4.2. Manufacturers' Test Methods: Correlation James D. Gwyn Libbey-Owens-Ford Company, Laboratory test procedures used by Libbey-Owens-Ford Co. to evaluate insulating glass are chosen to simulate the cyclic temperature and moisture effects which may occur for actual windows. Because the edge seal construction of the insulating glass contains no organic materials, ultraviolet radiation produces no deleterious effects and this item is not normally included in the testing. Testing is conducted in chambers capable of producing rapid changes in temperature and relative humidity when required. Many test-cycle configurations and test durations are used, depending upon the time available and other factors. Most commonly the procedure recognized by General Services Administration is employed. This procedure requires the insulating glass to be subjected to 175 continuous weathering cycles each consisting of alternate exposure to 48 hours at 0° F and 48 hours at 145° F and 95 percent relative humidity. At the conclusion the dewpoint of the air space must not exceed -18° F. During testing the glass is not moved, thereby eliminating the damage which may occur to the glass or seal when two separate test chambers are used. The long period required to conduct the above test precludes its use as a routine pro- Key words: Accelerated weathering; field experience; insulating glass units; outdoor 1. Introduction Before discussing our test methods and the experience we have had with our insulating glass units in the field, I believe it will be helpful if I briefly describe their construction. The edge seal construction is of especial interest. In the fabrication of the units, the glass is first washed, cleaned, and dried after which metallizing and tinning are applied to the glass surface around the periphery. This forms an integral bond much stronger than the cohesive strength of the glass itself. To the metallized and tinned glass is applied a lead calcium alloy separator strip formulated to withstand movement that may occur in an insulating glass unit. As a point of interest, this alloy is the same as that used for protecting outdoor telephone cables; therefore, its long term weathering properties have been thoroughly tested and are well recognized. The lead separator is soldered to the metallizing using a specially designed soldering iron and a compatible soldering material. The construction described to this point is that which was used for many years in our insulating glass. Field experience was excellent and trouble occurred only in cases where the sash was grossly misdesigned or other improper conditions were excessive in one way or another. Even though failures were rare, we felt improvements should be made so that the chances of failure were even further reduced. To arrive at this goal we developed several improvements. One of these is the application of a wax coating to the outside of sealing materials to 2. Testing Testing of our insulating glass units begins with the material suppliers who are required to furnish materials to rigid specifications and perform prescribed tests. We conduct similar tests in our laboratories to make certain that the materials meet specifications. We realize the important item for an insulating glass unit is not the performance of individual parts but the performance of the assembled unit and the majority of our tests are on this basis. In our laboratory we have two cyclic test cabinets for accelerated weathering. In these the temperature and relative humidity are automatically controlled and if desired can be preprogramed. The primary advantage of these cabinets compared with earlier ones used by ourselves and others is that the insulating glass is not manually moved from a cold chamber to a warm chamber during each cycle. If the tests involve a large number of cycles, there is a good chance of damaging the glass when it is moved negating the results of the tests. We have used various test cycles and have found no short duration test that remotely reflects the field performance. Most short duration tests tend to be unduly severe causing damage that could never occur in the field or are not severe enough thus giving misleading results. Of the many weathering cycles studies we have found two which we believe will result in failure of inferior insulating glass units. One of these consists of an 8-hr cycle with a dwell time of 48 min at 145 °F and 48 min at 0 °F, the relative humidity maintained at 95 percent when the temperature is above 40 °F. Heating and cooling is at a uniform rate. We believe that for a sampling of 6 test units, not more than one unit should have an air space dewpoint above 0 °F after 200 cycles. Passing this test does not necessarily mean the unit is adequate but if the unit is grossly inadequate it should fail. Another test similar in many respects consists of a 6-hr cycle with 30 min dwell time at 120 °F and 30 min at 20 °F. Again, the relative humidity is maintained at 95 percent when the temperature is above 40 °F. We believe a sampling of 6 units should withstand at least 600 cycles of this test with not more than one unit above 0 °F dewpoint. In the early stages of our testing we studied the effect of ultraviolet radiation on the edge seal of our units and found no effect. Therefore, units of our manufacture which we evaluate are not subjected to ultraviolet radiation. Of course with mastic type units ultraviolet testing is very necessary since these units are affected to some degree by extended exposure to ultraviolet radiation. Besides the various cyclic tests we conduct in our test cabinets, we also conduct what we call a "huff and puff" test. The apparatus used for this consists of two insulating glass units with a narrow air space between. To this air space is attached an air line and necessary pressure regulating and timing controls. The pressure is fluctuated within the space causing the units to bow inwardly and outwardly as the pressure is changed. The purpose of this test is to simulate gust wind loading to see if the edge separator materials are affected. The amount of pressure and the fre quency of fluctuation depend on the particular goal of the test. We also test units by exposing them to outdoor weathering. At present we have about 4,000 units in our two outside test areas. They are not glazed in openings but are open to the weather on both sides allowing rain and snow to reach the edge channel and exposing the edge seal to all weather factors. We have had units exposed up to 12 years in this type of testing. 3. Conclusion In attempts to correlate our laboratory tests with actual field experience we have found no conclusive correlation. We do know, however, that duplicate units of those which have weathered for 12 years and are still in good condition will withstand in excess of 1000 cycles of the 120 °F to 20 °F cycle test previously described. The two cyclic tests described earlier are, of course, much too long for quality control. The shorter one requires about 70 days to complete. Obviously there is a strong need for a test method requiring less time while accurately reflecting field experience. This is an overwhelming task to accomplish. After exhaustive testing we have a fair idea of the service life to be expected in climates similar to Toledo. We don't know precisely what should be expected for other climates such as might be found in Minneapolis, Miami, or Phoenix. Because of the lack of supportable correlation between laboratory tests and field experience, the expected service life of our insulating glass units is not definitely determined. We have manufactured insulating glass units for over 30 years and even the early units which lacked the many later improvements have performed well to the best of our knowledge. We have no precise records of these early units but do have information regarding units produced up to about 20 years ago. The largest installation of over 20 years ago was one containing 14,000 units and to date the units are performing satisfactorily. A small stock of replacement glass was ordered with the original glass and has been adequate for replacement for breakage and other types of failure. At present our insulating glass is warranted for 20 years. Based on our field experience, accelerated tests, and outdoor weathering of units, we are confident that the 20 year warranty is fully justified. 4.3. Test Methods and Field Experience With Double-Glazed Units R. J. Mazzoni and G. H. Bowser Glass Research Laboratory, Creighton, Pennsylvania 15030 Performance criteria for double-glazed units intended to assure building owners of a satisfactory period of performance in service must cope with field environments through accelerated testing procedures. Quantitative knowledge of performance variability and a minimum acceptable standard of satisfactory performance are essential ingredients of any program of this kind. The accelerated testing program we use includes: 1. Temperature cycling 2. Water vapor diffusion 3. Solar radiation 4. Outdoor exposure 5. Water immersion We have tested large numbers of units manufactured in our own plants and by others according to these procedures. A significant difference in performance among groups of units tested has been observed. Differences in the sealant composition and performance, in manufacturing procedures, in type and conditions of desiccant used are factors which explain the wide range of performance demonstrated by superficially similar units. Our service experience, because of the long guarantee which has been in effect for many years, is based both upon our replacement records and upon formal field exposure studies. Correlation between accelerated test procedures and field tests in recent years has been good. We believe that this correlation provides sufficient justification for establishing a minimum performance level in the accelerated tests. Key words: Accelerated lab tests; dewpoint measurement; double-glazed insulating units; exponential distribution; field performance; outdoor aging tests; organic seal units; pressure tests; polysulfide rubber sealants; temperature cycling test; water vapor diffusion. 1. Introduction During the past decade, serious attention has been devoted increasingly by government agencies to the problem of evaluating in-service life of double-glazed insulating units. I believe that the Canadian [3,4]1 and Scandinavian [1,2,6,7,8,9] Government Agencies first recognized this need several years ago. Perhaps this is explained because of the severe winter climates in their regions and the greater proportional use of double-glazing. Domestic interest is growing and it is apparent from the interest demonstrated in this Seminar that progress will be made. We hope in this paper to point out the importance of tight correlation between accelerated laboratory testing and orderly monitoring of field performance at the job site. The foundation for this work was described in an earlier ASTM paper [5]. From the very beginning of our participation in the Insulating Glass Market over twenty years 1 Figures in brackets indicate the literature references at the end of this paper. ago, it was apparent that a sophisticated and extensive testing and development program would be needed. While close correlation between accelerated laboratory tests and field performance has been acquired over a period of years, our initial experience, including both success and failure, taught us that product reliability on the job could be determined in advance by accelerated tests that simulate environment factors encountered in service. To make accelerated testing practical, it was important to evaluate test unit sizes. Based on careful studies of stresses in glass and the edge seal joints, we arrived at a standard test unit size of approximately 14 × 20 in with a 1/4-in air space and glass thickness of 18 in or less. We learned early that a great deal of time and effort could be saved if certain very rapid "screening tests" could be applied to eliminate test units of poor design or careless fabrication. These screening tests include dewpoint measurements and pressure differential tests of seal tightness. Generally, test units with air space dewpoints above 0 °F are obvious indications of seal leakage and should be discarded since their performance is so poor as to result in early failure. Figure 1 shows the 120 to 20 °F cycling test equipment. The units are mounted above a 1-in pool of water to maintain high humidity throughout the test. Within the cabinet, the atmosphere is cycled from 120 to 20 °F. This test is controlled automatically to give four 6-hr cycles each day, seven days a week. This test checks the ability of the sealant to withstand pressure loading caused by temperature fluctuations and expansion and contraction forces between dissmilar materials. A relatively simple test and one which can readily check the water vapor diffusion characteristics of the sealant is carried out at 110 °F and 90 percent relative humidity (see fig. 2). Air temperature is thermostatically controlled at 110 °F and a pan of water in the bottom of the cabinet maintains the high humidity throughout the test. The specimens are supported on wood frames and continuously exposed to conditions in the cabinet. The sunlamp test is used to obtain ultraviolet light exposure (see fig. 3). RS 275 watt sunlamps are positioned 14 in above and perpendicular to the specimens, and the light is directed on Test Unit corner areas. The aluminum surface below the specimens reflects the radiation to the oppo FIGURE 2. 90 percent relative humidity chamber in 110° F. room. site surface. We have found this test to be very effective when it is followed by the 120 to 20 °F temperature cycling test. The 130 to -30 °F cycling test, shown in figure 4, includes a circular table that rotates automatically according to a predetermined time schedule. Specimens are glazed into each of the five 4 × 4 ft panels which are mounted vertically on the table so that the exterior surfaces of the units are exposed to conditions within test chambers lo FIGURE 5. Specimens mounted in outdoor exposure rack. FIGURE 4. Indexing table for 130° F. to 30° F. cycling test. cated at fixed positions around the table. The opposite surfaces are exposed to prevailing room conditions. Each specimen is subjected to four complete cycles each day, six day a week. Specimens are exposed sequentially to (1) -30 °F, (2) room conditions (70 to 90 °F), (3) 130 °F, (4) water spray, and (5) room conditions (70 to 90 °F). Figure 5 shows how the specimens are mounted in the outdoor exposure rack. The units are facing south and inclined at a 45-degree slope to prevailing conditions at Harmar Twp., Pa., and Fort Lauderdale, Fla. Water immersion, Fadeometer, Weatherometer, elevated temperature and pressure tests are useful also. Our field testing program includes exposure of test units in the outdoor wall of our Creighton, Pa., laboratory building and full-size production unit installation in buildings selected geographically for exposure conditions. To provide statistical validity, a large number of units in many different installations is required. Let me show typical data for organic seal units obtained from specific field tests. For reference, let me remind you of the data first presented in our earlier ASTM paper [5]. The following data are in addition to those data. 3. Accelerated Test Results Curves in figures 6, 7, and 8 are for seven groups of units from five manufacturers obtained within the past four years. Most of the unit groups were sealed with polysulfide rubber sealants. These are representative of most organic sealed units. The minimum number of units in each group was 18 distributed among four to five accelerated tests. The data presented are from the 120 °F to 20 °F cycling, 110 °F -90 percent R.H. (relative humidity) and the combination ultraviolet and 120 °F to 20 °F cycling tests. Recommended exposure periods for these tests are made relative to 20-yr service life with 10 percent or less failure potential. The basis for these recommended periods will be discussed later. The significant difference in performance obtained among groups tested in the 120 °F to 20 °F cycling test can be seen in figure 6. The mean |