Teras çatılardaki çok katlı su tutucu tabakanın servis ömrünü tahmin için matematik model oluşturulması

Abstract

0ZET Bu çalışmada, teras çatıl ardaki çok katlı su tutucu tabakanın servis ömrünü tahmin için b.ir matematik model oluşturulması amaçlanmıştır. Su tutucu tabaka diğer tabakalarla birlikte ele alınarak, eskitme süreçleri ve kontrol deneyleri olmak üzere iki ana grupta deneyler yapılmıştır. Eskitme süreçleri, numunelerin saklama koşullarına göre, 1) Klima koşullarında saklanan, şahit grup, 2) dış ortamda saklanan, gerçek hasarın belirlendiği grup, 3) hızlandırılmış etkenlerin etkisinde saklanan ve hasar etkeni katsayılarının belirlendiği gruplardır. Kontrol deneyleri ise. 1) Çıplak göz ile gözlem, 2) su geçirimliliği deneyi, 3) çekme deneyleridir. Dış ortamın atmosfer etkilerinden sıcaklık ve bağıl nem değişimi, yağış ve güneş radyasyonu esas alınarak, bunların etkileri, ıslanma-kuruma (HG1), ıslanma-ısı-nem (HG2), donma-çözülme (HG3), yüzeysel ısınma-soğuma (HG4), ültraviyole radyasyon etkisi (HG5) olmak üzere beş grupta incelenmiştir. Bu etkenlerin etkisinde saklanan numünelerdeki hasar oranları belirli tekrar sayısı veya etkileme sürelerinden sonra kontrol deneyleri ile araştırılmıştır. Su geçirimliliği deneylerinden, sadece donma-çözülme grubun dan anlamlı sonuç alınabilmiştir. Çekme deneyi sonuçlarından elde edilen çekme dayanımı (veya tokluk özelliği) ile tekrar sayısı (veya etkileme süresi) arasındaki regresyon bağıntılarından, süreçlere ait birim hasar oranları (efet) belirlenmiştir. Matematik modelin oluşturulması için deneysel sonuçlardan başka, uzun süreli gözlemlerin ortalaması olan meterolojik verilerin tekrar sayısı (n.) veya etkileme süresine (hA dönüştürülmesi için bazı varsayımlar geliştirilmiştir. Teorik toplam hasar oranı ile gerçek hasar oranı arasındaki farkı gidermek için birinci aşama formülünde tekil düzeltme çarpanlarına gerek duyulmuştur. Deneysel sonuçlar, bunların belirlenme si için yeterli olamadığından mevcut deneysel veriler ile sadece İstanbul ili için global düzeltme çarpanları belirlenmiştir. İkinci aşamada, servis ömrünün tahmini için iki yöntem önerilmiştir. Birincisi, ortalama servis ömrünün tahmin edildiği determin- istik yöntem, ikincisi ise ortalama servis ömründen başka, belirli bir güvenirlik düzeyi ile maksimum ve minimum servis ömrü değerlerinin de tahmin edildiği stokastik yöntemdir. Her iki yöntemde de servis öm rünün tahmini için kritik hasar oranının veya servis yeteneğinin ka bul edilebilir minimum seviyesinin belirlenmesi gereklidir. Bu çalışmada, donma-çözülme hasar grubundaki su geçirimliliği (Wp), çekme dayanımı ve tokluktaki hasar oranları (Dgf, Dgt) ile tekrar sayısı (N3) arasındaki korelasyondan yarar! anı Tarak hasar oranları ile geçirimlilik arasında ilişki kurulmuştur. Bu ilişkinin, diğer gruplar için de geçerli olduğu kabul edilerek servis ömrü değerleri tahmin edilmiştir. xıv

SUMMARY THL PREDICTION OF SERVICE LIFE OF WATERPROOF LAYERS ON FLAT ROOFS Materials used under external conditions are confronted with deterioration caused by the environment as soon as they take their place within the structure. Depending on the properties of the materials and on the power of the environmental effect, materials become damaged gradually. As damages increase, materials lose their original characteristics and fail to perform their functions within the system. Although designers and constructors are well aware of the qua! il ati ve and quantitative characteristics of the materials used, they still are not able to know how these characteristics will change with the time and how long the service life will be. Damage analysis, durability studies and other studies are carried out to predict the service life of the materials and to obtain this information in advance. For this purpose, experiments should be performed both in laboratories and in open area, the nature-laboratory relations should be determined and the conditions which the materials may support during service should be estimated beforehand. Although the laboratory conditions can be taken under control, the control of natural conditions are not possible. Therefore, the results obtained from the laboratory are assumed as estimated values with a certain degree of reliability. Flat roofs which are structural elements open to external effects, are damaged in a short period of time and let to penetrate the water. This is the reason why waterproof materials on flat roofs are of special importance among materials used in external environment. The aim of the present study was to predict the service life of waterproof materials, whereby/ the waterproof layers have been considered either independently or together with the thermal insulation. During these studies, tests to predict the service life have been performed separately for each damage effect in the laboratory. According to these preliminary studies, relations between nature and laboratory, tests have been emphasized, but unfortunately no specific method has been proposed yet. The thesis is devided in five main sections. In the first section the importance of the subject, the objective and the aim of study have been explained. The second section deals with the general principles of studies carried out for the prediction of service life. For this purpose some main concepts on subjects, such as service life, durability, service capability and degradation factors have been defined. Various stages of studies regarding the subject have been xvlisted, certain hypotheses used throughout the experimental process have been explained and. criticized. Furthermore degradation factors are described and classfied. To evaluate the results of service life tests Value Analysis 1 and Value Analysis 2 Methods are described first. Then the mathematical model approach has been explained in general terms and concepts of deterministic and stochastic models have been given together with some related examples. In this section also previous research made on the subject has been summarized; types of flat roofs, flat roof layers, materials used for the layers and the properties of materials have been explained in general terms. Brief general information is given on the characteris tics of waterproofing cover materials, on hydrocarbon bonding and also on the rules to be followed in the application of these materials. In the third section the experimental studies are described. Information about the testing characteristics. of the materials, the materials used in the experiments, and the equipment or the tools especially developed for this study have been given. The tests may be divided in two categories ; the natural weathering processes and the laboratory control tests. Weathering out proceses have been realized in three main groups, depending on the environmental conditions in which the materials were kept. The first is so called control group which was kept under constant temperature and humidity in the laboratory for a period of 12 months. The second group is, that was kept in atmospheric conditions of the external environment of the laboratory for a period of 25 months. The third and the last group is, that has been kept in the laboratory under strengthened and accelerated atmospheric conditions than the natural weathering conditions. The effects of the variations in temperature and relative humi dity and exposure to the radi ati on of the atmosphere on the material properties have been investigated and their complex and interactive effect has been examined by linear superposition of the simple effects. These phenomena have been divided into five groups of damage factors, such as wetting-drying (HG1), wetting-heat+humidity (HG2), freezing-thawing (HG3), surface heating-cooling (HG4), and ultraviolet radiation (HG5). In the laboratory, equipment with similar effects like these factors have been set up, and separate groups of samples have been kept under these separate effects. After specified frequencies or durations, control tests have been performed to determine the amount of deterioration due to the effects. The details of the conditions of accelerated damaging processes were as follows : Only the surface of the waterproof layer samples in the wetting- drying damage group have been in contact with water of 20- 4 C in the wetting phase. In the dyring phase, these samples have been dried in oven at 65 C temperature. xviSimilary, the surfaces of the samples in the damage group subjected to the wetting-heat-huraudity effects, have, also heen in contact with water of 20*4 C in the wetting phase from their upper surface and then haye been kept in an ayerage tempareture of 70°C and a relative humidity of 70-80 % in a huraudity controled oven-. Samples of the freezing-thawing damage group have first been wetted with 20°C water and than frozen in a cooler at -15°C, finally they have been thawed in water at 20°C. This was repeated with each cycle having the same specified duration programme. For the surface heating-cooling damage group the waterproof layer samples were placed in a box especially designed for this damage process, and have been heated from their surface to about 70°C temperature for appraximately 8 hours a day. In the damage group with ultraviolet radiation, the samples have been placed under the effects of an especially developed equipment consisting of 16 ultra-violet bulbs. The bulbs have been kept on approximately 8 hours a day, and meanwhile, the surface tempareture of the samples has been raised to 70°C. In the combined application of wetting-drying, wetting-heat- humidity and freezing-thawing processes, the simple application of each phase was assumed as one cycle, also control tests have been performed after the 10 th, 20 th, 30 th, 40 th, 50 th, and 60 th cycles. In the combined application of surface heating and ultra violet radation exposure tests where the fundamental point was the duration of the damaging effect, control tests have been made after duration periods of 500, 750, 1000, 2000 hours. The same control tests have also been performed on the control group, after duration periods of 12 months, and on samples in the external environment after durations of 12, 18, 20 and 25 months.. For each group, three different control tests have been performed after specified cycles or damaging periods. First, visible changes on the sample surfaces coused by the damaging processes, have been examined visually. Secondly water penetration tests on the damaged elements have been performed under an atmospheric pressure of 0.15 at. Significant results could be obtained with these two methods only for the freezing-thawing process. For the other processes, the tests could not be extended in time to obtain significant results. Tensile tests have been made on samples, cut and prepared from the multi -layered waterproof ings. Results obtained from these tests have been evaluated according to the mean tensile strenghts (a.) and ultimate tensile strains (O. To eyaluate the specific toughness of the multi -layered waterproof ings, the areas under the stress-strain curves have been calculated. Variations of the mean tensile strength and the specific toughness versus cycles or duration of exposure were examined with linear regression analysis. xvi iThe coefficients of regression obtained for each damage process were designated as `damage rates` e^ or e^. They permi ted the estimation of loss of strength or toughness for one cycle per unit period of exposure. In the fourth section of the thesis » test results have been evaluated and, a mathematical model has been developed. First, a relation has been established between natural and laboratory conditions. The averages of a period of 42 years of meteorological observations have been used for this purpose. Some assumptions have been developed in order to obtain from this data? the annual frequency or duration of damaging events for a given region. An important difference is noticed between the real rate of damage in nature and, the theoretical annual total damage calculated from the assumptions and results obtained from accelerated and intensified processes. In order to eliminate this difference, separate correction factors for each damage group were necessary. But the ex perimental, results of this study were not sufficient to determine them. Instead of separate correction factors a global correction factor has been determined only for Istanbul. Thus, the following formula representing the first step of the mathematical model was developed and used in the calculation of the theoretical' annual rate of total damage: D f -ef -J.df,. n-, +ef 2-df o. n2+ef3 «dfo. n2*e^A *df4 * /*e^ 5 -^5 *n5 Dat=etl.dtvni +et2.dt2.n2+et3.dt3.n3+et4.dt4.h4+et5.dt5.h5 In the second step of the model, two methods have been proposed for the estimation of the service life. The f first of them is the determistic approach to calculate the average service life, ît is also possible to calculate the maximum and minimum service life values with a selected level of reliabilty. To determine service life, it has been necessary to determine first a critical level of damage. The critical damage level depended on water penetration, which was the main property of a waterproof material. But actual results of water penetration test have only been obtained for the freezing-thawing damage group. Thats why a relationship between tensile strength (or toughness) and water penetration could be set up for this damage group. Nevertheless with decreasing tensile strength a reduction of waterproofing level has been assumed for all damage groups, and for waterproofing which is the main function of material, the critial damage level has been replaced by the critical damage level of tensile strength (or toughness). In the determistic approach separate service lives have been determined, for each damage group, depending on the level of critial damage and on the meteorological characteristics of the region. xvmAlso an average theoretical service life has been calculated using the Value Analysis 1 Method. This theoretical service life is then divided with the global correction factor to find the actual service life. In the stochastic approach, the case proceeds in compliance with the Gauss' normal distribution law and aritmetic means and standart deviations are assumed to vary as linear functions of actual time. In the experiments there is no actual time, but there are frequencies or damaging periods. The transformation of these values into actual time could be done only, basing on the meteorological data of a specific region as follows: f= a.t+f0 where s= p.t*-q f: arithmetic mean of tensile strength s: standart deviation of tensile strength t: actual time Following assumptions have been made to determine the constants `a,f0, p, q` in the linear expressions of the aritmethic mean and the standart deviation versus time. The constants `fQ and q` may be determined for t=0 from the results of the control samples not worn out. The constant `a` has been obtained from the relation of the theoretical service life calculated by deterministic means to the theoretical service life, based on the critical tensile strength. The constant `p` has been obtained from the frequency distribution of the tensile strengths of samples kept in the external conditions. The first step formula of the model has also heen applied besides İstanbul, for Izmir, Antalya, Urfa, Trabzon and Ankara as well. The methods proposed in the second step have been applied for Istanbul, for which a global correction factor has been determined. In the fifth section of the thesis, the details of the flat roof element and the structure and properties of the model have been summarized. The results obtained have been presented and some proposals have been made to encourage the development and wider use of the model. xix