1. Introduction

Over the past 10 years a large number of manufacturers of factory-sealed double-glazing units have entered the Canadian market, and there has been an increasing use of these components in both residential and commercial buildings. In 1961, when the Dominion Bureau of Statistics first began to keep records on their use, the total value of annual production was about 9 million dollars; by 1965, the last year for which records are published, the value had risen to about 16 million dollars.

The development and availability of new organic sealing materials applicable to the construction of sealed glazing units has been one of the factors that has led to this growth, and all of the new manufacturers have utilized an organictype sealing arrangement.

The appearance of such large numbers of brands of sealed double-glazing for which there was no history of field performance presented a difficult problem to the Central Mortgage and Housing Corporation (CMHC). This Crown Company has responsibility for administering the National Housing Act of Canada, including the determination of requirements for acceptability of materials and components used in houses constructed under the Act. Because there were no published standards or test methods for sealed double-glazing units, the Corporation asked DBR/NRC to assist in developing a basis for

This paper is a contribution from the Division of Building Research, National Research Council of Canada, and is published with the approval of the Director of the Division.

1 Building Services Section, Division of Building Research.

establishing their acceptability as quickly as possible.

2. Developing a Test Program

During the early stages of the study that followed, discussions with experienced manufacturers provided much valuable background information. The provision and maintenance of a low dewpoint temperature in the air space was quickly identified as the major criterion of performance. A low dewpoint temperature is necessary to avoid condensation and eventual fouling of the glass surface from leaching of sodium salts, which are a normal component of soda-lime glass. A test method, described in Reference [1],2 to measure the relative dewpoint temperatures of the air space was established; initial measurements showed a wide variation among units, a number having high values of moisture content.

It was evident from calculations of the amount of moisture required to produce excessive dewpoint temperatures that only very small amounts of moisture transfer to the air space could be tolerated over the service life of a unit, even when desiccants were employed. Moisture is transferred to the space by diffusion of water vapor, or, if a leak exists, as a result of the movement of liquid water or air caused by pressure differences across the seal. These pressure differences are induced by temperature or barometer pressure changes, or by wind action, and result in the transfer of large amounts of moisture if leaks are present. Thus,

2 Wilson, A. G., and Solvason, K. R., Performance of sealed double-glazing units, J. Can. Ceram Soc., 31, 68-82 (1962).

the unit must be hermetically sealed with materials having a high resistance to water vapor diffusion and must remain sealed throughout its life. The primary problem of evaluation is, therefore, the determination of the adequacy of the seal.

In service, stresses leading to seal failures (and glass breakage) are imposed on double-glazing units in several ways: by pressure differences between the air space and surrounding air due to temperature and barometer pressure changes; by differential expansion or contraction of components caused by unequal thermal expansion coefficients and differential temperatures; by wind pressures; and by forces that may develop due to faulty installation. Sealing systems must withstand the repeated action of these forces and must also retain the necessary physical properties under normal conditions of exposure over their required service life.

The development of methods to determine the resistance to chemical degradation of the sealing system under service conditions was regarded as a long-term problem and efforts were, therefore, concentrated on developing methods of evaluating the ability of the sealing systems to withstand repeated cycles of stress without developing leaks. Attention was directed towards methods that could be applied to the units as a whole, rather than to the individual components, because the performance of the unit depends upon the interrelation of components and manufacturing techniques.

It was decided that for this purpose it was necessary to accelerate both the effects of various kinds of mechanical stresses that could occur in service and the moisture transfer process, particularly that due to total pressure differences across the seal, so that tests could be conducted in a reasonable time. It was also considered desirable to stress the sealing systems over the range of temperatures that could occur in service in order to expose weaknesses associated with temperaturedependent properties of sealants. The test established for this purpose consisted of exposing one side of the specimens to room conditions controlled to 73 °F (23 °C) and 50 percent relative humidity (RH) while exposing the other side to a simulated weather cycle of: heating to 125 ± 5°F (523°C) over a period of 90 min, air circulation alone for 25 min, water spraying at 755°F for 5 min, air circulation alone for 60 min, and cooling to -255 °F (-32 ± 3 °C) over a period of 60 min. The apparatus is shown in figure 1.

The dewpoint temperature of the air space after exposure to the weather cycle was taken as the criterion of seal adequacy.

Because of the possibility of wide variations in quality from faults in the assembly process, it was decided that several specimens of each brand should be tested. Owing to space limitations and the large numbers of specimens involved, there

FIGURE 1. Weathering apparatus for sealed double-glazing units.

was considerable incentive to use small specimens. A size of 14 by 20 in. (35.5 by 51 cm) was selected, somewhat arbitrarily as a practical mini

mum.

The size (the small dimension, particularly), the air space thickness, the glass thickness, and the rigidity of the edge, all influence the air pressure differences developed between the air in the space and ambient. The pressure difference, in turn, largely determines the stress imposed on the sealing system under the conditions of test. A rise in air temperature within the space results in an increase in pressure, a glass deflection and, hence, an increase in volume. The pressure rise and deflection are interrelated, so that on small units the deflection is relatively small and the pressure rise relatively large. A larger pressure rise occurs with thick glass than with thin, because of the smaller deflection. Both pressure rise and deflection increase as the air space thickness increases. The shape of the deflection curve on the glass is influenced by the rigidity of the edge arrangement. Larger pressure increases occur with rigid edges as this arrangement results in a smaller mean deflection (and hence a smaller volume increase).

Exposure in the weather cycling apparatus sometimes results in breakage of glass adjacent to the spacer on units having rigid sealing arrangements. As there was no evidence of such occurrences in the field, it had to be assumed that this effect was peculiar to the unit size, glass thickness, and rate of change in the cycle. The weather cycling apparatus was, therefore, deemed unsuitable for tests on units having all-glass edges or glass-to-metal seals.

The structural arrangements at the edges of various units utilizing an organic-type seal are very similar. Brands with several years of good field performance withstood many cycles in the apparatus, whereas units having poor field performance records failed in relatively few cycles. The apparatus, therefore, provides a good basis of comparison of different units of this type, provided that the unit size, glass thickness, and air space thickness are the same. Thirty-two oz (4 mm) glass and a 12-in. (1.3 cm) air space were selected for purposes of acceptance testing.

Some tests were conducted on 20- by 28-in (51 by 71 cm) and 28- by 40 in. (71 by 102 cm) (approx) units to assess the influence of size. It was possible to test only a few units because of space limitations. The larger sizes did, however, withstand many more cycles than the smaller ones.

There was no way of comparing combined effects of stresses and moisture transfer potentials produced in the weather-simulating apparatus with those in service, and it was not possible to relate laboratory exposure directly to field conditions. It was accepted from the beginning that the test provided only a basis for comparing the behavior of sealing systems under conditions of fluctuating mechanical stress and temperature comparable to those that might occur in service. Tests were therefore conducted on specimens of most of the units on the market, including a few for which there was some history of field performance. In addition, a simple initial screening test to identify gross leaks in the seal was adopted; this consisted of determining the ability of a unit to maintain a deflection of the two panes induced by a small change in ambient pressure.

While these initial laboratory tests were being conducted, a few specimens of each brand were exposed to outdoor weather, mounted in a vertical position on a plywood support facing south, and dewpoint temperatures were measured periodically (fig. 2). The primary purpose was to expose the specimens to ultraviolet radiation to determine whether the sealing systems were sensitive to failure from this cause.

The results of these initial studies have been reported [1]. Based on the results, CMHC established initial requirements for acceptability. In tests on 18 specimens submitted by the manufacturers, at least 17 were required to pass the initial screening seal test and to have dewpoints no higher than 30 °F (-1 °C). Twelve of the specimens were exposed to 320 cycles (2 months) in the weather cycling apparatus and at least ten were required to have dewpoints no higher than 30 °F at the end. Results of tests on 33 sets of units were as follows: units from 23 sources passed the initial seal test on first submission; units from 10 sources failed and 9 subsequently passed on re-submission. Of the 32 sets that ultimately passed the initial seal test, at least 17 failed the weather cycle based on the above re

quirements. Fourteen of 32 sets mounted on the outdoor racks had at least one failure after one year of exposure.

Following the establishment of these acceptance requirements in 1961 DBR began to conduct tests on a commercial basis for manufacturers, who were required to submit a detailed description of the units in applying to CMHC for acceptance. No attempt was otherwise made to ensure that specimens submitted by manufacturers for qualification testing represented typical production. Acceptance of products by CMHC was therefore based on the ability of manufacturers to meet the current test requirements rather than on any positive assurance that the units being marketed met these requirements.

At this time, development was begun on a further qualifying test procedure involving exposure of the units to an elevated temperature cycle (70 to 130 °F) (21 to 54 °C) and high humidity atmosphere. One of the purposes of the test was to provide a high average water vapor pressure, not present in the weather cycling apparatus, in order to obtain some indication of the resistance of the sealing systems to water vapor diffusion. In addition, there was need for a simpler qualifying test because of the large volume of testing and the limited capacity of the weather cycling apparatus; and for an inexpensive apparatus that could be reproduced by manufacturers for use in product development. The final form of the apparatus is shown in figure 3. Again, the dewpoint temperature of the air space, following exposure to the elevated temperature cycle, was taken as the criterion of seal adequacy.

During the development phase, an extensive series of tests was conducted to compare the performance of a number of sets of units exposed to

both the weather cycle and the high humidity cycle. In general, brands that failed in the weather cycling apparatus in less than 320 cycles failed in the high humidity cycling apparatus in less than 24 cycles (4 weeks); brands that withstood more than 320 cycles of the former, usually withstood over 8 weeks of exposure in the latter. Exposure to the high humidity cycle did not cause abnormal failures, such as breakage of the glass adjacent to the spacers, in units having rigid edges. The apparatus was therefore used in evaluating sealing systems of this type as well as those with organic seals.

In 1963 CMHC included exposure of six specimens to the high humidity cycle as a part of its acceptance requirements, and the number of specimens in the weather cycle was reduced to six. In 1964, the requirements for acceptance were reviewed in relation to the range of test results being obtained. It was apparent that the majority of manufacturers could produce units that provided initial dewpoint temperatures below -40 °F (-40 °C) and values after weather and humidity cycling below 0 °F (-18 °C). It was observed during the weather cycle that condensation sometimes occurred between panes with reference dewpoints above 0 °F (-18 °C). As a result, CMHC altered the initial and final dewpoint requirements to -40 °F (-40 °C) and 0 °F (-18 °C), and a new round of qualification testing was begun on this basis.

3. Development of a Standard

As a result of widespread recognition of the qualifying tests being used for CMHC acceptance, they were accepted as the basis of a national standard, preparation of which was begun in 1965 under the auspices of the Canadian Government Specifications Board (CGSB). The CGSB Committee on Sealed Double-Glazing Units consisted

of representatives of sealed glazing manufacturers, sealant suppliers, and government users. Consideration was initially given to establishing requirements for two grades of units, one based on the existing CMHC requirements and a second, higher grade based on initial and final dewpoint temperatures of -60 °F (-51 °C) and -40 °F (-40 °C). Results of the most recent qualifying tests for CMHC at that time indicated that a large percentage of the manufacturers were capable of making units that could meet the requirements of the higher grade. At the urging of the industry representatives, the Committee decided to include only the higher grade.

The Committee was concerned that the tests developed for CMHC acceptance did not include one to determine the likelihood of glass staining by the condensation of organic vapors evolved from the sealing system. Staining problems had been experienced with many early brands. Tests on individual components were considered, but preference was given to a single test on an assembled unit. The "Ultraviolet Exposure Fogging" test (fig. 4) was developed for this purpose. Test units are heated to about 150 °F (71 °C) so that if volatiles are present in the sealing system components or have been absorbed by the desiccant they will be driven off and condense on the glass area cooled by the cooling plate. An ultraviolet lamp is used for heating because it was suspected that a breakdown of components of the sealing system might occur under ultraviolet exposure. Very faint deposits can be detected if an appropriate lighting and viewing technique is used. Deposits appear to be produced by traces of oil on spacers, small amounts of resin binder on mineral wool used to retain the desiccant in spacers, certain glass cleaning agents, and some plastic inserts for spacer corners, as well as by the sealants used.