You are on page 1of 12

Correlation of Laboratory to

Natural Weathering

George W. Grossman
The Q-Lab Company

The basic stresses in weathering are known and quantified. The time a material is stressed and the
temperature during UV and water stresses are the primary variables in natural weathering. Reproducing
them exactly in the laboratory can provide acceleration by eliminating idle time. Further acceleration
is possible by maximizing UV, water and temperature stresses within natural limits. The tests reported
here indicate that the most important factor in accelerated laboratory weathering is the proper balancing
of these stresses.

Until recently, the weight of the evidence on accelerated weathering suggested that it is not a reality.
Thousands of attempts to reproduce natural weathering achieved only limited success at best. Our
inability to explain why weathering could not be reliably duplicated in the laboratory had led many
competent scientists to regard the entire concept as a myth, an unattainable goal.

In the past seven years, our company and other companies have been trying to erase some of the
mythology surrounding accelerated weathering with a new tool, the fluorescent UV-condensation
apparatus shown in Figure 1. In this apparatus, materials are alternately exposed to ultraviolet light
alone and condensation alone in a repetitive cycle. The UV source is an array of fluorescent lamps
with almost all of the lamp emission in the UV range. Condensation is produced by the mechanism
of the Cleveland Condensation Tester - exposing the test surface to a heated, saturated mixture of air
and water vapor, while cooling the reverse side with ambient room air.

The exposure conditions in fluorescent UV condensation apparatus may be varied by selection of
the fluorescent UV lamp, the timing of UV and condensation exposures, the temperature of the UV
exposure, and the temperature of the condensation exposure. The ability to vary all of these weather-
ing stresses is unique to the apparatus, and either not available or not convenient in prior weathering
apparatus or in natural weather exposures.

Our research has been directed to determining the limits for accelerating UV, water, and temperature
stresses and learning how various conditions of test correlated with specific Florida exposure condi-
tions on high gloss automotive coatings. Suggested limits and methods of operating the apparatus
have recently been standardized and published in ASTM G-53-77, "Recommended Practice for
Operating Light and Water Exposure Apparatus (Fluorescent UV-Condensation Type) for Exposure of
Non-Metallic Materials."

The data generated by repeated tests on the same materials was analyzed and compared with Florida
exposure results by graphic and statistical methods. Spearman rank correlation was found to be a
convenient method of determining which of several test conditions produced the best correlation to
outdoor exposure.

Florida exposures, in different years, or even at different sites in the same year, seldom result in the
same absolute values. The average gloss of identical series after a year's exposure can vary a great
deal. However, Florida exposure consistently produces good relative values. The objective of Florida
exposure is to determine the relative merits of materials. The assumption is made that rankings es-
tablished in a severe environment will prevail in more moderate weathering conditions.

Presented at the 20th Cleveland Society for Coatings Technology Symposium. Case Western Reserve University. Cleveland. Ohio.
March 1 1 , 1977. 26200 First Street. Cleveland, Ohio 44145.

we are asking whether pollution. Ery- weathering as that degradation which occurs when thema. These are local weathering stresses that the relative performance of a number of materials occur in some environments or on certain areas of in the laboratory correlates with relative perfor. water. relative merit standard to accelerated weathering and temperature. Before we combine these stresses in the laboratory. years. Even though with accelerated outdoor exposure or actual ser- our approach to weathering has been limited by vice conditions. Spearman rank correlation applies the same ultraviolet energy in sunlight. The response to these universal stresses can The purpose of this paper is to attempt to bridge take many forms. ceptable coatings for automobile wheels because UV and temperature stresses on wheels are at (2) Reporting what has been learned by repro- very low levels. what we call sunburn. begins to occur 2 a material is exposed to the universal stresses of . We are not asking "How many hours other stresses from salt water. ducing and varying UV.Fluorescent ultraviolet-condensation The 400nm upper limit for UV-A is the generally apparatus accepted boundary between visible light and ul- traviolet light. water. water as rain or dew. biological and air equal a year in Florida?". mance in the field. UV-B is the 315nm . This classification is very convenient and useful. we should understand how the indi- vidual stresses act in natural weathering. a structure or vehicle. and temperature stresses. The need to predict the resis- we conclude that accelerated weathering is a tance of materials to such local stresses has been reality. Tempera- ture can alter the rate of photochemical or oxida- (3) Proposing methods for evaluating how any tion reactions. Many solutions have been claimed as suc- cesses and then found wanting. Alkyd enamels are ac- tions. and photochemical attack can alter technique of accelerated weathering correlates the rate of oxidation during wetness. UV-A is the energy in wavelengths between 400nm and 315nm. Ultraviolet Stress in Sunlight The European weathering literature and the medi- cal profession divide the solar UV spectrum into three ranges. Figure 1 . which never reaches the earth's surface. The credibility of Our problems with predicting weatherability have any claim to the ability to accelerate weathering is always focused on reproducing the combined therefore suspect. and temperature and instances.290nm range. materials are not adversely affected by theorizing on the validity of the assumption that weathering until certain threshold levels of stress exposures in severe environments correlate with are attained. In other induced by UV. The response to localized stresses tends to follow Using the relative merit concept as our measure. quently act in this manner. it is still a complex multiple variable problem. It is recognized that there are techniques. stresses are the rule. UV-C includes the solar radiation below 290nm. Weathering can induce a stable. SO2 accelerated weathering has endured for over 65 and mold chambers. and temperature of accelerated weathering. Nevertheless. Low gloss paints can chalk at a this credibility gap by: relatively steady rate until the paint erodes away. definition. The 315 nm boundary between the Degradation Stresses in Natural UV-A and UV-B ranges defines the point at which Weathering ultraviolet energy begins to cause adverse effects The scope of this report is limited by defining and pigmentation changes in the human skin. water. UV and temperature stresses fre- performance in less aggressive weathering condi. today. effects of UV. water. and temperature stresses in the laboratory to illustrate the feasibility Interactions between UV. not the exception. adherent protec- (1) Examining the natural weathering stresses tive coating like the oxide on aluminum. the controversy over met by single property testing in salt spray. predictable patterns.

absorbing all solar radiation Essentially the same type and order of failure oc- below 290 nm so that energy in such wavelengths curs on both exposures. tion. A more precise confirmation light with relatively low cost. At solar 0. sharp cutoff of solar radiation at the earth's surface caused by ozone absorption (Figure 2). the maximum UV-B is only 0. The black lamp exposure required 1096 hr to reduce the gloss level to 11. the sunlight cutoff is 310 nm and. is repeatedly Figure 2 . sunlight cuts off at 303nm.Division and classification of solar wavelengths below 315 nm and reaches a peak Both exposures would be at essentially the same at 297 nm.2% of the total. The difference between UV-A and UV-B effects can also be examined in the fluorescent UV- The quantum theory explains the different effects condensation apparatus by substituting fluorescent of UV-A. The energy have conducted many experiments of this kind in photons is inversely proportional to wavelength. is a highly ef- fective UV absorber. We ted in discrete units called photons. the emission.315 nm band contains only at 40" altitude. the increase in energy level of light with decreasing wavelength. The sun emits radiation in wavelengths well below sun lamps on the other. it is amounts of UV-B reach the earth's surface and the about 0. with black lamps on one side of the apparatus. sunlight extends down to 320 nm. they also contributed to degradation.1% of the total energy in sunlight. Figure 3 . Although the 3 . Light energy is transmit. Ordinary glass starts to absorb strongly at 350 nm and excludes wavelengths below 310 nm. The UV-C boundary of 290 nm is a temperature. in summer. At a solar break bonds and cause photochemical degrada- altitude of 19o. is an excellent source of UV-B with some when the maximum UV-B intensities are present. sunlight cutoff is approximately 295 nm. it normally takes never reaches the earth's surface at sea level. altitudes between 60o and 90°. the observed reduction in The ozone cutoff in sunlight and quantum effect degradation could be partially attributed to reduced make it possible to reproduce the effects of sun- temperature stress. Photons in the UV-B 290 . in turn. However.05%of the total. black lamps for the fluorescent sun lamps. UV-B. 99% of enough to be completely absorbed by the atmo. reduced the 20" gloss of an alkyd enamel from 87 to 7 in 500 hr using the sun lamp. Even at an 1 1" solar altitude. The photons in UV-A are smaller and they are only partially absorbed by passing through the atmosphere. a paired exposure of UV-B in sunlight varies with solar altitude. The energy of photons in wavelengths emission in the UV-B range at 301 nm and 313 nm below 242 nm is high enough to dissociate oxygen mercury lines. In winter. Quantum effect. 2 to 4 times as long to produce degradation with Ozone partially absorbs UV-B so that the amount black lamps. A practical and convenient method of separating UV-A from UV-B is by filtering sunlight with window glass. The fluorescent sun lamp. UV-A is always present in sunlight. As an example. emission in the UV-A (Figure 3). The UV-B 290 . Black lamps have some 242 nm. demonstrated by the behavior of UV light in the violet spectrum atmosphere on human skin and on polymers. and UV-C. Since water stresses and temperature stresses were present during this test. changes can be obtained from paired direct and with peak emission at 3 13 nm and a low cutoff at under glass exposures in the Arizona summer 280nm.315 nm range are large Most of the light emission from the sun. However. It would be nearly impossible to sunburn human skin protected by window glass in any reasonable length of time. and create ozone.UV intensity between sunlight and fluo- Scott1 has shown what occurs when the UV-B is rescent sunlamp excluded from sunlight by wintertime atmospheric filtering. simply does not have the energy to sphere at solar altitudes below 14o. low energy demand that UV-B is responsible for most photochemical 40 watt UV sources. Ozone.

reproducing UV-A and visible light because these Such extreme variation in natural wetness time wavelengths are harmless to today's exterior and frequency obviously presents problems in materials. Water Forces in the Natural cutoff of 280 nm is 10 nm below the sunlight cut.1 have been observed in some been demonstrated for many years in the EMMA locations. The increased acidity is attributed apparatus where natural sunlight is intensified to sulfuric and nitric acids in the atmosphere. rain or dew weathering because it employed alternating UV condensed from an air-water vapor mixture are and condensation exposure in place of simultane.3 many times with reflectors. In any event. cold night sky and thus maintain the material Photochemical reactions occur instantaneously above the dewpoint. You might think that dew is innocuous but there Increasing UV stresses by employing UV-C radia. and other 4 materials. sistent all over the earth in rural areas. While the natural exposure inhibit condensation. is no reason to suspect that dew is any less tion is not acceptable for evaluating materials of acidic than rain. wood. there are no constraints on the vary in duration and frequency of wetness. spectral energy distribution of sunlight deserves Dew forms first on the insulated rear deck and a prompt and decent burial. plas- tics. vapor and air and typically saturated with dissolved summer sunlight on a clear day. of 4 to 20 hr in length at temperatures ranging from 40°C to 60°C have been evaluated.6 and rainfall is UV-condensation apparatus is above the intensi. However. Insulate duration of the UV exposure in the laboratory. There is no harm in hood of a car. Removing the filters from a xenon Rain can be present for only 15 min in a summer arc subjects the test specimen to UV-C radiation shower or continue for two or three days. usually acidic. of rain during 1972-73 was less than 4. rust-inhibiting oils. However. Winds can maintain a mate- with application of the energy and are directly rial at the ambient dry bulb temperature and also proportional to time. but it is a waste of energy.5 This apparatus is now in use all over the world to test the water resistance of paints. metals. the feasi. The "Dew Cycle"* remain near the ground may possibly make dew procedure was a step forward in accelerated more acidic than rain.S.3 and pH bility of increasing intensities in this manner has values as low as 2. In the Eastern U. There are duplication with laboratory cycles that are fixed in only two conditions to observe in reproducing the time and temperature. a surface so that it can lose heat to the sky without The myth that reproducing the effects of sunlight gaining heat from radiation from the ground and requires laboratory duplication of the entire it will be wet longer than an uninsulated surface. Our approach to this prob- effects of sunlight-UV-B wavelengths and good lem has been one of trial and error. The tendency of air pollutants to different chemical composition. no unusual UV effects have yet been observed The chemical quality of rain or dew is fairly con- in seven years of testing. is limited to about six hours of UV-B daily by the Identical materials exposed at the same site will earth's rotation. Water in equilibrium with atmospheric low wavelengths near 300 nm in the fluorescent carbon dioxide has a pH of about 5. Clouds at wavelengths down to 190 nm where light can can prevent a material from radiating heat to the dissociate oxygen. Materials exposed outdoors of UV-C radiation never found in sunlight on the are commonly wet with dew for 14 or 15 hr per day. the The duration and frequency of natural water cycles virtues of these advances were negated by use are extremely variable. . Environment off. Water cycles simulation of the atmospheric ozone cutoff. earth's surface.4. ous exposures with water sprays.. The intensity at oxygen. Both rain The UV stress obtained with the fluorescent sun and dew are condensed from a mixture of water lamp emitting UV-B is comparable to noon mid. electrolytes bearing dissolved oxygen. the average pH ties found in natural sunlight. The chemical properties of rain or dew can be easily and accurately duplicated in the labora- tory by employing the principles and apparatus of the Cleveland Condensation Test developed by the Cleveland Society for Coatings Technology in 1963.

At 50°C. Whenever air is introduced for heating good correlation to natural weathering can be ob. Further. It is postulated that the higher laboratory turers expose coated metal panels on plywood. with no wind and temperature. At 50"C. differences between dark and light materials in the Using these indicators. Abnormal whitening of acrylic lacquers at the test specimens. than a white panel. This requires the introduction 60°C was observed on Cleveland Condensation of cooling air which cools the panels as well as tests several years ago. The temperature spotting occurs. Abnor- Over the years. 45oS open rack exposure to 5oS open rack and provided that we do not go too high in temperature then to 5oS Black Box insulated exposures. Service temperature during nighttime wetness is similar for all colors and predictable from weather records. it is necessary to limit the time of exposure When hot arc sources are used in laboratory for some materials. the use of some ambient. temperatures of outdoor exposures to meet the Elevated laboratory condensation temperatures demand. can reach a surface temperature limits for accelerating the effects of water with 30°C above ambient.The Cleveland Society made a very significant Temperature Stress in Natural contribution to weathering science in designing Weathering and evaluating the Cleveland Condensation Tester. From a practical standpoint. of 40°C may be used with almost any material. At night. An automobile driven at 55 mph on a could be simply and reliably reproduced in the highway will attain' a surface temperature near laboratory at low cost. In the fluorescent UV-condensation apparatus. The current energy coatings can withstand 8 or even 16 hr exposure crisis will create more demand for temperature without abnormal effect. This is a limitation to accelerated cool cycles such as 20 hr at 40°C and short. range of 0°C to 20°C. tics are commonly exposed on plywood backing. or cooling. Heated air is in the fluorescent UV-condensation apparatus introduced to the chamber to raise the panel tem- because of the purity of the condensate. Experience with continuous con- paints. cycles such as 4 hr at 50°C can be used. It appears that reciprocity ex. White paints typically attain maxi- influence. When perature during UV exposure. Automotive paints have moved from accelerate water effects without abnormal results. UV stresses are also predictable but temperature in sunlight must be measured in 5 . This condition is difficult to reproduce in densation indicates that condensate temperatures the laboratory. White paints appear to fail more rapidly in hot and it is very difficult to match outdoor temperature long water cycles. Surface temperature is the most variable factor in This apparatus demonstrated how natural water weather. temperatures increase the rate of permeation and The most important piece of information needed the rate of oxidation reactions so that 4 hr at 50' C for designing accelerated test methods is the simulates 14 hr at 20" C. when resistance in materials as we add insulation. the 320 watts output from the UV There are other indicators to establish when water sources is not enough to raise the panel tempera- cycles are too long or too hot. locked and parked in 700 Cleveland Condensation Testers has provided direct sunlight. The lowest practical condensation tem- mum temperatures 10°C to 15°C lower than black perature is 40°C. Long. Nevertheless. ists between time of wetness and temperature of which is an excellent insulation. temperature differences between col- tained using a wide range of water cycles. we have established that laboratory. The Cleveland apparatus is Color is also a contributing factor in service a vented still using ambient room air as the cooling temperature. Some manufac- wetness. The same car. the surface temperature can drop 8°C Natural rain or dew usually occurs in a temperature below ambient. testing unknowns. the chamber. hot weathering with which we will have to live. some are insulated or enclosed. radiation heats the chamber along with occur. it may be attributed to the extrac- of a black panel will be only a few degrees higher tion of solubles from the coating or the substrate. service temperature stresses for mal effects on many materials may be caused at materials have gradually increased as structures 60°C condensation temperature. the time and temperature of the Materials scientists have gradually increased the condensation exposure must be limited. a clear sky. Spotting is rare ture to more than 55°C or 60°C. maximum service temperature in direct sunlight. At 60°C abnormal effects can chambers. ors are reduced. Plas- for too long a time.

racks. This is an intriguing theory but it varied to meet the need. CON. Testing different colors at the same temperature in the laboratory does not appear to be a seri- ous problem at this time. White paints are tested at higher relative temperatures causing more degradation. it is desirable to conduct laboratory evaluation of paint systems in several Table 1 . Uninsulated materials can be expected to attain temperatures around 50°C to 60°C while temperatures of 70°C to 80°C are common on dark. Sunlight drys off materials before the sun elevation reaches Methods of Correlation a point where UV-B is transmitted through the We have made correlation studies between atmosphere. All of these cycles. 6 UV-6 CON. The conditions necessary for reproducing natural weathering stresses in the laboratory are summa." all of the correlations relate 20o gloss after a period we have recommended time cycles of 8 UV-4 of time in Florida exposure to 20o gloss after a CON." We very good and visual discrimination differences of have concluded that "the answer" does not exist. about 5 points in 20o gloss appears to be possible. Conditions of accelerated weathering must be matched to the specific requirements of outdoor 6 . insulated horizontal surfaces. water. 5oS. is not consistent with real weathering. it was felt that customers respond reasonably well with natural weathering. Control of wetness tions and usually tested in Florida on insulated 5oS temperature is also a problem. The rank correlation with gloss values was that could be called "accelerated weathering. at various temperatures. Outdoor weather Temperature Forces test facilities expose materials at 45oS. ranking by five individuals not associated with the mental work was to find the one set of conditions testing. and exposure or service conditions. sure of relative appearance was tested by submit- ting two completed tests in separate color sets to For several years. Such alternate by 5oS open rack exposure. Natural weather. Each apparatus in the past. but our work indicates that the relative rankings of various colors are essentially unchanged. and 16 UV-8 period of time in fluorescent UV condensation test. Over the more to final gloss than to percent loss. are being used in fluorescent UV condensation The validity of using 20" gloss values as a mea- apparatus today to meet specific needs. and temperature cent loss of gloss was the value used in calcula- that satisfies these conditions can correlate tion. In simultaneous exposures. per- Any combination of UV. We should not be surprised to learn that effects that we have been unable to duplicate in laboratory weathering conditions must also be the laboratory. exposure. system has three identical colors-green. blue. various accelerated test conditions and Florida Nevertheless. In the first analysis. However. Five systems are used in insulated applica- keep the test specimen wet. The studies involved 27 automotive sures have been commonly used in weathering coatings in 9 proprietary paint systems. Therefore. verti- A persistent myth in weathering literature is the cal South or North in both insulated and open theory that the combination of sunlight and dew in racks to correlate with specific service environ- the early morning causes unexplained synergistic ments. Water. but color change and other effects were noted as they occurred. The other four are intended for use at lower ing occurs in separate UV and water exposures service temperatures and are normally evaluated that are at different temperatures. simultaneous UV and water expo. past seven years in our search for "the answer. Substrates were steel and flexible plas- tics. of racks. all 27 were exposed on both types weathering. Weathering Stresses in the Laboratory Combining UV.Conditions for Reproducing Natural colors ranging from light to dark. 4 UV-4 CON. 20" gloss. The sole property examined for correlation was rized in Table 1. and heat from the UV source makes it difficult to white. service. the basic thrust of our experi. However. For the purposes exposures can and should be used in accelerated of correlation.

20o Gloss Values After two insulated exposures. Such potential 7 .0 requires a linear relation. we can only speculate as to the reason for rate differences between the Table 3 . The 20o gloss values after one year of exposure are shown in Table 3. A claim that accelerated test methods are unreliable can be answered with the counter-claim that Florida expo- sures are not the same from year to year. Perfect correlation under the Spearman method can be achieved without a linear relationship. 11. but there were doubts that one exposure could be dupli- cated by subsequent exposures. 12. At one site. Spearman rank correlation is intended to handle nonparametic data such as visual rankings of appearance or chalking. tory methods and outdoor exposures. Using gloss readings of 47. Three simultaneous Florida exposures were subsequently conducted on the 27 coatings. It is recognized whether gloss values are true interval data and that another operator or another gloss meter could whether linear relationships exist between labora. applied to the analysis. Our company did not partici- pate in conducting or measuring the Florida expo- sure panels and. The Spearman procedure is thus similar to actual practice. How Consistent Are Outdoor Table 2 . rank these coatings differently. rankings are made for each exposure method assumes that the property under con. Our initial tests were compared to a single Florida exposure. series. It is questionable such a procedure is splitting hairs. The Spearman method was selected and all sets of gloss values converted to rankings as shown in Table 2. sure immediately raises the question of whether ship between two sets of data. therefore.45. No matter how much gloss and color difference data is generated.46. to rank coatings 10. Mitton6 has shown how Spearman rank correlation can be applied to evaluating the correlation of accelerated weather- ing methods to Florida for the property of chalking.Ranking of 20o Gloss Values After Exposures? Simultaneous Exposures The first question to be answered was the selec- tion of an outdoor exposure or exposures as the standard for comparison. Pearson linear correla- tion and Spearman rank correlation. Examination of the origi- Simultaneous Exposures nal gloss values at both stations showed the initial Two common methods of correlation could be readings were essentially identical. and 13 in the X expo- Perfect correlation of 1. Differences in rate of failure between exposures are immediately evident. Median gloss is perhaps a better expression of the rate difference but both median and mean indicate exposures Y and Z are the closest in rate. a 5oS open rack ex- posure was also made. and 43 sideration can be measured on an interval scale. The calculations are also simpler. Two were 5oS insulated exposures at different sites in the Miami area. The Pearson In Table 2. a final decision is almost always made by comparing the tests visually.

In order to established by such exposures are quite con- maintain the 1 to 27 rank difference between best sistent. It was concluded that the reason why coatings and relation calculation. all of The term "reversal" has often been used to de- the laboratory tests were correlated to exposure X scribe variation in weathering tests.95 obviously indicate excellent open rack Z exposure. Rank differences below 4 or 5 Florida exposures X and Z (rho X-Z) was 0. Z for the five best systems only X and Y. is a quantitative way to express the magnitude outdoor tests. The classification of the data materials from all over the world are sent to Florida into broader groups will not influence the correla. but the overall rank correlation is quite good. rank correlation comparisons correlation between the two insulated exposures. In Table 5. The rank correlation coefficient derived from this E Blue dropped from 53 to 22 in the insulated X type of relative merit analysis is also relative. Duplicate Florida exposures do contain and worst. Attaining the same level little in a series of 27. Calculating rank differences. "reversals" and differences in rate. 8 (n is the number of items ranked) .90 to 0. of exposures X. the variation in performance of the system E Blue coating. or only that slightly different results are The value of rho between separate Florida tests is obtained. One can hardly expect laboratory of reversals.90. The Spearman rank correlation coeffi- are shown. exposure and dropped only from 68 to 64 on the Values of 0. influence the resulting correlation coefficient very while rho Y-Z was 0. At lower levels. It can mean with some confidence that it represented "reality. The impact of the rank difference on tests to correlate with Florida better than Florida the final correlation is proportional to the square of correlates with itself. This is a comparison of systems that cient is calculated by the formula: are of the same general type used for the same purpose. Y. as in Table a limit for the value of rho between laboratory and 4." that the best coating in one test is the worst in an- other test. and Arizona for exposure is that the relative merits tion coefficient to any great degree. the principal value of rho is in comparing two or more sets of data to an- The completion of the calculation shows better other set. correlation. ties were ranked as shown in Table 2. The difference between the of correlation in a few weeks of laboratory testing open rack and insulated exposure is centered on could be considered real progress. In the last six months of the exposure. Table 4 . In that context.Spearman Rank Correlation for Ranked 20o Gloss Values experimental error is compensated for in the cor.88. The correlation between the rank difference.

rho X-Z from 0. some tests were nate between similar materials.Spearman Rank Correlation for Ranked 20o Gloss Values of 15 Similar Coatings Table 6 . The coefficient is expected to weekly basis. the conclu- sion is the same. Most from 0. Table 6 illustrates how the 27 coatings in 9 Condensation Tests to Florida systems responded to different temperature and Twenty-one separate tests of the 27 coatings were water stresses. # 16.95 to 0. 9 . Rho X-Y drops 168 and 336 hr because of abnormal effects. simply because it's harder to discrimi. The tests were concluded when the be lower when materials of similar durability are median gloss reached a level near the 37 median compared. However. However. concluded at 1000 hr without reaching that level of degradation. A few tests were discontinued at this manner reduces the coefficient.90 to 0.49. Correlation of Fluorescent UV.60. Table 5 . gloss of exposure X. The largest rank differ- was exploratory with 14 of the tests outside of the ences in each test as compared to exposure X are limits now published in ASTM Recommended designated by an asterisk. The program tests #18. The UV source was the same in conducted over a two-year period. 3 to 6 weeks. and #7.88. rho Y-Z from 0. and tests had a length of 500 to 1000 hr.Rank Correlation of Fluorescent UV-Condensation Exposures to Exposure X It may be observed that limiting the comparison in Practice G-53-77. exposures X and Y have the Twenty degree gloss values were taken on a closest correlation.88 to 0.

The 10 principles learned with high gloss coatings were place coatings in the same order without demand- . System E still had a higher ranking than Florida. Estimating the ratio for the systems and colors responded. and temperature stresses may be violated found to be one or two ranks in nine. We are aware standing of the reason for this difference. relationship in fluorescent UV-condensation tests. The was observed. The This test program was initiated by observations of temperature difference between Green and White rank differences between Florida and fluorescent in accelerated testing was less than the difference UV condensation tests on two of the systems. ing stresses must be balanced to compare even slightly different systems. exhibited A test trend of nonlinearity with Florida exposure better rankings in the Green and Blue colors. the magnitude of the reversal was water. A me- Repeated tests on these 27 coatings at various dian gloss of 37 in 500 hr for the entire set would levels of UV. in the study exhibited a fairly consistent rank dif- ference of 1. Cleveland Condensation tests using water ing 18 hr of condensation at 50°C. each individual coating produced acceleration ratios of perhaps 8:1 for the most durable coatings (1) None of the nine systems were appreciably changed in ranking by reducing UV stress using and ratios of 25:1 for the least durable. How- Poor correlation was obtained in test #7 by us. frequently within a week. The Spear- correlation coefficients around 0. environments. Maximizing the levels of stress by Three systems were resistant to such water insulated exposures in Florida and in laboratory stresses. It is doubted that such a difference (5) The color relationship in Florida may gener. (3) Three systems exhibited varying degrees of temperature sensitivity at temperatures within the The magnification of differences by maximizing range encountered in insulated service. did not do so. cannot be drawn for all materials and coating System A was reduced in rank along with system substrates. D. wetness in test #7 was less than in test #16 and Hard and fast rules for accelerating weathering 70°C UV temperature was the same in both tests. black lamps. The tests seemed to magnify the differences between coatings. The time of alone produced the desired change in properties. However. fluores- that they are being violated to reduce test time. test program has never ranked these two systems exactly the way they rank in Florida insulated expo- It may be concluded that the general limits for UV.85. Two almost identical systems expected in service did not influence rankings. Spearman rank correlation has proved useful in handling the observed nonlinearity between accel- By the time we reached tests #16 and #18 with erated testing and Florida exposure. With under- when testing specific systems.62 for X to test #7 is a fair appraisal of the graphically for linearity. in most instances. stresses can be both an advantage and a disad- (4) Three systems. water. and temperature stresses has indicate an acceleration ratio of about 17: 1 when provided enough information to generalize on how compared to exposure X. expected to show temperature vantage. The of other properties could alter relative merit. sure. Acceler. tests could be expected to degrade weaker coat- ings much faster than good coatings. ever. differences between similar coatings ated tests at temperatures higher than might be can be overvalued. System J. such procedures sort out the sensitivity because they are used at moderate weaklings very rapidly. cent UV-condensation tests on both systems are However. However. Good correlation was obtained on test # 18 and applied to a test of low gloss white coatings on test #16 but with different test conditions. ings degraded much faster. of test on fluorescent UV-condensation apparatus in another laboratory gave the same result. the least durable system. The best coatings held poor correlation.26. It was assumed that degra- required twice as much time and did not reduce dation would result in a few weeks but six weeks of the gloss of the five best coatings as much as test only produced a little chalking. would be revealed in service. was Green > White > Blue. and examination ally be expressed as Green 2 White > Blue. Obviously. and every test was examined rho of 0. It can be speculated that nonlinearity also exists (2) Two systems were changed in rank by water with respect to moderate and severe weathering stresses beyond those encountered in service. it is more than evident that weather- being used. Twelve weeks Florida. our confi- man method valued a test procedure's ability to dence had reached corresponding levels. Correlating systems A through gloss longer in relation to Florida and weak coat- E only to exposure X gave an rho of 0. The in outdoor exposure. temperatures in service. Test #16 hardboard substrates.

accelerated weathering variables in natural weathering. News. each should be prepared important. the test established limits for combining in four or five colors.. F. intermittent 49. seasonal. tain levels. ranking more than 30 coatings is cumbersome. No. . B. G. Thresholds of sensitivity References to UV. J . If five systems are under evaluation. (1963). water. With today's materials. natural evolutionary process. there are even duplicate exposures in one area can be answers. It is difficult to conceive of a simple number of samples. As few as seven The work of the Cleveland Society for Coatings samples can be handled by the Spearman method Technology gave a new dimension to weather- but the significance of rho is proportional to the ing research. correlations between different environments can The fluorescent UV-condensation apparatus has be obtained. Controls. A consensus exists that such tests the combined stresses. of water.. 490 time. some are degraded by natural weathering can be accelerated by various water alone. Some materials such as Florida and Arizona. D. Further acceleration is possible by maximiz. is wasted because no deterioration is happening. 633. humid envi- procedures can match the various realities of ronments cause the most degradation. stresses. 35. and most materials are weathered by techniques. Hot. a linear time relationship. W. apparatus in a wood frame as a breakthrough. coatings of known time and temperature to accelerate water effects. No. Conclusion It would be difficult to rank the relative importance Thousands of tests in aggressive environments. natural weather. J. ing UV. 467 quantified. sion is rendered.. water. N. Weathering research in past years has concen- trated almost exclusively on a single weathering stress. 27. B. 27 (1977). It is hoped that the theories and practices proposed in this paper can serve as a Today's coatings and plastics are the products of a framework for such tests. & Richards. The time a material is stressed and the tempera- We conclude that accelerated weathering is a ture during UV and water stresses are the primary reality. and temperature stresses within (6) Mitton. Official DIGEST. but the Cleveland Condensation test showed how However. P. Little attention was given to balancing these stresses. Journal of Paint Technology. water. temperature. Water and temperature stresses were relegated to a second- ary status. Technology. 43. Chem. The basic stresses in weathering are known and (4) Foecking. natural weathering is not a (1) Scott. More precisely. When annual. There isn't one answer. It is a sporadic. A. Reproducing them exactly in the labo. durability. ratory can provide acceleration by eliminating idle (5) Higgins. 15 values can be handled easily and 27 to reproduce the quality of natural water. Official DIGEST. P. and temperature have been raised. E. (3) Likens. Journal of Coatings Technology. Our tests show how violating these limits can cause abnormal degradation. 563 (1971). No. Eng. No. have useful correlations to weathering in more The tests reported here indicate that the most moderate environments. 48. and UV. Maximum UV. correlation procedures. 492. process that occurs only when stresses reach cer. From a practical viewpoint. The weaklings have disappeared from use. More was not an unreasonable number. 29 (1976). 11 . Much of the time on outdoor test racks No. the ultraviolet energy in sunlight. should always be included if available. (2) Stieg. An analysis of natural important factor in accelerated laboratory weath- weathering has indicated the reasons why such ering is the proper balancing of these stresses. 54.. demonstrate that are degraded by UV alone. The best answer for a particular pur- expected to vary in rate. L. NO. and temperature the capacity to vary the balance of weathering stresses are similar everywhere. (1965). Journal of Paint natural limits. continuous process. pose can be determined by relative merit proce- and location rate differences are excluded by rank dures such as Spearman rank correlation. However. 29 (1966). natural accelerated weath- ering is shown to consistently and reliably predict Many years of testing will pass before a final deci- the relative merits of materials.

OH USA info. .Q-Lab Corporation www. Ltd. AZ USA China 中国上海 Tel: +1-305-245-5600 Tel: +1-623-386-5140 电话: +86-21-5879-7970 q-lab@q-lab. Q-Lab Deutschland. the Q-Lab logo. All Rights Reserved. England Saarbrü and Q-PANEL are registered trademarks of Q-Lab Q-Lab Headquarters Q-Lab LU-0824 © 2011 Q-Lab vertrieb@q-lab. FL USA Buckeye. GmbH Westlake. Germany Tel: +1-440-835-8700 Tel: +44-1204-861616 Tel: +49-681-857470 info@q-lab. Q-Lab Florida Q-Lab Arizona Q-Lab China 中国代表处 info.