solve issues of product compatibility and address consumer safety and health concerns. According to Amy Marasco, "Standards also allow for the systemic elimination of non-value-added product differences (thereby increasing a user's ability to compare competing products), reduce costs, and oRen simplify product development."[2] Two of the most important standards organizations in the U.S. are the American National Standards Institute (ANSI) and the ASTM International. The American National Standards Institute was founded in 1918 with the purpose of ensuring that U.S. voluntary standards minimize waste, duplication of efforts, and conflict. The ANSI process is based on determining whether a standard meets the necessary criteria to be approved as an American National Standard. The approval process verifies that the principles of openness and due process have been followed and that a consensus of all interested parties has been reached.[3] Standards consider the needs of producers, users, and other interest groups. An appeals process and a requirement for balance assure that no one interest can manipulate the process unfairly.



ASTM traces its roots back to the 19th century and the driving force of Charles Dudley, a chemist with the Pennsylvania Railroad. Dudley issued standard material specifications for the company's suppliers of oil, paint, steel, and other materials. To alleviate problems, he organized technical committees to discuss the specifications and testing procedures and form consensus.[4] Today there are more than 80 technical committees involved in a wide range of activities. ASTM standards include material standards that cover quality of a building product, engineering standards that cover product design, and testing standards covering the product performance. In 2002, ASTM changed its name to ASTM International and its focus from that of a national standards organization to one with an international scope. According to ASTM President Jim Thomas, "ASTM's method of developing standards is based on consensus without borders. Our process ensures that interested individuals and organizations representing academia, industry, product users, and government alike all have an equal vote in determining a standard's content. Participants are welcome from anywhere on the globe."[5]
International Standards

In the 1980s and 1990s, global manufacturers began to demand international standards to minimize confusion caused f~om many proprietary and regional standards. These standards are beneficial because they often reduce time-to-market and lower product development costs. In addition, global standards facilitate the introduction of products to abroad range of countries, including developing countries.[6] Consumers also benefit from products that are safer and of a higher quality. The glass industry is actively developing international standards through the International Organization for Standardization (ISO), a non-gnvemmental organization established in 1947. ISO is a worldwide federation of national standards bodies from some 140 countries, one from each country.J7] National bodies are characterized as Participating (P) members, Observer (O), or Liaison members. "P" members are responsible for submitting votes and/or comments on all technical matters coming before the committee. They send delegates to meetings, offer candidates for leadership positions, and host meetings. "O" members monitor the technical work, but do not actively participate. They have no power of vote within the committee, although they may attend meetings. Liaison members have no power of vote, but are able to attend meetings and receive documents. They are typically other committees within ISO with related interests or other international organizations. The United States is a "P" member of many ISO Technical Committees through ANSI. While there are other national standards organizations in the U.S., ANSI is responsible for submitting the U.S. vote on ballots issued by ISO. In addition, ANSI offers U.S. participants training and support on international procedures and standards writing. ANSI interfaces at the international level, regional level, and national level with a variety of standards organizations. (Fig. 1).



Public Sector

Private Sector


Intl. Standards Bodies


I I io--.i
Standards Bodies


Standards Bodies








ron I





Figure 1 - - Standards







American ArchitecturalManufacturers Association American National Standards Institute American Society for Civil Engineering British Standards Institute European Committee for Standardization Pan American Standards Commission European Telecommunications Standards Institute Institut Beige de Normalisation International Electrotechnical Commission International Organization for Standardization National Fire Protection Association Ente Nazionale Italiano di Unificazione



Glass interests are organized under ISO's Technical Committee (TC) 160, Glass in Buildings. The technical committee (TC) has two subcommittees, one on glass properties and the other on glass uses. Within the subcommittees are smaller working groups consisting of national experts who collaborate on the development of intemational standards. These standards may contain requirements found in other national standards, but in many cases they are totally new standards developed through the expertise of working group members. There are fifteen TC 160 working groups. Under subcommittee one, there are working groups on basic glass products, toughened glass, laminated glass, insulating glass units, mirrors, coated glass, glass blocks and glass paver units, and curved/bent glass. Under subcommittee two, working groups are developing standards on the design strength of glazing, light and energy transmission properties and thermal properties of glazing, airborne sound insulation, fire resistant glazed assemblies, assembly rules and structural sealant glazing, safety glazing tests, and security glazing tests. Within each group, work is underway to produce international standards. One working group, for example, has produced seven draft documents on security glazing tests. These drafts include test methods and classifications for destructive windstorm resistant glazing material, glazing subject to arena airblast load, explosion resistant glazing (shock tube loading), bullet resistant glass, and forced-entry resistant glazing products (tests include repetitive ball drop, repetitive axe and manual attack). Another is working on a test method for safety glazing. Each of these standards must eventually go through a ballot review process in order to become a recognized ISO standard. The ISO standards development process is based on consensus. This means that there is general agreement, but does not imply unanimity. ISO adheres to established target dates. Once a New Work item has been approved, a working draft must be approved within six months. A Committee Draft is required 18 months after the working draft has been submitted for review. Once a Committee Draft has been reproduced, 36 months are allowed to take the draft through the ballot process to final publication. Approval of a final standard is based on acceptance by a two-thirds majority of Pmembers voting and not more than a quarter of the total votes cast being negative. The glass industry participates and responds to ISO ballots through a Technical Advisory Group (TAG) that is administered by ASTM. The TAG is recognized as Task Group C14.92 under ASTM C14 Glass and Glass Products. Interested U.S. companies, organizations, and individuals can become members of the U.S. TAG. Experts are appointed from the TAG to represent the United States at interuational Working Group, Subcommittee, and Full Committee meetings oflSO TCI60. In this way, the U.S. position is heard around the world.
Recognizing the Need for Codes and Standards

On August 24, 1992, Hurricane Andrew hit Dade County, Florida causing $20 billion in property damage. This catastrophic weather event resulted in a change to the South Florida Building Code requiring all external glazing material to be either capable of resisting windborne debris or to be protected by shutters.[8] Three other South Florida counties and areas along the Gulf Coast in Texas enacted similar requirements. The



building officials used Australian data supplied by the glass industry to develop these windborne debris requirements. Soon after South Florida building officials adopted requirements for windborne debris, as ASTM Working Group organized to develop a consensus test method and specification that addressed requirements for glazing subjected to the severe effects of wind events. This resulted in two standards, ASTM Test Method for Performance of Exterior Windows, Curtain Walls, Doors and Storm Shutters Impacted by Missile(s) and Exposed to Cyclic Pressure Differentials (El 886) and ASTM Specification for Performance of Exterior Windows, Curtain Walls, Doors and Storm Shutters Impacted by Windborne Debris in Hurricanes (El 996). Once these standards were adopted as consensus standards, proponents moved them into the building code arena where they were eventually adopted into the International Building Code. In the area of glass quality, ISO TC 160 SC 1 Working Group 1 has reviewed the CEN, U.S. and Japanese quality standards in order to dratt an ISO standard on physical and mechanical properties of soda-lime silicate float glass and stock sized and cut sizes of fiat glass. Since basic glass products are sold around the world, these international standards will facilitate international trade and communication by defining clear and unambiguous provisions. By allowing input from key producers, the standards will be consistent and accurate, and will represent the state of the art in float glass production capabilities.
Federal Standards

In the late 1960s, there was a pattern of social activism in the United States that gave rise to a grassroots consumer rights movement. Many industry standards were developed at this time. One such standard is the American National Standard Z97.1. This standard was initiated by the glass industry as a means of reducing glass-related injuries. Despite its adoption as an industry standard, it became clear that it would only be effective if it were adopted as a regulation. In 1972, the Consumer Products Safety Act was passed, establishing a federal commission, the Consumer Product Safety Commission (CPSC), with the power to promulgate consumer product standards. With the support of industry, labor, safety, and general interest groups, the federal commission granted a petition to develop a federal safety standard for architectural glass, and in 1977, the CPSC standard 16 CFR Part 1201 was enacted by the federal government. The standard, like the voluntary ANSI standard, was designed to reduce or eliminate the unreasonable risks of injury associated with architectural glazing materials.J9] Though unlike ANSI Z97.1, CPSC 16 CFR 1201 was mandatory for all parts of the United States. Initially, the CPSC standard applied to glazing in doors and other glazed panels in hazardous locations, such as sidelites and panels adjacent to walkways. However, in 1981, CPSC withdrew its glazed panel provisions to permit regulation and enforcement of glazings in those locations by state and local building code authorities. State and local building code officials were expected to impose criteria for the use of glass, subject to human impact, that were consistent with the regulations of CPSC. As required by the federal preemption mandate, the three regional model code bodies enacted safety glazing provisions for all hazardous location applications conforming to CPSC standards. When



the three regional codes united to produce the Intemational Building Code, these requirements were incorporated into the new building code. Today hazardous locations requiring labeled safety glazing materials in the model building codes are defined to include: [ 1O] * 9 Glazing in swinging doors except jalousies. Glazing in fixed and sliding panels of sliding patio door assemblies and panels in other doors, including walk-in closets and wardrobes. 9 Glazing in storm doors. 9 Glazing in unframed swinging doors. 9 Glazing in doors and enclosures of hot tubs, whirlpools, saunas, steam rooms, bathtubs and showers. 9 Glazing in any portion of a building wall enclosing these compartments. 9 Glazing in an individual fixed or operable panel adjacent to a door. 9 Glazing in individual fixed or operable panels where the exposed area of an individual pane is greater than nine square feet and the exposed bottom edge is less than 18 inches above the floor, the exposed top edge is greater than 36 inches above the floor, and one or more walking surface(s) are within 36 inches horizontally of the plane of glazing. 9 Glazing in guards and railings, including structural baluster panels and nonstructural in-fill panels. 9 Glazing in walls and fences enclosing indoor and outdoor swimming pools and spas. 9 Glazing adjacent to stairways, landings and ramps.

The model codes refer to the Consumer Product Safety Commission standard 16 CFR Part 1201 for impact test parameters. With the exception of polished wired glass, all safety glazing products must meet Category I or Category II requirements of the CPSC standard.

Safety Glazing and Glass Quality Standards, the North American Experience
Safety glazing regulations, standards and code requirements have been in existence for decades in the United States, but only recently has Mexico adopted a safety glazing/quality standard for glass. It is an official Mexican standard, NOM-146-SCFI2001, Productos de vidrio - Vidrio de seguridad usado en la construcci6nEspecificaciones y m6todos de prueba. Although many parts of the Mexican standard were modeled after U.S. standards, there are some notable differences. For one, the scope of the Mexican standard is broader than the U.S. standard. It does establish a minimum level of safety in order to reduce the threat of injury to people from glass breakage caused by human impact, but it also covers physical attack, accidental and natural events, and acts of aggression and vandalism. The Canadian safety glazing standard is similar to the U.S. standard. (Table 3)



TABLE 3 --

Comparisonof Safety GlazingRequirements.

CAN/CGSB 12.1" CANADA Tempered or Laminated Glass Type l=Laminated Type 2=Tempered NOM-146-SCFI-2001** MEXICO Tempered or Laminated Glass Class l=Tempered Class 2=Laminated Type A=PVB Type B=Resin 1 = 0.46 m (18 inch) 2 = 1.22 m (48 inch) 3-3.0m (9.8 ft) 4 = 6.0 m (19.6 fl) 5 = 9.0 m (29.5 t~) 86.3 cm x 193 cm (34 76 inch) Included in safety standard Single impact Multiple impacts for Levels 3-5 Boil, humidity, accelerated weathering Yes 3


CPSC 16 CFR 1201' U.S. Safety Glazing None


Cat 1=18 inch drop < 9 square feet Cat 1I--48 inch drop


Largest manufactured up to 34 x 76 inch N/A Single impact

Cat I = 460mm (18 inch) drop < .8 square meter Cat II = 1,220 mm (48 inch) drop > .8 square meter (9 SQ FT) Up to 865 x 1,930 mm (34 x 76 inch) CAN/CGSB-12.2 and 12.3 Single impact


Boil, weathering None Not specified, for nonsymmetrical equal number of specimens from each side No opening>3 inch/4 lb sphere can pass 10 largest particles < 10 sq. in. Does not remain in frame and no break No break Permanent label or paper certificate: Standard reference Date of manufacture Place of manufacture Name of manufacturer

Boil Yes 4 Asymmetrical, alternate impacts


No opening > 3 inch/4 lb sphere can pass 10 largest particles < 10 sq. in. No break

No open > 3 inch 10 largest particles < l0 sq. in. Does not remain in frame and no break No break


Legibly and permanently marked: Manufacturer name or logo Classification and/or designation of the glass, level of resistance "Made in Mexico" NOM- 106-SCFI *Safety Glazing Testing Comparison, GlassMagazine,September 2001, pp. 60-61. **Norma Oficial Mexicana NOM-146-SCFI-2001, Productos de vidrio-Vidrio de segnridad usado r la construeci6n-Especificaciones y m~odos de prueba.

Legibly and permanently marked: Manufacturer name or logo CAN/CGSB- 12. I-M M-I for Cat I only



Certification Product certification demonstrates compliance with standards and regulations and is mandatory to place a product on the market in many countries. Certification programs are designed to assure high quality and performance. In some countries, the government regulates the certification process. In other countries, manufacturers are able to selfcertify or test and certify through third party, independent testing laboratories. A permanent label usually identifies program compliance. The Safety Glazing Certification Council (SGCC) offers manufacturers in the United States an opportunity to test, certify, and label products to both the federal safety glazing standard CPSC 16 CFR, Part 1201, and the voluntary ANSI Z97.1 safety standard. The SGCC is a nonprofit corporation established in 1971 by manufacturers, building code officials and others interested in public safety. The Council is responsible for conducting independent routine sampling and the testing program, approving and registering the form of a Licensee's label, and withdrawing authority to use that label if products do not meet specifications.[11] Over the years, many North American companies have participated in this certification program. The National Fenestration Rating Council (NFRC) provides a framework for testing, certification, and labeling of fenestration products. While the NFRC program is broader in scope than SGCC, it essentially provides the same benefits. Manufacturers test their products to determine thermal and solar performance. Once testing information is available, product performance can be determined through simulation. Participants in the NFRC program are required to follow specific labeling guidelines. NFRC standards are now referenced in the International Building Codes, as well as in the ASHRAE energy standards. Trade associations have also recognized the value of certification programs in the United States. The American Architectural Manufacturers Association (AAMA) has maintained a certification program since 1962. This ANSI-accredited Certification Program has given manufacturers a way to independently demonstrate product performance. AAMA's Certification Label tells customers that products have been verified as conforming to ANSI/AAMA/NWWDA 101.I.S. 2 standard. The Hallmark Certification Program developed by the Window and Door Manufacturers Association (WDMA) consists of a series of inspections and tests to determine that products are being manufactured in the same way in which they were tested. Products are evaluated by performance requirements in one of the WDMA standards or test methods. The European community has embraced certification of products with its Construction Products Directive (CPD). This legislation calls for all products used in buildings to satisfy six basic criteria: 1. 2. 3. 4. 5. 6. Mechanical Resistance and Stability Safety in Case of Fire Hygiene, Health and Environment Safety in Use Protection Against Noise Energy and Heat Retention



In order to comply with the CPD, performance tests are required to demonstrate the ability of a product to perform a particular function, such as fire resistance, as well as factory production control tests to demonstrate that the product continues to pass the required tests. The CPD requires the product to have a CE mark in order to be "placed on the market."[ 12]

Building codes and standards rely on test methods, performance and quality specifications, and practices that define product usage. Building codes provide regulations for adoption and enforcement. Industry standards provide definitions, classifications, procedures, measurements of quality and many other important requirements or conditions to which a product or material must conform. Together, they raise the bar on quality and safety of building construction; and the quality and performance of manufacturing in the United States. As more products are sold globally, the number of international standards will increase, offering the same benefits and providing a common basis for product usage and understanding around the world.

[I] Nickson, R., "Consensus Codes-Does It Matter?" ICC Newsletter, June 2001, p.4. [2] Morasco, A. A., "Standards Development: Arc You at Risk?" ASTMStandardization News, June 2000, p.22. [3]Morasco, A. A., "Standards Development: Are You at Risk?" ASTMStandardization News, June 2000, p.22. [4] "Innovation by Consensus: ASTM's First Century," http://208.211.80/ANNIVERJconsensus.htm. [5] "A New Name A Longstanding Commitment," ASTMStandardization News, January, 2002, p.25. [6] Sterling, J., "Going Global," ASTMStandardization News, June 2001, p.27. [7] "What is ISO?" International Organization for Standardization, [8] Smith, W. D., "Hurricane Glazing Building Codes Continue to Evolve," Glass Magazine, September 2001, p.68. [9] Block, V. L., "What the Safety Regulations Are for Glass," Glass Magazine, September 2001, p.68. [ 10] For complete code requirements see Chapter 24 of the International Building Code. [11] CertifiedProducts Directory, July 2001, Safety Glazing Certification Council, P.O. Box 9, Henderson Harbor, New York 13651, p.7. [12] Colvin, J., "The Effects of European Standardization on the Smaller Company," Glass ProcessingDays Conference Proceedings, Tampere (Finland), 2001, pp.715-718.

Christopher J. Barry, t and Thomas O'Day:

The Impact of Self-Cleaning Glass

Reference: Barry, C. J. and O'Day, T., "The Impact of Serf-Cleaning Glass," ASTM STP 1434, Use of Glass in Buildings, V. Block, Ed., ASTM International, West Conshohocken, PA, 2002.

Abstract: Today there is yet one more invisible coating available to improve the properties of window glass. At first there were the nearly invisible, low-emissivity coatings, which admit daylight. Some of them can also admit beneficial passive solar gain. These prevent winter heat loss by reflecting, or not emitting, long-wave (10 micrometer wavelength) infrared thermal radiation. These low-emissivity coatings also enhance the effectiveness of heat absorbing solar control tinted glass, and reflective coatings, by preventing absorbed solar heat from radiating towards the room side of a window. Now clear, self-cleaning coatings are available for the outer surface of the window. These coatings act in different ways to prevent the deposition and build-up of dirt. Some can rinse inorganic dust offthe glass with rain or water by their hydrophobic, or hydrophilic properties. Some of them can break down deposited organic dirt using a catalytic action powered by the ultraviolet component of daylight.
Keywords: glass, self-cleaning, hydrophobic, hydrophilic, photocatalytic

Self-cleaning glass has to deal with organic and inorganic dirt. The former is composed of molecules containing a carbon atom that can be broken down by chemical reactions. Inorganic dirt is found as dust and grit from fine earth or sand particles from road dust. Inorganic materials are not broken down by chemical means but must be prevented from sticking to glass, or must be removed from it by breaking down the adhesive which holds it in place. A third form of dirt on windows is seen when salt from sea spray, or minerals and inorganic salts from lawn sprinklers, create deposits on glass as the water evaporates.

1Director of Technical Services, Pilkington North America Inc., Toledo, OH 43697. 2 Sales and Marketing Dept., Pilkington North America Inc., Toledo, OH 43697.

Copyright9 by ASTM International



Clear coatings on glass for self-cleaning or dirt-resisting properties can be temporary, hand applied, on installed glass, or permanent; applied while the glass is being made, either by vacuum deposition or by the pyrolytic chemical vapor deposition (CVD) processes. This paper addresses only the permanent dirt resisting and self-cleaning coatings and not those hand-applied solutions which need to he reapplied periodically.

Benefits of Self-Cleaning Glass

The immediately obvious benefits, besides convenience, are as follows:

Economic Benefits
For commercial buildings where professional window cleaning services are used and their cost is known, it is very easy to calculate an economic case for self-cleaning glass. Under normal conditions one could expect the need for manual cleaning to he at least one half to one quarter as frequent as for plain glass.

A study of"Worker Deaths by Falls" [1] over a 15-year period, was undertaken by the Department of Health and Human Services, National Institute for Occupational Safety and Health, in September 2000. They found that in the United States there were 88 reported falling accidents involving window cleaners, of which 62 were fatalities.

The continuous cleaning action of the self-cleaning window means that under normal weather conditions, where rain can occasionally rinse the glass, the level of visible dirt can he expected to stabilize. This is in sharp contrast to ordinary glass where the level of dirt continues to accumulate until someone decides it's time to clean the windows.

Self-Cleaning Methods
Self-cleaning is partly effected by controlling the action of rain water. This can be done either by repelling water (hydrophobic), or by attracting it (hydrophilic).

Water Repelling
Automotive applications can use hydrophobic coatings which increase the wetting angle of water drops. This is aided by the higher wind speeds for cars as compared to buildings, which help carry off dirt containing water drops.



105~ a wetting

Figure 1 - Hydrophobic Wetting Angle

Water Attracting
Perhaps surprisingly, the action of a water attracting (hydrophilic) coating is very effective as a self-cleaning method. When water droplets coalesce on a vertical or sloped surface they form a sheet of water which slides down under the influence of gravity, pulling with it inorganic dirt particles. At the top edge of the water sheet a thin section shows interference colors as it slides down. This 'Invisible SqueegeerM' action leaves behind a relatively clean and dry glass surface, without droplets or rivulets. These are the droplets and rivulets which dry on ordinary, non-coated, glass and leave spots behind as the mineral content and inorganic particles are concentrated in one location. The action of solar ultraviolet light, either direct, or indirect as reflected from clouds, can charge a titanium dioxide (TiO2) coating by raising the outer electrons to a higher band. In its charged state it becomes hydrophilic, with a wetting angle in the 10 degree range.

10~ w

Figure 2 - Hydrophilic Wetting Angle The build-up of salt deposits from sea water spray, or mineral deposits from hard water in lawn sprinklers on a hydrophilic coating will not be different from ordinary glass because of the salts' inorganic nature. But it is expected that the hydrophilic action of an activated coating will make it much easier to rinse off salt deposits with a hose. Mineral deposits from hard water should be prevented from building up. If it is necessary to rinse glass with hard water then a few drops of liquid dish washing detergent can be added as a surfactant to prevent droplet formation. The best method is to rinse off dust with a portable, 1 or 2 gallon, hand-pump pressurized, garden spray bottle with plain distilled water from a hardware store.



Photocatalyfic Breakdown

A titanium dioxide (TiO2) coating acts, in a simplified description, by first having

its electrons raised to a higher level where they react with water vapor molecules to
create OH radicals. The direct or indirect, reflected UV light from the sun, present in all outdoor daylight, performs the charging action on the electrons. The created OH radicals react with organic dirt on the glass, breaking it down into carbon dioxide (CO2) and water vapor (H20) gases in an accelerated version of naturally occurring decomposition. One manufacturer of self-cleaning glass calls their product "PhotoActivrM'' to illustrate this action. It should be noted that indirect UV reflected from clouds and buildings is sufficient to activate the'coating. Fully activated coatings have been seen on north elevations, behind insect screens and under roof eaves.

Figure 3 - CVD Pyrolytic Process

A reactive gas mixture is presented to the freshly formed hot glass ribbon while it is still in the float bath. The higher temperature of the glass causes a reaction to occur forming a TiO2 coating on the glass.

Manufacturing by Vacuum Deposition Process

Sputter coating various materials on glass, in a vacuum chamber, can also create dirt resistant coatings. But the nature of the vacuum deposition process appears to give these coatings only hydrophilic (or hydrophobic) properties, without any effective photocataiytic activity.

Applications of Self-Cleaning Glass in Buildings

An invisible self-cleaning coating is now added to the list of available glass options to control window appearance, heat loss and heat gain, amongst other design characteristics. This new coating must be properly fabricated and installed to achieve the desired results.



Glazing Details
Self-cleaning glass must be protected from contaminants which could smother or otherwise inhibit the catalytic and hydrophilic actions. Such contaminants include metal oxide run down from copper or lead roofing and silicone oils which can be leached out, by water, from silicone sealants.

Fabrication and Glazing Tools

It is obviously vitally important that the coating is correctly installed in a window. Placing the coating incorrectly on the room side surface would not only deny it most of the UV radiation needed to activate the TiO2, but there would also be no rinsing action available from rain to remove inorganic dirt. Handheld portable detectors are available to correctly identify the photocatalytic coating. These work by emitting and detecting the reflection of UV light from the TiO2 coating. Such light would not be reflected as strongly from clear glass. Handheld tools are also available to detect low-emissivity coatings by measuring their electrical conductivity but unfortunately they do not detect the self-cleaning coatings.

The Impact of Self-Cleaning Glass The new self-cleaning coating offers yet one more option to the window designer. By intelligent selection of products, a window can now be designed and manufactured that is strong enough to resist hurricane winds and the associated debris, it can keep out unwanted solar heat gain, it can reduce winter nighttime heat loss, admit sufficient daylight, provide acoustic insulation, have a color and appearance selected from a wide range of tints and reflectivities, and finally have an exterior self-cleaning surface. Self-cleaning glass significantly reduces the labor needed to maintain the exterior surface of windows and represents a considerable cost saving to a building owner over its useful life, but it also offers reduced accident potential in an area where OSHA has reported many accidents and fatalities from falls. The self-cleaning coatings available today do not mean that windows will never have to be cleaned again. The coatings work with daylight and rain. In dry areas a light hosing will be needed to remove dust. Where large deposits of organic dirt, such as bird droppings, are involved the coating is overwhelmed by the amount of material to be broken down. Such deposits are quickly loosened by the coating and can normally be removed by hosing, but the organic breakdown of the complete deposit by the coating alone will typically take an unreasonably long time. Finally, it has been seen that most silicone sealants can release silicone oils which are not organic and do not take part in the photocatalytic self-cleaning reaction. These oils can cover the sell-cleaning coating for a band a few centimeters wide all around the window and permanently inhibit photocatalytic reaction.



The window fabricators and installers now have an added complication to their work. When the typical double-glazed window includes a clear low-emissivity and a clear self-cleaning coating, there are up to 8 or 12 different assembly combinations for positioning the coatings: only one is correct; the others are all wrong. While this problem may appear to be trivial, its occurrence has already been seen in the field. There is now a requirement upon the window fabrication and glazing industries to recognize the availability of these significant glass improvements and to promote, fabricate and install them correctly. Conclusion It is less than 20 years since clear and essentially colorless low-emissivity glass coatings have been readily available. The benefits of such coatings are now universally recognized and they can be found on commercial and residential windows in the cold north and the warm south where they give improved comfort and energy savings by day and by night, in both summer and winter. When the performance and benefits of the clear and colorless self-cleaning coatings are recognized, it is suggested by the authors that in less than 20 years these coatings too will become similarly ubiquitous and will be used in windows as readily as the low-emissivity coatings. References [1] "Worker Deaths by Falls," Department of Health and Human Services, National Institute for Occupational Safety and Health, September 2000.

Alex S. Redner l

PC-Based Stress-Measuring System for On-Line Quality Control of Tempered and Heat-Strengthened Glass Reference: Redner, A. S., "PC-Based Stress-Measuring System for On-Line Quality Control of Tempered and Heat-Strengthened Glass," Use of Glass in Buildings, ASTMSTP 1434, V. Block, Ed., ASTM International, West Conshohocken, PA 2002. Abstract: A new PC-based stress measuring system was developed for measuring edge-stress in tempered glass. The system is based on the Spectral-Contents Analysis method, modified to permit the very high-speed data acquisition needed for measuring stress in a moving item, within a region where the stress gradient is high. Keywords: Stress, tempered glass, quality control Introduction
Heat-strengthened and fully tempered glass can be as much as 3 and 7 times as strong as annealed glass, depending upon aspect ratio and other factors such as the actual residual surface compression present [1]. The additional strength is due to the presence of residual compressive stress in the surface and edge layers, offsetting tension due to handling, impact and service loads. Because the increased strength is the product's main characteristic, there is an obvious need to implement systematic Quality Control stress testing programs to assure conformance to US and foreign specifications [2]. Quality Control procedures are designed to assure the producer and the user that the product meets material specifications, has the desired strength and that test documentation supporting the test data is available. In addition, the testing should help the manufacturer maintain uniformity and production economy. ASTM C1048-97b, "Specification for Heat Treated Flat Glass - Kind HS, Kind FT Coated and Uncoated Glass," clearly defines the surface and edge pre-stress levels that must be met to satisfy "Fully Tempered" (FT) or "Heat-Strengthened" (HS) glass specifications. In addition, the C1048-9To specification includes a reference to ASTM C 1279-94, "Test Method for Non-Destructive Photoelastic Measurement of Edge and Surface Stresses in Annealed, Heat-Strengthened, and Fully Tempered Flat Glass," for measuring surface and edge compression stresses. Historically, the specification changes have been closely related to development of new stress measuring instruments. An example of this influence can be seen in the specification C1048-97b where the residual stress range allowed in Heat-strengthened glass was narrowed as a result of the availability of more accurate test methods described in C1279-94.

Copyright9 by ASTM International



These surface and edge stress levels are not explicitly stated in the European Norms prEN1863, part 1 and part 2, and prEN 12120, part 1 and part 2, [2] where only the strength is specified, and correlation between the measured surface compression and the strength is the user's burden. Assuming, however, that the surface compression increases the strength due to service loads, the European standards are nearly identical to ASTM C1048-97b. Product specifications published by automotive industries define an additional test parameter: the maximum average tensile stress occurring near the edges. In automotive glass, edge stress testing has higher emphasis, justified by the frequency of installation and service failures. In both architectural and automotive applications, conformance verification requires an accurate and economical stress measuring test procedure.

Testing Procedures
The fragmentation test is suitable only for "safety" glass. This is a destructive test seldom practiced in strict accordance with the test method. The European standard [2] also includes the required frequency of fragmentation testing and documentation. Measuring surface compression using surface polarimetry [3] [4] and edge stress using transmitted light are described in the ASTM C1279 test method. The Grazing Angle Surface Polarimeter (GASP*) is a non-destructive test instrument extensively used in architectural, automotive and "IV glass industries [4], [5]. While the GASP is extremely valuable for Quality Control, the method requires glass contact and is not adaptable to an "On-Line" process control. Measurement of edge stress offers a viable solution for this purpose. The new method described below permits "On-Line" measurements, assuring 100% quality control.

Edge Stresses in Heat-Strengthened and Tempered Glass

As shown in Figure 1, the edge of tempered glass is, in reality, just another free surface (E). The glass edge is exposed to air quenching in the tempering process [6] and develops edge-surface stress related to the temperature gradient developed during the quenching process in the layers adjacent to the edge-surface (E).

Figure 1 - Development of Residual Stresses on Edge-Surfaces



Surface compression (-) is balanced by tensile (+) stresses in the mid-layers. In regions distant from the edges, a typical parabolic distribution develops.
3.00 Compression







2.00 Tension 4.00 6.00


Figure 2 - Average Stress Measured in Transmitted Light Similarly, the edge surface compression (-) is balanced by tensile (+) stress in the region adjacent to the edge. The tensile stresses balancing the surface-compression add to the edge-balancing tension, creating a region where the average or integrated stress is tensile (+). As a result, tempered and heat-strengthened glass is substantially weakened near the edge [7]. This weakened region is of concern to the automotive industry and maximum tensile average stress is stated in automotive glass specifications. The edge stress also reveals the strength and service performance of the product. Edge stresses are routinely measured in transmitted light. As result of edge finish geometry, edges are not transparent and an extrapolation method is needed to obtain the "real edge" stress. A simple, (but not necessarily most accurate), linear extrapolation technique using results of measurements at 2 points, xl and x2 (Figure 2) is included in the ASTM C1279-94 test method. These calculations make the procedure cumbersome. The PC-based SCA method [8] automates the test and eliminates the difficulties of manual extrapolation. The speed of data acquisition of the SCA method makes it possible to implement On-Line edge stress measurements.
Automated On-Line Production Control of Prestress

The edge stress gradient is very high in a very narrow region. The average stress, measured in transmission, decreases rapidly as the distance from the edge increases. The Figure 2 shows experimental results acquired using several samples [6]. It should be noted that a 4 th degree polynomial provides an excellent fit to experimentally acquired data points. A minimum of 10 points are needed to fit the polynomial with a suitable confidence level within the critical region 2 to 5 mm from the edge. In addition, glass exiting the tempering furnace moves with a linear velocity ranging between 100 and 500 ram/see. Combining the linear speed and the length of the measured re-



gion, one finds that data acquisition near an edge must be completed in less than 0.01 second, requiring a measuring speed of 1000 data points/second. The SCA sensor described below, installed at the exit of a tempering furnace, measures stress at a speed in excess of 2000 points/see. The system yields an over-abundant number of data points to accurately establish the leading and trailing edge-stress, as well as the maximum average tensile stress in the near-edge region.

SCA Measuring Method [8]

When polarized light crosses a sheet of glass exiting a tempering furnace, the transmitted light intensity is modified. The light source (Figure 3) projects polarized light of intensity Io on the glass. The transmitted light acquires a retardation 8, related to the stress S. 8=txCxS where: fi t C S Retardation related to the Stress Thickness of material Material Constant Stress (1)

The method is using white light containing a broad spectrum of wavelength g. For each wavelength 2,., the transmitted light intensity I becomes: Ii =Iosin2~rfl/~i where: (2)

/o 2i

Intensity of light measured at ~ wavelengths Source intensity Wavelength

The light transmittance Ii/Io becomes a function of stress and of the wavelength L. Using a spectrophotometer, the light intensity Ii is measured at several wavelengths providing a set of simultaneous equations (2) sufficient to retrieve the retardation 8 and compute stress. The basic principles and applications of this method are extensively documented in several publications. This SCA method has been used extensively during the last 10 years for On-Line stress measurement in float glass. The system schematic (Figure 3) illustrates the components used to implement this method On-Line. An SCA based edge-stress scanning system used to measure edge stress in automotive glass is shown in Figure 4. A light-weight sensor is used to scan the edge of automotive glass, yielding 1000 data points/sec and a spacial resolution of 0.1 mm, demonstrating the method capability. The selection the spectral range determines the speed of





Io sin2- ~



Light Source Iii ~ l J Polarizer Analyzer i MeasuredSample ] SpectralAnalyzer I

Io sm2~rS

Figure 3 - Schematic of the SCA Measuring System

Figure 4 - Portable Stress Scanner using PC-Based SCA Method



data-acquisition, resolution and maximum retardation measuring range. The SCA systern designed for measuring edge stress was evaluated up to 8,000 nm, capable of measuring 150 MPa stresses in glass thickness up to 20 mm thick. The measuring sensitivity of the SCA sensor shown in Figure 4 was 1 nm (0.02 MPa in 2.5 mm thick glass). On the other hand, in a 2 mm thick heat-strengthened glass (stress is 30-40 MPa), the retardation to be measured is less than 200 nm, requiring a resolution of I nm. At a small distance from the edge, the average stress measured in transmission decreases to zero. A typical edge scan result is shown in Figure 5. The software permits automated calibration, eurve fitting, verification of"zero" and performs a scan based on selected scan length and scan speed.

Figure 5 - Stress Scan Graph


A new stress measuring system was developed. The speed and spatial resolution of the method permits On-Line monitoring of tempered and heat-strengthened glass. The system is PC based allowing 100% inspection and documentation, not possible to obtain using present Quality Control methods.




[1] [2] [3] [4] [5] [6] [7] [8]

Minor, J.E., "Basic Glass Strength Factors," Glass Digest, pp. 52-57, 8-90. European Standards prEN1863 (part 1, part 2) and prEN12120 (part 1, part 2) draft, 9-2000, CEN, rue de Stassart 36, B-1050 Brussels, 2000. Guillemet, C. and Acloque, P., "New Optical Method for Determination of Stresses Near the Surface," 2nd GAMAC Conference, pp. 157-163, Paris, 1962. Redner, A.S. and Bhat, G.K., "Precision of Surface Stress Measurement Test Methods and Their Correlation to Properties," Proceedings, GPD, pp. 169-171, June 1999. Redner, A.S., "Stress Measurement in TV Production," GLASS, 74 (6), pp. 218-219, June 1997. Redner, A.S. and Voloshin, A.S., "Surface and Edge Stress in Tempered Glass," Proceedings, 9th International Conference on Experimental Mechanics, Copenhagen, 1990. Gulati, S.T., et al, "Delayed Cracking in Automotive Windshields," Material Science Forum 210-213, pp. 415-424, 1996. Redner, A.S., "Photoelastic Measurements by Means of Computer-Assisted Spectral Contents Analysis," Experimental Mechanics 25(2) pp. 148-153, ffune 1985.


George P,. ToroL ~Wemer Lichtenberger,2 and Allan Majorz In-Situ Dew-point Measurement to Assess Life Span of Insulating Glass Units

Reference: Torok, G. R., Lichtenberger, W., and Major, A., "In-Situ Dew-point Measurement to Assess Life Span of Insulating Glass Units," The Use of Glass in Buildings, ASTM STP 1434, V. Block, Ed., ASTM International, West Conshohocken, PA, 2002.

Replacement of insulating glass (IG) units in buildings is expensive. Replacement costs can be estimated fairly accurately. However, timing is less certain. In the author's experience, time estimates for replacement are o/~en based on poor understanding of the causes of IG unit "failure" (water vapour condensation on glass surfaces facing the IG unit cavity) and previous negative experience, and thus are reactive rather than predictive. The life span of insulating glass units in service is not well known. Insulating glass units have been made in North America since the late 1950s. Laboratory test methods developed in Canada in the late 1950s and early 1960s, subsequently used as the basis of most IG unit test methods worldwide, were intended to assess the likelihood of successful performance through the IG unit manufacturer's warranty period, not to determine service life span. In the 1980's, based on in-situ testing for the "Field Correlation Study" by the Sealed Insulating Glass Manufacturers Association (SIGMA) in the USA, Spetz proposed that one component of the laboratory test method, dew-point measurement of cavity gas fill, could be used to estimate time to failure oflG units in service. This technique is examined in this paper and modifications are suggested to improve accuracy. Keywords: Insulating glass units, service life, longevity, dew-point

i Project Manager, Gerald g. G-engeBuilding Consultants Inc., 27 Main Street North, Newmarket, Ontario, Canada, L3Y 3Z6. 2 Special Projects Manager,.Tmseal Technologies Ltd., 260 Jackson Street West, Hamilton, Ontario, Canada, LgP IM5. 3 Senior Technologist, Insulating Glass Laboratory, Bodycote Materials Testing Canada Inc., 2395 Speakman Drive, Mississauga, Ontario, Canada, L5K 1B3. 35 Copyright9 by ASTMInternational



North American Laboratory Test Methods to Assess IG Unit Performance Existing North American IG unit laboratory test methods, and indeed many of the IG unit laboratory test methods world wide, are based on research by Solvason, Wilson and Nowak at the Division of Building Research (DBR), National Research Council Canada in the late 1950s and early 1960s [1, 2, 3]. The DBR test protocol was developed at the request of what is now the Canada Mortgage and Housing Corporation (CMHC) as a tool to evaluate the suitability oflG units promoted by manufacturers for installation in new housing funded under the Canadian National Housing Act (NHA) and administered by CMHC. The DBR test protocol consisted of the following components: 9 An initial seal test to test the integrity of the hermetic seal; 9 Repeated cycles of heating, water spray, drying, and cooling primarily to test mechanical strength of the perimeter seal; 9 Repeated cycles of exposure to high humidity with pressure change to test the water vapour resistance of the seal (other standard test methods do not include cycling of pressure and humidity); and 9 Outdoor exposure to "natural" weather cycling, to provide some correlation to "real" life including exposure to UV radiation (currently, neither the Canadian CGSB-12.8-97 Insulating Glass Units standard nor the American ASTM Test Methods for Seal Durability of Sealed Insulating Glass Units (E 773) and ASTM Specification for Sealed Insulating Glass Units (E774) laboratory test standards include outdoor exposure or UV testing, although ASTM E773 / E774 does include UV exposure in the lab during weather cycling). Throughout testing, the dew-point temperature of the IG unit cavity gas fill was measured, and units with a dew-point warmer than 30~ were considered to have "failed" (water vapour had condensed on surfaces bounding the air space; in a double-lite IG unit, on surface 2 or 3). This temperature was arbitrarily selected by the DBR researchers as a likely temperature at which building occupants might consider such condensation, or "fogging," to be objectionable [2]. However, given that winter temperatures in much of Canada often fall below 30~ the value was not considered sufficiently severe and in the CGSB-12.8-97 Insulating Glass Units standard, the maximum (warmes0 dew-point was set at -40~ (-40~ Correlation of the DBR test protocol and the subsequent CGSB and ASTM protocols with service life is limited. In their 1962 paper, Wilson and Solvason noted a "rough correspondence" between failures of units subject to varying lengths of laboratory testing (fi'om 0 to 880 cycles) and of units of the same type subject to one year of outdoor exposure testing [2]. The number of cycles of heating, water spray, drainage and cooling were eventually standardized to 320, about the same number of thermal cycles experienced, on average, across Canada in a 5 year period. This was also (and still is) the length of the industry standard IG unit warranty against seal failure. Thus was established the supposed correlation of laboratory testing to 5 years of"real" service life and the industry standard 5 year warranty period. In the late 1970s, the Sealed Insulating Glass Unit Manufacturers Association (SIGMA) in the USA embarked on a "Field Correlation Study" to confirm the apparent correlation between the lab test protocol and field service life. At the time, it was



generally understood that IG unit constructions tested successfully in the laboratory were capable o f much longer service lives than 5 years. Field studies began in 1980 and terminated 15 years later. The study has its limitations: it is a comparison of field exposure to the American ASTM laboratory test protocol ASTM E773 and accompanying specification E774, a modified version of the DBR. test protocol and thus somewhat different than other laboratory test standards, such as CGSB-12.8-97; 2,400 IG units e r a population of 40,000 in 40 buildings in 14 cities in the continental USA were studied, most of which faced south or southwest; some of the units were "lost" during the test period because of demolition, renovation or subsequent denial of access; and the units studied were made with available sealant products, desiccants, etc., and installed in accordance with practices of the day. Thus the results are, perhaps, unique to the USA, to units of that vintage, to units with those orientations [4,5]. Within these limitations, the SIGMA "Field Correlation Study" revealed that failure oflG units made to the highest performance level of the ASTM E774 specification ("CBA"), installed so that the perimeter seals were not subject to prolonged wetting, was about 2.9% after 15 years [5]. This result was anticipated by Wilson and Solvason. In their 1962 paper to the Canadian Ceramic Society, they remarked that diffusion of water into an IG unit cavity and saturation of the desiccant was unlikely to lead to failure within the industry standard 5-year warranty period but it might within the anticipated service life of a unit [2]. But what is the anticipated service life of a unit? We cannot rely on the results of the SIGMA study because it was terminated after 15 years, when most of the units had not yet failed. We are therefore left with estimates based on personal experience, on reports in the glazing media, and on learned dissertations in technical papers. For example, Francis [6] reported that IG units manufactured in accordance with current industry standards and properly glazed should achieve life spans of 20 years or more. Others have made similar estimates [I, 7]. The authors have encountered many IG units still in service, fog free, up to about 26 years in age in service, and in one exceptional case (given the technology available at the time of manufacture), after 46 years of service. In the building to be discussed later, after about 25 years many of the original IG units are still performing without even transient evidence of failure (fogging on cold nights).

Dew-point Testing to Assess IG Unit Performance

Existing Technique
An outcome of the SIGMA "Field Correlation Study" was a proposal by Spetz (the auditor of the units) that in-situ dew-point measurement could be used to assess performance and, under limited circumstances, to estimate remaining life span [8, 9]. This was based on analysis of dew-point measurements made during the first 10 years of the study. By relating dew-point measurements to desiccant manufacturer's isostere charts (plots of desiccant saturation as a function of desiccant temperature and dew-point of the air exposed to the desiccant), it was possible to estimate desiccant moisture content. Spetz found that units with desiccant close to saturation were likely to fail within a short time.



Spetz proposed the following evaluation scheme for IG units in service [8]: 9 Dew-point < -80~ there is almost no moisture in the IG unit cavity, thus the IG units can be expected to have a "very long expected future clear life"; 9 Dew-point between -80~ and 0~ there is some moisture in the cavity, thus the IG unit can be expected to have a future clear life less than units with a dew-point < -80~ 9 Dew-point between 0~ and +32~ there is "considerable" moisture in the air space, thus the IG units will have a relatively short future life. Estimation of remaining life span requires knowledge of the construction of the units; 9 Dew-point > 32~ permanent fogging of glass surfaces bounding the IG unit cavity can be expected to develop within two years. The method used by Spetz to measure in-situ dew-point measurement was formalized as ASTM E 576, Test Method for Frost Point of Sealed Insulating Glass Units in the Vertical Position. This standard addresses only the method of measurement of dew-point temperatures for IG units. It does not include Spetz's proposed assessment scale or other methodology for evaluating the performance of an IG unit and its remaining service life. Apart from the two articles by Spetz in the 1980s previously referenced, there does not appear to have been any further, formal development of the technique of in-situ dew-point measurement to evaluate remaining life span of IG units.

Critique of the Existing Technique

Moisture adsorption capacity of desiccants varies with temperature of the desiccant, and different desiccants adsorb different quantities of moisture at a given temperature [12, 13, 14, 15, 16, 17]. If we assume that the scale given by Spetz is based on a desiccant temperature of 72~ (22~ the standard temperature for IG unit laboratory testing), then as for the original DBR test protocol, a dew-point temperature of32~ (0~ is probably too warm to be useful as an indicator of an impending failure problem, because the suggested time period to permanent fogging, 2 years, is too short for building owners to begin a reasonable savings plan to fund replacement. The next lowest dewpoint temperature range given, 0~ (-18~ may be sufficient, but that depends on the rate of water vapour gain of the IG unit cavity gas fill. The rate of increase is a function of several factors, including [7, I0, I1, 14, 15, 16]: 9 Water vapour adsorption capacity of the desiccant; 9 Sealant Moisture Vapour Transmission Rate (MVTR, or water vapour permeance); 9 Sealant Moisture Vapour Transmission Path (MVTP) length (from outside the I(3 unit to the cavity) and area (perimeter length x width from spacer bar shoulder to adjacent glass face); 9 Construction techniques; 9 Water vapour pressure differential across the perimeter seal; 9 Workmanship; and 9 Service environment



The water vapour adsorption capacity of a desiccant is, in turn, a function of several factors, including [12, 13]: 9 Rate of diffusion through the binder (for bead forms) or matrix (for extruded forms) that holds the desiccant crystals; 9 Size of openings of water molecules versus desiccant crystal pore size; 9 Strength of the attractive force between water molecules and the surface of the desiccant; and 9 Temperature of the desiccant. The effect of many of these factors is dynamic, not static. For example: 9 The water vapour adsorption capacity of both silica gel and molecular sieve desiccants varies with temperature (generally, capacity decreases at low and high temperatures, distinctly more so for silica gels than for molecular sieves); 9 Water vapour permeance (MVTR) varies with changes as the water vapour pressure gradient differential across the sealant changes (increasing as relative humidity of the IG unit cavity gas fill increases); 9 MVTP length and area changes as the cavity volume increases and decreases (in response to ambient air, temperature change, solar gain, and general atmosphere pressure changes), causing the perimeter sealants to extend or compress; and 9 The service environment may change (from summer to winter, from occupant to occupant, or from one occupancy type to another). Thus the rate of increase in the water vapour content of the IG unit cavity gas fill may not be constant over time. Therefore, from a single measurement of dew-point temperature it is not likely that an accurate assessment of remaining fog-free life span can be made. However, if measurements are repeated over time, and if care is taken to reduce the effects of the factors listed, we propose that it is then possible to make such an assessment.

Proposed Technique
When "failure" of an IG unit due to fogging is considered to occur is relative, depending on location and sensitivity of the building occupants. For example, wintertime temperatures in Edmonton, Alberta, are colder than in Toronto, Ontario, so for the same IG unit construction, fogging is likely to occur first in Edmonton. Many building occupants probably wouldn't be too concerned if fog formed in an IG unit during a very cold night then evaporated shortly after the sun went up, although it would raise some questions. However, if fog lingers well into daylight hours when unobstructed vision is generally desired, most occupants would likely complain. Thus the first step in assessment of remaining service life of an IG unit is to determine the outdoor ambient air temperature range in which fogging is likely to be considered objectionable. One could choose, for example, the average of mean daily minimum temperatures for the winter months, or perhaps for a sensitive client, the mean daily minimum temperature for the coldest winter month. These temperatures could be selected from historical climate records available from government meteorological agencies.



As noted, the water vapour adsorption capacity of a desiccant in an IG unit is dependent upon its temperature. Therefore, the moisture content of the IG unit cavity gas fill to which the desiccant is exposed, and thus the dew-point temperature of the gas fill, is dependent upon the desiccant temperature. The desiccant temperature of an IG unit installed in a window in a building will lie somewhere between the outdoor ambient air temperature and the indoor ambient air temperature. Thus if dew-point temperatures are to be measured over time and compared to determine a rate of increase, and from that an estimate of time to failure is to be made, the dew-point measurements from year to year should be made when outdoor ambient air temperatures arc about the same. The outdoor ambient air temperature at which we measure dew-point temperature for time to time comparison should relate to the range of outdoor ambient air temperatures in which fogging is considered unacceptable. The two could be the same. Although some error would be introduced by measuring dew-point over a range of temperatures, it would be practical for field measurement. The relationship between moisture content of the IG unit cavity gas fill, the outdoor ambient air temperature, and the measurement of dew-point temperature over time is shown in Fig. 1. When plotted together on a psychrometric chart, the relationship between cavity gas fill dew-point temperature and the range of outdoor temperatures in which, for a particular building, location, and client, fogging is unacceptable, becomes clear. The relationship of time to dew-point temperature rise is, perhaps, not as clear because the time scale is distorted to fit the temperature scale. For example, as shown in Fig. 1, the rate of dew-point temperature rise is decreasing with time.

Timeperiods between dew point measurements are of equal length

Timeperiodfrom last measurement of dewpoint temperature to cold weather temperature is / estimated, basedon [ r.--.- previous time period ] \ lengths and dewpoint / \ temperatureris, /

Dewpoint measurements made from timetotime

Time, years

~/~/ % ~ / Maximum IG y unit.cavitygas fill

mo~,tu~ content,

i / l ,/
Outdoorcold weather temperaturerange in whichfogging is unacceptable Temperature, ~

Figure 1 - Relationship Over Time Between Cavity Moisture Content, Outdoor Ambient
Air Temperature, and Dew-point Measurement



From Fig. 1, it can be seen that in order to estimate time to failure, that is, the length of time that may be required for the dew-point temperature to rise from the last recorded measurement into the outdoor cold weather temperature range in which fogging is unacceptable, dew-point measurements must be made when those temperatures are likely to be well colder than the outdoor temperature range. The intent is to make several measurements and yet leave sufficient time from the last measurement to the lower end of the cold weather temperature range for the owner to make reasonable financial arrangements to fund IG unit replacement. We propose that a measurement program begin as soon as possible after occupancy. With respect to those factors, one of the advantages of measuring dew-point temperatures consistently at the same outdoor ambient air temperature range is the practical elimination of desiccant temperature variation as a potential source of error in estimating time to failure. Since measurements are made at similar temperatures (the smaller the range, the better), and at the temperatures at which fogging is not acceptable, it is not necessary to determine the quantity, type, and manufacturer of the desiccant and to obtain the manufacturer's isostere charts for estimating moisture content. In essence, moisture content of the desiccant becomes irrelevant; the symptom of increasing moisture

Timeperiods between dew point measurements are of equal length ~.


Timeperiodfrom last measurement of dewpoint temperature to cold weather temperature is estimated, based on previous time period lengths and dewpoint temperature rise. l l


Time, years Frequency distributions I I ] [ ofdewpoint mue er n ces ~_ m t made ] A \[ \ / l d-r so,,, time to time ~ / ~ 1 ~' I ,

.. A ,4,-- ,,.,.,.,o,,,.osl 'cu-,ve--'!






t l ', ~


Outdoor cold weather temperature range in whichfogging is unacceptable


Figure2 - Relationship over time between cavity moisture content, outdoor ambient air temperature, and dew-point measurement frequency distributions.



content, the rise in dew-point temperature of the cavity gas fill, is of greater practical importance. This greatly reduces the complexity and therefore, the cost of testing. Even with the practical elimination of dew-point temperature as a factor in dewpoint temperaatre rise, there remain sufficient other factors such that within a given population of IG units, at the same desiccant temperature, it is likely that there will be some variation in measured dew-point temperature from one unit to the next. Therefore, it can be expected that measurements of a random sample of IG units would yield a distribution of dew-point temperatures. Over time, tracked by repeated measurements, the temperatures of this distribution would increase, eventually into the outdoor temperature range in which fogging is unacceptable. This is illustrated in the Figure 2. Figure 2 is a blend of frequency histograms of temperatures measured at several different times, with the psychrometric chart from Fig. 1. From this diagram is can be seen that, over time, as the IG units age, it can be expected that all of the units will not fail at once; thus funding for replacement can be arranged to be available over some time period. These concepts will be demonstrated in the following case study.

Case Study: Municipal Building

Figure 3 - Municipal Building, South Elevation (North Elevation Similar) Curtain Wall Repairs and IG Unit Replacement in Progress, Summer 2002.



Dew-point testing to estimate remaining life span of insulating glass units was required for a municipal building in Toronto, Ontario, Canada (Figure 3). A previous consultant had made a preliminary investigation of water leakage problems and had determined, correctly, that removal of IG units in curtain wall cladding in the north and south elevations of the building was necessary to correct defects in the original installation (missing framing joint seals) that were responsible for chronic water leakage. Because the majority of the IG units were original (fabricated in 1977) and now aged (23 years at the time of the preliminary investigation), the consultant suggested that removal and reinstallation could result in stress to the IG unit perimeter sealants, possibly hastening failure (fogging). Dew-point testing was recommended to assess the condition of units and determine timing for repair. If testing revealed that failure was likely to occur within three years then the repairs could take place immediately; if testing revealed that failure was likely to occur beyond three years, then the repairs could be delayed until some future date (although not stated in the consultant's report, presumably, the future date would be established by repeated dew-pint measurements). This was a feasible approach because the water leakage being experienced had virtually no impact on day-today building operations.

Figure 4 -Dew-Point Testing with ASTM E547 Test Unit.



Twenty (20) IG units in the north elevation curtain wall and 16 IG units in the south elevation curtain wall, for a total of 36, were tested in the late winter of 2001. This represented 15% of the total of 240 IG units in the two curtain walls, a good sample size. Dew-point measurements were made in 10~ brackets below 0~ Temperature brackets were used rather than precise temperatures to speed measurements. The bracket boundaries were based on the range of mean daily temperatures for the month of January for Toronto, from -2.5~ to -11.1~ This range was obtained from the "Climate Normals" records available from Environment Canada for the 1937-1990 period. Ideally, dew-point temperature measurements should have been made in January, but this could not be arranged for a variety of reasons. Outdoor temperatures at the time were about -3~ at the high end of the range of January mean daily temperatures; ideally, temperature measurements should have been made at the low end of the range, at -11.1"(2 or the low end of the first bracket, -10~ It is therefore expected that dew-point measurements made would be somewhat high and a resultant estimate of remaining time to fog at -I~ would be less. The indoor temperature was a constant 21~ Dew-point measurements were made with ice cubes, the test apparatus as described in ASTM E 547 (Fig. 4), and the electrically powered "Cryocool" unit from the insulating glass laboratory of Bodycote Materials Testing Canada Inc. The ice cube was used to check for dew-point temperature above 0~ and the E 547 and the "Cryocool" equipment were used for testing at temperatures of-10~ and below. Most measurements were made with the E 547 equipment, which proved to be more durable. Unit No. Field (ln-Situ) Measurements Dew-PoimTemperature <0oC 0oC -10oC -20oC -30oC to to to to -10*C -20~ -300C -400C Fog Fo8 Fo8 Fos Fog Fog Fo8 Fo8 Fo~ Fo8

1 2

4 5 6
7 8

9 10

12 13 14 15 16 17 18 19 20

Fo[ Fo8 No Foa Detected Fog Fog Fog Fog No Fo~ Detected I IF~

Figure 5 - Dew-Point Test Results, North Elevation.

TOROK ET AL. ON INSULATING GLASS UNITS Laborator~Measuro...ments Moisture Comments Content. % by Weight


Unit No.

1 2 3 4 5 6 7 8 9 10 11 12

Field {ln-Situ)Measurements DewDew-Point Temperature <0*C 0*C - 1 0 * C -200C .30oc Point. tO ~ to to to -10*C -200C -300C -400C No Foil Detected ] F~ Visible Condensation {Failed Unit) No Fog Detected Fog Fos Foil Visible Condensation (Failed Unit) Fog Fo$ Fog

,o 1



13 14 15

Fog Fo8 Fog



Local sealant failure at setting blocks. Blend of molecular sieve and silica gel desiccants, filled two long sides. Blend of molecular sieve and silica gel desiccants, filled two lons sides. 11.5 era long sealant failure at setting block. Blend of molecular sieve and silica gel desiccants, two long sides.





Figure 6 - Dew-Point Test Results, South Elevation, Results for the 1977 units are tabulated in Figs. 5 and 6. When graphed in the form shown in Figure 2 (frequency histogram combined with a psychrometric chart), separately for the north and south elevations, two distinct bell-shaped curves emerge, as expected (Figure 7). The modal value for the north elevation IG units was in the -10~ to -20~ bracket and the modal value for the south elevation was in the 0~ to -10~ bracket. Clearly, water vapour had entered the units and the desiccant was becoming saturated. This should not be surprising for 24 year-old I(3 units. The difference in modal values suggests that IG units in the south elevation can be expected to fail before IG units in the north elevation. This corresponds well with observed IG unit failures, all but one being in the south elevation. Clearly, more testing was required to confirm the trends that emerged from the dew-point measurements made. However, the client required immediate direction so it was decided to remove several units of the original 1977 units for further examination and testing in the laboratory. Three units were successfully removed from the south elevation and delivered to the Insulating Glass Laboratory at Bodycote Materials Testing Canada Inc. Aider conditioning at standard temperature, dew-point measurements were again made. As expected, lab dew-point measurements were warmer than those



measured in the field (Fig. 6). This is to be expected because of desiccant reflux, increasing the moisture content of the cavity, and thus raising the dew-point temperature. However, the difference between field dew-point and lab dew-point measurements was not consistent for the 3 units tested. This could have been caused by error in field measurement of dew-point temperature or rupture of the perimeter seal during removal and transportation, causing additional moisture gain. The latter was felt most likely, because discolouration and partial loss of adhesion of the perimeter sealant was noted at setting block locations.

Time for peak o f south elevation measured dew point frequency distribution curve to move into outdoor cold weather temperature range is unknown.

North Elevation measured dewpoint . _ temperature Jrequency /~ / : t

x=? Soath Elevation I measured dew point 4t I ] - ' - temperatureJrequency[ I/ distribution / 'I I

t ~ s

;",'"[",, 'I
cavity gas fill moisture


..'" / ,'

I',I',./ I l.X

content without . . . . . . . . . /~ . . . . . . . . . . . ~ ' I



I ~





I I~


Outdoor cold weather temperature ranEe Temperature, ~




Figure 7 -

Results o f ln-Situ Dew-point Testing at Municipal Building

The perimeter sealant for the original 1977 units was polysulfide (single seal). Desiccant was a blend of silica gel and molecular sieve, and filled two long sides (unit dimensions were approximately 1.83 m (6 fl) x 0.76 m (2 1/2 ft), 6 mm (1/4 inch) thick glass lites, 13 mm (12 inch) wide cavity). Desiccant was removed from each unit, weighed, then dried and weighed again. The technique followed was similar to that described in the proposed European insulating glass unit test standard, prEN-1279-2, Glass in Buildings, Insulating Glass Units, Part 2: Long Term Test Method and Requirements for Moisture Protection. Desiccant saturation was determined to vary from 16.1-17.9% by weight (Fig. 6). Based on available desiccant manufacturer's data sheets



for the types of desiccant found (molecular sieve and silica gel), this indicated the desiccant was very close to saturation(between 18% and 20% by weight) [10, 13, 14]. Given the warm dew-point temperatures measured in the field and in the lab, and discolouration and adhesion failure of perimeter sealants at setting blocks, it was reasonable to concur with the consultant who made the preliminary review of the building, that removal and reinstaUation could result in increased seal stress, increasing the rate of moisture gain and reducing the time to failure. Because the south elevation IG unit in-situ dew-point curve modal value was in the ambient outdoor temperature range in which fogging, if it occurred, might be considered objectionable, we advised the owner to schedule repairs (and IG unit replacement) to that elevation as soon as funds permitted. However, because the north elevation IG unit insitu dew-point curve was colder, we advised the owner that if funds for replacement were limited, repairs (and replacement of IG units) could be delayed. Timing could be estimated by further testing. As it turned out, sufficient funds were made available to make repairs to both the south and north curtain walls. Unfortunately, this meant that repeated testing to confirm the notion of advancement of the dew-point curve and estimate to failure could not, therefore, be confirmed. Fortunately, there were some administrative delays in the approval of the work so in the meantime, another winter passed. During that winter, we visited the building and observed that the number of IG unit failures (fogging) in the south elevation doubled, whereas there were no new failed IG units in the north elevation. This generally confirms our findings, and suggests that the proposed test methodology is valid. Conclusions In-situ dew-point testing has been proposed as a method to estimate performance after IG units have been in service for some time. The basic technique was outlined by Spetz in the mid 1980s and the method of field determination of IG unit cavity gas fill dew-point temperature was formalized in ASTM E 576. Because of the many factors that contribute to water vapour ingress into an IG unit, the rate of moisture gain is not likely to be constant such that estimation of remaining life span is not likely to be accurate based on a single measurement. More accurate estimation could be made if dew-point measurements were taken periodically, over many years. The results could be plotted as frequency histograms, and the rate of temperature increase of, for instance, the peak of the histogram curves vs. time could be calculated. This rate could then be used to estimate when the dew-point temperature of any point in the curve would coincide with the selected range outdoor ambient air temperatures in which fogging, .if it occurred, would be considered unacceptable. The range of mean daily temperatures for the coldest month of the year is proposed as a guideline for this temperature range. Further field testing is required to validate the technique. An opportunity exists with recent legislative changes in Ontario, Canada. The recently revised provincial Condominium Act requires boards of directors for condominium corporations to establish reserve fund plans to fund future capital repairs and replacements, which must be periodically updated to maintain them current and accurate. Dew-point testing oflG units in buildings with large expanses of glazing could be done as part of on-site reviews for reserve fund planning. If repeated periodically, performance oflG units could be



tracked and it may be possible to estimate and eventually confirm if the technique is accurate.

[ 1] [2] [3]


[5] [6] [7] [8]

[9] [10] [11] [ 12] [13] [14] [15] [26]


Burgess, J. C., "The History, Scientific Basis and Application of International IGU Durability Tests," Building and Environment, Vol. 34, 1999, pp 363-368. Wilson, A. G., Solvason, K. R., "Performance of Sealed Double-Glazing Units," Journal of the Canadian Ceramic Society, No. 31, October 1962, pp 62-68. NRCC 7042, DBR-RP-168. Wilson, A. G., Solvason, K. g., Nowak, E. S., "Evaluation of Factory-Sealed, Double-Glazed Window Units," Symposium on Testing Window Assemblies, ASTMSTP 251, ASTM International, West Conshohocken, PA, 1959, pp. 3-16. NRCC-5270, DBR-RP-85. "SIGMA Field Correlation Study," SIGMA-GRAM Technical Bulletin TB-2000520, Sealed Insulating Glass Manufacturers Association (SIGMA), Chicago, August 2000. "IGMA Unit Longevity Statement," Technical Bulletin TB-4000-01, Insulating Glass Manufacturers Alliance (IGMA), Ottawa, August 2000. Francis, G. V., "Zeroing in on Premature Failure of I(3 Units," Glass Magazine, August 1996, pp. 22-27, 42, 43. Lichtenberger, W., "Field Performance of Insulating Glass," Proceedings of Window Innovations '95, Toronto, Ontario, June 5th and 6 th, 1995. Spetz, J. L., "Frost Point Measurement: How a Frost Point Tester Can Be Used to Predict the Future Service Life of Insulating Glass Units in Buildings," Glass Magazine, June 1986, pp. 38 ft. Spetz, J. L., "Desiccant Works with Temperature to Prolong the Life of Insulating Glass Units," Glass Digest, November 15, 1987, pp. 66--67. Dangieri, T. J., and J. S. Amstock, "Desiccants," Handbook of Glass in Construction, McGraw-Hill, New York, 1997, pp. 297-321. Gallion, W. R., "The Role of Desiccants in Insulating Glass Units," Insulating Glass Manufacturers Association of Canada (IGMAC) Technical Seminar, 1991. Kromminga, T., "The Role of Desiccants in IG Units," 1GMAC Technical Seminar, 1996. "UOP Molecular Sieves," UOP Inc., 1990 "UOP Molsiv@Adsorbents," UOP Inc., circa 1995. Giangoirdano, R. A, and Harris, N., "Send it Back! Know How to Spot the Signs of a Badly Fabricated IG Unit," Glass Magazine, November 1990, pp 42-47. Spetz, J. L., "Design, Fabrication and Performance Considerations for Insulating Glass Edge Seals," Science and Technology of Building Seals, Sealants, Glazing and Waterproofing, American Societyfor Testing and Materials (ASTM) STP 1168, C. J. Parise, Ed., ASTM, West Conshohocken, PA, 1992, pp 67-81. Spetz, J. L., "Examining the Factors that Determine Insulating Glass Longevity," Glass Digest, August 15, 1996, pp. 28 ff.

Daniel J. Wise and Bipin V. Shah

Evaluation of the Condensation Resistance Rating as Determined Using the NFRC 500 Procedure Reference: Wise, D. J. and Shah, B. V., "Evaluation of the Condensation Resistance Rating as Determined Using the NFRC 500 Procedure," Use of Glass in Buildings, ASTMSTP 1434, V. Block, Ed., ASTM International, West Conshohocken, PA, 2003.

Abstract: This paper presents a comprehensive and detailed assessment of the National Fenestration Rating Council (NFRC) 500 Procedure. This procedure emphasizes simulation calculations and allows the utilization of computer software in analyzing and calculating the Condensation Resistance Rating for comparative purlx)ses. For those products that cannot be simulated, a test only option is also offered in the standard. This paper analyzes the results obtained through NFRC annuai round robin tests run from 1997 to 2001. Each round robin specimen was modeled per NFRC 100 (1997): Procedures for Determining Fenestration

Product U-factors and NFRC Test Procedure for Measuring the Steady-State Thermal Transmittance of Fenestration Systems using NFRC-approved computer software tools WINDOW 4.1 and THERM 2.1. The simulated results were used as a benchmark for the analysis of the testing data. All of the NFRC round robin test specimens were tested at NFRC-accredited testing laboratories.

NFRC 500 [1] provides a method of determining a Condensation Resistance rating for fenestration products; including windows, entrance doors, sliding glass doors, skylights, door-lite, and curtain wail. Fenestration products include residential, commercial and sitebuilt applications. For the purposes of rating, the product is modeled and rated for a specific size at a net zero air leakage: meaning the product is sealed and air leakage effects on condensation index rating are not taken into consideration. The total product is evaluated for condensation, and the rating is a total product rating. By using NFRC-approved software tools, a Condensation Resistance (CR) Rating can be modeled and calculated simultaneously with U-factor, Solar Heat Gain Coefficient (SHGC) and Visible Transmittance (VT) ratings.

Copyright9 by ASTM International



Simulation Requirements for Condensation Resistance

Since both temperature and surface film coefficient affect the results, standardized conditions were used for the evaluation and rating of the Condensation Resistance rating. The conditions were as follows: 9 interior ambient temperature of 21.1 ~ (70 ~ 9 exterior ambient temperature of-17.8 ~ (0 ~ 9 relative humidity (RH) of 30%, 50%, and 70% providing dew point temperatures of approximately 2.9 ~ (37.2 ~ 10.3 ~ (50.5 ~ and 15.4 ~ (59.7 ~ 9 wind speed of 6.7 m/s (15 mph) to generate film coefficient of 28.97 W/m2-K (5.11 Btu/hr-ft2-F) 9 sky condition of 100% cloud cover The THERM 2.1 [2] computer program was utilized in this study to obtain center-ofglazing heat transfer.

Physical Testing for Condensation Resistance

Physical testing is used to obtain a Condensation Resistance Rating only in cases where a product cannot be accurately simulated using currently approved NFRC software tools. The test can be performed simultaneously with the U-factor testing using the NFRC

Test Procedure for Measuring the Steady-State Thermal Transmittance of Fenestration Systems. For condensation evaluation, thermocouples are attached to the interior surface of
the test specimen in pre-specified locations.

Calculation of Condensation Resistance Rating

A Condensation Resistance Rating for a fenestration product is determined by analyzing each section of the fenestration system (i.e., frame and sash, edge-of-glazing, dividers, edge-of divider and center-of-glazing). For a specific cross-section, the indoor surface is subdivided into smaller segments that are no larger than the size of mesh or grid used by the simulation program. Once the indoor surface is subdivided, calculations were made to determine the following: 1) total length of each 2-D cross-section (Equation 1). 2) frame areas and glazing areas that have surface temperatures at or below the three prescribed dew point temperatures at 30%, 50% and 70% relative humidity. (Equation 2) 3) Condensation Resistance rating of the frame, CRf, center-of-glazing, CReog,and edge-ofglazing, CRag (Equation 3)



S= i (t,, -to)*L


t4,p = dew point temperature + 0.5 F t,. = temperature of the surface segment, i Li = length of the surface segment, i L = total length of the surface

SS= j


*A* }1/3t.100


where Ak = area of each fenestration section A = total area of flame, center-of-glazing (includes dividers and edge-of-divider), or edge-of-glazing Details on how to determine each Condensation Rating for each section and the whole product are given in NFRC 500. When the computer modeling is performed with NFRC-approved software, the temperature measurement at locations, measurement of segment areas at or below the dew-point temperature, and subsequent Condensation Resistance calculations are an automatic routine of the software. The simulation also accounts for detailed radiation exchange between the product and the baffle, and it includes the self-viewing effects between products components. The relative humidity called out in NFRC 500, 30%, 50%, and 70%, represents a range with 30% and 70% at the extreme ends beyond which human comfort is affected and 50% is considered the optimal humidity level for a home or workplace. The final Condensation Resistance Rating is the minimum value of the CRf, CP~og,and CR~og.



Ratings and Sizes Size is an important characteristic when comparing U-factor, Solar Heat Gain Coefficient (SHGC), Visible Transmittance (VT), and Condensation Resistance (CR) ratings from product to product. Size is important because of the area ratios of the frame, edge-ofglazing, and center-of-glazing. NFRC has addressed the size issue for each fenestration type and provides referenced sizes for comparison purposes. The current size for each type can be found in NFRC 1002001, Table 1.[3] One size for each fenestration type has been designated for the following general categories: windows, entrance doors, double doors/glazed wall systems/sloped glazing, and sidelites/transoms. The same product size will be utilized for all reported indices, including Condensation Resistance. There is a difference between U-factor and Condensation Resistance values, namely, the former is an average index and the latter is the minimum of three areas. The minimum of three areas is taken for Condensation Resistance because condensation is local in nature. It forms in areas that are at or below the dew point. Thus, an average rating for condensation is meaningless, because extremes may be present in the same product (e.g., low performance frame and high performance glazing, or vice versa). An average number would indicate acceptable performance, while in reality, one significant part of the window could be under condensation in the form of water or even frost, potentially damaging the window and/or surrounding wall structure.

Analysis of the Round Robin Tests The proposed Condensation Resistance equations, as found in the NFRC 500, have been used to analyze the three round robin products. In all cases, the products were also simulated using NFRC-approved sot~ware per NFRC 500. The product tested in 1997/1998 was composite an aluminum/wood frame window, incorporating high performance double glazing. In 1999/2000, the product tested was an aluminum (no thermal break) frame horizontal slider window with very high performance triple glazing, also known as HeatMirror product. The third product, tested in the 2001 test round robin, was a thermally-broken aluminum fixed window with high performance glazing. WINDOW 4.1 [4] computer program was utilized to calculate center-of-glazing thermal performance for both windows. These center-of-glazingmodels were then imported into THERM 2.1, a two-dimensional finite element computer program, for analysis of thermal performance of frame and edge-of-glazing cross sections. The THERM software program incorporates detailed multi-element radiation heat transfer model, which is utilized on indoor window surfaces. This is done because on the interior (indoor) side, radiation heat transfer accounts for almost two thirds of the total surface heat transfer coefficient, while on the exterior side, radiation heat transfer is only a small fraction of the total surface heat transfer coefficient. The Condensation Resistance model in THERM also incorporates a



convection heat transfer model in the glazing cavities, as well as a multi-element radiation heat transfer model. Table 1 summarizes the calculated Condensation Resistance values from the 1997/1998, 1999/2000 and 2001 round robin testing and one simulation run for each of the robin test specimens. It is significant to note that the data obtained on tests done on identical units from one year to the next produced different results. For example, the round robin Condensation Resistance value for the edge-of-glazing (reported value, in this case) shows a difference of almost 8 points from 1997 to 1998. The simulation values have the advantage of producing consistent results, even when run by different simulators. When compared to test results, simulation results show good agreement for the center-of-glazing values in all three cases. The 1997/1998 results show good agreement in the CRf and CP,,ogareas of condensation evaluation, with the exception of the CP-~ in the 1998 test round robin. The 1997 Cr value was in close agreement with the simulated value. The 1999/2000 and 2001 test round robin specimen CR values showed variance from the simulation results for the CRf and C P ~ .



Table 1 - Condensation resistance average values. Round Robin Year 1997/1998 1999/2000 2001 CRf CRcog CR~og CRf CR~og CReog CRf CReos CReog 69.9 68.9 50.9 1 8 . 3 67.9 31.9 49.0 69.7 45.6 71.0 71.8 68.7 67.9 43.1 52.2 1 8 . 8 68.9 6.7 66.7 32.0 48.2 N/A 41.8 N/A 70.0 N/A 59.4

First Year Test Second Year Test Simulated Value

Note: The values in the tables are the averages of the laboratories that participated in the round robins for the years stated.

Condensation resistance tests done in adjacent years on identical units have produced different results. The simulation values have the advantage that they produce very consistent results, even when run by different simulators. When compared to test results, simulation results show consistently good agreement for the center-of-glazing values. It was observed from the 1999/2000 and 2001 round robin data simulated CRf was considerably lower than the tested value and the CR~ s was considerably higher. This outcome is attributed to the fact that simulation accounts for every segment on the cross section of frame and edge of glazing, while testing relies on pre-determined single point thermocouple locations, which may or may not capture an average performance of that region. Thus, it can be concluded that the NFRC 500 simulation method produces more consistent, accurate Condensation Resistance ratings than that obtained through testing.

[1] NFRC 500. Procedure for Determining Fenestration Product Condensation Resistance Values, 2002. [2] THERM 2.1. A PC Program for Analyzing Two Dimensional Heat Transfer Through Building Products Procedure. [3] NFRC 100. Procedure for Determining Fenestration Product U-factors,1997, revised 2001. [4] WINDOW 4.1. A PC Program for Analyzing Windows Thermal Performance in Accordance with Standard NFRC Procedure.


Stephen J. Bermison, ~C. Anthony Smith, Alex Van Duser, and Anand Jagota

Structural Performance of Laminated Glass Made with a "Stiff' Interlayer

Reference: Bennison, S.J., Smith, C.A., Van Duser, A. and Jagota, A., "Structural Performance of Laminated Glass Made with a "Stiff" Interlayer," Glass in

Buildings, ASTMSTP 1434, V. Block, ed,, ASTM International, West Conshohocken, PA, 2002.
Abstract: The growing demand for laminated glass in building facades and interiors is

driving the development of new interlayers that extend the performance of laminated glass. In this contribution we discuss the structural performance of laminates made with such a new interlayer: SentryGlas | Plus. This new interlayer is significantly stiffer and tougher than traditional PVBs and provides enhanced structural performance in many applications. The mechanical properties of a stiff interlayer lead to benefits in three main areas: 1) glass strength; 2) stiffness and creep resistance, both before and after glass breakage; 3) temperature performance. Strength benefits are realized in applications where significant bending stresses develop during loading, such as: laminates with two sided support, balustrades (cantilevers), bolted glass and laminates supported on three sides. Enhanced post glass breakage performance expands the design envelope for glass in horizontal and sloped applications such as overhead glazing, glass floors and stairs. All of these structural properties are maintained to higher temperatures as compared to conventional PVB. Keywords: creep resistance, glass strength, temperature performance

Architectural laminated glass is dominated by the use of Polyvinyl Butyral (PVB) interlayers, such as Butacite | This domination can be attributed to the long successful history of PVB use in the automotive industry for safety glass. The primary structural requirements of automotive safety glass are: 1) impact resistance from external objects, 2) sufficient compliance to ensure minimal head trauma for a passenger that strikes the windshield. This latter feature is a major reason why plasticized PVB was developed; an elastomeric interlayer with the correct elastic response to cushion passenger impact was needed. However, if we look to the requirements of architectural laminated glass we find that high compliance and an elastomeric response to be unnecessary in many applications. Indeed, a compliant interlayer can hinder structural performance in some cases such as in applications where low glazing deflection is specified. If we relax the requirement of high compliance after glass breakage, many more polymer systems now look attractive for laminated safety glass. In this contribution we consider a new polymer system for i All authors:E.I. DuPontde Nemours& Co. Inc.,Wilmington,DE 19880-0356.


by ASTM International



laminated glass. The polymer, SentryGlas | Plus (SGP), is significantly stiffer than conventional PVB and displays elasto-plastie stress-strain behavior. It has a totally different molecular architecture to PVB and for most applications SentryGlas | Plus (SGP) is below its glass transition temperature (T8 ~ 55~ We describe several examples of the performance benefit of laminated glass made with SentryGlas | Plus (SGP). This structural interlayer is particularly effective in applications where: 1) glazing is subjected to bending loads, such as 1 and 2-sided and point support conditions and/or concentrated loading; 2) post-glass breakage performance is important, e.g., in overhead glazing; 3) enhanced performance under long duration loading or at elevated temperatures is required.

Strength Performance
The benefits of enhanced interlayer stiffiaess for laminate strength can be readily appreciated from a simple experiment in which glass stress during loading is measured directly as a function of polymer type and thickness. Figure 1 shows the results from one such experiment.

FIG. 1--Measurement of glass stress as a function of uniform applied pressure for 4sided support monolithic glass plates and several laminate plate builds. Note that for the glass design stress specified by the horizontal dashed line all laminates display greater load bearing, or strength characteristics.
In this experiment a grid of 15 3-gage rosette strain gages were attached to the~lass surface of a series of laminates made with either PVB (DuPont Butacite ) or SentryGlas | Plus. The laminate plates were supported on four sides and loaded with uniform pressure. Maximum principal stress was measured at each gage location



along with plate deflection during loading. Figure 1 shows the development of maximum principal stress in the glass during an experiment. Several features may be noted from these data. First, the upper curve consists of observations from a nominal 6 nun monolithic glass and shows non-linear stress-pressure behavior that is characteristic of large deformation of a 4-side supported plate. Second, a laminate made from 3 mm glass / 2.29 mm Butacite| PVB / 3 mm glass shows significantly less glass stress development at a specified applied pressure; i.e. the laminate is stronger than the equivalent monolithic glass plate. This result is consistent with much research demonstrating that 4-side supported, uniform pressure loaded laminated glass often displays equivalent or enhanced strength over monoliths [1-7]. Third, a laminate made from 3 mm glass / 2.29 mm SentryGlas| Plus / 3 nun glass shows significant strengthening over monolithic glass and laminated glass made with PVB. For example, at the glass design stress denoted by the horizontal dashed line the SentryGlas| Plus -based laminate can sustain almost 2x the applied pressure than the equivalent monolith. Of course the overall laminate thickness is greater than the monolith but in many design codes around the world this laminate would be considered to be structurally inferior to the equivalent monolith. As a guide to design with SentryGlas| Plus -based laminates we have constructed strength charts that aid the selection of polymer type and thickness, glass thickness, for specified loading/support and rate/temperature conditions. These charts have been computed using a procedure described in detail elsewhere [8]. Briefly, the procedure consists of: 1) establishing a constitutive model for the interlayer by dynamic mechanical analysis [9]; 2) carrying out finite element analyses of glass stress development [10]; 3) validating selected analyses against controlled loading experiments [8]; 4) combining stress analyses with a statistical (Weibull) glass breakage model [11-13] that incorporates a time-dependence for glass strength [14,15]. A design chart for a SentryGlas| Plus -laminate is shown in Figure 2.
o 50
' ,

-3 mm11.52mm SGIP/3'ram'/ 9 Load(kPa) zao~. Pb = o.oos / \ . 3-Second =~

150 i







Length (ram)

FIG. 2--- Strength design chart for a SentryGlas| Plus -based laminate. The chart plots allowable applied uniform pressure contours for various plate sizes. Note the specified probabilily of breakage (0.008), load duration (3s) and temperature (50 ~



This chart has been constructed in ASTM El300 [16] format using the glass strength parameters and minimum glass thickness specified in that standard [17]. The chart plots allowable pressure loading contours as a function of plate dimensions for a breakage probability of 0.008. Note that the chart is constructed for a 3 mm glass / 1.52 mm SentryGlas| Plus / 3 mm laminate under 3 s wind loading at 50~ A series of such charts for common laminate constructions is used in an iterative fashion to select the optimum laminate build for specified wind-load conditions. Comparison of this chart with one for an equivalent 6 mm monolithic glass shows strengthening over the whole field of plate sizes. Interestingly, the degree of strengthening depends on plate size and aspect ratio with the greater strength benefits appearing at higher aspect ratios. At these higher aspect ratios a greater degree of bending deformation occurs and suggests that the benefits of SentryGlas| Plus may be realized in loading conditions where bending stresses dominate. Accordingly, a series of beam bending experiments have been carried out in which glass-stress development as a function of loading has been measured directly using attached strain gages. Figure 3 shows results of one set of experiments.


5 ram/2,29 mm Sem~yC.du(a)P]us 5 mm / I0 nun Monolithic Gins 12 mm Monolithic Gtass



60O .~

400 300

Glass Edge Design Stress= 14 MPa T ~ 23~







Displacement Rate, dS/dt (ram/s)

FIG. 3---Measurement of glass stress development for monolithic glass beams and SentryGlas | Plus -based laminate beams tested in three-point bending. For a specified glass design stress (14 MPa) the figure plots the applied force as a function of rate. Note that the 5 m m / 2.29 mm SentryGlas Plus / 5 mm laminate demonstrates greater strength than a 10 mm monolithic glass and essentially equivalent strength to a 12 mm monolithic glass.
In this figure the force required to produce a specified glass stress has been measured as a function of loading rate in a three-point bend geometry. The main feature to note from these measurements is that a beam made from a 5 mm glass / 2.29 mm SentryGlas| Plus / 5 rnm laminate significantly outperforms a 10 mm monolithic glass beam in strength properties. Again the total laminate thickness is greater than the equivalent monolith but many building codes around the world neglect this and



penalize laminates against equivalent monoliths. When comparing strength performance of the laminate against a 12 mm monolithic glass it can be seen that the monolith and laminate are essentially equivalent. Of course the laminate contains significantly less glass and has a weight advantage over the monolith. In bendingdominated loeding/support conditions the interlayer shear properties play a significant role in glass stress development as pointed out by Hooper [18,19]. Under such conditions SentryGlas| Plus-based laminates are expected to significantly outperform PVB-based laminates.

Post-Glass Breakage Performance

SentryGlas| Plus was originally developed to enhance post-glass breakage performance in a cycling test component of new hurricane performance standards that are being adopted in certain regions of the USA. We demonstrate the benefits of enhanced interlayer stiffness of SentryGlas| Plus -based laminates in the following test. Laminate plates, 0.762 m x 1.219 m, consisting of 3 ram heat strengthened glass / 2.29 mm interlayer / 3 mm heat strengthened glass have been dry glazed into a frame, loaded under uniform pressure to glass breakage, and then the pressure cycled to study the maximum (center) deflection response of the fragmented laminate. Figure 4 shows the results of a pressure deflection test during loading after glass breakage.
50 i ! i i i i i

9 9 40

3mm/2.29mm Se~tryGlas Flus/3mm t cR) 3mm/2.29mm PVB/3mm








Applied Pressure, P (kPa)

FIG. ~'--Deflection-appliedpressure behaviorfor two laminates tested under uniform pressure, 4-side support. The PVB-based laminate displays significantly greater compliance than a $entryGlas| Plus -based laminate, after glass breakage.
Basic plate analysis [20] of the experiment, assuming that the plate is acting as a pure membrane, reveals that the effective modulus of the PVB laminated plate is on the



order of 400 MPa. Consider now the SentryGlas| Plus laminate in Figure 4. Analysis of this test reveals that the laminate modulus post-,glass breakage is on the order of 12 GPa. Thus the stiffness properties of SentryGlas Plus result in a 30-fold increase in plate stiffztess, post-glass breakage. The reduced plate compliance reduces the tendency for interlayer tearing during loading of a cracked laminate and diminishes in-plane motion and tendency for glazing pullout from the framing system. It is expected that the stiffness of a cracked laminate will scale with interlayer modulus. However, other factors contribute also to the resulting laminate compliance. These other factors include fragment size, which is determined by glass strength and glass type, and glass/polymer adhesion. Qualitatively, small glass fragments or high glass strength will reduce laminate stiffness and low glass/polymer adhesion will reduce laminate stiffness. Note that the full potential stiffening effect of SGP is not realized due to these other factors and glass fragments play a significant role in plate deformation characteristics.

Temperature Effects
The polymer structure of SentryGlas| Plus results in a glass transition temperature, Tg, on the order of 55 - 60~ the stiffness advantage of SentryGlas | Plus versus traditional PVBs is maintained up to and exceeding the glass transition temperature. The advantage of this higher Tg is demonstrated in a design example for an overhead glass canopy in which laminated glass is required for safety performance. The question addressed in the example concerns the long-term creep performance of the point supported overhead structure under self-weight at 40~ We ask the question: how will the canopy deflect over time and what effect will changing the interlayer from PVB to SGP have on performance?

FIG. 5--- Computed deformed shape of a point-supported laminate canopy after 30 years at 40 %7under self-weight. We approach this problem using our t'mite element design methodology used to construct wind load design charts discussed earlier. Figure 5 shows the final



deformed shape of a PVB-based laminate canopy after self-weight loading for 30 years at 40~ One advantage of our finite element based approach is that we have constructed constitutive models for our polymers that have full time-temperature superposition capabilities that allow efficient calculation of rate and temperature effects [8,9]. Figure 6 shows the predictions deflection behavior. Loading: self-weight, 40~ 12 mm glass/1,52 mm interlayer/12 mm glass DuPont Laminate Design Engine 50
--O-P V B - D u P o n t B u l a c i t e TM

40 =" .o 30

- - < 3 - - S G P - D u P o n t S e n t r y G l a s TM Plus


I0U I01 102 I(P I& I0-s 10r 107 !0s 10'~ Time, t (s) FIG. 6----Computed maximum deflection o f the canopy shown in Figure 5 as a function o f time under self-weight at 40 ~ Note that the SentryGlas | Plus-based laminate canopy is predicted to display essentially constant deflection over time as compared to the PVB-based laminate, which is predicted to steadily deflect over time. Consider first the upper curve for a PVB-based laminate. It is predicted to deflect over time until reaching a steady state condition after a million seconds or so, The lower curve in Figure I5 shows the predicted response for a SentryGIas| Plus -based laminate of the same build. Note that overall deflections are lower than those predicted for PVB-laminates and that the deflection response is essentially stable with time for these conditions. Again the enhanced stiffness of SentryGlas| Plus versus PVB results in significant performance enhancements in deflection response over time at elevated temperatures.

We have shown that a new interlayer, DuPont's SentryGlas| Plus, demonstrates significant structural performance advantages over traditional PVB in many applications. The enhanced structural performance results from increased interlayer stiffness and higher glass transition temperature compared to conventional PVB. The primary attributes resulting from a stiffer interlayer include: I) enhanced strength, particularly where bending stress states dominate laminate deformation; 2) enhanced laminate stiffness both in pre and post-glass breakage conditions; 3) enhanced



temperature performance. Such performance attributes present architects and engineers with more design options for optimum performance glass facades and structures. It is expected that these attributes will also extend the design possibilities for laminated glass.

1. Behr, R.A., Minor, J.E., Linden, M.P., Vallabhan, C.V.G., "Laminated Glass Units Under Uniform Lateral Pressure," Journal of Structural Engineering, 111 [5] 1037-50 (1985). 2. Behr, R.A., and Linden, M.P., "Load Duration and Interlayer Thickness Effects on Laminated Glass," Journal of Structural Engineering, 112 [6] 1141-53 (1986). 3. Vallabhan, C.V.G., Minor J.E., and Nagalla, S., "Stresses in Layered Glass Units and Monolithic Glass Plates," Journal of Structural Engineering, 11311] 36--43 (1987). 4. Das, Y.C. and Vailabhan, C.V.G., "A Mathematical Model for Nonlinear Stress Analysis of Sandwich Plate Units," Mathematical Comput. Modelling, 11 713-19 (1988). 5. Behr, R.A., and Norville, H.S., "Structural Behavior of Architectural Laminated Glass," Journal of Structural Engineering, 11911] 202-22 (1993). 6. Vallabhan, C.V.G., Das, Y.C., Magdi, M., Asik, M., Barley, J.R., "Analysis of laminated glass units", Journal of Structural Engineering, 119[5] 1572-1585 (1993). 7. Van Duser, A., Jagota, A., Bennison, S.J. "Analysis of Glass/Polyvinyl Butyral (Butacite~) Laminates Subjected to Uniform Pressure" Journal of Engineering Mechanics, ASCE, 12514]435-42 (1999). 8. Beunison S.J., Davies P.S., Jagota A., Van Duser A., Smith C.A., Foss R.V., "Structural Performance of Laminated Safety Glass", presented at Glass Tech Asia 2000, Singapore. 9. Ferry, J.D., Viscoelastic Properties of Polymers, 3'd edition, John Wiley & Sons, (1980). 10. ABAQUS| version 5.8, (1998) Hibbit, Karlsson & Sorensen, Inc., Pawtucket, R.I. 02860-4847. 11. Weibull, W., "A Statistical Distribution Function of wide Applicability," J. Appl. Mech., 18 293 (1951). 12. Davidge, R.W., "Mechanical Behavior of Ceramics," chpt 9, p 132-39, Cambridge Solid State Science Series, Cambridge 1979. 13. Beason, W.L., Morgan, J.R., "Glass Failure Prediction Model", Journal of Structural Engineering, 110 [2] 197-212 (1984). 14. Brown W.G., "A Practicable Formulation for the Strength of Glass and its Special Application to Large Plates", Publication number NRC 14372, National Research Council of Canada, Ottawa (1974). 15. Reed D.A., Fuller E.R. (Jr.), "Glass Strength Degradation Under Fluctuating Loads", Journal of Structural Engineering, 111 [7] 1460-1467 (1984). 16. ASTM Standard E 1300-97 "Determining Load Resistance of Glass in Buildings" in 1997 ASTM Annual Book of Standards, American Society for Testing and Materials, West Conshohocken, PA. 17. Norville, H.S., Minor, J.E., "Strength of Weathered Window Glass", American Ceramic Society Bulletin, 64 [11] 1467-1470 (1985).



18. Hooper, J.A., "On the Bending of Architectural Laminated Glass", International Journal of Mechanical Sciences, 15, pp. 309-323 (1973). 19. Norville, H. S., King, K. W., Swofford, J.L., "Behavior and Strength of Laminated Glass", Journal of Engineering Mechanics (ASCE), 124 [I] 46-53 (1998). 20. Timoshenko, S., Woinowsky-Krieger, S., Theory of Plates and Shells, McGraw-Hill, New York (1959).

Mostafa M. EI-Shami ~and H. Scott Norville2

Development of Design Methodology for Rectangular Glass Supported on Three Sides to Resist Lateral Uniformly Distributed Loads

Reference: E1-Shami, M. M. and Norville, H.S., "Development of Design Methodology for Rectangular Glass Supported on Three Sides," The Use of Glass in Buildings, ASTMSTP 1434, V. Block, Ed., ASTM International, West Conshohocken, PA, 2002. Abstract: The forthcoming version of Practice to Determine the Load Resistance of Glass in Buildings (E 1300-02) will contain several new features. These include new procedures to determine load resistance of laminated glass, revised procedures to determine load resistance of insulating glass, and procedures to determine load resistance of rectangular glass supported on 1, 2, and 3 sides. This paper discusses the development of load resistance charts for rectangular glass supported on three sides with one side free. It presents the nonlinear finite element model used to determine stresses and deflections. The load resistance charts use maximum edge stress as the criterion for determining load resistance. The value of maximum edge stress derives from a failure prediction approach for glass edges rather than surface stresses. The paper describes load-induced stresses along the edge of an unsupported side and over the glass surface for glass supported along three sides. It compares probability of breakage based upon surface stresses with that derived from the maximum edge stress criterion. Keywords: glazing design, three-sided support, finite element analysis, maximum stress Introduction
The next revision of ASTM Practice to Determine the Load Resistance of Glass in Buildings (E 1300-02) contains nonfactored load (NFL) charts for glass supported along one, two, and three sides. This paper describes the development of NFL charts I Post-Doctoral Research Associate, Glass Research and Testing Laboratory, Department of Civil Engineering, Texas Tech University, Box 41023, Lubbock, Texas 79409. 2 Director, Glass Research and Testing Laboratory, and Professor, Department of Civil Engineering, Texas Tech University, Box 41023, Lubbock, Texas 79409. 66
Copyright9 by ASTM International



and corresponding deflection charts for rectangular glass simply supported along three sides with the fourth side free. It describes the nonlinear finite element model ffEM) used for the analysis, the boundary conditions, the maximum edge stress criteria for determination of NFLs, and a simple example that illustrates the use of the NFL charts and compares them with results obtained using a linear analysis.
The Finite Element Model

In this section the authors describe the stiffness matrix formulation for the nonlinear FEM for thin plate analysis. The authors employ Mindlin plate theory [1]. Mindlin plate theory makes the same assumptions as does Von Karman theory [2] except that the final rotations of lines normal to the plate's middle surface are obtained by adding the respective derivatives of the lateral displacement function w(x,y) with respect to x and y with the corresponding vertical shear deformations. The displacements used at any point (x,y,z) are:

u.(x,y,z)=u(x,y)+ zOr(x,y ) v.(x, y,z) = v(x, y)- zO,(x, y)

w. (x, y, z) =w(x, y)


where u(x, y), v(x, y) and w(x, y) denote the displacements of the middle surface and Ox and Oy denote the rotations of the normal in the undeformed plate in they-z and xz planes, respectively, after deformation. Figure la shows an element before deformation. Figure Ib shows the same element after deformation and illustrates that, according to Mindlin plate theory; lines normal to the middle surface before bending do not necessarily remain normal to the middle surface aider bending. Mindlin plate theory represents the relevant nonlinear strain vector as:

s = 7"~


to j



denotes the linear component of membrane strain,



Z~ W

--0 X

~ !

v=-ZO x

~'YZ=!~ W,y-x~ "'~


.~W,y , y,v


l a-Undeformed Element

Z~ W~


v=-zO x

'wYy-~~ "~
kL..f Figure lb.

, y,v

Deformed Element: Solid Lines Indicate Rigid Body Displacement and Dashed Lines Indicate Final Position After Element Deformation



{~o}=~y,.-0~.y denotes the bending strain,




{~:)~={,, +~. -o, +w/

denotes the shear strain, and mr I w 1 }r



denotes the nonlinear component of membrane strain. In terms of the shape function for a 9-node quadrilateral element, the displacements and rotations are:
9 9 9

u = E N,(~,~)~,,

~ = EN,(~,~),,,
ifl 9 i=l

~ = E N,(r



where N,(r 17), i = 1,9 denotes the shape functions of the 9-node quadrilateral elemem [3]. The terms r andr] denote coordinates used to transform the quadrilateral element into a square. The nodal displacement vector


w, o., o,


o.. o.}~


has 5 degrees of freedom [dot'] per node. Hence, with 9 nodes, each element has 45 dof. Since, both displacements and rotations are interpolated using quadratic functions, with three nodes on the boundary, both displacements and rotations have complete compatibility along the inter-element boundaries of the quadrilateral elements. Figure 2 shows a quadrilateral elemem with 9 nodes and indicates the dofat one node. The following equations define the relationships between the strain vectors and nodal displacement vectors.

{~:)f ,




z ~




x Figure 29-Node Quadrilateral Element with Degrees of Freedom Indicated at Node 1.


The equations above combine to form a new matrix for linear relationships: 0

Thus, the complete relationship between the strain and nodal displacement vectors is the sum of the linear and nonlinear parts, [B]= [Bol+ [B,] and the stress-strain relationship is (13)



{r = [D]{ 6}, in which[D] denotes the elasticity matrix


[D]= In the elasticity matrix

E'"I E,~
+ D~


[zr]- (i_,,2)
and "D, [D h] = 0 0 0 /)12 0 0 0


0 (1- v ) / 2

0 0 D33

0 0 0

0 0 0 (17)

D21 D22

0 D44 0 0 0 /955

E=71.7 GPa (10.4 xl06 psi) denotes Young's modulus of elasticity, v=0.22 denotes Poisson's ratio, and t denotes glass thickness. The individual elements of [Db] are as follows:
Et 3

Dl~ = D22 = 12(1 - v2) '

vEt s

(1 8)

D~2 = D2' = v ~12(1

Gt 3



Da3 = and



D44 = D55 = - - . 1.2 For linear elastic materials, such as glass,





E G = ) 2v-1 ~ ( denotes the modulus of rigidity. The equilibrium equation in its incremental form is:




where [Kr] denotes the tangent stiffness matrix, {AU}denotes the nodal displacement increment vector, and [AF] denotes the nodal force increment vector. The tangent stiffness matrix, [Kr], consists of three parts: the linear stiffness matrix, [K,], the nonlinear stiffness matrix, [Kj], and the initial stress stiffness matrix, [K~][3]. Therefore,

in which


[Ko]-- (.[BoY[D][BoldV.




Several authors have described Mindlin plate theory and solution techniques associated with it, including E1-Shami et al [4] and Vallabhan and EI-Shami [5]. The authors use Newton-Raphson techniques [1] in their solutions. Several authors [4-6] have demonstrated the FEM's efficacy in analyzing thin monolithic plates.

Boundary Conditions
Since the glass is simply supported along three sides with one side free, the model allows the simply supported sides to slip in the direction of the undeformed plane of the glass, it allows the glass to rotate at the supported sides, and it prevents any deformation normal to the unreformed plane of the glass along the supported sides. Figure 3 shows the boundary conditions for a plate supported along three sides with one side unsupported in detail, noting that the authors conducted the analysis on half the plate and employed symmetry about the glass centerline that lies perpendicular to the unsupported side.



Simply Supported

A w,OyOB/

w, Ox=O



at po=t a; ,~, o~,o, = o At p, lnt Cr v, o~ =o At point D; v, w, Ox= O

Figure 3-

Boundary Conditions for a Plate Simply Supported Along Three Sides with One Side Free.

Edge Stress Criteria

The maximum edge stress criterion used for these calculations comes from work performed by Walker and Muir [7] in which they performed bending tests on glass louvers to determine edge failure strengths. Walker and Muir [7] developed a failure prediction model based upon edge stress analyses. Application of their failure prediction methodology to the general case of an edge under stress indicates that a 60second equivalent failure stress that would initiate fracture at the edge with a probability of 0.008 is approximately 13.8 MPa (2000 psi). Converting this slress to a 3-second duration equivalent stress gives a value of approximately 16.6 MPa (2400 psi) as the maximum edge stress that would initiate fracture with a probability of breakage of 0.008. The authors used this maximum edge stress criterion to develop monolithic load resistance (LR) charts for glass supported along one, two, and three sides.



Stresses in Rectangular Glass Supported Along Three Sides

Figure 4 shows contours of maximum tensile stress for a nominal 6 mm (1/4 in.) glass lite having rectangular dimensions of 965 mm x 1930 mm (38 in. x 76 in.) with one 965 mm (38 in.) dimension unsupported under a lateral load of 1.0 kPa (20.9 psf). The authors observe that, as in the linear case, nonlinear analysis indicates that the maximum tensile stress occurs at the mid-length of the unsupported side. The maximum lateral deflection also occurs at this location.

Figure 4-

Contours of Maximum Principal Tensile Stress for a 965 x 1930 x 6 mm (38x 76x 1/4 in.) Lite Supported Along Three Sides with One Side Free.

The LR for this lite is very close to 1.0 kPa (20.9 psi'), based upon the maximum edge stress criteria of 16.6 MPa (2400 psi). The authors used the stress distribution depicted in Figure 4 in conjunction with the failure prediction methodology with m = 7 and k = 2.86x 10"53N'Tm to compute the probability of breakage based upon a ~2 fracture originating on the surface. This is the methodology to compute probability of breakage for glass supported along four sides. For this lite, the probability of breakage under a 1.0 kPa (20.9 psf) loading having 3 second duration is P~ = 1.65x10 "s. This strongly indicates that fractures occurring along the edge of the unsupported glass side provide a limiting criterion for determining LR.



Design Example

The authors would now like to determine the required thickness for an annealed glass lite having rectangular dimensions of 965 mm x 1930 mm (38 in. x 76 in.) with one 965 nun (38 in.) long side unsupported to resist a uniform loading of 2.0 kPa (41.8 psf). They also wish to determine the approximate maximum deflection for the lite. The process of determining the required thickness using the NFL charts from ASTM E 1300-02 is iterative. The designer picks a trial thickness designation and determines its LR for the given dimensions. Depending on the LR determined, the designer earl (a) use the trial thickness, (b) try a larger thickness, or (c) try a smaller thickness. Experience with ASTM E 1300-02 significantly hastens this process. From comments above, 6 mm (1/4 in.) glass will have insufficient LR, hence the authors try 8 mm (5/16 in,) glass. Figure 5 shows the nonfactored chart for this nominal thickness with a horizontal line projected from 965 mm (38 in.) along the vertical axis and a vertical line projected from 1930 mm (76 in.) along the horizontal axis. The lines intersect at about 1.6 kPa (33.4 psf). The glass type factor (GTF) for annealed glass under short duration load is 1.0. The LR is:
LR = GTF x NFL = 1.0 x 1.6kPa = 1.6kPa(33.4psf).
Length of Parallel Supported Edges (in.) 0 100 20 40 60 80 1O0 120 140 ' " 1 ' ' ' " / I ,~" ~'~' '" ' '~'" , ~ . " ~' '~' '" " "' 0.75~1-I-//T" 160 180 200 2500


80 70 !
tU 0 12.

/ " - - ! 'i ........ Three Sides Simply Supported f , ! i "Pb =0-008 ........ ...i./ lkPa = 20.9 psf ............ i / / ! 3-Second Duration
........ ................~"~- :'.................... ~-~..~---~i

-J--..i .

Nonfaetored Load (kPa)

2000 E



~.-~---~--.~ . . . . . .

i ~

. ~

1500 "~



I1 i







..:~ .'~.-..{-~. ~


10 i '
0 ~ ' ' '

50 0






Length of Parallel Supported Edges (mm)

Figure 5-

Nonfactored Load Chart for 8 mm (5/16 in.) Glass Simply Supported Along Three Sides.



The value of LR for 8 mm (5/16 in.) annealed glass is insufficient. The authors next try glass with a thickness designation of 10 mm (3/8 in.). Figure 6 shows the appropriate NFL chart with the lines projected on it. The value of NFL is approximately 2.20 kPa. Hence, the LR is:

LR = 1.0 x 2.20kPa = 2.20kPa(41.8psf)


Length of ParalleI Suppofled Edges(in.) 110 100 90 80











200 2500


70 60 50 40 30 20 10 0


LL tr-.J









0 5000

Length of Parallel Suppofled Edges (mm)

Figure 6-

Non-Factored Load Chart for 10 mm (3/8 in.) Glass Simply Supported Along Three Sides.

This LR is greater than the specified loading; so annealed glass with 10 mm (3/8 in.) glass will suffice. The authors could have used a smaller thickness designation and specified heat strengthened or fully tempered glass with a smaller thickness. Each load chart in ASTM E 1300-02 has a corresponding deflection chart. For glass supported on three sides, the maximum deflection occurs at the center of the unsupported glass side. While developing the NFL charts, the authors also developed deflection charts. The horizontal axes of the deflection charts contain various forms of "redimensionalized" loading. The particular redimensionalization depends upon the support conditions. Presentation of deflections in this form saves some calculation on the part of the designer but it also requires some familiarization for efficient use. For glass supported on three sides, the redimensionalized load takes the form:



}} = w x L4


in which ~ denotes the redimensionalized load, w denotes the uniform load, and L denotes the length of the unsupported side of the glass. The sloping lines, curving for small thickness designations, relate the redimensionalized load to maximum deflection. For this example, the redimensionalized load is:
{v = w x L 4 = (2.00kPa) * (0.965m) 4 = 1.73kN - m ~ = 4.20kip - fl2


Figure 7 shows the deflection chart with a vertical line projected from 1.73kN'rrl2 (4.20 kip't~) and a horizontal line projected from the intersection of the vertical line with the deflection line for AR > 1.5. The authors "read" the approximate maximum deflection from the intersection of the horizontal line with the vertical axes as 4.2 mm (0.17 in.).
x L4

(kip,ft 2) [L Denotes Length of Free Edge]

30 ' 40 50 60 70 80 I~ ' F L~[ ! '1 I ' I f 90 100 110 '1 ' I i ' j ~ ,

45 40




. . . . . . . . . ! - . - ..............









iI. . . . .

1,8 1.6


/ r/

,',. y
. . . . . .



E E 30


, ...................




~'/ ...... r Z /i y .," .,,,r




10 5

,J............:=:.4..................... I.


10.0 mm (3/8 in.) Glass Three Sides Simply Supported Deflection va (Load x L 4)

0.4 .0.2 0.0









Load x L4 (kN, m2) [L Denotes Length of Free Edge]

Figure 7. Conclusions

Deflection Chart f o r 10 m m (3/8 in.)Glass Simply Supported A l o n g Three Sides.

The authors described the development of the three sided NFL charts for ASTM E 1300-02. Their discussion included descriptions of the FEM used for analyses and the design failure criteria for glass with one or more unsupported sides. Previous



analyses [4] of thin glass lites used Kirchhoff plate theory coupled with nonlinear von Karman relationships to obtain a solution. The formulation presented herein uses Mindlin plate theory coupled with nonlinear von Karman relationships. The Mindlin plate theory accounts for shear deformation which is insignificant in thin glass lites but becomes significant for thick glass lites. For thin glass lites supported on four sides, the results obtained using Mindlin plate theory are identical to those obtained using Kirchhoff plate theory [4]. In addition the authors presented an example that illustrates the use of the NFL charts and deflection charts for rectangular glass with three sides simply supported and one side free.

[1] Cook, R. D., Malkus, D. S, and Plesha, M. E. "Concepts and Applications of Finite Element Analysis," John Wiley & Sons, New York, 1989. [2] Dym, C. L., and Shames, I. H., "Solid Mechanics: A Variational Approach." McGraw-Hill, New York, 1973. [3] Zienkiewicz, O. C., "The Finite Element Method," McGraw-Hill, London, 1977. [4] E1-Shami, M. M., Vallabhan, C. V. G., and Kandil, S. A., "Comparison of Nonlinear von Karman and Mindlin Plate Solutions." Proceedings of the American Society of Civil Engineers, Texas Section, Spring Meeting, April 49, Houston, Texas, 1997. [5] Vallabhan, C. V., G., E1-Shami, M. M."Comparison of Rectangular Glass Plates with Area-Equivalent Trapezoidal Plates." Proceedings of the American Society of Civil Engineers, Texas Section, Spring Meeting, March, San Antonio, Texas, 2001. [6] E1-Shami, M.M. and Norville, H.S., in review, "A New Model for Analyzing Architectural Laminated Glass," Journal of Engineering Mechanics, American Society of Civil Engineers. [7] Walker, G. R. and Muir, L.M., "An Investigation of the Bending Strength of Glass Louvre Blades," Proceedings: 9th Australian Conference on the Mechanics of Structures and Materials, August, Sydney, Australia, 1984.

H. Scott Norville, l Mostafa M. El-Shami, 2 Ryan Jackson, 3 and George Johnson4

Wind Load Resistance of Large Trapezoidal Glass Lites

Reference: Norville, H. S., EI-Shami, M. M., Jackson, R., and Johnson, G., "Design Methodology for Large Trapezoidal Window Glass Lites," The Use of Glass in Buildings, ASTMSTP 1434, V. Block, Ed., ASTM International, West Conshohocken, PA, 2002. Abstract: Large trapezoidal annealed window glass lites glaze airport control tower cabs. The authors have developed a design methodology for these lites as discussed in this paper. In developing the methodology, the authors conducted failure tests of large, trapezoidal, weathered window glass lites, obtained from the U.S. Federal Aviation Administration, under uniform pressure. They used a nonlinear finite element analysis technique for thin plates coupled with failure prediction techniques that underlay ASTM Practice for Determining the Load Resistance of Glass in Buildings (E 1300-00) to develop their design methodology. The paper summarizes the experimental data, loadinduced stresses in large trapezoidal window glass lites, and the load resistance of large trapezoidal lites relative to that of rectangular window glass lites. It presents the design methodology and illustrates it with an example. Keywords: glazing design, airport control towers, finite element analysis, failure prediction, maximum stress Introduction
Large trapezoidal window glass lites and window glass constructions glaze control tower cabs at all U.S. airports. They lean outward to provide relatively unobstructed and distortion-free views of runways and taxiways for ground traffic controllers. The fact that they lean outward requires the lites to have a trapezoidal surface area. These window 1Director, Glass Research and Testing Laboratory, and Professor, Department of Civil Engineering, Texas Tech University, Box 41023, Lubbock, Texas 79409. 2 Post-Doctoral Research Associate, Glass Research and Testing Laboratory, Department of Civil Engineering, Texas Tech University, Box 41023, Lubbock, Texas 79409. 3 Research Assistant, Glass Research and Testing Laboratory, Department of Civil Engineering, Texas Tech University, Box 41023, Lubbock, Texas 79409. 4 9 9 9 Senior Project Engineer, Federal Aviation Administration, Atlanta NAS Implementation Center, Terminal Platform, ANI-340, PO Box 20636, Atlanta, Georgia, 30320.

Copyright 9 by ASTM International



glass lites may consist of annealed monolithic glass, laminated glass (LG) with annealed plies, or insulating glass (IG) units fabricated with annealed lites. Ground traffic controllers require annealed glass because it provides nearly distortion-free viewing. At present, in U.S. airport control tower cabs, framing members support all four sides of the trapezoidal window glass lites and window glass constructions. New towers, presently under design, may use large trapezoidal window glass lites supported only along their top and bottom edges. Currently, the U.S. Federal Aviation Administration (FAA) has no regular design methodology to guide in the selection of proper glass thickness for the large trapezoidal lites to resist design wind loadings. Instead, FAA window glass designers use approximate methods that typically result in a thickness larger than necessary to resist design wind loads. The Glass Research and Testing Laboratory (GRTL) has undertaken a project to develop a methodology for the FAA to use when designing window glass in control tower cabs. In producing the design methodology, GRTL staff conducted nonlinear stress analyses in conjunction with destructive tests of large, weathered trapezoidal window glass specimens under uniform load. The design methodology relates uniform loading having a specified time duration to probability of breakage. This paper presents: (a) experimental results from failure tests, (b) analysis results, (c) strength comparisons between load resistance data for the sample of large trapezoidal lites and load resistance data from published research, and design load resistance from Weibull parameters in ASTM E 1300-00, "Standard Practice for Determining the Load Resistance of Glass in Buildings," (d) contours of maximum stress in large trapezoidal lites, and (e) a simple design example to illustrate the procedure for a trapezoidal lite supported on four sides.

Experimental Procedure and Data

The first step in this project consisted of obtaining samples comprised of several large trapezoidal window glass specimens from airport control tower cabs undergoing demolition or replacement. The FAA provided numerous specimens from which GRTL selected two relatively small but consistent samples. One sample consists of nine IG units that had been in service approximately 21 years. The other sample consists of eight relatively new large trapezoidal LG lites. GRTL staff began with the IG units since they arrived first. The authors note the removal from a control tower cab and subsequent storage of these IG units occurred some time before their shipment to GRTL facilities. The FAA shipped them to GRTL approximately six months before this project began. GRTL stored them until inception of the project. For both window glass samples, GRTL had no role in their removal and realizes that the lites underwent significant handling that probably caused some surface damage.

Monolithic Specimens Separatedfrom IG Units

GRTL staff separated the individual lites from the IO units and tested them to failure under uniform pressure. The top and bottom sides of the lites were parallel. The bottom side had a length of 2310 mm (90 in.). The top side had a length of 2705 mm (106.5 in.).



The vertical dimension of each lite was 2110 mm (83 in.). The lites had a nominal thickness of 10 mm (3/8 in.). Researchers placed each lite in a test chamber that provided support conditions and glazing stops in compliance with ASTM E 997-00 "Standard Test Method for Structural Performance of Glass in Windows, Curtain Walls, and Doors Under the Influence of Uniform Static Loads by Destructive Methods." Researchers evacuated air from the vacuum chamber at a controlled rate to produce a uniform pressure difference, i.e., a lateral load, acting across the window glass lite. They increased the lateral load linearly with time from the inception of loading at a rate of 1.82 kPa/min (38.0 psf/min) until the specimen fractured. Researchers monitored pressure using a piezoelectric pressure transducer calibrated against a mercury manometer before each test. They monitored center of glass deflection using a linear variable displacement transducer. Analog signals from the pressure transducer and the linear variable displacement transducer went to an analog/digital (A/D) converter. The A/D converter sent digital signals to a personal computer at a rate of 2 Hz. The computer simultaneously wrote the data to a file and graphed it on the computer monitor. The graphing allowed GRTL staff to monitor the load and control its rate on a continuous basis during each test. Following fracture, researchers carefully measured and recorded the location of the fracture origin. The authors summarize the failure pressure data for the monolithic specimens as follows: the mean failure pressure for 13 specimens was 2.38 kPa (49.7 psi), the standard deviation of the failure pressures was 0.754 kPa (15.7 psi), and the coefficient of variation for the failure pressures was 32.1%. The authors converted the raw failure data to 60-second duration equivalent loads as defined in ASTM E 1300-00. The mean 60second equivalent failure load for the specimens was 2.14 kPa (44.6 psi). The standard deviation for the 60-second equivalent failure loads was 0.730 kPa (15.3 psf). The coefficient of variation for the 60-second equivalent failure loads was 34.2%. The authors note that the variability for the loads was high. Fracture origins occurred in the following locations. Three fractures originated in the obtuse comer of the trapezoid, one fracture originated along the sloping edge, two originated in central areas of the trapezoidal lite, and the remaining fractures originated in the right angle comers. The authors again point out that GRTL staff did not supervise the lites' removal from the airport control tower cabs, subsequent handling, or shipping prior to their delivery to GRTL facilities. Handling during this period most certainly inflicted significant damage to the glass surfaces that resulted in lower load resistance values. In addition, the process of separating the lites from IG units induced evert more damage. Surface damage results in experimental data falling below typical load resistance values and possibly having higher variability.

Laminated Glass Specimens

The series of tests on LG specimens followed the same fundamental procedure as described above except that this series required no separation of lites. In addition, GRTL staff located fracture origins for both the inside and the outside glass plies, and then measured their locations from a common reference at the lower left comer of the glass when mounted in the test frame. This test series contained eight specimens. The top and



bottom sides of the lites were parallel. The bottom side had a length of 2030 mm (80 in.). The top side had a length of 2260 mm (89 in.). The vertical dimension of each lite was 2260 mm (89 in.). The lites had a nominal thickness of 38 mm (1-1/2 in.). GRTL staff tested the LG specimens using a load rate 15.3 kPa/min (319 psf/min) sampled at a rate of 2 Hertz. In summary, GRTL staff removed the LG specimens from their packing crates, mounted them to the test chamber, and loaded them. All other steps remained the same as described for the monolithic test series. The authors summarize the failure pressure data for the LG specimens as follows: the mean failure pressure for eight specimens was 22.5 kPa (469 psi'), the standard deviation of the failure pressures was 4.15 kPa (86.7 psf), and the coefficient of variation for the failure pressures was 18.5%. The authors converted the raw failure data to 60second duration equivalent loads as defined in ASTM E 1300-00. The mean 60-second equivalent failure load for the specimens was 20.2 kPa (422 psf). The standard deviation for the 60-second equivalent failure loads was 3.86 kPa (86.4 psf). The coefficient of variation for the 60-second equivalent failure loads was 20.5%. GRTL staff classified all fractures on the tension surfaces of the LG specimens as occurring in the center of the specimens. None occurred in any proximity to the obtuse comer or other high stress areas.

Glass Deflections
GRTL staff measured deflections during all tests using a linear variable displacement transducer (LVDT). They compared deflection data with finite element model (FEM) predictions. The authors discuss the comparison below.

Stress Analysis
While GRTL staff destructively tested the individual lites of the IG units, one of the authors conducted a nonlinear FEM analysis for the trapezoidal plates. Several references describe the model [1, 2, 3,4]. It uses Mindlin plate theory [5] that considers shear deformations through the thickness of the plate. In Mindlin plate theory, a plane normal to the middle surface before deformation remains straight but not necessarily normal to the middle surface after deformation. Figure 1 presents contours of maximum principal tensile stresses on the surface of the trapezoidal lite associated with a uniform lateral load of 1.80 kPa. Acting for a 60second duration, this load corresponds to a probability of breakage of 8 lites per 1000 (Pn = 0.008), using the failure prediction methodology in ASTM E 1300-00. Numbers along the sides of the graph in Figure 1 indicate distances in mm. Values of maximum principal tensile stresses on the contour lines in Figure 1 are in MPa. The authors observe that the highest magnitude of the maximum principal tensile stress, approximately 22.8 MPa (3300 psi), occurs in the area near the obtuse angle of the trapezoid. High values of maximum principal tensile stresses also occur near the other comers. Ten of the 13 fracture origins from the monolithic specimens occurred in areas of relatively high maximum principal tensile stress. The nonlinear FEM predictions of deflection agree with experimental measurements to within about 5% for both monolithic



and LG specimens. This observation provides a high degree of validation to the nonlinear FEM analysis technique.
500.00 1000.00 1300.00 2000.00 2500.00


Figure 1-Contours of Maximum Principal Tensile Stress (MPa) in a Ttrapezoidal Lite.

Failure Prediction

In the previous section, the authors produced maximum principal stress contours using equivalent loads that, if acting over a 60-second duration, correspond to a probability of breakage of 0.008. To determine the probability of breakage, the authors coupled the failure prediction methodology [6, 7,8] with the nonlinear finite difference stress analysis. The failure prediction methodology simply represents a Weibull probability distribution for window glass strength as given in the following equation:

P. =l-exp[-B]


in which Pa denotes probability of breakage and B denotes a risk function. The risk function takes the following form:

s--, Hcc(x,y)v~ (x, y)r d-~dy



in which m and k denote the Weibull parameters, c(x,y) denotes a stress correction factor at location (x,y), and ~m~(x,y) denotes maximum equivalent principal tensile stress at



location (x,y). For the probabilities calculated herein, the authors used values of the Weibull parameters consistent with those in ASTM E 1300-00, namely m -- 7 and k = 2.86 x 10"53N7m 12(1.37 x 10-29in) 2 lb'7). Figure 2 shows curves representing probability of breakage versus 60-second duration equivalent loads for various


J 0.0~6

~ 0.004





2 IkPII)



Figure 2- Probability of Breakage Curves versus 60-Second Equivalent Load for

Rectangular and Trapezoidal Lites.

geometries of lites based upon the Weibull parameters in ASTM E 1300-00 and the thickness of the monolithic specimens. The curve farthest to the left (0-90 ~ represents probability of breakage versus 60-second equivalent load for an encompassing rectangular lite. The next curve, moving to the right (0=78.5~ represents probability of breakage versus 60-second equivalent failure load for a trapezoidal lite with the dimensions of the monolithic specimens tested in this research. The two curves on the fight ((0-=45~ and 0=39.5 ~ represent probability of breakage versus 60-second equivalent load for trapezoids with the acute angle of the sloping side decreasing. As the lite moves from a rectangle towards a triangle, its area decreases. Although the magnitude of the principal stress in the obtuse comer increases, the failure prediction methodology indicates that the probability of breakage decreases or, in other words, the load resistance associated with a given probability of breakage increases. It indicates that designing for an encompassing rectangular shape should be eonservative, at least for trapezoids and triangles. The conclusion is not exhaustive for all conceivable polygons that the rectangle might encompass.



The authors next used the statistics from the failure loads to estimate WeibuU parameters for the experimental data from each test sample described above. This procedure consisted of an iterative procedure of trying a set of values for m and k, constructing a curve representing probability of breakage versus 60-second equivalent load, and then using a Kolmogorov-Smirnov goodness-of-fit test to determine how well the constructed probability of breakage distribution modeled the experimental distribution. Jackson [9] gives the details of this analysis. Jackson [9] found that the Weibull parameters from ASTM E 1300-00 adequately described the load resistance distribution of the LG sample. Jackson [9] also found that the Weibull parameters m = 4.8 and k = 7.98 x 10"36N sm7"6(1.37 x 10-20lb'4"Sin.T6)adequately described the load "4' resistance distribution of the monolithic specimens separated from the IG units. The authors constructed a probability of breakage curve versus 60-second equivalent failure load for a lite with rectangular dimensions 1524 x 1219 mm (48 x 60 in.) and nominal 6 mm (1/4 in.) thickness using these Weibull parameters. NorviUe and Minor [7] used a comparison similar to this. Figure 3 shows this curve along with curves based on Weibull parameters from other weathered glass samples [7] and those that form the basis for the non-factored load charts in ASTM E 1300-00. Figure 3 indicates that the monolithic specimens discussed above display lower load resistance than do any previously tested samples. It also indicates that the Weibull parameters in ASTM E 1300-00 result in significantly higher load resistance than any data from weathered window glass samples would justify. This observation indicates that significant numbers of window glass lites designed using ASTM E 1300-00 should fracture during every occurrence of a design wind event. That breakage does not occur. The authors feel that this breakage does not occur for several reasons. The most prominent reason is that the design wind load, i.e., a load having constant magnitude for 60 seconds never occurs, even during a design wind event. Furthermore, the authors believe that the handling window glass samples undergo in testing, regardless of the degree of care taken, significantly reduces the load resistance of the test specimens. Finally, the authors surmise that magnitudes of wind loads used for design exceed magnitudes of wind loads that design wind events produce on in-service window glass lites.

Design Procedure
General Design Criteria Figure 2 indicates that basing the design of trapezoidal window glass, simply supported along all edges, on the dimensions of the encompassing rectangle should prove conservative. In addition, the Weibull parameters from the trapezoidal monolithic sample indicate that it might be weaker than rectangular samples for whatever reason. Finally, an airport control tower cab is a critical facility. Hence, the design of trapezoidal lites for airport control tower cabs simply supported along all four sides will be based on the load resistance of an encompassing rectangle, ASTM E 1300-00 Weibull parameters, and a probability of breakage of 0.001.



Pressure (kPa) 0 2 ,

4 1

6 I

8 I

10 I

Monolithic Trapezoidal Specimens

. . . f ~ f /~.. ~/ ~ / /

o= 0O

"~ ./.'>~ ~
0,6 0.4

Weathered Glass Window [7]

d3 O


0,0 0.0

i t i




0,8 Pressure (psi)





Figure 3- Probability of Breakage versus 60-Second Equivalent Load for a Common

Lite Size.
For trapezoidal lites supported only along the top and bottom sides, the maximum tensile stress along the edges of the unsupported sides will control the design. For 60second duration loadings the maximum design stress based on an edge stress analysis as Walker and Muir [10] presented is 9.31 MPa (1350 psi). For 3-second duration loads, this value is 11.2 MPa (1620 psi). Currently GRTL personnel are producing load charts and computer software to facilitate this design methodology for the FAA. The charts are not ready for publication at this time.

Design Example
The authors wish to determine the thickness designation for a LG trapezoidal life, simply supported along all four edges, to resist a 60-second duration wind loading of I kPa (20.9 psl). The lite slopes outward at I 5~ from the vertical. The horizontal sides have lengths of 2540 mm (I00 in.) and 2290 mm (90 in.). The length oft_he side perpendicular to the horizontal sides is 2540 mm (I00 in.). Select a trial thickness designation of 19 mm (3/4 in.). The non-factored load associated with a probability of breakage of 0.001 is 2.32 kPa (49.4 psi). The component



of glass weight acting normal to the glass is 0.122 kPa (2.55 psi). The aspect ratio is 1.0 and the b/t ratio is 134. Therefore, the glass type factors (GTF) are 0.75 for short duration loading and 0.45 for long duration loading. This leads to a short duration load resistance of: Short Duration Load Resistance = 0.75x2.32kPa = 1.74 kPa (36.3 psi) and a long duration load resistance of: Long Duration Load Resistance = 0.45x 2.32kPa = 1.04 kPa (21.8 psi). (4)


The short duration load resistance is greater than the combination of the wind load and the weight component acting normal to the glass, which is 1.12 kPa (23.4 psi). The long duration load resistance far exceeds the component of glass weight acting normal to the glass. Consequently, the authors select a new trial thickness designation of 16 mm (5/8 in.). The non-factored load associated with a probability of breakage of 0.001 is 1.62 kPa (33.8 psi). The component of glass weight acting normal to the glass is 0.103 kPa (2.15 psi). The aspect ratio is 1.0 and the b/t ratio is 159. Therefore, GTF are 0.90 for short duration loading and 0.45 for long duration loading. Going through the computations for this thickness designation: Short Duration Load Resistance = 0.9x1.62kPa = 1.46 kPa (30.4 psi) and Long Duration Load Resistance = 0.45x 1.62kPa = 0.729 kPa (15.2 psi). (6) (5)

Without showing the calculations, a trial thickness designation of 12 mm (1/2 in.) will not have adequate load resistance. Hence the required glass thickness designation is 16 mm (5/8 in.). The approximate maximum glass deflection will be approximately 8 mm (0.3 in.) under the design wind loading.

This paper describes the development of a design methodology for large trapezoidal window glass lites. It presents the basis for the design methodology, which will function along the lines of the methodology used in ASTM E 1300-00. It also presents a design example. In developing the design methodology, the authors found that the encompassing rectangle provides a bounding geometry for design of glass with trapezoidal and triangular surface areas, although the authors do not wish to generalize to other geometries at this time. The authors also note that the Weibull parameters, m = 7 and k = 2.86 x 10.53N'Tm12(1.37 x 10.29 in. 12lb'7), in ASTM E 1300-00 tend to predict weathered window glass load resistance higher than that indicated by data reported in the literature. They observe that, even though ASTM E 1300-00 tends to predict higher window glass load resistance, wind loads do not fracture window glass during design



wind events. They offer opinions as to why design wind events putting loads on inservice window glass do not cause the predicted failure rate of 8 lites per thousand.

The authors gratefully acknowledge the U.S. FAA for providing monetary support for this work.

[1] E1-Shami, M. M. and Norville, H. S., "Development of Design Methodology for Rectangular Glass Supported on Three Sides to Resist Lateral Uniformly Distributed Loads," The Use of Glass in Buildings, ASTM STP 1434, V. Block, Ed., American Society for Testing and Materials, West Conshohocken, PA, in review. [2} EI-Shami, M. M. and Norville, H. S., "A New Model for Analyzing Architectural Laminated Glass," Journal of Engineering Mechanics, American Society of Civil Engineers, in review. [3] EI-Shami, M. M., Vallabhan, C. V. G., and Kandil, S. A., "A Comparison of Nonlinear von Karmen and Mindlin Plate Solutions." Proceedings, American Society of Civil Engineers, Texas Section Spring Meeting, Houston, Texas, April, 1997 [4] Vallabhan, C. V. G., and El-Shami, M. M., "Comparison of Rectangular Glass Plates with Area-Equivalent Trapezoidal Plates," Proceedings, American Society of Civil Engineers, Texas Section Spring Meeting, San Antonio, Texas, March, 2001. [5] Cook, R. D., Malkus, D. S., and Plesha, M. E., "Concepts and Applications of Finite Element Analysis," John Wiley & Sons, New York, 1989. [6] Beason, W. L. and Morgan, J. R., "Glass Failure Prediction Model," Journal of Structural Engineering., American Society of Civil Engineers, 110(2), pp. 197212, 1984. [7] Norville, H. S. and Minor, J. E., "The Strength of Weathered Window Glass," Bulletin of the American Ceramic Society, 64(11), pp. 1467-1470, 1985.. [8] Beason, W. L. and Norville, H. S., "Development of a New Glass Thickness Selection Procedure," Journal of Wind Engineering And Industrial Aerodynamics, Elsevier Science Publishers, 36, pp. 1135-1144, October, 1990. [9] Jackson, R., "Load Resistance of Trapezoidal Window Glass," Master of Science Thesis, Submitted in Partial Fulfillment of Degree Requirements, Department of Civil Engineering, Texas Tech University, Lubbock, "IX, December, 2001.



[10] Walker, G. R. and Muir, L. M.,"An Investigationof the Bending Strength of Glass Louvre Blades," Proceedings." 9thAustralian Conference on the Mechanics of Structures and Materials, Sydney, Australia, August, 1984.

Stephen M. Morse l Window Glass Design Software

Reference: Morse, S. M., "Window Glass Design Software," The Use of Glass in Buildings, ASTMSTP 1434, V. Block, Ed., ASTM International, West Conshohocken, PA, 2002. Abstract: Standards Design Group, Inc., (SDG) produces software to aid engineering and design professionals in the building industry with lengthy and often tedious design procedures. The paper describes recent rapid improvement in computing technology and the corresponding increase in complexity of engineering design standards. It provides justification for development of software that incorporates long computation procedures with logical decisions to achieve designs with just a few simple input operations on the part of the designer and engineer. The paper goes on to discuss specifically modifications that SDG incorporated into its window glass design software in conjunction with the forthcoming revision of E 1300. It presents design examples to illustrate some of the differences that will result with this revision. Keywords: glazing design, computer software, design standards, computer technology, ASTM El300

Introduction Over the years advances in technology have significantly altered the role of engineers and designers. For instance, engineers and designers have employed many tools to assist them in performing calculations. Each advance in technology has led to increased productivity for engineers and designers. This paper discusses the impact of technological advances on window glass design methodology. It addresses those changes that have made possible software that reduces the computations in ASTM Practice for Determining Load Resistance of Glass in Buildings (E 1300) to a few simple input operations.

1president, Standards Design Group, Inc., 3417 73rd Street, Suite K-3, Lubbock, TX 79423.

Copyright9 by ASTM International



A Brief History of Engineering Computing

Prior to about 30 years ago, engineersand designersused sliderulesto perform basic mathematical calculations, includingmultiplication, exponentiation,and trigonometric calculations. The slideruleproved quicker and more versatilethan eithertables combined with pencil and paper or mechanical adding machines for performing calculations. The sliderule leftplentyof room for errorand provided no assistance whatsoever in interpretingdesign procedures. Sliderulessimply helped engineers and designers perform calculations. Floatingpoint and scientific calculatorsmade theiradvent in the early 1970s. These proved much faster,more accurate,and considerablymore versatilethan sliderulesin performing calculations.As programmable calculatorsappeared in the mid-1970s, they provided the engineer and designerwith the ability accomplish more than simple to calculations. Programmable calculatorsprovided engineersand designerswith the ability to perform chains of computations and make simple decisions. While toolsto perform computations progressed from sliderulesand mechanical adding machines to programmable calculators, electroniccomputer was becoming the more accessibleand easierto use, In the late 1960s and early 1970s, largecorporations and universities access to electroniccomputers. Electroniccomputers in those days had were huge machines thatfilledentirerooms. Initially, programmers could only communicate with thesehuge devices through machine language. They used card decks to provide input instructionsand data. The computers produced output in printedform. Despite being cumbersome to use, computers enabled engineersand designersto perform huge seriesof calculationsrapidly. Electroniccomputers also had the ability, within the scope ofprograrns,to make numerous decisions,and on the basisof these,to alterthe calculationpaths. In the late 1960s, programming languages such as BASIC and F O R T R A N appeared and provided an interfacethatengineers and designerscould use in place of machine language. In the middle to late 1970s, data input changed from card decks to electronic communications through keyboards and monitors. Engineers and designerscould view output on monitors ratherthan waiting for printedoutput,thereby increasingthe speed of input and output. Computer sizesbecame smaller,and more programming languages became available. In the 1980s, desktop computers began to gain prominence. These machines had computing power thatexceeded thatof the huge machines from the preceding decade. The earlydesktop versionsused a disk operatingsystem thatwas not particularly userfriendly. But by the middle to late 1990s, desktop computers became compact and reasonablypriced,with the resultthatone of thesemachines occupied nearly every designer'sor engineer'sdesk. As desktop,and laptop computers became more prominent, methods of communicating with these machines rapidlyadvanced. What has allof thisdone to computational abilities engineers and designers? of Simply put, ithas removed them from the onerous tasksof performing stringsof simple calculationsto accomplish analyses and designs. In the modem world, engineers and designers have several options. Most significantly they can write programs that embody numerous calculations with decision trees that allow the programs to accomplish complicated analysis and designs. This frees the engineer and designer to devote their time to other considerations, enabling them to become more efficient.



A General Overview of Standard Development

At the same time as computer hardware and software underwent these significant advances, design practice for window glass began changing. Research on window glass strength and behavior indicated that strength models relating failure load to window glass thickness failed to adequately describe window glass strength. Other parameters, including stress duration, state of stress, window glass geometry, temperature, relative humidity, and window glass age proved to have measurable effects on window glass strength. Beason and Morgan [1] published a paper on failure prediction for window glass. In addition, Norville and Minor [2] published a paper that used failure prediction methodologyto assess the strength of weathered window glass. The failure prediction methodologyincorporated all factors known to affect window glass strength. The failure prediction methodology required significant computational power, for its time, at least, to provide assessments of window glass strength. The research underlying both of the papers relied heavily on advancing computer technology in experimental research as well as analysis techniques. The impact of computers on research efforts revolved around the ease with which computers allowed researchers to gather and record experimental data, Improved computing technology allowed the development of more sophisticated models of physical phenomenon. Reliability methods replaced straightforward working stress approaches in concrete and steel analysis and design. The failure prediction methodology replaced maximum stress theories and aforementioned simpler methods of assessing window glass strength. In wind load computations, load magnitude became higher, but duration shortened as improving technology facilitated improved methods to measure wind speeds and their resultant pressure on buildings. As models of physical phenomena gained in complexity, standard design methodologies followed a parallel path. Window glass design in the 1970s and early 1980s required one glass strength chart and a table of factors. With the advent of the first edition of E 1300, window glass design required 12 charts and a table of factors. In 1998, E 1300 grew into 12 charts, 5 tables embodying many faotors, and special procedures for different window glass types and constructions. The 2002 edition will contain approximately 84 charts, 6 tables, and 15 procedures to accomplish window glass design. As complexity increases, the time required for the professional's learning curve increases, Furthermore, the time needed to read charts, look up factors, and produce designs increases, too, even when the user is familiar with the standard. The author feels that a good design standard should produce the same results for the same input conditions and professional judgments of the designer. In other words, a good standard is simply an algorithm for the mathematical calculations necessary to produce a design. E 1300 fits into this definition. The author also feels that while engineers and designers must know and understand the design standards they use, they need not go repeatedly through all the machinations entailed in these design standards. Consequently, Standards Design Group, Inc., (SDG) set about formulating software that performs the computations in various standards for the design industry. SDG's first venture consisted of Comprehensive Window Glass Design, software that performed the calculations in E 1300 1998 and 2000. The beauty of the SDG software lies in the fact



that it can keep pace w/th changes in standard design practices such as E 1300 and incorporate all complications that accrue as standards become more complex in nature. Changes in the 2002 Edition ofASTM E 1300 The revision for the 2002 edition includes the addition of 1, 2 and 3-sided supported single glazed lite designs. Also, design methods for asymmetric laminated IRes, and laminated over laminated double glazed insulating units have been added. These changes are Incorporated in over 42 charts. Additionally 42 deflection charts have been added, one for each single glazed lite support and construction combination replacing the complex polynomial method. Table I summarized the additions in the 2002 edition.
Table I -

Summary ofchangesbetweenE 1300-00and E 1300-02

E 1300-00 Design Methods l-sided 2-sided



x x

Single Glazed Lite Monolithic Laminated (Symmetric) Double Glazed Insulating Unit Monolithic - Monolithic 1 Monolithic - Laminated E 1300-02 Design Methods 1-sided 2-sided Single Glazed Lite Monolithic Laminated (Symmetric)
Laminated (Asymmetric)

3-sided x x x

4-Sided x x x

x x x

x x x

Double Glazed Insulating Unit Monolithic - Monolithic Monolithic - Laminated Laminated- Laminated


' Monolithic - Monolithic represents an double glazed insulating glass unit constructed with two monolithic lites. The subsequent double glazed insulated unit combinations are represented in the same format.



Design Procedure for E 1300 - 00

For Monolithic Single Glazing:

1. Determine the non-factored load (NFL) from the appropriate chart for the glass thickness and size. 2. Determine the glass type (GT) factor for the appropriate glass type and load duration. 3. MultiplyNFL by GT to get the load resistance (LR) of the life.

For Single-glazed Laminated Glass (LG) Construction of Two Glass Plies of Equal Thickness and Glass Type Bonded Together with PVB Interlayer:
1. Determine the NFL from the appropriate chart for the glass thickness and size. 2, Calculate the aspect ratio (AR) of the lite by dividing the long side length (a) by the short side length Co): AR = a/b. 3. Calculate the flexibility ratio (b/t) by dividing the short side length (b) by the laminated glass thickness designation (t). 4. Determine the GT factor for the appropriate glass type and load duration. 5. Multiply NFL by GT to get the LR of the lite.
Design Procedure for E 1300 - 02

For Monolithic Single Glazing Simply Supported Continuously Along Four Sides:
1. Determine the NFL from the appropriate chart for the glass thickness and size. 2. Determine the glass type factor (GTF) for the appropriate glass type and load duration. 3. Multiply NFL by GTF to get the LR of the lite.

For Single-glazed Laminated Glass Constructed with a PVB Interlayer Simply Supported Continuously along Four Sides where In-Service LG Temperatures do not Exceed 50 oC (122 ~
1. Determine the NTL from the appropriate chart for the glass thickness and size. 2. Determine the GTF for the appropriate glass type and load duration. 3. Multiply NFL by GTF to get the LR of the lite.
Procedure for Calculating the Approximate Center of Glass Deflection

The maximum glass deflection as a function of plate geometry and load may be calculated from the following polynomial equations.

w=t*exp(ro +r1 * x-I- r2 *X 2)

where w = center of glass deflection (ram) or (in.), and t = plate thickness (ram) or (in.).




ro = 0.553 - 3.83 * (a / b) + 1.11 * (a / b) 2 _ 0.0969 * (a / b) 3 r 1 = -2.29 + 5.83 * (a /b) - 2.17 * (a /b)2 + 0.2067 * (a/b) 3 r 2 = 1.485 - 1.908 * (a / b) + 0.815 * (a / b)2 _ 0.0822 * (a / b) 3
x = ln{ln[q*(a/b) 2 ~E,t4]}

(2) (3) (4) (5)

q = uniform lateralload (kPa) or (psi). a = long dimension (ram) or (in.). b = short dimension (ram) or (in.). E = modulus of elasticityof glass (71.7 x 106 kPa) or 00.4 x 106 psi).

Design Examples
The author will illustrate some of the differences between E 1300-00 and E 1300-02 through two examples. The first will be a monolithic single glazed lite supported on all four sides. The second will be a laminated single glazed life supported on all four sides. For each case the design will be performed, first using E 1300-00 and then E 1300-02.
Example 1: Monolithic Window Glass Design

In this example, the author wishes to determine the appropriate glass thickness required to glaze a vertical opening having rectangular dimensions of 1600 mm (63 in.) by 1600 mm (63 in.) to resist a 3-second duration design loading of 2.2 kPa (45 psi). Additionally, after the appropriate glass thickness has been determined the author will also calculate the approximate center of glass deflection under that loading.
Using E 1300-00 - Following the steps outlined in the procedure section, for a monolithic single glazed lite design. The author starts the procedure by trying a monolithic annealed window glass lite having a nominal thickness designation of 6.0 mm (1/4 in.). E 1300-00 contains 12 non-factored load charts, one for each nominal lite thickness. Figure 1 presents the non-factored load chart for a nominal thickness of 6.0 mm (1/4 in.).

To determine the load resistance for this lite, the author projects a horizontal line from the 1600 mm (63.0 in.) point on the vertical axis and a vertical line from the 1600 mm (63.0 in.) point on the horizontal axis. These two lines intersect on the 1.50 kPa (31 psf) load line; therefore, the basic NFL is 1.50 k/'a (31 psf). For an AN monolithic lite under short duration load, the glass type factor equals 1.00. The LR for the 6.0 mm (1/4 in.) AN glass is computed by the product: LR = GT x NFL = 1.0 x 1.5 = 1.5 kPa (31 psi) (6)

Since the design load is a 3-second duration load, and E 1300-00 is b a s h on a 60-second load, the design load must be converted from a 3-second duration to a 60-second duration



Plate Length (inl 0 50 8.0 m (5/IB ~n) S m Nonfactored (Pro) PD = O.OOB I00 150 200

La oO



I kPe = 2 0 , g p s f



r, 50

a. 5(

,d rC

i0 O0
0 1000 2000 3000
(ram) 4000

0 P l a t e Length

0 5000

Figure 1 - Non-factored Load Chart for Nominal 6-ram (1/4 in.) Glass (E1300) in order to compare it to the load resistance. The author finds in E 1300 the load duration factor for a 3-second load to a 60,second load is 1.21 for annealed glass. The equivalent 60-second duration design load is the 3-second duration load divided by the load duration factor for a 3-second load: Q6o = 2.2 kPa / 1.21 = 1.8 kPa (38 psf) (7)

Since LR is less than the specified design loading, the AN glass thickness is insufficient. To increase the LR, the author will perform the same procedure using the next greater lite thickness, which is 8.0 mm (5/16 in.). The load resistance for the 8.0 m m (5/16 in.) [ire is determined to be 2.0 kPa (42 psf), which is greater than the equivalent 60-second duration design load, therefore, the 8.0 m m (5/16 in.) lite will work. The approximate center of glass deflection is calculated using the polynomial method as follows: a/b = 1600/1600 = 1.0 (8) Subsequently, ro = -2.284; rl = 1.577; r2 = 0.31 x = ln{ln[(2.2.kPa* ( 2 . 5 6 . 1 0 ' .mm2)~)/(71.7"10 ~ .kPa*(')]} = 1.43 w = 7.42. mm * exp(-2.284 + 1.577 * 1.43 + 0.31 * 1.435) - 13.7 nun (0.54 in.) (9) (10)



Using E 1300-02 - Using the steps outlined in the procedure section, for a monolithic single glazed simply supported continuously along four sides design, the author starts the procedure by trying a monolithic annealed window glass lite having a nominal thickness designation of 8.0 mm (5/16 in.). E 1300-02 contains 12 non-factored load charts for monolithic 4-sided supported lite, one for each nominal lite thickness. Figure 2 presents the non-factored load chart for a nominal thickness of 8.0 mm (5/16 in.).

Plate Length (in.)





140 8.0 mm (5116in.) Gla:ts Nonfactored Load 120 : Pb = 0.008 I kPa = 20.9 psf 1.o0


9 9 3000

9 oo
8o 60
40 20 0

3-Second Duration


1 N I r'-k I.
z0o/l I~l
A,, I I~'III


L~! ,

I ~

..'1 Ill

5~~.oo ,~ I~t'q I ~A"qkl ~-."~

7,ooAdhL~,--~k~ , .~ ~ I~ 2~ 3000 == i 4~


0 5000

Plate Length (mm)

Figure 2 - Non-factored Load Chart for Nominal 8-ram (5/16 in.) Glass (E1300-02) To determine the load resistance for this lite, the author projects a horizontal line from the 1600 nun (63.0 in.) point on the vertical axis and a vertical line from the 1600 mm (63.0 in.) point on the horizontal axis. These two lines intersect between the 2.5 kPa (52 psf) and the 3.0 kPa (63 psi') lines. Since the intersection is between two load lines, the author will "eyeball" interpolate the non-factored load. Therefore, the basic NFL is 2.7 kPa (56 psf). For a AN monolithic lite under short duration load, the GTF equals 1.00. The LR for the 8.0 mm (5/16 in.) lite is computed by the product: LR = GT x NFL = 1.0 x 2.7 = 2.7 kPa (56 psi') (ll)

Since the design load is a 3-second duration load and E 1300-02 is also based on a 3second load one does not need to factor the design load. Therefore, the LR for the 8.0 mm (5/16 in.) lite is greater than the design load, and the design is acceptable. E 1300-02



provides a new set of deflection charts for each lite thickness. Figure 3 presents the deflection chart for nominal thickness of 8.0 mm (5/16 in.).
Load x Area 2 (kip" ft 2)






40 35


I _~,

" ii








25 2O

9 s ~

10 84 5


.-.jr..7 I/

" t

,' S

~ /


, ........ i


o, _0.4


~ / ~

l~" / _ r

.... /

0~ 0


I Oe,,eo,,o.v.Ooa,...~ , I , I
120 160

Four Sides Simply Supported

~io.2 I Ioo

Load x A r e a 2 (kN" m 2)

Figure 3 - Deflection Chart for Nominal 8-mm (5/16 in.) Glass (E1300-02) One can determine the approximate center of glass deflection using the deflection chart in a similar fashion to the load charts. To determine the deflection for this lite, the author projects a vertical line from the point on the horizontal axis corresponding to the value of Load x Area 2 in kN-m2 (kip-ft'). Load x Area 2 = 2.2 kPa x (1600 mm x 1600 ram)2 = 14.42 kN'm 2 (34.88 kip'ft2) (12)

The deflection is the point on the vertical axis where a horizontal line projected from the intersection of the aspect ratio and the vertical line meets the vertical axis. Therefore, the approximate center of glass deflection is 14 mm (0.55 in.).

Using Window Glass Design 2002 - The following figures illustrate the so,ware interface. Figure 4 presents the main design input window, where the designer defines the glass construction, rectangular dimensions, and loading conditions. Once the designer enters in the desired parameters he or she simply clicks on the calculate button. Figure 5 shows the results window returned by the software which includes the design load, the load resistance and the approximate center of glass deflection.



Figure 4 - Window Glass Design lnput Window (Window Glass Design 2002)

Figure 5 - Results Window (Window Glass Design 2002) Two additional windows are available to the designer. The sketch window shows a dimensioned drawing of the glass construction and the details window shows specific factors used in calculating the load resistance, The software also provides a detailed two page report that summarizes all of the information contained on the four design windows.

Example 2: Laminated Window Glass Design

In this example, the author wishes to determine the appropriate glass thickness required to glaze a vertical opening having rectangular dimensions of 965 nun (38 in.) by 1930 mm (76 in.) to resist a 3-second duration design loading of 2.4 kPa (50 psf).



Additionally, after determining the appropriate glass thickness, the author will also calculate the approximate center of glass deflection under that loading. Using E 1300-00 - Following the steps outlined in the procedure section, for a single-glazed laminated glass construction of two glass plies of equal thickness and glass type bonded together with PVB interlayer, the author starts the procedure by trying a glass lite having a nominal thickness designation of 8.0 mm (5/16 in.). Figure 6 presents the non-factored load chart for nominal thickness of 8.0 mm (5/16 in.).

Plate Lengt}~ (in) 5O

i i i r i l # J i




B.O mm

(5/15 in) Crass


-'/]' '!'"

:J )J"')'i

Nonfmctonea Loar




1 kPa = 20.g Dsf

/q I I I I I I L~I o.75.,~11 I I I I I J./] I /'11~ I I I l.~11

r'. "0
Z al


I I g 4 " l I I I .. 25/] I P I J -/1 T I I I.~V1 I 1.5(/~] k l I "F./rl I P I.~V1 I I I. 2000 /1%Fl-d J/Itl I I..~'1 I I . L - ~ I 2 . o o / I II ~L L ~ I I~t"1 P ~ I IJ2,5o/]%jll ~ I%L.-4-I~1 J J r ~ ~ r ' l ~ L - ~ l I 3-oo~I~1 ~YLI~b,~I~L-~L~T--~..-; ,~! I-Jr~T I 1000



I [ I IIIII Ii iJl~r-I I I I In,I I I J I I 1 I I I I I I I I o 1000 2000 3000 4000 5000

Plate Length (mm}

.424 ~


_l .I I. ! ! ! ! ! ! !l!

Figure 6 - Non-factored Load Chart for Nominal 8-ram (5/16 in.) Glass (El300) To determine the load resistance for this lite, the author projects a horizontal line from the 965 mm (38.0 in.) point on the vertical axis and a vertical line from the 1930 mm (76.0 in.) point on the horizontal axis. These two lines intersect on the 2.5 kPa (52 psf) load line; therefore, the basic NFL is 2.50 kPa (52 psf). For a AN laminated lite under short duration load, the glass type factor equals 0.75. The LR for the 8.0 mm (5/16 in.) AN glass is computed by the product: LR -- GT x NFL = 0.75 x 2.5 kPa = 1.9 kPa (40 ps0 (13)

Since the design load is a 3-second duration load, and E 1300-00 is based on a 60 second load, one must convert the design from a 3-second duration to a 60-second in order to compare it to the load resistance. The author finds the load duration factor for a



3-second load to a 60-second load is 1.21 for annealed glass. The equivalent 60-second duration design load is the 3-s~,,ond duration load divided by the load duration factor for a 3-second load: Q6o = 2.4 kPa / 1.21 = 2.0 kPa (42 psf) (14)

Since LR is less than the specified design loading, the AN glass thickness is insufficient. To increase the LR the author will perform the same procedure using the next greater IRe thickness, which will be 10.0 mm (3/8 in.). The load resistance for the 10.0 mm (3/8 in.) lite is 3.5 kPa (73 psf), which is greater than the equivalent 60-second duration design load. Therefore, the 10.0 mm (3/8 in.) 1Re will with stand the design load. The approximate center of glass deflection, calculated using the polynomial method, is 4.6 mm (0.18 in.). Using E 1300-02 - Following the steps outlined in the procedure section, for singleglazed laminated glass constructed with a PVB interlayer simply supported continuously along four sides where in-service LG temperatures do not exceed 50* C (122 ~ F), the author starts the procedure by trying a glass 1Re having a nominal thickness designation of 8.0 mm (5/16 in.). E 1300-02 contains 12 non-factored load charts for laminated 4sided supported IRes, one for each nominal 1Re thickness. Figure 7 presents the nonfactored load chart for laminated glass with a nominal thickness of 8.0 mm (5/16 in.).



Plate Length(in) 100


i ' .'. ,~ i


8mm(5116in.)PVBLiminate NonfactoredLoad ~ FourSidesSimplySul:ported 1.0(], , , Pb = 0.008 4 ,~= / / 1 kPa = 20.9 psf e- 100 3 SecondDuration . 5 0 , / / ~ . ~ 50oc (122oF)
. 2.0 /--IN. '

.... ~" E ~"

~0oo .~

m n 5O


T ,--q--l-I-




2000 3000 Plate Length(ram)



Figure 7 - Non-factored Load Chart for Nominal 8-ram (5/16 in.) Laminated Glass (E1300-02)



To determine the load resistance for this late, the author projects a horizontal line from the 965 mm (38.0 in.) point on the vertical axis and a vertical line from the 1930 mm (76.0 in.) point on the horizontal axis. These two lines intersect between the 3.0 kPa (63 psf) and the 4.0 kPa (84 psf) lines. Since the intersection is between two load lines the author will "eyeball" interpolate the non-factored load. Therefore, the basic NFL is 3.2 kPa (67 psf). For a AN laminated late under short duration load, the GTF equals 1.00. The LR for the 8.0 mm (5/16 in.) late is computed by the product: LR = GT x NFL = 1.0 x 3.2 = 3.2 kPa (65 psf)


Since the design load is a 3-second duration load, and E 1300-02 is also based on a 3second load, one does not need to factor the design load. Therefore, the LR for the 8.0 mm (5/16 in.) late is greater than the design load so the design is acceptable. E 1300-02 provides a new set of deflection charts for each late thickness. Figure 8 presents the deflection chart for laminated glass with a nominal thickness of 8.0 mm (5/16 in.).

Load x Area2(kip.ft 2) 0 40 , 100 i 200 , 300 ,


400 1.4 1.2 1,0 0.8


,r /


E 20




8 mm (5/16 in.) PVB Laminate --., 0.4 Four Sides Simply Supported Defllmtlon vs. (Load x Area= 0.2 500C (122"F) -"

30 60 90



0.0 180

Load x Area =(kN,m =)

Figure 8 - Deflection Chartfor Nominal 8-ram (5/16 in.) Laminated Glass (E1300-02) One can determine the approximate center of glass deflection using the deflection chart in a similar fashion to the load charts. To determine the deflection for this late, the author projects a vertical line from the point on horizontal axis corresponding to the value of Load x Area 2 in kN-m 2 (kip-t~). The deflection is the point on the vertical axis where



horizontal line projected from the intersection of the aspect ratio and the vertical line meet the vertical axis. Therefore, the approximate center of glass deflection is 9.5 mm (0.37 in.).

Using Window Glass Design 2002 - The following figures illustrate the software interface. Again the designer simply enters in the desired parameters and clicks the calculate button. Figure 9 presents the main design input window and Figure 10 shows the results window returned by the software.

Figure 9 - Window Glass Design Input Window (Window Glass Design 2002)

Figure 10 - Results Window (Window Glass Design 2002)




As standards continue to change and evolve to become more accurate, the calculations comprising them increase in complexity. In general, standards either will include large amounts of complex calculations or include many charts and tables which summarize the complex calculations. Both methods have pros and cons. For the standards that employ complex sets of calculations, many designers and engineers may have difficulty becoming proficient with the design procedures which are very time consuming. A benefit of using this type of standard is that they will provide, when used correctly, the most accurate designs that the current research and technology support. On the other hand, standards that consist of many charts to be "eyeball" interpolated, although faster to use, provide opportunity for variation in results due to individual interpretation. As with E 1300 the non-factored load charts are easy to use, but the values of the loads may vary by more than + 5% due to the way the designer "eyeball" interpolates. Depending on the design, this may be the difference between using a smaller economical lite and one that is larger and more costly. This is where the advantage of software comes in. Through software, the designer gets the best of both situations. The software performs the calculations of the complex design procedures, while only requiring minimal input from the user. Since the software does the actualcalculationsand includes the underlying data from the charts,the designs are more consistentand accurate. Window Glass Design 2002, the lastedversion of SDG's window glassdesign software includes the numerical data that forms the basis for the non-factoredload chartsas well as allcalculationsused in E 1300. This software produces consistentresultswhere everyone obtainsthe same designs for the same input,alleviatingthe differencesdue to "eyeball"interpolationof charts. An additionaladvantage in using soRware for design calculationsis the reduced learningcurve for designers and engineers due to significantrevisionsin the standards. Since the program willperform allthe calculationsvery quickly,the time spent on calculationsby engineers and designerswill be greatlyreduced.

[1] Beason, W. L. and Morgan, J. R., "Glass Failure Prediction Model," Journal of Structural Engineering, Vol 110, No. 2, February 1984. [2] Norvillo, H. S., and Minor, J.E., "The Strength of Weathered Window Glass," Bul. Bulletion of the American Ceramic Society, Vol. 64 No. 11, pp 1467-1470, November 1984.

W. Lynn Beason t and A. William Lingnell ~

A Thermal Stress Evaluation Proeedure for Monolithic Annealed Glass

Reference: Beason, W. L., and Lingnell, A. W., "A Thermal Stress Evaluation Procedure for Monolithic Annealed Glass," Use of Glass in Buildings, ASTM STP 1434, V. Block, Ed., ASTM International, West Conshohocken, PA, 2002. Abstract: Thermal breakage caused by the uneven heating of glass by solar irradiance is one of the most common causes of annealed glass breakage in buildings. The classic thermal breakage situation occurs when the edges of a glass plate are shielded by the window frame system or significant building protrusions, while the central portions of the glass are not. This uneven heating situation causes tensile stress to develop along the edges of the glass. This paper presents a theoretically based procedure that can be used to estimate the level of tensile stress induced in a glass plate exposed to this typical thermal loading situation. This procedure is based on the results of a finite element parametric study and engineering judgment. In addition, an edge strength failure prediction model for annealed glass is presented which allows a designer to determine if the level of risk associated with an estimated edge stress is acceptable or unacceptable. Keywords: glass, thermal stress, windows, design Introduction

One of the most common factors that leads to the breakage of window glass in buildings is thermal breakage. Thermal breakage involves the occurrence of thermal gradients induced by uneven heating of glass by solar irradiance or other heat sources. When sunlight impinges upon a glass plate, some of the energy is reflected from the surface of the glass, some of the energy is absorbed by the glass, and some of the energy is transmitted through the glass. The energy that is absorbed by the glass increases the temperature of the glass above the previously existing equilibrium condition. If the glass is uniformly heated, and if the support system can accommodate the thermally induced expansion of the glass, no stresses will be induced by a uniform temperature increase. However, if part of the glass is shielded from the direct effects of the sunlight by the edge support system or by a shadow pattern, the glass will be heated unevenly. The uneven heating of the glass will give rise to thermally induced in-plane tensile and compressive stresses. When these thermally induced tensile stresses interact with critical edge flaws, thermal breakage can result.

' AssociateProfessor,CivilEngineeringDepartment,TexasA&M University,CollegeStation,TX 77843. 2Principal,LingnellConsultingServices, 1270ShoresCourt,Rockwall,TX 75087.



by ASTMInternational



In a typical thermal breakage situation, the edges of the glass are subject to higher tensile stresses than the surfaces of the glass away from the edges. In addition, the stress concentrating flaws along the edges of a glass plate are generally more severe than surface flaws away from the edges. These circumstances combine to create a situation where thermally induced glass breakage usually initiates at the edges of glass plates. Although the generalities of the solar induced thermal stress problem have long been understood by those working in the glass industry, there is very little information available to explicitly describe the development of thermal stresses in glass. As a result, most thermal stress evaluation procedures are based more on empirical methods and experience than explicit engineering formulations. The purpose of this paper is to present a rational procedure that can be used to evaluate the thermal stress effects of sunlight on rectangular glass plates. A generalized theory to explain the breakage of glass subjected to thermal stresses must first incorporate a procedure to calculate the magnitude of the maximum tensile stress along the edges of the glass. This is accomplished through the use of finite element analysis (FEA). Secondly, the procedure must include a method to account for the occurrence and severity of edge flaws. This latter objective is accomplished through the use of an edge strength failure prediction model (ESFPM). These two items are combined to develop a procedure that allows a designer to determine the risk associated with a particular thermal stress situation.
Solar Induced Thermal Stresses in Glass Plates

In-plane or membrane stresses are introduced in glass by uneven heating of the glass plate. In the current discussion it is assumed that the glass plate is installed in a typical frame that provides four sides of continuous lateral support with an edge bite (glass engagement into frame system) that protects a perimeter strip of glass from direct exposure to the sunlight. It is also assumed that the edges of the glass are free to slip within the support system so that normal thermal expansions and contractions may be accommodated. If the glass is supported in some other manner, the character of the stress distribution may be significantly different than that discussed herein. Thermal stress breakages tend to originate at the edges of glass plates as the result of the interaction of thermally induced tensile stresses with edge flaws or edge damage. While not as common, severe surface damage or flaws can also trigger thermal stress fractures. Thermal stress breakage is exacerbated by the heat absorption characteristics of the glass, the magnitude of the edge bite, interior shading devices, exterior shading conditions, and other factors. The thermal stress evaluation procedure presented in this paper focuses only on shading caused by the edge bite. In general, thermal stress breakage is associated with relatively low levels oftensile stress. Hence, the level of residual edge and surface compression inherent in either heat-strengthened or fully tempered glass tends to preclude thermal stress breakage problems. Therefore, the majority of thermal stress problems are experienced by annealed glass that has a low level of residual edge and surface compression. One of the most commonly discussed examples of thermal stress breakage occurs on clear cold winter days. During the cold night the glass cools to an equilibrium condition that depends on the temperature inside the building, the temperature outside of the



building, the interior surface film coefficient, and the exterior surface film coefficient. When sunlight begins to shine on the glass, the exposed area of the glass is heated. As the exposed area of the glass warms, it expands. As the exposed area of the glass expands, the unheated edges of the glass around the perimeter are forced into tension. Ifa small block of thermally loaded glass is examined along the edges as shown in Figure 1, it becomes clear that the edge block must be in uniaxial tension or compression. This is due to the fact that the stress components on the free surfaces of the block are all zero as shown.

/,[ O~

Figure 1 --Thermally induced edge stress. Because thermally induced stresses are membrane stresses, both the top and bottom surfaces of the glass are exposed to the same stress. Because of the orientation of the edge stress, breakage of the glass at the edge of a thermally loaded glass plate usually results in a crack that is normal to both the vertical and horizontal projections of the edge of the glass as shown in Figure 2. As the crack propagates inward, the stress distribution in the glass plate becomes more complicated and the crack may branch into different directions depending upon the transient state of stress within the glass as it fractures.



Figure 2 --Thermally induced crack. While the above scenario is the most frequently quoted example of the occurrence of thermal stresses, the initial temperature of the glass has little to do with either the occurrence or severity of the thermal stress. Rather, the severity of the thermal stress associated with a particular situation is primarily dependent on the amount of energy absorbed, the convection/radiation conditions, and the nature of the support conditions. The initial temperature of the glass has little effect as long as the glass is in a steady state condition when it is first subjected to the sunlight. The most probable reason that thermal breakage tends to be associated with winter conditions is that solar irradiance in the northern hemisphere is at a maximum during winter months, and the elevation angle of the sun in winter is more favorable for loading vertical glass plates than it is in the summer. While winter provides excellent conditions for inducing thermal loads in glass, thermal stresses can become critical at any time when the proper conditions exist. Finite Element Analysis of Thermal Stresses FEA provides a convenient mechanism for standardizing a procedure that allows a quantitative analysis of thermal stresses induced in glass by uneven heating caused by solar irradiance. In most FEA codes the general thermal stress analysis procedure is conducted in three major steps. The first step is the development of the FEA model that consists of nodes and elements that represent the geometry of the glass plate being studied. Then, the model is used to perform a transient or steady-state analysis of the heat flow in the glass. The result of a thermal analysis is the variation of temperatures throughout the body of the glass at discrete points in time. In the third step, the temperature data corresponding to a given point in time is transferred from the thermal analysis to a stress analysis. The stress analysis allows the stresses associated with the particular temperature distribution to be calculated. Although there are several different FEA codes avail-



able for use to analyzethermal stresses in glass, the FEA data presented in this paper were generated using the ALGOR version 12 program [ 1].

Formulation of the FEA Model

The first step in formulation ofa FEA model to represem a particular situation is to formulate the geometry of the problem. In this process discrete nodes throughout the geometry of the plate are graphically defined in the context of a three dimensional orthogonal coordinate system. Then, these nodes are used to create elements that define the extent of the glass plate under consideration. This collection of nodes and elements is referred to herein as the finite element model.

Thermal Analysis
The first step in formulation of a thermal analysis is to define the properties and characteristics of the materials involved in the problem under study. The material properties of interest include the specific heat, thermal conductivity, and density. In this paper all units are expressed in SI units. The surface film coefficient is used to model the transfer of heat energy between a glass surface and its surrounding environment based on the temperature differential betwoen the two. Thus, the surface film coefficient has a significant effect on the rates of temperature change and the magnitudes of the temperatures that develop throughout a glass plate that is subjected to given set of conditions. It is the magnitudes of the temperature differentials throughout the body of the glass plate that ultimately determine the magnitudes of the thermal stresses within the glass plate. Therefore, it is necessary to define realistic surface film coefficients for exposed glass surfaces. Following the precedent established by American Society of Heating, Refrigerating and Air-Conditioning Engineers (ASHRAE), the surface film coefficients employed in the analyses discussed herein, represent the combined effects of radiation and convection [2]. The surface film coefficients are expressed in terms ofunits of W/(m 2 .K). The writers have made estimates of the interior and exterior surface film coefficients for glass based on information presented by ASHRAE and engineering judgment [2]. These values are 13.55 W/(m2-K) for the exterior surface and 8.04 W/(m2-K) for the interior surface. Under some conditions, actual surface film coefficients can be greater or less than these assumed values. However, it is believed that FEA analyses conducted with these assumed coefficients provide conservative results for the glass design situations discussed herein. Thermal analyses can be conducted for both steady-state and transient conditions. A steady state analysis is used to determine the equilibrium conditions in a long-term heat flow problem with constant conditions. A transient analysis is used to determine the variation of the temperatures in glass as a function of time. In the current effort, a steadystate analysis was conducted for each situation with the glass exposed to constant indoor and outdoor temperatures to establish an equilibrium condition with no solar irradiance. The resulting equilibrium temperatures were then used as initial conditions for a transient analysis of the glass plate exposed to solar irradiance. The results of the transient thermal



analysis provided the variation of temperature in the glass as a function of time for the given set of conditions.

Stress Analysis
Formulation of a stress analysis for glass requires a definition of its material properties and a specification of appropriate boundary conditions. The required material properties of interest are the modulus of elasticity, E, expressed in GPa, Poisson's ratio, g , and the thermal coefficient of expansion, a , expressed in m/m/K. In the current situation, the modulus of elasticity for glass was taken to be 71.7 GPa, the Poisson's ratio for glass is taken to be 0.22, and the thermal coefficient of expansion for glass is taken to be 8.82 x 10"6 m/m/K. The magnitude and distribution of the thermal stresses induced in a glass plate is extremely sensitive to the type of edge support conditions. In most instances, the boundary conditions associated with the edge of a glass plate are such that the edge is prevented from motion perpendicular to the surface of the glass plate, but is allowed to slip freely in the plane of the plate. Therefore, these boundary conditions were used to generate the results presented in this paper. If the edges of a glass plate are prevented from slipping in the plane of the plate, the resulting stress conditions will be substantially different. To calculate the thermal stresses for a particular situation, the variation of temperatures through the glass as a function of time are used as input for a linear static stress analysis. The linear static stress analysis allows the calculation of the distribution of stresses induced in the glass given the distribution of temperatures and the prescribed support conditions. This process is conducted for different times to determine the variation of the maximum edge tensile stress as a function of time. By examining a wide range of times, the magnitude and time of occurrence of the maximum edge tensile stress can be estimated.

Edge Strength Failure Prediction Model

Most current glass thickness selection charts presented in the United States are based on the glass failure prediction model (GFPM). The GFPM recognizes that laterally loaded rectangular glass plates with four-edge support fail as the result of the interaction between surface tensile stresses and stress concentrating surface flaws [3]. However, the maximum tensile stresses and the critical flaws associated with thermal breakage are almost always located along the edges of the glass. Thus, a modified failure prediction model must be used to model the thermal stress situation. The thermal stress failure prediction model is patterned after the GFPM, except that it relates the probability of failure to the distribution of edge stresses and the characteristics of the edge flaws, instead of surface stresses and surface flaws. The thermal stress model thus developed is referred to herein as the edge stress failure prediction model (ESFPM). Both the ESFPM and the GFPM are based on a statistical failure theory for brittle materials that was originally presented by Weibull [4]. The Weibull theory suggests that the probability of breakage, Pb, of a brittle material can be expressed as follows:



Pb = 1-e-B


where B is a risk function that is evaluated by integrating the combined effects of flaw severity and tensile stresses experienced by the glass plate. The following expression for the edge strength risk function, B, assumes that the risk o f failure in a thermal loading situation is controlled by the length of the edge subjected to tensile stress:

No. es Le



where t~ is the duration of the thermal stress expressed in seconds, k, and m are the edge flaw characteristics, and rr,~, (x) is the maximum principal tensile stress along the edge of the glass. As indicated, the integral is summed along all glass edges that are subjected to tensile stresses. In the case of thermally loaded glass plates, the edge stress is zero at the comer and increases to a maximum value that remains relatively constant along the edge. Experience suggests that it requires a distance of approximately 150 mm from the comer for the edge stress to reach its maximum value. Thus, in application of the ESFPM, it is assumed that all of the perimeter except for the 150 mm zones around the corners are subjected to the same maximum principal edge stress, rr.~ x . Thus, the effective length of the perimeter of the glass exposed to thermal stress is found by subtracting 1.2 m (150 mm x 8) from the total perimeter. This allows equation (2) to be rewritten as follows:

(3) where p is the length of the perimeter of the glass expressed in meters, (p -1.2) is the effective length of the perimeter, and the other factors are as previously defined. It is recommended that plates with a perimeter length less than 1.5 m be conservatively treated as though they have a 1.5 m perimeter. This means that the recommended minimum effective perimeter length is 0.3 m. Equation (3) is the form of the ESFPM that was used in development of the information presented in this paper. To use the ESFPM for the design of thermally loaded glass it is necessary to have edge flaw characteristics that are representative of the glass edge conditions that are to be expected in a given installation. It is recommended that the m edge flaw parameter be set to a value of 7 for glass design purposes based on expected coefficients of variation associated with glass failure strength data [3]. Using this value of m, and results of a large number of edge strength experiments, it is recommended that a value of k, = 8.93 x 10.54 m u. N'7be used to model glass edges. It is believed by the writers that these edge flaw parameters are reasonable for most thermal design situations associated with glass thicknesses up to and including 6 mm.



Finally, before the ESFPM can be used for thermal design it is necessary to establish a reasonable duration for thermal loadings. While it is reasonable to use a 60-second duration loading for wind loading situations, thermally induced edge stresses usually last more than 60 seconds. While no exhaustive effort has been made to establish a definitive duration for thermal loading, it is the writers' opinion that 60-minutes is a reasonable duration to assume. This judgment is based upon the writers' previous experience observing thermal loadings situations and FEA results. Therefore, a 60-minute load duration is used in the formulations presented herein.

18 " ' t ~ ' ~ 16 0 . 0 0 ~ o.oo'~ 14 --'-t

POB = 0.008 POB = 0.004 POB = 0.002

. . . . . . POB = 0.001

POB = 0.0001



i.,i,,, \

~ '

----- -__._._____~

0 0 5 10 15 Pedmeter (m) 20 25 30

Figure 3 --Probabilityof Breakage (POB)Chart If the assumed edge flaw characteristics and load duration are substituted into equation (3) the following relationship results: B = 5.36 x 10-53(p -1.2)[trm~ ]7 (4)

This equation can be solved for the maximum edge stress, tr~.,~, as a function of the glass perimeter, p , and the edge strength risk function, B, to yield the following:


1 13

tYm~ =I1"87X 1052 (p-l.2)B ]~


Equation (5) was used to develop Figure 3, which presents the variation of the maximum edge stress as a function oforobability_ofbreakage (POB) and perimeter length. As shown in Figure 3, the allowable edge stress varies from about 5 to 17.5 MPa as the POB varies from 0.0001 (1 lite per 10 000 lites) to 0.008 (8 lites per 1 000 lites) and the perimeter varies up to 30 m. To determine the maximum allowable edge stress for a particular situation simply enter the horizontal axis of Figure 3 with the glass perimeter and project upward to the desired POB and then project to the left to estimate the allowable edge stress.
Thermal Stress Evaluation Procedure

The purpose of this section is to present a thermal stress evaluation procedure (TSEP) that can be used to evaluate the thermal stress performance conditions associated with annealed glass. The TSEP is accomplished by performing three steps. First, it is necessary to estimate or calculate the level of edge stress induced in the glass subjected to a defined set of conditions. Then, an acceptable POB that is consistent with the project expectations must be determined. Finally, a rational decision as to whether or not the glass meets expectations must be made. The details of the TSEP are discussed below.

Determining Thermal Stress

The best method to calculate the edge stresses for a given situation is to use FEA techniques as described in the previous section to model the specific situation under consideration. To do this, it is necessary to develop a detailed FEA model that accurately represents the situation under consideration including a detailed description of the geometry and exposure conditions. It is recommended that a detailed FEA approach be used for all significant projects, particularly those projects that involve large glass plates, a large number of glass plates, or unusual glazing and building conditions. While developing a detailed FEA model to calculate the edge stress is the most desirable approach, it is not always economically feasible or practical to expend the resources necessary to accomplish this for each individual project. Therefore, results of three simplified FEA models are presented herein. These results allow a reasonable estimate of the magnitude of the edge stress to be made without performing a detailed FEA for each specific problem. These three models were selected to represent the most favorable edge conditions, the most unfavorable edge conditions, and a set of edge conditions that are believed to represent many conventional applications. The most favorable edge conditions occur when the glass edge in the glazing pocket is insulated so that no heat flows through the glass boundary within the edge bite. This means that the only mechanism for heat flow into or out of the glass in the edge bite region is through conduction within the glass. This condition occurs when the glass is perfectly insulated in the edge bite area.



The most unfavorable edge conditions occur when the glazing pocket is assumed to have a large thermal mass (heat sink condition) that effectively prevents the edges of the glass from warming when exposed to a solar heat load. In development of the model that represents the most unfavorable conditions it was assumed that the glass boundary within the edge bite remains at its initial equilibrium temperature throughout the entire thermal exposure. This closely represents the case where the edge of the glass is in contact with a massive heat sink. Such a situation occurs if the edge bite is encased in concrete. This case represents the most critical situation for the development of thermal stresses. 35


f" F





--Worst Case Assumed ( Conditions

- - - Best Case

E d g e Bite, m m


Figure 4 --Thermal Stress Factor Chart The conventional conditions selected for the current analysis are based on generic characteristics of glazing systems commonly used in commercial construction situations. These characteristics consist of aluminum framing members with rubber perimeter gaskets on the interior and exterior surfaces of the glass in the glazing pocket. In this conventional model it is assumed that the edge of the glass is supported by the rubber gaskets on both sides for the full extent of the glazing pocket with proper setting and edge blocks so that the glass does not directly contact the ahtminum. The rubber gaskets are assumed to be 4.76 mm thick. The rubber and aluminum surfaces facing the exterior and interior


1 15

building environments are assumed to have the same surface film coefficients as the exterior and interior glass surfaces, respectively. It is believed by the writers that these conditions represent realistic conditions for many practical glazing situations. Results from FEA models for each of these three cases were used to develop Figure 4. Figure 4 presents the variation of the thermal stress factor (TSF) in terms of edge bite for thicknesses up to 6 mm glass. The units associated with the TSF are kPa/(W/m2). The maximum thermal stress experienced by the edge of the glass is found by multiplying the TSF by the total solar load (TSL) on the glass given in W/m 2 . The TSL is obtained by multiplying the total solar absorptance of the glass expressed as a decimal fraction by the intensity of the solar irradiance on the glass surface expressed in W//m .

Acceptable Probability of Breakage

The concept of acceptable POB for glass subjected to uniform wind loads is well accepted within the glass design community. However, a clear definition of the acceptable POB for glass subjected to thermal loadings is not as well defined. Logically, the acceptable POB for a particular situation should be based on a number of factors including the number oflites in the building and the consequences associated with thermal failures. The design professional should determine the acceptable POB for each application based on each unique situation. For wind design applications an acceptable POB associated with the occurrence of the design wind event typically ranges from 0.001 (1 per 1 000) to 0.008 (8 per 1 000). However, it seems reasonable to the writers that the designer might wish to contemplate the use of POBs as low as 0.0001 (1 per 10 000) depending upon the application. The final selection of the acceptable POB is the responsibility of the glass designer. Formal Thermal Stress Analysis Procedure The first step in evaluating a thermal stress situation is to determine the thermally induced stresses affecting the plate. Then, a decision must be made as to whether or not the thermally induced stress for the current situation presents an unacceptable risk or not. Both of these steps are explained below. In addition, a brief example demonstrating the use of the procedure is presented.

Evaluation of the ThermalStress

The most accurate way to estimate the thermal stresses for a particular glass plate is to perform a detailed FEA of the situation under consideration. In lieu of this process, the information presented earlier in this paper can be used to estimate the thermally induced stresses. The following steps have been organized into a step-by-step procedure to assist in application of the information presented in this paper. 9 Select the type of glass to be analyzed. Determine the size of the lite, width, and height.

11 6


Determine the total solar absorption (As) from the solar optical properties of the glass. As is found by subtracting the total solar transmission (Ts) expressed as a decimal fraction and total solar reflection (Rs) expressed as a decimal fraction from 1.0 as follows: A, = 1.0-(T~ +R,) (6) 9 Determine the solar load by multiplying the maximum solar irradiance (SI) expressed in W/m s by the total solar absorption as follows: SL = SI x As 9 (7)

Determine the glazing type that best represents the glass under consideration based on information presented earlier in this paper. 9 Determine the thermal stress factor, TSF, for the selected edge conditions and edge bite using Figure 4. 9 Finally, determine the estimated thermal stress by multiplying the thermal stress factor, TSF, by the solar load as follows: o't,,.,.~t = TSFx SL (8)

Evaluation of Risk of Breakage

The first step in the evaluation of the risk of breakage is to establish an acceptable probability of breakage (POB). The POB is to be established by the designer. As stated earlier, selection of the proper POB is based on a range of variables and engineering judgment. Then, the allowable stress, cro,ow,is determined using Figure 3. If the calculated thermal stress, crt,n , t , is less than or equal to the allowable stress, the risk of breakage is judged to be acceptable. The following steps describe the evaluation procedure. Select a defensible POB that is consistent with the goals of the project. Determine the perimeter, P, of the lite by summing two times the width, W, and two times the height, H, of the lite as follows: P=2xW+2xH (9)

9 Enter the horizontal axis of Figure 3 with the perimeter of the glass plate. Project a vertical line up to the curve that represents the POB selected for the application. At the intersection of the vertical line with the acceptable POB curve extend a horizontal line to the left to find the allowable stress, O'o,ow. 9 If the calculated thermal stress, cr,h,,,,o~, is less than the allowable stress, cra,,~, then the risk of breakage is judged to be acceptable.


1 17

If the calculated thermal stress, cr,h~,,ot, is greater than the allowable stress, o'aUow, then the risk of breakage is judged to be unacceptable. If the risk of breakage is judged to be unacceptable, it might be desirable to perform a more accurate estimate of the thermally induced stresses through a detailed application of FEA. The more detailed FEA might result in a more acceptable outcome than the more conservative procedure discussed above. Otherwise, the glass being examined should be heat-treated or another type of glass should be selected.

Example of Thermal Stress Analysis

In this example, a 6 mmx 1.5 m x 2.4 m glass plate with a total solar absorptance of 0.64 is evaluated. It is assumed that this glass plate is subjected to a thermal load of 630 W/m 2 . Further, it is assumed that it has an edge bite of 19 ram. The total solar load experienced by the glass is then calculated to be 403 W/m 2 . It is found that the TSF is about 14 kPa/(W/m 2) for the best case situation, about 26 kPa/(W/m 2) for the conventional glazing system, and 32 kPa/(W/m 2) for the worst case situation. The COl'responding edge stresses are then calculated to be 5.6 MPa for the best case, 10.5 MPa for the conventional case, and 12.9 MPa for the worst case. The allowable edge stress is found to be about 11.3 MPa for a probability of failure of 0.008 (8 lights per 1 000) from Figure 3. This means that the best case situation is acceptable by a wide margin, the conventional case is acceptable by a narrow margin, and the worst case situation is not acceptable. If a lower probability of failure is desired because this glass is installed in a major structure with hundreds of windows, or if there are exacerbating circumstances such as shadows, the probability of breakage might be unacceptable.

If the middle area of a glass plate is heated by sunlight while the edges of the glass are shielded by edge support conditions, then the plate experiences differential heating. The thermal gradients that develop in the glass induce tensile stresses along the edge of the glass that can he sufficient to initiate failure of annealed glass in some cases. While this general phenomenon has been widely recognized within the glass industry, the mechanics of the development of the thermal stresses have not been well understood. Thus, designers have been forced to deal with thermal breakage using anecdotal and empirical results. This paper presents the basis for a thermal stress evaluation procedure (TSEP) for annealed glass. The TSEP incorporates results of FEA to estimate the magnitude of the thermally induced stresses and an edge strength failure prediction model (ESFPM) that can be used to judge the significance of the calculated thermal stresses. Procedures and guidance are presented in this paper that allow a designer to estimate the magnitude of the thermal edge stresses induced in a glass plate. This procedure is based upon assumptions and idealizations that are believed to be reasonably representative of many situations. However, it is the responsibility of the designer to examine these assumptions, and to make sure that the assumptions are compatible with the application under consideration. If the assumptions are judged to be inappropriate for a particular



application, or if the simplified procedure yields results that are marginal, a designer should consider formulating a complete FEA for the situation under consideration. Once the thermal stress in a particular situation is determined through the application of the simplified procedure or through a rigorous application of FEA, the significance of the calculated stress must be evaluated. This is accomplished through the application of the ESFPM. The ESFPM was used to develop a relationship between allowable thermal edge stress, perimeter of the glass, and the acceptable POB. This relationship is summarized in the POB chart presented in Figure 3. This chart allows a designer to make a decision as to whether or not a particular thermal stress is acceptable or not. The proposed thermal stress evaluation procedure presented in this paper allows a designer to make a reasonable decision for different glass applications. It is believed that the assumptions and parameters that are incorporated in the analyses are reasonable, however, it is recognized that this new approach is largely untried. While the general procedure presented herein is a major step forward, the thermal stress procedure does not address the effects of shadows on the development of thermal stress, the effects of interior shading devices, the effects of interior heat traps, the effects of the use of low emissivity coatings, and the effects associated with the use of insulating glass. The writers are currently in the process of incorporating these and other factors into the TSEP.

The writers gratefully acknowledge support provided by Cardinal IG and Visteon Float Glass Operations for some aspects of the research reported herein.

[ 1] "ALGOR Finite Element Analysis Software - - Version 12,"ALGOR, Inc., 150 Beta Drive, Pittsburgh, PA, 2001. [2] "ASHRAEHandbook, Fundamentals," American Society of Heating, Refrigerating and Air-Conditioning Engineers, Inc., Atlanta, GA, 2001. [3] Beason, W. L., Kohutek, T. L., and Bracci, J. M., "Basis for ASTM E 1300 Annealed Glass Thickness Selection Charts," Journal of Structural Engineering, ASCE, Vol. 124, No. 2, 1998, pp. 215-221. [4] Weibull, W., "A Statistical Theory of the Strength of Materials," Ingeniors- vetenskapsakademiens, Handiingar NR151, Stockhom, Sweden, 1939.


Paul E. Beers, t Mark A. Pilcher, 2 and Jeffrey C. Sciaudone 3

Retrofitting Commercial Structures with Laminated Glass to Withstand Hurricane Effects

Reference: Beer, R E., Pilcher, M. A., Sciaudone, J. C., "Retrofitting Commercial Structures with Laminated Glass to Withstand Hurricane Effects," ASTM STP 1434, Use of Glass in Buildings, V. Block, Ed., ASTM International, West Conshohocken, PA, 2002. Abstract: There is an enormous inventory of buildings in the United States. The majority of these buildings have not been designed to resist hurricane effects, namely high winds and windbome debris. Commercial structures have a history of damage during previous hurricanes. Many large buildings in downtown Houston, Texas suffered extensive glass breakage during Hurricane AlMa in 1982. This was the first documented instance of windbome debris causing significant glass breakage in an urban area. Before Alicia, it was thought that high winds alone were responsible for hurricane damage. However, recently there have been several more instances of glass breakage during hurricanes have occurred, including the Kendall area in Dade County, Florida during Hurricane Andrew in 1992.

While mitigation efforts for newly constructed buildings are very important, this activity only benefits a small percentage of the built inventory. The vast majority of the built commercial building inventory is not protected. Without retrofitting, it will take many years before a significant percentage of commercial buildings are protected, even if all newly constructed buildings were designed for hurricane effects. Protecting commercial structures from hurricane elements, namely wind and windbome debris, requires unique considerations. Many buildings have glass areas too large to accommodate shutters. Even if shutters can be employed it may be impractical to

Chief Executive Officer; Glazing Consultants Inc., 5700 Lake Worth Road, Suite 100, Lake Worth, FL 33463. 2Underwriting/Operations Superintendent, State Farm Fire and Casualty Company, 22 State Farm Road, Monroe, LA. Associate Director of Engineering, Institute for Business and Home Safety, 1408 N. Westshore Blvd., Suite 208, Tampa, FL 3361 I.


by ASTMInternational



install them in advance of a storm. Often the best solution for a commercial building is to use impact-resistant laminated glass. With laminated glass, there is no special preparation required in advance of a storm as the protection is always in place. And, it provides invisible protection because it appears as ordinary window glass. The use of laminated glass for hurricane protection requires special designs. Laminated glass is often thicker than ordinary window glass and it must be attached or anchored to the window frame so it remains in place, if broken. The challenge for retrofitting commercial windows with laminated glass is to find a cost-effective method that does not require replacement of the framing as well. This paper will present methodology for a cost-effective retrofit using laminated glass and the existing window frames. Keywords: cyclic pressure cycles, hurricanes, hurricane protection, impact resistant glass, laminated glass, missile impact test, retrofit, storm shutters.

Communities can survive natural disasters only if their businesses survive. The Insurance Information Institute reports that, on average, 25 percent of businesses that close following a natural disaster never reopenvl. This phenomenon, in turn, can have a dramatic effect on people being able to go back to their jobs and maintain "normal" lifestyles, which diminishes the tax base. The key to keeping these businesses from closing is limiting the damage they experience as the direct result of a storm. However, most commercial building envelopes are not designed to resist hurricane effects, namely high winds and wind-borne debris.

Figure 1 - Houston Skyscraper that Experienced Extensive Glazing Damagefrom Windborne Debris During Hurricane Alicia



When Hurricane Alicia struck Houston, Texas in 1982, many large commercial buildings suffered extensive glass breakage. This was the first documented instance of wind-borne debris causing significant damage in an urban area2. Figure 1 shows some of the damage that occurred during Hurricane Alicia. Before this event, most people assumed that high winds alone were responsible for hurricane damage to commercial buildings. The damage from Alicia and several more recent hurricanes, including Hurricane Andrew in 1992, have shown that wind-borne debris plays a major role as well3. The International Building Code now refers the designer to the procedures of ASCE 798 for the calculation of wind loads on structures 4. Along the hurricane coasts, ASCE 7 requires the designer to either provide protection of glazed openings from wind borne debris or design for internal pressures. While keeping the structure intact is important, it is more important to protect the building envelope to prevent destruction of the interior and contents of businesses. If the envelope stays intact, the business will be able to reopen sooner. While these code developments are very important, they only protect new buildings. In any given year, new buildings typically represent about 2% of the building stock. It will take many years before a significant percentage of commercial buildings are protected, even if all new buildings include impact-resistant features. Retrofit solutions are necessary to reach the massive inventory of existing commercial windows vulnerable to wind-borne debris.

Need for Retrofit Methodology Coastal cities like Houston, Tampa, New Orleans, West Palm Beach, Norfolk and Jacksonville are vulnerable to hurricanes. Yet few of their most important structures hospitals, government buildings, corporate offices and multi-family residences - have wind-borne debris protection.
Storm shutters are a viable option for protecting low-rise buildings, but are not a perfect solution for several reasons. First of all, many commercial buildings have glass areas too large to consider using protective shutters. Second, the addition of shutters may dramatically alter the appeararlce of the building. Third, deploying shutters in the face of an approaching storm may prove difficult, especially if the windows are above the ground floor. Fourth, because they are not frequently opened and closed, hardware can become rusted and the shutter difficult to operate. And fifth, the people who are supposed to ensure that the shutters are properly closed and ready for the storm may be more interested in securing their own homes and evacuating the area. In order to alleviate these concerns, building owners may choose to protect their businesses with impact resistant window and door products. These impact resistant products consist of laminated glass; strong but flexible sealant to hold the glass in place, stronger frames and more secure anchorage back to the structural system of the building. These systems require no preparation in advance of a storm - the protection is always in place. And laminated glass offers invisible protection that does not detract from the



building's appearance. However, new impact resistant window systems.can be prohibitively expensive to install in existing buildings. In response to repeated problems, engineers are beginning to develop retrofit solutions to incorporate laminated glass into existing window frames. The laminated glass used is often thicker than ordinary window glazing, and as such, presents some unique design considerations. The challenge for retrofitting commercial windows with laminated glass is to find a cost-effective method that does not require replacement of the framing as well.

Overview of Methodology Most existing commercial buildings were built following the requirements of earlier building codes. Many do not comply with current building code requirements.
Recent changes in building codes, such as The South Florida Building Code and The International Building Code now require that windows, doors and glass comply with higher wind loads and impacts from windbome debris. Wind pressure is calculated using ASCE 7 Minimum Design Loads for Buildings and Other Structures or by conducting a wind tunnel study where a scaled building model is subjected to a simulated wind field. Missile impact criteria is based upon performance tests where materials intended to provide hurricane protection are tested in a laboratory by subjecting them to missile impacts followed by a series of wind pressure cycles. A typical test sequence includes impacting three specimens twice with a nine pound two by four traveling at fifty feet per second, followed by subjecting the impacted specimens to 9,000 inward and outward (4,500 inward and 4,500 outward) acting pressure cycles6. Outside of recently constructed buildings in south Florida, few commercial buildings provide protection from impact from windborne debris. Thus, most existing buildings require retrofit or renovation to provide meaningful hurricane protection. To retrofit an existing building to provide hurricane protection, the glazing system must be evaiuated for resistance to wind pressure and protection from windborne debris. This includes analysis of the window frame, anchors and the glass product. The first step in the retrofit process is to determine appropriate design criteria. This includes calculating the wind design pressure and selecting missile impact performance criteria. The South Florida Building Code, International Building Code, Southern Building Code Congress SSTD 12-97: Test Standard for Determining Impact Resistance from Wind-Borne Debris and ASTM Standard Test Method for the Performance of Exterior Windows, Glazed Curtain Walls, Doors and Storm Shutters Impacted by Missile(s) and Exposed to Cyclic Pressure Differentials (E 1886) and ASTM Standard Specification for the Performance of Exterior Windows, Glazed Curtain Wails, Doors and Storm Shutters Impacted by Missile(s) and Exposed to Cyclic Pressure Differentials (E 1996) provide guidance with the selection of appropriate missiles, impact speed and cyclic pressure loading.



Some critical facilities such as hospitals, police and fire rescue stations, corporate data centers and buildings to be used as shelters require design criteria that exceed minimum building code requirements. ASTM E 1886 and ASTM E 1996 provide design criteria for critical facilities. Enhanced design criteria can also be developed through the use of a site-specific wind risk analysis by modeling the historic wind records for the site and selecting an enhanced missile based upon the debris threat at the site and precedent for similar projects. Once the design wind pressure has been calculated, the existing window frames and anchors must be analyzed. This can be done using theoretical calculations of the existing frames and anchors to determine their strength as compared to the design wind pressure. Normally the existing frames will not meet the current project design wind pressure requirements because they were built under an earlier version of the building code. Additional structural bracing for the window frames can be designed to strengthen the framing to the current project requirements. This often includes the application of structural steel elements to the interior or exterior of existing vertical framing members. The new steel usually must be anchored to the structure at the top and bottom. The steel and anchors can be painted to match the existing frame color or clad with matching aluminum sheet break metal. Existing anchors also are typically not compliant with the new project design wind pressure requirements. This is often due to inadequate sizing and spacing or corrosion. A new anchor schedule to comply with the current project requirements can be developed through calculations. A conservative anchor design would be to abandon the existing anchors and design new anchors without relying upon or using the benefit of the existing ones. Most building code jurisdictions require submittal of a full set of calculations illustrating the additional structural elements necessary to comply with the project design wind pressure. These calculations must be signed and sealed by a Registered Professional Engineer. Existing buildings predominately utilize annealed, heat-strengthened or tempered glass in monolithic or insulated configurations. These materials do not pass the minimum missile impact tests prescribed by building codes and standards 7. Therefore, it is necessary to replace the existing glass with a material that is capable of passing the project's missile impact test criteria. The new glass is typically a multi-layer laminated glass unit and is thicker than the existing glass it is replacing. The challenge is how to provide enough space in the existing frame to accommodate thicker glass and a silicone anchor bead to affix the glass to the frame. This is accomplished by utilizing an aluminum frame adaptor that fits into the space where the existing glass was and provides a wider glass pocket for the new glass and silicone. The adaptor decreases the daylight area of the glass by about 3 mm all the way around, but is not noticeable if painted the same color as the existing frame



material, Figure 2 shows examples of impact resistant glazing systems with and without the adapter.

Figure 2 - Retrofit GlazingSystem Building codes require performance testing of the new glass adaptor in the existing frame to show compliance with code and project requirements. Sometimes it is not possibte to get an exact replica of the existing framing, if the system is now obsolete. In these instances, framing that replicates the glass pocket condition should be used for the laboratory test. The combination of engineering calculations and laboratory testing of the retrofit assembly provides a system that can be accepted by building code authorities. More importantly, with proper design criteria, these systems will provide reliable performance during hurricanes at a lower cost than new windows and some shutter systems. Buildings that have been retrofitted with laminated glass provide additional benefits to the owner and occupants. The laminated glass system is always in place and requires no advanced preparation in advance of a storm. The retrofit can be designed to have little visual changes from ordinary windows and is not noticeable to most building occupants. Laminated glass also provides additional benefits beyond hurricane protection to include enhanced security from intruders, reduction of ultra violet rays, better acoustical performance and it is safer than ordinary window glass because it remains in the frame when broken.
Case Studies

Tarpon Springs Chamberof Commerce; TarponSprings, Florida



The Institute for Business and Home Safety and Tampa Bay Regional Planning Council selected this building as a demonstration project for commercial glazing retrofit in Florida. Tarpon Springs Chamber of Commerce is an important organization within the local community and it is important that they are ready to support local business during the aRermath of a storm.

Figure 3 - Tarpon Springs Chamber of Commerce The building contained glazing along the street front and in an upper clerestory area in the rear of the office reception area. It is in the downtown section of Tarpon Springs that is considered to have high windborne debris potential. The building will not be used as a shelter during a storm. The requirements of ASTM E1886 and ASTM E1996 were used as the design criteria. Evaluation by a Florida Registered Professional Engineer determined that the window frames and anchors did meet current wind load requirements and no additional bracing or anchors were required. A retrofit laminated glass panel similar to a tested system was designed and installed. The cost of the project was deemed to be similar to the cost of accordion or roll down storm shutters.

Luxury Hotel," St. Thomas, United States Virgin lslands

This project was renovated following heavy damage from Hurricane Marilyn in 1995. The project consisted of multiple buildings serving functions including reception, meeting space, restaurants and guest rooms. The selected design criteria was the wind pressure requirements for Miami, Florida using ASCE 7, which at the time were more stringent than the loads for St. Thomas and the missile impact criteria from The South Florida Building Code. The common area buildings contained custom shop built wood framed windows and doors. They were heavily damaged during the hurricane and many exhibited wood rot. They were not suitable for retrofit and were replaced with South Florida Building Code approved assemblies.



Figure 4 - Luxury Hotel, St. Thomas, United States Virgin Islands The guest rooms contained aluminum sliding glass doors, which fared better in the storm. A retrofit design was developed to add aluminum clad external steel bracing and re-anchor the frames. A new aluminum framed laminated glass panel was installed to replace the existing tempered glass that was not shattered during the storm. The cost for the retrofit was less than several cost proposals for accordion shutters. Additionally, the hotel operators did not want shutters because the appearance and the logistics of deploying them as a storm approached.

Owen Roberts Airport," Grand Cayman, Cayman Islands, British West lndies
The Cayman Islands Government decided to retrofit the airport to serve as a hurricane shelter in the event a hurricane struck the island and all visitors could not be evacuated. Additionally the airport is a critieai part of the country's infrastructure as it is the point of entry and departure for residents and visitors to the island. The airport facility must remain operational immediately after a hurricane. A wind risk analysis was commissioned for shelter design criteria in the Cayman Islands. The study recommended enhanced wind speeds be used to calculate the wind pressure design loads using ASCE 7 and an enhanced large missile. The specified large missile is a 15 pound 2 x 4 timber traveling at 65 feet per second. The terminal building was built in the 1970s and has had alterations and additions since. Most of the glazing systems are similar and there are many operable doors of various types. Engineering calculations determined the need for additional structural bracing and anchors. Steel reinforcement tubes clad in tubular aluminum extrusions were installed behind each vertical framing member and anchored at top and bottom to the structure. Additional anchors were installed as required.



Figure 5 - Large Missile Testing

Retrofit glazing panels for the glass and doors were designed and installed. Prior to final approval of the design, the panels were tested using various laminated glass types to determine which would perform satisfactorily. A laminated glass with a polycarbonate core was selected. There were some areas at the airport, such as a large breezeway, that were not suitable for the glazing retrofit. Accordion storm shutters were installed at this location. Upon completion, a majority of the terminal will contain laminated glass systems that are designed and tested to the enhanced Cayman Islands shelter criteria in combination with a few storm shutters. Conclusions Windows in existing buildings in hurricane-prone areas can be cost-effectivelyretrofit with laminated glass to provide protection from windbome debris. The case studies presented show that commercial buildings can be cost effectively retrofitted with laminated glass using the retrofit method described in this paper. In addition to saving money, there is much less disruption to the facility than when replacing entire window and door systems. This work can usually be performed while the building is occupied and operational and does not damage interior or exterior finishes. Building owners need to be educated about this option. Most owners are not aware of the retrofit option. Because shutter application is not practical at many commercial buildings, retrofit is the only way a significant portion of these buildings will be protected from hurricane damage. As demonstrated in the case studies, the methods presented in this paper have proven to be cost effective and practical. [1] IRobertHartwig,ChiefEconomist,InsuranceInformationInstitute,New York,NY, Personal Communication,July2000.



[2] 2Minor, J.E. (1985), "Window Glass Performance and Hurricane Effects," Proceedings, Hurricane Alicia: One Year Later (Galveston, TX, August 16-17, 1984), ASCE, New York, pp 151-164. [3] 3Sparks, P.R., Schiff, S., Reinhold, T., "Wind Damage to Envelopes of Houses and Consequent Insurance Losses", Proceedings of Wind, Rain, and the Building Envelope Invitational Seminar, University of Western Ontario, 1994. [4] 41nternationalCode Council, International Building Code, Falls Church, VA, ICC, 2000. [5] 5ASCE 7-98, "Minimum Design Loads for Buildings and Other Sl~'uctures,"American Society of Civil Engineers, Washington, DC, 1998, Section [6] 6Institute for Business & Home Safety, "Natural Hazard Mitigation Insights, Industry Perspective, Impact Resistance Standards," IBHS, Boston, MA, 1999. [7] ~Institute for Business & Home Safety, "Natural Hazard Mitigation Insights, Industry Perspective, Impact Resistance Standards," IBHS, Boston, MA, 1999.

Bruce S. Kaskel 1, John E. Pearson 2, Mark K. Schrnidt 3, and Roger E. Pelletier4

Testing of Annealed Glass with Anchored-Film Glass Retention Systems for Fallout Protection After Thermal Stress Cracking

Reference: Kaskel, B.S., Pearson, J.E., Schmidt, M.K., and Pelletier, R.E., "Testing of Annealed Glass with Anchored-Film Glass Retention Systems for Fallout Protection After Thermal Stress Cracking," Use of Glass in Buildings, ASTM STP 1434. V. Block, Ed., ASTM International, West Conshohocken, PA, 2002. Abstract

Recently adopted ASTM Standard Test E 1886-97, "Standard Test Method for Performance of Exterior Windows, Curtain Walls, Doors, and Storm Shutters Impacted by Missile(s) and Exposed to Cyclic Pressure Differentials" and its accompanying specification, E 1996-99, "Standard Specification for Performance of Exterior Windows, Glazed Curtain Walls, Doors, and Storm Shutters Impacted by Windbome Debris in Hurricanes" are used to determine the the performance of exterior fenestration elements when subjected to hurricane-like conditions. The intent of these standards is to minimize property or personal loss due to the breaching of fenestration elements in a hurricane. Protective systems for both new and existing buildings have been developed in response to these new standards. One possible application for existing building glass is an anchored-film system. This system relies on a polyester film adhered to the existing glass. The film is anchored around the glass perimeter to the window frame. An existing building with a history of thermal stress glass breakage required a system to safeguard against the possibility of glass fallout. Testing was necessary to determine the effectiveness of an anchored-film system for this application. The aforementioned ASTM standards were consulted; however, the cause of cracking, the nature of anticipated wind loads, and the pass/fail criteria were significantly different from those presemed in these standards. Test criteria were developed by the authors to simulate the cracked glass, to apply appropriate loads and to measure the degree of fallout protection. This paper summarizes the development, implementation, and results of the test program, thereby demonstrating the feasibility and effectiveness of anchored-film systems for this unique application. J Senior Consultant, Chicago Branch, Wiss, Janney, Elstner Associates, Inc., 120 North LaSalle Street, Suite 2000, Chicago, IL 60602 2 Senior Engineer, Structures I Department, Wiss, Janney, Elstner Associates, Inc., 330 Pfingsten Road, Northbrook, IL 60062. 3 Consultant, Structures I Department, Wiss, Janney, Elstner Associates, Inc., 330 Pfingsten Road, Northbrook, IL 60062. 4 Senior Instrumentation Specialist, Wiss, Janney, Elstner Associates, Inc., 330 Pfingsten Road, Northbrook, IL 60062.


by ASTMInternational



Keywords Anchored-film, fenestration, annealed glass, fallout protection, temperature differential, load test, glass retention

The windows of a Midwest high-rise building consist of gray-tinted annealed glass set into aluminum frames that are recessed from the face of the building's steel cladding. Because of the recessed configuration, shadows are cast across these windows, causing large temperature differentials between the center and edges of the glass. Figure 1 shows window temperature differences measured at the center and edges during a typical sunny day during the spring. As can be seen in Figure 1, temperature differentials as high as 25~ (45~ were measured. Temperature differentials of this magnitude can be large enough to produce edge thermal stresses greater than 2000 psi. This may exceed the edge strength of annealed glass in some cases. Window glass has cracked on occasion and a system to safeguard against the possibility of fall-out of the cracked glass was investigated and tested. For this study, various anchored-film glass retention systems were tested. The windows tested consisted of annealed glass measuring approximately 203 cm (80-in.) high by 292 cm (115 in.) wide and glazed in aluminum window frames. The glass thickness was 10 mm (3/8-in.) and 12 mm (V2-in.), which is the same thickness as the glass in the building.

"~dowT~ Data Apdl- 33d Roar- ~xeh 50 40 ~_30


-20 -30


8-.~.00 g-~ir-00 io.~r-00 n-~r-00 t~Vr-00 134g.oo 144or-oo

Figure 1 - Temperature difference between center and edge of window



Industry Testing of Cracked Glass

Annealed glass of the size and thickness used on the subject building will typically remain in the window frame after a thermal stress crack. One condition that could cause glass to fallout involves the application of an external force to the cracked glass. This external force can be due to normal pressure differences between the interior and exterior of the building by the air distribution system, vibrations, and wind pressures. Wind pressure is the most critical, since it potentially produces higher magnitudes and more cycles of load application than the other forces. Load tests to simulate the critical cyclical wind pressures were performed on the test specimens after the glass was intentionally cracked. ASTM has recently adopted standard test E 1886-97, "Standard Test Method for Performance of Exterior Windows, Curtain Wails, Doors, and Storm Shutters Impacted by Missile(s) and Exposed to Cyclic Pressure Differentials" to specify a protocol for testing glass that has been cracked. In the case of E 1886, the crack is induced by projecting either a 10 g. or a 40 g. missile at high velocity at the glass specimen. Following this impact, the glass is tested through repetitive cyclic static loading, intended to replicate the type of forces that can occur during a hurricane event. Another ASTM standard, E 1996-99, "Standard Specification for Performance of Exterior Windows, Glazed Curtain Walls, Doors, and Storm Shutters Impacted by Windbome Debris in Hurricanes" provides pass/fail criteria for the performance of glass subjected to the E 1886 test method. The intent of these ASTM standards is to minimize property or personal loss due to the breaching of fenestration elements that can commonly occur in a hurricane event. These standards consider the wind-borne debris that could impact the fenestration during the storm and the cyclic nature of the hurricane wind forces. The passing criteria is intended to ensure that the fenestration will maintain the building enclosure. Protective systems for both new and existing buildings have been developed in response to these new standards. One possible application for existing building glass is an anchored-film system. This system relies on a polyester film adhered to the existing glass. The film is anchored around the glass perimeter to sound construction, such as the window frame. The use of anchored-film at the subject building was intended to safeguard against the possibility of fall-out of cracked glass, The aforementioned ASTM standards do not address the subject conditions related to the cause of cracking, the nature of anticipated wind loads, nor an appropriate pass/fail criterion. Test criteria were therefore developed to simulate the cracked glass, to apply appropriate loads and to measure the degree of fall-out protection.

Test Specimens and Load Test

The load test for the subject building consisted of: 1. Constructing test specimen windows that replicate the existing windows. 2. Applying an anchored-film glass retention system to these test specimen windows and allowing 21 days to cure.



3. Intentionally cracking the test specimen glass to simulate the type of initial thermal crack observed on the building. 4. Subjecting the test specimens to cyclical loads representative of code not clear at this point but later wind loads for the building. 5. Measuring the degree to which the anchored-film system prevents the cracked glass from falling out of the frame.

Test Specimens
Seventeen specimens were constructed utilizing monolithic annealed glass sheets set into an aluminum window frames. Twelve specimens had 10 mm (Vs-in.) thick glass and five had 12 mm (89 thick glass. The type of window frame used for the test was a readily available off-the-shelf system selected to resemble the glazing conditions of the subject building windows. A section of the test window is shown in Figure 2.

203 CM By 292 CM By 10 MM or 12 MM
MOaoU~c A t o n e d

Glazing SealantOr

Glass (1/2 In. Shown)

~ie~ f ' .

Glazing Gasket

T~icad "InFt~'ramoSide

Figure 2 - Section through test specimen at jamb before the application

of the anchored-film glass retention system (head and sill similar) Glass Retention System Application
The anchored-film glass retention system consisted of a clear polyester composite film from 0.15 mm (0.006 in.) to 0.20 mm (0.008 in.) in thickness, adhered to the inside of the glass and anchored to the window frame. Each test specimen received an application of one of four different systems tested. Experienced and licensed installers approved by the glass retention system manufacturers performed all applications. Three of the four glass retention systems were "mechanically anchored." For these systems, the polyester film was secured at the perimeter with an aluminum bar that was screwed to the



aluminum window frame (Figure 3). The fourth glass retention system was "adhesively anchored." In this case, structural silicone sealant was installed around the window perimeter to bond the polyester film to the window (Figure 4). In all cases, the film was allowed to cure for at least 21 days before load testing. Due to the limits of the maximum width of a roll of polyester film, all four glass retention systems contained a center vertical butt joint on each test specimen. DashedLine Indicates Polyec, Film et

~ -





Figure 3 - Section through mechanically anchored glass retention system DashedLineIndicates PolyesterFilm AdhesiveSealAnt Anchorage





Figure 4 - Section through adhesively anchored glass restraint system



The mechanically anchored systems were installed with either two-sided or foursided anchorage. The two-sided condition consisted of the anchorage system installed at the top (head) and bottom (sill) of the test specimen. At the sides (jamb) the polyester film was extended to the sight line but was not anchored to the window frame. The foursided condition had the anchorage system installed along all four sides of the window frame. The adhesively anchored system was always anchored on all four sides.

Test Protocol
Cermak, Perterka, Petersen, Inc., a wind engineering consulting finn, determined the magnitude and number of cycles of the test pressures to be apply to each specimen. Their analysis to determine the magnitude took into account the temporary nature of a cracked, filmed window condition [1]. Determination of the cyclical application of the load was developed from a review of current cyclical tests that demonstrate the ability of a building panel to withstand fluctuating wind loads [2-3]. As previously discussed, these standards are developed for hurricane-prone areas and with design loads greater then proposed for this test protocol. A reasonable test profile was established as a "scaled-down" application of these standards. From these analyses, three test protocols were developed for this project as tabulated below. Table 1 - Negative Pressure Test Protocol Load Range 0.2 P - 0.5 P 0.0P-0.6P 0.5 P - 0 . 8 P 0.3 P - 1.0 P 0.5 P - 0.8 P 0.0P-0.6P 0.2 P - 0.5 P Total Cycles Number of Cycles 575 9 180 17 180 9 575 1545

Pressure, P = -1.05 kPa (-22 lb/flz) for windows near the comer or edges of the building. P = -0.58 kPa (-12 lb/fl2) for the middle portion or field of the building. All testing using the negative pressure loads was performed using P = -1.05 kPa.

KASKEL El" At.. ON FALLOUTPROTECTION Table 2 - Positive Pressure Test Protocol LoadRan~ 0.2P-0.5P 0.0P-0.6P 0.5P-0.8P 0.3P-1.0P 0.5P-0.8P 0.0P-0.6P 0.2P-0.5P Total Cycles Numb~Cyeles 6~ 53 103 34 103 53 6~ 1546 P = +0.58 kPa Table 3 - Combination of Positive and Negative Test Protocol LoadRange 0.0P--0.25P 0.0P-+0.25P +0.25P--0.5P ~.25P-+0.5P Total Cycles Numb~ofC~ 1~ 1~ 1~ 1~ 1545


P = -1.05 kPa and +0.58 kPa The duration of one cycle for any of the test protocols was approximately 15 seconds. Each test took approximately 6 hours to perform. At the completion of a test, a static proof load of 150% of P was applied for I0 seconds. An overload pressure of 250% of P was then applied after the proof load for I0 seconds.

Test Equipment
A steel test frame was constructed to support each window in a similar fashion as in the building. Loading was achieved by a fan blower connected to the test chamber. The fan evacuated air from or forced air into the test chamber to create a press~e difference. A reinforced plexiglass panel was bolted to the back portion of the chamber. An opening was cut into a section of the plexiglass panel. A motor controlled sliding door was used to allow or restrict airflow through the opening in the reinforced panel (Figure 5). The interior test chamber pressure varied as the airflow through the opening changed. A programmable controller was used to monitor the pressure and communicate with the slide door motor to increase (close slide door) or decrease (open slide door) the interior pressure. Two electronic pressure gages, one for the low and one for the high pressures, were used to measure the internal pressure. During each test, an electronic data acquisition system recorded the test pressures and center of glass deflection (Figure 6).



Figure 5 -

Sliding door to adjust the test chamber internal pressure

Figure 6 -

Computer controlled data acquisition system



Test Procedure After a cure period of at least 21 days from the date of installation of the anchored-film glass retention system, each test specimen was placed into the steel-framed test chamber with the filmed glass surface (interior side in the actual building) facing the exterior o f the test chamber for visual examination (Figure 7). The test chamber had a plexiglass back panel to allow for visual inspection (exterior side in the actual building) of the specimen during testing. The test specimen was seated into the chamber creating a pressure seal between the test specimen and the test chamber. The specimen frame was clamped to resist displacement under application of the test load.

Figure 7 - Positioning test sample into test chamber



Once the test specimen was situated in the chamber, a crack was created in the glass with a glass cutter and impact tool. The pattern and length of the crack was patterned after the initial thermal crack typically observed at the building (Figure 8). As test loads of increasing magnitude were applied to the specimen, additional cracks would develop. A test specimen during cyclical load application is shown in Figure 9 and a sample pattern of cracking at the end of a test is shown in Figure 10).
Id INDICIIA ~ - N T E I q l . INE

S E A M IN ~ L ~


Figure 8 - Test specimen indicating initial crack pattern

Figure 9 - Test specimen during testing





Figure 1 0 - One test specimen indicating final crack pattern

During the test, the specimen was closely observed. Repeated load cycling caused some grinding of the glass at the crack interfaces, resulting in small glass shavings collecting at the bottom of the test chamber. At the end of the entire test, the total amount of glass shavings at the bottom of the test chamber was weighed and documented. Evidence of large glass fragments not retained by the anchored-glass film glass retention system were considered failure of the test. The application of the test loads to the test specimen was conducted by air pressure differences in general conformance with American Society for Testing and Materials (ASTM) test method, E 1233-97 "Standard Test Method for Structural Performance of Exterior Windows, Curtain Walls, and Doors by Cyclic Static Air Pressure Differential."



Test Results Test results for each sample are presented in (Table 4). These tests demonstrate the following regarding the anchored-film glass retention systems: 1. Twelve of the completed tests were performed on glass specimens with an anchored-film glass retention system that was mechanically anchored to the window frame. Four of the completed tests were performed on adhesively anchored systems. Eight of the twelve completed mechanical anchorage tests had two-sided anchorage and the other four had four-sided anchorage. The first of these tests was stopped shortly after beginning to check test equipment. 2. Thirteen tests were performed with the specified negative pressure load for comer conditions (P = - 1.05 kPa). Three tests were performed with the specified positive pressure load (P = +12 psf). After two of these three positive pressure load tests were completed, the same test specimens were additionally subjected to the positive/negative pressure test. Most tests were conducted with the negative pressure load, since this was the more severe loading. 3. Eleven mechanically and adhesively anchored systems passed the cyclic load and proof test requirements that were established before beginning the tests. Two two-sided mechanical anchorage (Test No. 6 and 7) and one four-sided mechanical anchorage (Test No. 16) did not pass during the cyclic load test. Two additional two-sided mechanical anchorage tests, after passing the cyclic load test, failed the proof load test at 120% (Test No. 11) and 130% of the test load (Test No. 12). All adhesive anchorage tests passed both the cyclic load test and the proof test. 4. Two test specimens had a 10-cm. vertical strip of the film removed from the window at the center of the test specimen. The intent of this modification was to simulate the condition in the building where the film would not be applied continuously across the glass due to obstructions from existing partition walls, which typically abut the glass at mid-width of the window. Both of these tests passed the cyclical load test, although one of the tests failed the proof test. 5. At the completion of all testing, the accumulated glass shavings and fallout pieces were weighed. Between 100 g and 1 kg of glass were weighed for each test, except for test No. 14 where a piece weighing 1.75 kg fell out only during the 250% load test. 6. Deflections increased as loads increased. This relationship was not linear and was influenced by the number of cracks in the test specimen. At overload test pressures, mid-point deflections up tol 5 cm (6-in.) were recorded.



Table 4 - Load Test Results Test No. 1 2 3 4 5 6 Glass Thickness 10 mm 10 mm 10 mm 10 mm 10 rnm 12 mm Extent o f Pressure Cyclic 150% Proof Anehoralge Direction Load Load 2 SidedtL) Test stopped to check equipment 4 Sided(2) Negative Passed Passed 2 SidedO) Negative Passed Passed 4 Sided (2) Negative Passed Passed 2 Sided(l) Negative Passed Passed 2 Sided(l) Negative Fall-out at --cycle 655



2 SidedO)


Fall-out at cycle 843

Fallout piece weighed 480 g. Fallout piece weighed 40 g.

8 9 10

12 mm t0 mm I0 mm

2 Sided0) 2 Sided0) 2 SidedCt)

Negative Negative Negative

Passed Passed Passed

Passed Passed Fall-out at 120%

I 1"

I 0 mm

2 SidedO)



Fall-out at 130%

Fallout piece weighed 170 g. Fallout piece weighed 90 g.

12" 13 14 15 16

10 mm 12 mm 10 rnm 10 mm 12 mm

4 4 4 4

Sided(z) Sided(l) Sided(1) Sided~

4 Sided 0)

Negative Negative Positive Positive Pos./Neg. Negative

Passed Passed Passed Passed Fall-out at cycle 1379

Passed Passed Passed Passed --Fallout piece weighed 370 g.


12 mm

4 Sided(2)

Positive Pos./Neg.



* 10 cm. wide film strip removed at the center to simulate partition wall against glass. (~)Mechanically anchored (Z)Adhesively anchored




Laboratory testing of anchored-film glass retention systems was performed to determine the degree to which such a system could retain cracked glass under simulated wind loading conditions that were determined from a probability analysis considering the building's location and duration of exposure. The testing was conducted using annealed glass of the same size and thickness as actually used on the subject building. The test protocol used concepts developed in ASTM E 1886 and E 1996. However, the new test protocol testing established different criteria with regard to the induced glass crack, the number and magnitude of applied static cycles, and the interpretation of the test results. These criteria were modified to more accurately test the performance of cracked glass with an anchored-film glass retention system at the subject building. The tests determined that an anchored-film glass retention system can provide an adequate level of glass fallout protection for a window that incurs a thermal crack and is exposed for one to eight days prior to being replaced.

[1] Boggs, D. W. and Peterka, J.A. "Winds Speeds for Design of Temporary Structures," Structures Congress 1992 Compact Papers, ASCE, 1992. [2] American Society for Testing and Materials, "Standard Test Method for Performance of Exterior Windows, Curtain Wall, Doors and Storm Shutters Impacted by Missile(s) and Exposed to Cyclic Pressure Differentials," ASTM E 1886-97. [3] American Society for Testing and Materials, "Standard Specification for Performance of Exterior Windows, Glazed Curtain Wall, Doors and Storm Shutters Impacted by Windborne Debris in Hurricanes," ASTM E 1996-99.


Michael Betten ~and Henri Berube

The Advantages of Glazing in an Overall Security Strategy

Reference: Betten, M., Berube, H., "The Advantages in an Overall Security

Strategy," The Use of Glass in Buildings, ASTM STP 1434, A. B. Smith and C. D. Jones, Eds., ASTM International, West Conshohocken, PA, 2002.

Keywords: glazing, security, windows, burglary The prevailing public sentiment is that glazing materials in a business or home are extremely vulnerable to penetration. One homebullder recounted his personal feelings on the topic of residential security by saying, "if you want a fortress, don't put any windows in it." A seemingly simple solution? However, as individuals who routinely consult with citizens on security issues, we must weigh actual threats while addressing the fear of clients. The challenge is to balance the heightened need for security against the fortress mentality. The concerns people have over glazing materials in their homes or businesses often result in them taking actions that result in fire hazards, a sacrifice in aesthetics, and a statement about the community. Burglar bars and double cylinder deadbolts are becoming more common in residential applications creating fire hazards and business owners frustrated with their inability to protect their assets from crime are resorting to unsightly, ineffective alternatives. Research, criminal interviews and crime prevention practitioners suggest that if you want to increase security, strategic use of glazing materials may be the answer. Our role as crime prevention/security practitioners must focus on educating the public on effective security products while at the same time maintaining suitable aesthetics, egress capabilities, and enhancing the overall quality of life. The goal should be to increase security and not trade to protect against one threat at the expense of other life safety systems and strategies. The perceived vulnerabilities of glazing are self-evident, but what evidence is there to support the fact that glazing materials can actually enhance an overall security strategy? Many theories that try to explain criminal behavior make mention of or reference an absence of witnesses or suitable guardians as providing the opportunity for crime to flourish and escalate the likelihood of criminal activity. The general concept being that if criminals perceive that they may be observed, the risk of detection / apprehension becomes unacceptable to the malfeasant. This is most evident in a crime
Overland Park Kansas Police Department, 12400 Foster, OverlandPark, KS 66213.


by ASTM International



prevention philosophy known as Crime Prevention Through Environmental Design (CPTED). The three main components of CPTED are: natural surveillance, access control, and territorial reinforcement (Crowe 1991:). Key to the performance of these criteria is the concept of natural surveillance. "When CPTED is applied to surveillance, prominence is given to no obstructions, low landscaping, expansive windows, and raised entrances that encourage observation by the entire population, creating a palpable crime deterrent and a sense of proprietary ownership in the residents... " (McKay 1997) CPTED practitioners rely heavily on natural surveillance because without it, traditional unattractive target hardening approaches such as bars, barbwire, and concrete barriers could be possible alternatives. These measures, if not applied in conjunction with quality of life considerations can signal the beginning of urban decay and fear, a problem that can feed on itself. Traditional target hardening measures send a definite message about an area or neighborhood and can affect the overall quality of life. Interviews with burglars further support the claims made by CPTED practitioners. One significant study on burglary found: "The location and type of windows both at the target site and at neighbors" houses were considered critical by almost all informants." One informant stated: "Notice how that picture window looks out onto the street. The curtains stay open all the time and both the houses across the street can see straight into the living room. I wouldn't do this place." (Cromwell, Olson, Avary 1991:35) Other research further supports this assertion. (Decker, Wright 1994:110) Burglars perceive windows as a vulnerability of detection where possible witnesses can observe their activity and relay that information to the proper authorities. Other research seems to contradict the assertion that surveillance prevents burglaries. Wright and Decker (1994:1244/1250) found that burglars are not afraid to break glass. "Many of the offenders turned to the second method; breaking the window." Although burglars perceive windows to be a risk, they are not opposed to breaking glass to gain entry. The National Burglar and Fire Alarm Association indicates burglars gain entry into homes 23% of time through a first floor window. No specific information was mentioned regarding attacks on glazing. Depending on what part of the country is referenced, glass breakage can be unusually high. For example, on a survey offered to many law enforcement agencies, Plano, TX listed a significantly high number of residential burglaries where glass was broken to gain entry. 2 The police department suggests glass breakage is a problem since most of the year, resident's windows are closed due to climate conditions. Therefore, the risk of detection is relatively low since the noise of breaking glass is mitigated by many closed windows. Glass breakage is more likely when the potential entry point is isolated from witnesses or neighbors. For example, residences that back up to woods, parks and isolated or unincorporated areas are at a more significant risk to glass breakage. But the question has to be asked, why the contradiction? 2Griffin,Gary,Detectivewiththe Piano,TX PoliceDepartment(2002).



How then do we explain the seemingly contradictory evidence of windows being both a deterrent and a vulnerable point of entry? The answer probably resides in the context of the overall situation target selection. If occupancy (or lack of) can be easily established through assessment of occupancy clues, the windows of the intended target are vulnerable to attack while the windows of overlooking properties and other capable guardianship cues would act as a deterrent. The assessment of these variables along with other factors not relating to windows will invariably result in a 'desirability assessment' on the part of a burglar. This assessment would be established on a case-by-case basis. By using appropriate security glazing materials, the advantages of windows can by maintained (or even enhanced) while the disadvantages (ease of penetration) can be minimized. Over time, there have been relatively few incidents where burglars have encountered security-glazing materials during the commission of their crimes. At most of those crime scenes (anecdotally), it was quite obvious the criminal wasn't prepared for the additional work required to penetrate such security glazing materials. If the glazing material was able to withstand several impacts, it was typically enough to deter most criminals from completing the crime. Burglars do not like multiple impacts on glass fearing that the second and third blow will attract to much attention to their actions (Wright and Decker, 1994:124) Unfortunately, the use of security glazing materials is not

This brings us back to the prevailing sentiment: Can overall security strategies be enhanced with the use of appropriate glazing materials? Before answering this question, let's examine some of the other security alternatives often used by both homeowners and businesses. The Uniformed Crime Reports (UCR) in 2000 reported 2,049,946 burglaries nation wide, with nearly two-thirds being residential in nature. The average reported dollar loss per residential burglary was $1,381 while a non-residential burglary was $1,605. Adding to the effect of a residential burglary is the fact that this is where people eat, sleep, keep valuables and raise their children. "Home" is where people generally feel "safe." The psychological harm of a burglary to the occupants is severe and difficult to measure. Commercially other tangibles must be considered when a burglary occurs. Repairs to the facility, down time, insurance premiums, and perceptions by customers, just to name a few. A common concern is, how much will the glazing improvements cost compared to their worth? Many homeowners and businesses perceive security glazing materials as a questionable investment. Will the additional expense pay dividends in the end and can security-glazing materials compliment overall security strategies? Alarms systems, another popular burglary prevention measure for both businesses and homeowners, have a few vulnerabilities, one being false activation. Nation wide 9598% of all alarms are false (IACP: 2002). To combat the false alarm problem, some law enforcement agencies have stopped responding to alarms until the activation has been verified by private security services, an additional expense for alarm owners. Furthermore, most municipalities have enacted false alarm ordinances that fine alarm users for every false activation - another alarm expense. According to the UCR, it is not uncommon for an alarm user to accumulate more in false alarm fees over one year than the average dollar loss per burglary (Austin, D.: 2002). The most crucial downfall of an alarm system is, they do not keep intruders out! Most systems are designed to activate



after entry is achieved at which point the alarm is only as good as the response it generates. With a false activation rate in excess of 95%, it is understandable why law enforcement does not respond to them as 'emergency' calls. Further, to be effective, evidence suggests that police need to respond to alarms within three minutes of the activation for apprehension (deterrence) to be probable and that such a response is unlikely (Berube 2001:56, Small Business Administration, 1968:29). Using post alarm delay tactics effectively can increase deterrence while allowing for more reasonable police response times and thus provide a powerful deterrent to burglary (Berube, 2001, 64). With the use of an appropriate security glazing material, the alarm system can be a significant deterrent. For example, a business that invests in laminated glass and incorporates glass breakage detectors into their alarm system has a strategy where all components compliment one another. Once the glass is attacked, the glass breakage sensor is activated, triggering the alarm. With the alarm activated, the burglar is confronted with a situation where they are probably on the exterior looking at glass that is still blocking their entry. The burglar hears the alarm and knows the police are being notified. The increased penetration time makes apprehension more likely forcing the burglar to choose to flee while the going is good or assume additional risk that may not have been considered during the original target selection process. Lighting is another security tool that could benefit from the expanded use of glazing materials. Lighting is primarily designed to enhance the visibility of an area. But, what good is the lighting with nearby buildings or houses without the ability of people to witness events? The purchases of closed circuit television (CCTV) systems have boomed in recent years. Yet, CCTV has its limitations and may not compare to the surveillance capabilities provided by the use of security glazing in homes and businesses. The premise behind Neighborhood Watch programs, organized by law enforcement, is to encourage residents to watch out for one another. How are surveillance opportunities to be maximized if security-glazing materials are not used? Criminals and crime prevention practitioners clearly recognize the need to increase surveillance opportunities to develop a comprehensive security strategy. CPTED practitioners claim that with appropriate CPTED strategies, quality of life can be improved (Crowe 1991:28-29). However, since the fallout of September 1lth, there may be a trend to minimize the use of glazing materials in buildings because of it's perceived vulnerability. But, what impact could this have on local crime trends? The concern of terrorist bombing attacks must be weighed along with the more prevalent crimes within a community. The effective use of glazing strategies includes the consideration of window location of both proprietary and adjoining facility windows, supported by other security strategies (capable guardianship, alarms etc.). This will significantly enhance the overall security of a building (burglary prevention, emergency egress) while avoiding many of the signals of fear and urban decay. Architects, developers, planners and law enforcement need to consider the impact glazing materials have on the environment and it's impact on crime and perceptions of safety and security. Glazing materials have an impact on criminal behavior. This impact must be considered in the development of an overall security strategy. Appropriate use of glazing materials based on threat and risk assessment is called for. The outcome may very well be that the question to ask is not



"can we afford appropriate security glazing materials?" But, "can we afford not to use security glazing materials?" Following the assumption that there will always be crime, it is absolutely essential to take appropriate measures to minimize its effects. And while there are positives to any crime-prevention approach, none are fool-proof, some are obviously better than others and some can actually increase risk (double cylinder deadbolts/bars etc.). Ideally, every home and business would incorporate a CPTED philosophy, utilizing proper lighting techniques, constantly monitored CCTV, activate an alarm system and utilize securityglazing materials. Unfortunately there is a perception that by applying only one of these security measures you can dramatically increase security. The reality is that each one of these measures is part of an integral strategy that needs to be catered to individual sites based on a threat assessment. That said, by working together in each of the related fields we will facilitate the development of these integrated strategies. Window and glazing products are an integral pan of that strategy. Architects, developers, planners and law enforcement need to consider the impact glazing materials have on the environment and it's impact on crime and perceptions of safety and security. The question shouldn't be "can we afford appropriate security glazing materials, but can we afford not to?"


Austin, D., (2001) Overland Park (KS) Police Department False Alarm Statistics. Berube H. (2001) An Examination of Alarm System Deterrence and Rational Choice Theory: The Need to lncrease Risk. MSc Thesis, University of Leicester, Leicester, England. Cromwell, P. F., Olson, J. N. and Avary, W. D. (1991), Breaking and Entering- An Ethnographic Analysis of Burglary. Sage Publications, London, U.K. Crowe T. D. (1991), Crime Prevention Through environmental Design: Applications of Architectural Design and Space Management Concepts. National Crime Prevention Institute, Louisville, KY. Federal Bureau of Investigations, (2000) Uniformed Crime Reports McKay T. (1997), Security Architecture: The Right Designfor Reducing Crime, Security Management Magazine, December. Small Business Administration (1969), Crimes Against Small Business: A report transmitted to the Select Committee on Small Business - United States Senate.



International Association of Chiefs of Police (1993) False Alarm Perspectives, Alexandria, Virginia, USA. Wright R. and Decker S., 1994. Burglars On The Job - Streetlife and Residential Break. ins. Northeastern University Press. Boston, MA.

Jerry Razwick 1

The Relationship Between Sprinkler Systems and Glass

Reference: Razwick, J., "The Relationship Between Sprinkler Systems and Glass,"
Use of Glass in Buildings, ASTM STP 1434, V. Block, Ed., ASTM International, West

Conshohocken, PA, 2002.

Abstract: This paper explores the particular relationship between sprinklers and glass in
a building. There are many misconceptions about sprinklers' ability to prevent non-rated glass from breaking during a fire, when in fact sprinklers can actually cause some glass to break. Laboratory testing has shown that unless the conditions can be strictly controlled, sprinklers and non-rated glass are not an adequate combination to prevent the spread of a fire. Specific testing examples and illustrations are cited in the paper. Sprinklers are also "active" systems, depending on proper maintenance and activation sequences in order to work properly, whereas fire-rated glass is a "passive" system providing compartmentation around the clock.

Keywords: sprinkler systems, deluge, fire-rated, glass, thermal shock

Since the first automobile air bag was introduced in 1973, there have been many advances in the technology for car safety. Today, air bags are standard equipment in nearly all cars being manufactured in the United States, and the track record continues to grow for how effective the devices are in saving lives. Imagine for a moment that auto manufacturers became so enamored with air bags that they decided to abandon other safety features. Would you want to buy a car that had great air bags but a weak frame? Would you believe strongly enough in that one safety device to give up your anti-lock brakes? Probably not. And hopefully, you would not even consider a car that substituted air bags for safety belts - particularly since air bags without safety belts can cause serious injury when inflated during an accident. Such scenarios sound absurd in the context of automobile safety. But in the construction arena, something similar is happening on a regular basis. Sprinkler systems have become so popular as fire safety devices that in many cases architects are abandoning other important fire protection systems. Code officials are allowing "tradeoffs" - approving the use of sprinklers where the code would normally require fire-rated building materials.

1president, Technical Glass Products, 2425 Carillon Pt., Kirkland, WA 98033.



by ASTMInternational



Is this a safe practice? If sprinklers can offer equivalent levels of fire protection, it would be logical to conclude that other fire safety measures might be unnecessary. Given sprinklers' track record of saving lives and property, this concusion seems logical. However, closer examination of this position reveals that it is not as cut and dried as it may appear on the surface. There are many reasons to be skeptical of putting all the eggs in one basket when it comes to something as critical as fire and life safety. There are three basic components to a comprehensive fire protection program: Detection, Suppression and Compartmentation. The first two categories require activation, while the third category is generally passive. For instance, smoke alarms (which provide detection) and sprinklers (which provide suppression) must first be triggered in order to be effective. In contrast, fire walls, doors and ceilings compartmentalize and contain smoke and flames without any activation process. They offer around-the-clock protection in their passive state by acting as a physical barrier to fire and smoke. When active systems become the primary mode of fire protection, there is always the danger of mechanical failure, human error or poor maintenance interfering with the way the systems function. As seen with the recent massive recalls, sprinklers are no exception. The NFPA Journal has cited numerous additional causes that have rendered sprinklers inoperable in real world fires -- valves painted over, systems shut down during construction, fire burning through PVC supply pipe, fire fighters diverting water, etc. [I]. The point is not to negate the value of sprinklers, but instead to recognize their limitations. Relying exclusively on a single method of fire protection may create a false sense of security that is unwarranted. This becomes even more apparent when you examine the relationship between sprinklers and fire-rated building materials. For instance, misunderstandings abound regarding the interaction between sprinkler systems and glass. Deluge systems and non-rated glass are sometimes recommended as a combination instead of fire-rated glass and traditional sprinklers. Yet such configurations can pose a hazard. The problem centers around an issue known as thermal shock. Most glass cannot tolerate drastic variations in temperature on its surface. When one area is hot and another cooler, the glass doesn't know whether to expand or contract, and therefore it typically shatters and falls from the opening. If you have ever seen water sprayed on the glass doors of a fireplace when a fire is going, you may have seen this principle at work. Even without water involved, ordinary window glass breaks at about 250 ~ F, and tempered glass at about 500~ F. In contrast, fire-rated glass is often capable of withstanding temperatures above 1600 ~ F.



*...~'~ 1,500
mllmmm B mm B

1,000 m










Time(in minutes)

Figure 1 - Standard time.temperature curve for furnace testing of glass for fire rating.
1/4" (3 mm) *F *F Max *F Max

Thick Materials

Normal Service 1500

Thermal Shock 1400



Fire-Rated Glass Ceramic


Non-Rated Tempered Float Glass Non-Rated Float Glass







Figure 2 - Comparison of thermal properties for monolithic glass types.

When fire-rated glass is used, sprinklers pose no threat. However, when attempts are made to get by with non-rated glass, the situation becomes tricky. It is quite possible that the sprinklers could actually cause the glass to vacate the opening during a fire, leaving a breach for flames and smoke to spread.



This is a crucial issue. When sprinklers suppress a fire, they can generate large volumes of deadly smoke. If glass windows have shattered, the smoke will be free to escape into other areas of the building [2]. Laboratory tests conducted over the last several years demonstrate the complexity of the sprinkler-and-glass relationship. One such test was conducted in 1995 [3]. Officials at Factory Mutual Research Corp. observed the test to determine if a non-fire-rated window assembly and sprinkler "system" could provide equal protection to that of a firerated assembly. The system (which combined specially designed sprinklers and tempered or heat-strengthened glass) was exposed to fire, with the hope that the glass would be able to stay intact. The researchers discovered that the glass could survive the test if two conditions were met: First, the fire had to start far away from the glazing assembly, in this case approximately 2.44 m (8 ft.). Second, the sprinkler needed to activate very soon after the fire started. When the heat source was brought closer to the non-fire-rated glass, the glass fell out of the frame in less than 5 minutes. Apparently, the close proximity of the flames caused the temperature of the glass to rise too quickly, outpacing the sprinklers' ability to cool the glass surface. Nearly a decade prior to that test, Lawrence Livermore National Laboratory (LLNL) conducted a similar experiment to find out what would happen when a fire started near the surface of non-fire-rated (tempered) glass [4]. They used two different sizes of fire (250 kW and 40 kW), and sprinklers were also installed for the test. When the larger fire was started, the sprinklers activated early and the glass remained intact. But in both tests conducted with the smaller fire, the glass fractured and fell out of the test assemblies in less than 4 minutes - even before the sprinklers activated. Looking at the results of both tests, we can conclude that when fires are relatively large and distant, non-rated, tempered glass may perform as needed. The overall room temperature rises rapidly enough to activate sprinklers before the glass becomes "stressed." However, when a smaller fire is concentrated close to the glass surface, it may not activate sprinklers early enough, generating sufficient stress to shatter the nonrated glass. One commonly proposed method of dealing with this problem is to utilize special deluge sprinklers that will bathe the glass with water before such stress can occur. This was the concept scrutinized in the "Hospital for Sick Children" test conducted at Canada's National Fire Laboratory in Toronto [5]. A propane burner was ignited across the room from the tempered glass, roughly 2.13 m (7 ft) from the glass surface. Deluge sprinklers were installed on the bum side of the glass, and were carefully positioned to ensure uniform water coverage. As in the tests mentioned previously, the non-rated glass was able to survive the test under these conditions. However, the tests left too many questions unanswered to be considered definitive. For instance, officials adjusted the water flow rate from the sprinklers during the test when "dry spots" appeared on the fire-exposed face of the glass. In a real-world fire, no one would be on hand to monitor water flow. Any dry spots on the surface of hot glass cause heat stress and can be a primary cause of glass fracture. Adjusting water flow during the test prolonged the endurance of the glass.



Sprinkler activation time was also critical to the success oftbe test. The sprinkler system used was a "quick response" type that activated two to three times faster than standard sprinklers. These test limitations were noted in an article entitled "Fire Resistant Wall Assemblies with Glazing" in the Society of Fire Protection Engineers Bulletin. "The location and response time of the sprinkler must be such that activation will occur before the glazing reaches critical temperature levels... Should sprinkler activation be delayed so that the temperature of tempered glass is in the range of 250 ~ C (approximately 482 ~ F), glass failure could possibly occur [5]." Underwriters Laboratories (UL) tested one manufacturer's deluge type "system" that combined specially designed sprinklers and non-fire-rated glass. Recognizing the importance of heat source location, UL conducted four tests in which the heat source was placed close to the glass. The glass failed three out of four times. In two of the tests, the sprinklers activated early, but the thermal shock proved too much for the glass, causing it to break. The UL test report states that in all three cases, "...large pieces of glass fell to the floor" after an average of just 4 minutes [6]. The Canadian Construction Materials Center reviewed UL's test results. They concluded that the window sprinkler system would work as long as flammable materials could be kept away from the glass surface. To accomplish this, they recommended construction of a 36-inch-high pony wall, in addition to restrictions on curtains and blinds [7]. This solution has its own problems. First, a pony wall does not prevent flammable objects such as desks, file drawers, coat racks, etc., from being placed near the glass. Second, the ledge created by a pony wall frequently becomes a convenient place to stack flammable books, papers, etc. To prevent this, the National Evaluation Report No. NER516 requires that when using this system, "all combustible materials shall be kept 2 in. (50.Smm) from the front face of the glass [8]." Once a facility is occupied, however, enforcement of such instructions becomes unworkable. Curtains, blinds or other window coverings can also affect the performance of sprinklers. When these materials are placed between the glass and sprinkler, the water is unable to cool the glass, thus causing the glass to fail early on during a fire. Recognizing this fact, one sprinkler manufacturer states in its literature: "Blinds or curtains must not be between sprinkler and glass [9]." With sprinkler heads located up to 30 cm (12 in.) away from the windows, this presents an awkward challenge. Again, once tenant improvements are made at a later date to a building, initial advisories regarding window coverings are likely to-be ignored. Two additional conditions are worth mentioning as well. Since the source of potential fires cannot always be identified ahead of time, the sprinkler and tempered glass systems require deluge sprinklers on both sides 0fthe glass. Also, the commonly seen window designs with intermediate horizontal mullions are not allowed, since the mullions would interfere with the water's ability to evenly bathe the glass surface.

Clearly, sprinklers have improved the standard of fire safety. Yet as the National Fire Protection Association (NFPA) acknowledges in their evaluation of U.S. experience with sprinklers, sprinklers are not adequate in and of themselves - they must be part of a more



comprehensive fire protection program [10]. There are ample glass products on the market today that offer outstanding fire-ratings and the ability to withstand thermal shock. From a life safety standpoint, there is no reason to compromise. Combining sprinklers with fire-rated glazing offers the better solution.

[1] Journal Subject List, NFPA Journal, URL: http://www.nfpa.orl~, NFPA International, Quincy,M A February2002. [2] Mawhinney, J. R. and Tamura, G. T. "Effect of Automatic Sprinkler Protection on Smoke Control Systems," ASHRAE Transactions: Research 3785 (RP-677). [3] Test report J.I. 0Z5Q2.AM (4510), Factory Mutual Research, Norwood, MA, May 1995. [4] Beason, D. G. "Fire Endurance of Sprinklered Glass Walls," Fire Journal, Vol. 80, No. 4, July 1986, pp. 43-45. [5] Richardson, J. K and Boehrner, D. J. "Fire Resistant Wall Assemblies with Glazing," SFPE Bulletin No. 87-3, Society ofF[re Protection Engineers, July 1987, p. 6. [6] File Ex683, Project 94NK27353, Underwriters Laboratories, Northbrook Illinois, August 1995. [7] Canadian Construction Materials Centre Evaluation Report, CCMC 12752-R, June 24, 1996. [8] National Evaluation Report for Central Window SprinklerTM Model WSTM, Report No. NER-516, National Evaluation Service, Falls Church, VA, April 2001. [9] Central Sprinkler, Model W S TM Specific Application Window SprinklersTM Data Sheet No. 6-2.0, Care & Maintenance Figure G, Tyco Fire Products, Lansdale, PA, March 2001. [10] Rohr, K. D., "U.S. Experience with Sprinklers," National Fire Protection Association, Quincy Mass, 2001, p. 50.

H. Scott Norville ~and Edward J. C o u r a t h 2 Design Procedure for Blast-Resistant Laminated Glass

Reference: Norville, H. S. and Conrath, E. J., "Design Procedure for Blast-Resistant Laminated Glass," The Use of Glass in Buildings, STP 1434, V. Block, Ed., ASTM International, West Conshobocken, PA, 2002. Abstract: When an explosive threat exists, building owners should strongly consider using blast-resistant glazing in windows and curtain walls. Architects and engineers have few, if any, publicly available tools, procedures, or formal guidelines to aid in designing blast-resistant glazing. This paper presents a design procedure for blast-resistant laminated glass. The procedure finds its basis in an empirical relationship between air blast pressure, positive phase impulse, and 60-second equivalent design loading for laminated glass and window glass constructions fabricated with laminated glass. The procedure is relatively simple to use. The paper also addresses framing considerations for the blast-resistant glazing. It gives a design example and shows comparisons with experimental results. Keywords: air blast pressure, glazing design, laminated glass, insulating glass, blast resistance The Purpose of Blast-Resistant Glazing Design When explosions occur in populated areas, air blast pressure typically fractures windows, causing catastrophic results. In the worst scenarios, the shards flying and failing from fractured window glass injure and kill persons [1-6], even in the absence of building collapse. At the same time, air blast pressure entering buildings can cause severe damage to ears that can result in diminished hearing ability, loss of balance, and headaches [3]. Relatively small explosions can cause significant window glass breakage, requiting window glass replacement and substantial cleanup. Blast-resistant glazing should minimize and possibly eliminate flying and falling glass shards in any explosion. In addition, under air blast pressure loading, blast-resistant glazing should maintain closure of its fenestration, significantly reducing air blast pressure-related injuries and cleanup costs. Even with blast-resistant glazing, air blast pressure will fracture windows, necessitating replacement. However, blast-resistant z Professor and Director, Glass Research and Testing Laboratory, Department of Civil Engineering, Texas Tech University, Box 41023, Lubbock, TX 79409. 2 Structural Engineer, Protective Design Center, U.S. Army Corps of Engineers, 12565 W. Center Road, Omaha, Nebraska 68144-3869.


by ASTM International



glazing should remain in its openings, thus reducing the urgency for immediate replacement. Explosive threats in today's uncertain world require architects and engineers to produce designs that afford protection from air blast pressure. To accomplish this, they must make a prediction of a potential blast threat to a building. They can define the potential blast threat in terms of (1) an amount of explosive and (2) a standoff distance from a building. They design and install blast-resistant glazing based upon this predicted design threat. Should an explosion larger than the design threat occur, provided it does not cause building collapse, properly designed blast-resistant glazing should (1) minimize flying and falling glass shards and their associated lacerative hazard and (2) maintain closure of most of the glazed openings. The more probable scenario for blast-resistant glazing arises when no explosion occurs during its in-service life. In this case, the blast-resistant glazing must perform the functions of standard glazing, i.e., providing a barrier between environments inside and outside a building while allowing light to enter the building and building occupants to observe the outside world. Blast-resistant glazing must perform these everyday glazing functions economically, without maintenance beyond that which standard glazing requires. In summary, the primary goal of blast-resistant glazing design consists of protecting people inside or near a building subjected to an air blast pressure loading. To afford this protection, blast-resistant glazing must not contribute to the hazards associated with the blast. Blast-resistant glazing accomplishes this by remaining in its frame following fracture and eliminating or greatly reducing the number and sizes of flying and falling glass shards. In the vast majority of its applications, blast-resistant glazing will never experience loading from air blast pressure. Consequently, blast-resistant glazing must economically perform the functions of standard glazing.

Window Glass Design and Blast-Resistant Glazing Design

For design of vertical glazing, window glass must usually resist only wind loading. For design of sloped glazing, window glass must resist loading from snow, its own weight, and wind. Consequently, window glass design consists of determining the appropriate window glass type, construction, and thickness designation(s) to resist uniform loads from wind, snow, and its own weight, as appropriate. Designers assume these loads act in a quasi-static manner. The failure prediction methodology [ 7-9] provides the theoretical model that describes load resistance of window glass for US design procedures. The failure prediction methodology addresses all factors known to affect window glass strength [8]. It relates a uniform load having constant magnitude over specified time duration to a probability of breakage. Under this theory, any uniform loading having finite time duration that acts on an annealed window glass lite induces a non-zero probability of breakage. Within traditional window glass design, any breakage occurring in a window glass lite, i.e., a crack or fracture, constitutes failure. Standard Practice for Determining Load Resistance of Glass in Buildings (E 130000) uses the failure prediction methodology as a basis for design procedures. This document defines the load resistance, i.e., strength, of a window glass lite, in terms of the



magnitude of the uniform loading which acts over a time duration of 60-seconds to produce a nominal probability of breakage of eight lites per thousand, PB = 0.008, at its first occurrence. Currently, US model building codes [10-12] have adopted the E 130000 methodology in some form to facilitate design to resist uniform loading. E 1300-00 presents load resistance for annealed glass, termed nonfactored load, in 12 charts one for each nominal thickness designation. Figure 1 shows a nonfactored load chart for window glass having nominal 6-mm (1/4 in.) thickness, similar to that contained in E 1300-00. Plate Length(in,)
0 140 20 40 60 80 100 120 140 160 180 200

' ' ' ' ' 8 . 0 . (1/4 in.)'. . . mm . . . . Glass' . . . .

' ' ' ' ' ' ....

' ' ' '/"


Nonfactored Load (kPa) ~Pb = 0.008

1 kPa = 2 0 . 9 psf / l J j

' " ' ' ' ' ' " '~' " ' ' ~ /~/"3000 J
,. . . , .


v I



60 I1. 40

i i N.

/" / I

,-L.-.-- ""






I ooo





P l a t e L e n g t h (rnm)

Figure 1-NonfactoredLoad Chart. E 1300-00 uses glass type factors that relate the strength of other monolithic window glass types (heat strengthened and fully tempered) and constructions (insulating glass and laminated glass) to that of monolithic annealed window glass. Current window glass theory does not precisely define the relationship between loading, time duration, and probabilities of breakage for window glass constructions and heat treated monolithic window glass. When designed using E 1300-00, the probability of breakage for heat treated monolithic window glass and window glass constructions with any glass types has nominal values less than or equal to 0.008 at the first occurrence of the design loading. Traditional window glass design methodology assumes that loads act quasi-statically with durations measured in seconds or longer periods. When an explosion occurs, air blast pressure loads window glass lites dynamically over very short time durations. Figure 2 shows the approximate relationship between stress duration and stress magnitude at which fracture occurs for annealed window glass [13]. Figure 2 indicates



that under short duration loading the stress at which fracture is initiated, which somehow correlates with window glass load resistance, increases dramatically.

9 I
== ==
m 10,000 5,000

1 second 1 hour

1 day

1 week

1 month

Duration of Stress
Figure 2 -

ApproximateRelationship Between Magnitude of Fracture Stress and Stress Duration.

On the other hand, the dynamic air blast pressure loading associated with an explosion excites higher mode shapes in a window glass lite causing much larger deflections and stresses than would a quasi-static loading having the same magnitude of pressure. Because of the excitation of higher modes, the stress distribution for a dynamically loaded window glass lite differs significantly from the stress distribution under quasi-static loading of the same magnitude in that stresses having high magnitudes occur over large regions of a window glass lite [14]. In addition to their dynamic nature, air blast pressure loadings tend to have much larger magnitudes than wind and snow loadings that typically govern window glass design. Incorporating these factors, the failure prediction methodology indicates that under air blast loadings, the probability of breakage for typical monolithic window glass lites or window glass constructions approaches 1.0 even for relatively small air blast pressure loading [14]. In short, the distribution and severity of the load-induced tensile stresses in a window glass lite subjected to loading from air blast pressure typically overcome any increase in resistance resulting from the relatively short duration of the loading. Some designers attempt to devise "strong" monolithic window glass lites or window glass constructions to resist a design air blast pressure without fracture. Many factors



tend to make this approach undesirable. First, any window glass construction designed to withstand even an air blast pressure loading of low magnitude would be very thick and would most probably involve heat treated window glass. Such window glass constructions would have prohibitive costs. Furthermore, a window glass construction with sufficient strength to resist without fracture an air blast pressure loading would transfer a large portion of the loading into the structural frame. This load transfer would require a frame design that could resist such loading without collapse. Finally, if the design employs monolithic window glass, regardless of its load resistance, it would have a finite probability of breakage under air blast pressure loads. When monolithic glass fractures under air blast pressure dire consequences ensue. The authors recommend that blast-resistant glazing constructions using glass should fracture under air blast pressure loading. Following fracture, they should rely on post breakage behavior characteristics to eliminate flying and falling glass shards and maintain closure of fenestrations. Because window glass constructions that fracture transfer much less load into the structural frame, the designer should find this approach desirable when considering the effect of air blast pressure on an entire building. The designer must base blast-resistant glazing designs on maintaining closure of fenestrations and eliminating flying and falling glass shards to the greatest extent possible. Blast-resistant glazing that performs these functions will minimize damage to building contents and maximize safety to building inhabitants. For these reasons, laminated glass makes an excellent blast-resistant glazing material. Laminated Glass and Insulating Glass in Blast-Resistant Glazing Laminated glass consists of two or more plies of monolithic glass bonded together using an elastomeric interlayer. Laminators use polyvinyl butyral (PVB) most commonly as the interlayer material in fabricating laminated glass. The glass plies can consist of annealed, heat strengthened, fully tempered, or a combination or window glass types. Also, a single laminated glass lite can consist of glass plies having different thicknesses. Aesthetically, annealed plies make the best, distortion-free laminated glass. Its post breakage behavior characteristics make laminated glass an excellent blastresistant glazing. When laminated glass fractures, the majority of glass shards adhere to the PVB interlayer. Small shards may spall from laminated glass in a blast, especially if the PVB tears. If the blast-resistant glazing must completely hold all shards, no matter how small, the designer should consider using a plastic film over the inner glass ply. Manufacturers provide commercially available laminated glass constructions with a layer ofpoly(ethylene terephthalate) (PET) laminated to the inside glass ply using PVB. PET can be scratched easily although it does not degrade from ultraviolet exposure in these constructions nearly as rapidly as it does in retrofit window film applications. In properly designed blast-resistant laminated glass, the interlayer material should not tear under loading from air blast pressure. In addition, laminated glass should remain in its frame after fracture. The designer should strive to ensure that blast-resistant laminated glass behaves in this manner. Insulating glass consists of two window glass lites with a sealed air space between them. The two window glass lites can be monolithic glass or laminated glass. As its name implies, insulating glass provides thermal insulation far superior to that of single



window glass lites. When fabricated using laminated glass, insulating glass provides excellent sound insulation. Norville, et al [2] observed that insulating glass fabricated with two lites of monolithic glass could provide some minimal additional protection over monolithic glass under air blast pressure loading in the Oklahoma City bombing (Figure 3). The authors feel that insulating glass fabricated using two laminated glass lites provides one of the most economical and effective blast-resistant glazing constructions available.

Figure 3-Insulating Glass with Outside Lite Fractured in the Oklahoma City Bombing.

Design Recommendation for Blast-Resistant Glazing

In designing a blast-resistant glazing system, the architect or engineer must consider four factors: the windowglass type and construction, the framing system, the attachment of the window glass construction to the framing, and anchorage of the window frame to the structural system. The authors observe that most blasts in the US come from relatively small quantities of conventional explosives. In window glass design discussions encompassed herein, small amounts of explosives produce blast waves that will have passed the windows before they fracture. The positive phase duration of the air blast pressure occurs on the order of milliseconds. In other words, by the time the window has fractured the blast pressure is long gone.



Window Glass Type and Construction

The authors feel that laminated glass and insulating glass fabricated with laminated glass comprise the most effective and economical blast-resistant glazing materials. The authors offer a simple chart (Figure 4) that provides an empirical relationship between the weight of a hemispherical TNT charge detonated on the ground surface and its standoff distance from a window glass lite to a 60-second duration static design load. In developing this chart the authors considered magnitude of reflected air blast pressure, magnitude of positive phase impulse, and experimental results [15] from blast tests involving laminated glass and insulating glass fabricated using laminated glass [16].
Standoff Distance (m)

~. 175 "o 150 O~ o "J 125 .m Or) 100 .o 90





50 60 70 80 100 120


I I!

9.00 ~" 8.00 ,,.., 7.00 "o


5.00 ~


e- 70
45 I 50

= 0


4.00 ~ 3,(X~ ! e2.00 ~ 1.50 ~

40 50 75 100 125 150 200 250 300 400

8 4O

30 20
Figure 4-


Standoff Distance (ft)

Chart Relating Equivalent TArTCharge Size and Standoff Distance to 60Second Equivalent Load.

Designers should use this chart to obtain 60-second duration equivalent design loads for laminated glass and insulating glass fabricated using laminated glass. The values in the chart do not apply to any monolithic glass type, insulating glass fabricated with monolithic glass lites only, or to monolithic glass with retrofit window film. In comparing designs obtained using values from this chart, the authors observe that designers using this approach will obtain laminated glass thickness designations equal to or slightly higher than they would obtain using more esoteric procedures. To use the chart, the designer defines the design threat explosion in terms of an equivalent weight of a hemispherical TNT charge and a standoffdismnce. The designer enters the chart in Figure 4 by projecting a vertical line from the horizontal axes that represent standoff distance to the sloping line that represents the charge weight. From the intersection of the vertical projection and the sloping line, the designer projects a



horizontal line to the vertical axes and reads the equivalent 60-second duration equivalent design load. For charge sizes other than those shown, but less than 231 kg (500 lb), the designer can interpolate between the sloping lines. The designer then uses procedures in E 1300-00 to select the laminated window glass type(s) (annealed, heat strengthened, or fully tempered) and to determine the required thickness designation(s) to resist the 60second duration equivalent design load. If the wind load for a given design exceeds the magnitude of the 60-second duration static design loading determined from the chart in Figure 4, then the designer should use wind load to design the laminated glass or insulating glass fabricated using laminated glass. The PVB in laminated glass should have 1.58 mm (0.060 in.) thickness in blastresistant glazing [16], although 0.762 mm (0.030 in.) PVB will usually suffice for smaller explosions. Thicker PVB will result in larger forces that the window frame must accommodate the laminated glass plies should be annealed or heat strengthened glass. Upon fracture, annealed and heat strengthened glass plies tend to produce larger shards than do fully tempered glass plies. The larger shards adhere well to the PVB thus reducing flying and falling glass shards and giving the laminated glass some stiffness following fracture, helping to maintain it in its frame. If blast-resistant glazing is designed using values obtained from this chart, glass will fracture and require replacement. Laminated glass and insulating glass fabricated with laminated glass and designed using values from this chart will more than satisfy GSA Level 3B Criteria in that it fractures "safely," producing minimum hazard. The chart in Figure 4 coupled with standard window giass design procedures provides a means to arrive at a laminated glass design suitable to resist the specified design threat explosion. The chart in Figure 4 also indicates that the best protection from a bomb is standoff distance, i.e., the 60-second equivalent design loading associated with a given bomb goes down rapidly with increasing standoff distance.
Framing System

Under air blast pressure loading, the designer should design the window glass framing system and its anchorage to the surrounding structure to resist the maximum loading that the window glass would transfer to its supporting frame at fracture. The authors interpret this as the frame loading that would result ifa 60-second equivalent loading associated with a probability of breakage of 0.5 acted on the window glass lite [16]. The designer can also use dynamic analysis techniques with a design blast loading to determine the loading that the lite would transfer to the frame if it never fractured, although this might prove overly conservative, Procedures in E 1300-00 present load resistance values based on maximum nominal frame deflections of L/175. The authors note that blast tests on curtain walls indicate that more flexible frames supporting window glass constructions tend to provide better glass performance under blast loading

Attachment of Window Glass Construction to Framing

For blast-resistant glazing, the designer should avoid "dry glazing," in which gaskets alone hold the blast-resistant glazing in its frame. Standard glazing bites with gaskets



will not restrain fractured laminated glass under air blast pressure loading and the entire lite could fly from the frame. The use of very deep bites with gaskets might restrain the blast-resistant glazing but could lead to other problems. Blast-resistant glazing should attach to the frame using either structural silicone sealant or adhesive glazing tape. The bite depth should not exceed standard depths any more than necessary to facilitate the width of the structural silicone bead or the glazing tape, where the width is measured parallel to the plane of the glass. When using structural silicone sealant, the width of the bead forming the structural connection should equal the thickness designation of the blast-resistant glazing material with which it is in contact. This thickness will usually be less than the thickness of the entire blast-resistant glazing construction. For example, if the blast-resistant glazing construction consists of an insulating glass unit with two nominal 6 mm (1/4 in.) lites and a 12 mm (1/2 in.) air space, the authors reeornmend a 6 mm (1/4 in.) structural silicone bead. In the event of an explosion, this width should result in some tearing of the silicone bead before the PVB interlayer tears but should maintain the laminated glass in its frame. This mode of failure will tend to eliminate flying and falling glass shards while maintaining the blast-resistant glazing in its frame, especially insulating glass units. Glazing tape has more flexibility than structural silicone and the designer should use a width of glazing tape 2 to 3 times the thickness of the blast-resistant glazing material with which it is in contact.

Design Example Compared With Test Result

Design Procedure
Consider selecting the thickness of a single laminated glass lite required to glaze a fenestration having rectangular dimensions of 1190 x 1640 mm. The design blast loading results from a 25 kg TNT charge situated 25 m from the window. To use the chart in Figure 4, the designer projects a line down from 25 meters, the standoff distance, to a point corresponding to the 25 kg, determined by interpolation. From this point, the designer projects a horizontal line to the vertical axis. This intersection indicates that the lite should resist a 60-second equivalent design loading should of approximately 2.05 kPa. Going through procedures in E 1300 -00 to design the laminated glass gives a thickness designation of 10 mm. For this size opening the load resistance of a 10 mm laminated glass lite is 2.39 kPa. As mentioned above, the authors recommend using a PVB interlayer with 1.52 mm thickness and using a 10 mm structural silicone bead to attach the glass to the frame.

Comparison with an Experiment

Figure 5 shows a nominal 6 mm thick laminated glass test specimen with rectangular dimensions of 1190 x 1640 mm after being subjected to a blast loading that corresponds to the design loading in this example. The laminated glass specimen has a 0.76 mm thick PVB interlayer. Clearly the specimen fractured in a safe manner, producing minimal hazard. The design approach presented herein would require a larger thickness designation. Under the same blast conditions, a laminated glass lite designed with this procedure would have broken in a similar manner.



Figure 5-Specimenfrom Blast Test. Conclusions The authors present this design approach to give guidance to the window glass and engineering community on how to size laminated glass and laminated window glass constructions to resist blast loadings. They note that in formulating the empirical relationship between air blast pressure, positive phase impulse, and 60-second equivalent loading, the authors looked at results f~om numerous blast tests with laminated glass and laminated glass constructions. Use of this approach results in laminated glass designs



that will produce minimal hazard in the occurrence of an air blast pressure loading of design size or smaller. For explosions with higher than design air blast pressures or positive impulses, laminated glass designed according to this approach will still provide a significant level of protection.

[1] Norville, H. S., Smith, M. L., and King, K. W., "Survey of Window Glass Broken by the Oklahoma City Bomb on April 19, 1995," Glass Research and Testing Laboratory, Texas Tech University, Lubbock, TX, 1995. [2] Norville, H. S., Swofford, J. L., Smith, M. L., and King, K. W., "Survey of Window Glass Broken by the Oklahoma City Bomb on April 19, 1995, Revised." Glass Research and Testing Laboratory, Texas Tech University, Lubbock, TX, 1996. [3] Norville, H. S., Harvill, N., Conrath, E. J., Shariat, S., Mallonee, S., "Glass-Related Injuries in the Oklahoma City Bombing. "Journal of Performance of Constructed Facilities, American Society of Civil Engineers, 13(2), 1999. [4] Blocker, V. and Blocker, Jr., T. G., "The Texas City Disaster: a Survey of 3000 Casualties." American Journal of Surgery, 78, 1949, pp. 756-771.

[5] Brismar B. and Bregenwald, L., "The Terrorist Bomb Explosion in Bologna, Italy,
1980: an Analysis of the Effects and Injuries Sustained." Journal of Trauma, 22(3), 1982,pp. 215-220. [6] GRTL, "Misty Picture Data: Window Glass Experiment, Final Data Report, '" Glass Research and Testing Laboratory, Texas Tech University, Lubbock, TX, 1987. [7] Beason, W. L., "A Failure Prediction Model for Window Glass, "NTIS Accession Number PB81-148421, Institute for Disaster Research, Texas Tech University, Lubbock, TX, 1980. [8] Norville, H. S., and Minor, J. E., "The Strength of Weathered Window Glass." Bulletin of the American Ceramic Society, 64(11), 1984, pp. 1467-1470. [9] Beason, W. L. and Norville, H. S.,"Development of a New Glass Thickness Selection Procedure," Journal of Wind Engineering and Industrial Aerodynamics, Vol. 36, Elsevier Science Publishers, 1990, pp.1135-1144. [10] International Code Council (ICC), "International Building Code," Falls Church, Virginia, 2000. [11] Southern Building Code Congress International (SBCCI), "Standard Building Code," Birmingham Alabama, 1999.



[12] Building Officials and Code Administrators International, Inc., (BOCA), "The BOCA National Building Code/1999," Country Club Hills, Illinois, 1999. [13] Minor, J. E., "Basic Glass Strength Factors." Glass Digest. 69(9), 1990. [14] Norville, H. S., "Dynamic Failure Prediction for Annealed Window Glass Lites," Glass Research and Testing Laboratory, Texas Tech University, Lubbock, TX, December, 1990 [ 15] WinDAS, "Window Design and Analysis Software." Protective Design Center, Omaha District Corps of Engineers, 215 N. 17th Street, Omaha, NE, 2000. [16] Norville, H. S., and Conrath, E. J., "Considerations for Blast Resistant Glazing Design," Journal of Architectural Engineering, American Society of Civil Engineers, Vol. 7, No. 3, 2001, pp. 80-86. [17] Smilowitz, R., Termant, D., Rubin, D., and Lawver, D., "Las Vegas Courthouse Curtain Wall Project; Blast Prediction, Test Planning, and Post-Shot Analysis." Report No. WANY 99-01, Weidlinger Associates, Inc., New York, NY, 1999.

STP1434-EB/Dec. 2002

Author Index

Mazula, Ted W., 3 Morse, Stephen M., 90 N Norville, H. Scott, 66, 79, 159 O O'Day, Thomas, 20

Barry, Christopher J., 20 Beason, W. Lynn, 105 Beers, Paul E., 121 Bennison, Stephen J., 57 Berube, Henri, 147 Betten, Michael, 147 Block, Valerie L., 8 C

P Conrath, Edward J., 159


Pearson, John E., 131 Pelletier, Roger E., 131 Pilcher, Mark A., 121

E1-Shami, Mostafa M., 66, 79


Hennings, Iva~ 3

Razwick, Jerry, 131 Redner, Alex S., 26


Jackson, Ryan, 79 Jagota, Anand, 57 Johnson, George, 79


Schmidt, Mark K., 131 Sciaudone, Jeffrey C., 121 Shah, Bipin V., 49 Smith, C. Anthony, 57 T Torok, George R., 35

Kaskel, Bruce S., 131

Lichtenberger, Werner, 35 Lingnell, A. William, 105 M Major, Allan, 35

V Van Duser, Alex, 57 W Wise, Daniel J., 49



by ASTMInternational

STP1434-EB/Dec. 2002

Subject Index
A Air blast pressure, 159 Airport control towers, 79 Anchored-film glass retention systems, 131 Annealed glass, 131 thermal stress evaluation, 105 ASTM C 1036, 3 ASTM C 1048, 26 ASTM C 1279, 26 ASTM E 1300, 66, 79, 90 ASTM E 1886, 131 ASTM E 1996, 131

G Glass, breakage during fire, 131 Glass inspection, 3 Glass lites, 79 Glass quality, 8 Glass retention, 131 Glass specification, 3 Glass storage, 3 Glass strength, 8, 57 Glazing, security, 147 Glazing design, 66, 79, 90 blast resistance, 159 wind load resistance, 79

Blast resistance, 159 Building codes, 8 Burglary, 147 C Coatings, 20 Computer software, 90 Computer technology, 90 Condensation resistance rating, 49 Creep resistance, 57 Cyclic pressure cycles, 121

Heat-strengthened glass, 26 High winds, 121 Hurricane resistance, 121, 131 Hydrophilic, 20 Hydrophobic, 20

Impact resistant glass, 121 Insulating glass, 35 blast resistance, 159 Interlayers, 57 International standards, 8

Damaged glass, 3 Design standards, 90 Dew-point measurement, 35


Laminated glass, 57, 66, 121 blast-resistant, 159 Load resistance, 66 Longevity, 35 M Missile impact test, 121

Edge strength failure prediction model, 105 Edge stress, 66 F Failure prediction, 79 Fallout protection, 131 Fenestration, 131 Finite element analysis, 66, 79, 105 Fire-rated glass, 131

National standards, 8 NFRC 100, 49 NFRC 500 Procedure, 49


Photocatalytic, 20 Polyvinyl butyral interlayers, 57




Quality control, 26

Stress maximum, 66, 79 measurement, 26 Surface damage, 3 T

Retrofitting, hurricane resistant glass, 121

Safety glazing, 8 Scratched glass, 3 Security, 147 Self-cleaning glass, 20 SentryGlas Plus, 57 Serivce life, 35 Skylights, 8 Sprinkler systems, 131 Standards Design Group, Inc., 90 Stiff interlayer, 57 Stiffness, 57 Storm shutters, 121

Temperature differential, 131 Temperature performance, 57 Tempered glass, 26 Tensile stress, 105 THERM 2.1, 49 Thermal shock, 131 Thermal stress cracking, 13 i evaluation procedure, 105 Three-sided support, 66

Windbome debris, 121, 131 WINDOW 4.1, 49 Wind resistance, 79