Yapı sistemlerinin karakteristik parametrelerinin tanımlanması

Çağlayan, Barlas Özden
Süreli Yayın başlığı
Süreli Yayın ISSN
Cilt Başlığı
Fen Bilimleri Enstitüsü
Yapı sistemlerinde, üzerlerindeki servis yükleri ile birlikte çevre koşullarının neden olacağı, diğer yüklere maruz kaldıklarında, hasarlar oluşabilir. Oluşan bu hasarlar ilerisi için daha büyük hasarlara yada göçmelere neden olacaktır. Bundan dolayı günümüzde hasar tespit yöntemlerine oldukça rağbet vardır. Bu çalışmada, taşıyıcı sistemlerdeki rijitlik değişimini ve bu değişimin yerini saptayabilen bir yömtem geliştirilmiştir. Bu yöntemin etkinliği ve güvenirliği bir prototip üzerinde sınanmıştır. Elde edilen datanın geliştirilmiş olan yöntem ile değendirilmesi sonucu, prototip taşıyıcı üzerinde sonradan kaynaklanmış olan levhanın neden olduğu rijitlik artışı ile yerinin saptanması mümkün olmuştur.
Structural systems such as buildings, offshore structures, and aerospace structures accumulate damage under service loads, including environmental conditions. Damage may cause more damage or failure if it is not detected and not repaired in time properly. In the last several years, application of the recursive identification methods for structural parameters in the tune domain has attracted considerable interest [5,6]. This is due to the limited economic resources available all over the world to rebuild the old structures instead to rehabilitate. An adequate policy concerning both maintenance and rehabilitation can only be achieved with important investments. Thus, when a decision concerning rehabilitation or replacement or any maintenance has to be made,the inspection and evaluation of existing structure to determine the actual resistance and the other parameters are very important. Of course, damage will result degradation of system characteristics, such as stiffness or damping. As an example, from 50% to 70% stiffness degradation was observed based on the data obtained from the Mülikan Library Bunding, located on the campus of the California Instate of Technology during the 1971 San Fernando earthquake,Udwadia and Jerath and Lemura and Jennings [1]. Several authors [2] have also indicated that stiffness degrades in both full-scale structures and small models as a result of seismic damage. Ogawa and Abe and Carydis and Mouzakis attampted to correlate stiffness degradation to the severity of damage. Based on all these observations, structures under strong enviromental loads are thus expected to undergo nonhnear and time-dependent degrading behaviour. Hence,time-varying behaviour of system parameters can occure, and on-line identification becomes a real issue under these conditions to permit real-time corrective action, repair, and control to rm'nimize the possibility of further damage. To assure structural safety, a real-time, rapid, and also remote structural-damage-detection capability is required. In this study an efficient and reliable real - time identification procedure is developed to identify the stiffness changes in any structure whose efficiency and validity are then examined by employing prototype tests. The prototype structural system selected for this purpose is a simply supported single scan steel beam having 230 cm. span length and 2.18 cm. by 10 cm. cross sectional dimension. The Servo(Terra Technologies, Inc.) accelerometers were used to measure accelerations on the prototype structure from predetermined points. They have a referace sensitivity at 50 Hz, 2.5 volt/g scale factor and damping coefficient of 0.7. The accelerometer were mounted on a heavy steel base having three short pointed legs and fastened to the beam at predefined points with clamps (Fig. A). Figure A. Accelerometer The response coming from the transducers while the dynamic tests was taken by lead cable back to the center. The cables were in standard length of 50m and terminated with switchcraft connectors at each end. It was a simple matter to plug a cable connector into the mating connector on an instrument, lead the cable through a small hatch in front of data collection center. As the data acquisition system, Keithley-500 series were used. AMM1A, Master Analog Measurement Module combines three important Series 500 functions into a single module. Firstly, the AMM1 A functions as a standard analog input module will accept up to 16 single-ended or eight differential analog input signals. It contains signal conditioning and switching circuitry for these channels. Secondly, the AMM1A selects and conditions analog signals from other analog input models in a Series 500. Lastly, the AMM1A serves as a 12-bit A/D converter for its own analog input channels, as well as any other analog signals which have been processed by the global select/conditioning circuitry. After analog conditioning, signals are routed to the A/D converter section of the module for the analog-to-digital convertion process. These analog input channels can be conditioned with programmable local gains of either xl or xlO. Global conditioning consists of a high-speed software-controlled gain amplifier with programmable xl, x2, x5, xlO gain values. For A/D convertion, the AMM1A uses a 12-bit successive approximation converter. A maximum conversion time of only 20 mic. sec. allows sampling rates as high as 50 kHz. To maximize resolution, the AMM1A has 0-10V and 10V A/D converter ranges which are software selectable. Also a programme was written for data capture by using 500 Series functions. VI Ivnuı S?İ£*MdUijr Figure B. Test System Collected data were recorded in a portable computer (Toshiba 3200) which has 8MB RAM, 40MB hard disk, 80387 16 MHz Math co-processor and intel 80386 16MHz CPU. A typical configuration of this equipment used during the test is described below. A set of preliminary studies were started and completed before tests. A two dimensional computer model of the beam was set up. In this computer model beam is assumed to be composed of 5 two dimensional beam type elements. To obtain the condenced stiffness matrix of the beam, according to the measured direction,unit forces were applied to the system (to the needed joints). Using COSMOS/M finite element analysis program, static analysis have been performed employing this computer model to determine the joints' displacements caused by the applied forces. Using the following well-known relation, inv(d) = k where d is the displacements of the system caused by the related force k is the stiffness matrix condenced stiffness matrix was calculated. During the test, impulsive load is used as an excitation mechanism as is shown in Figure c. Data were captured from four different points and saved to the hard disk in the form of voltage differences, and as its scale factor is 2.5 V/g, a programme was written to convert data from voltage to acceleration. The system parameters identification produced in this study needs correct time histories of accelerations, velocities and displacements of the considered structural system. Actually, to obtain these values is not possible by employing the standard filtering and trend removing techniques and considered as a up-to-data research topic by several researchers. As the first step in this study, an efficient algorifhym is developed for this porpose whose efficiency and validity were checked by performing laboratory Figure C. System Excitation mechanism prototype tests employing a displacement transducer besides the acceleration transducer located on same point of the prototype beam considered ( Figure C ). Equation of the motion for the free vibration of any system canbe expressed as, [m]x"+ [c]x'+ [k] x =0 (1) in which [M], [C] and [K] are diagonal mass, damping and stiffness matrices of the system respectively. On the other hand x, x, x represent system acceleration, velocity and displacement vectors. For i-th degrees of freedom of the system Eq.l can be written in the following form, * n ?, n Xi == 2j cijXi X kijxj mij J J mij J J in which n indicates number of degrees of freedom considered for the system. (2) Thus, after having obtained correct velocity and displacements from the measured acceleration data collected for the considered degrees of freedom of the system it is possible to calculate system accelerations from Eq.2 with the correctly known system mass, damping and stiffness values. Any error in prediction of these values results in the diferences between measured and calculated accelerations from Eq.2 viu E = xim - xk (3) From this point in the analysis, methodologies from the system identification literature are applicable. The criterion function for our application is defined to be E2 =£ (xkn -'İle)2 (4) This represents the total squared error. The values of mass or stiffness parameters sought, are those which minimize Eq.4. The criteria for minimizing E is obtained by taking its partial derivatives with respect to each structural parameter to be obtained and setting it to zero, BE2 = 0 k=l....N (5) in which & denotes k-th structural parameter to be defined, N is the total number of parameters to be defined. Eq.5 leads to a set of N-coupled nonlinear equations in £*.. These nonlinear equations possess different roots, only one of which is the root sought. Since to identify relative stiffness upgradation or degradation is quite enough to predict the occurrance and its location, in this study, damping coefficient matrix is eliminated. Thus, depending on the acceleration data collected from the dynamic tests performed on the prototype considered for this study ( see Figure C ) and employing the identiflcationprocedure developed herein, the stiffness parameters of the prototype beam considered in this study was identified within reasonable accuracy margin. After having locally stiffened the prototype beam (see Figure C ), similar tests and procedure were employed and the stiffness parameters were identified for this beam which helped to define stiffness upgradation and its location properly. Obviously, the above mentioned procedure can be equally employed for the large structures such as bridges, multistory buildings or offshore structures. Thus, it is planned to use these procedures on the data collected from the dynamic bridge tests performed on the seven old railway bridges located in the various parts of the Turkish Railways Network. These tests were performed in 1991 and 1992 summer in the frame of TU-850-BRIDGES Research Project which is still being carried out with the colaboration of ITU and TCDD, under the sponsorship of NATO Science For Stability Research Programme. The equipment, software and computer facility employed in this study were provided by above mentioned NATO grant. This is why, the valueable contributions of above mentioned institutions are gratefully acknowledged. rx BÖLÜM 1 GİRİŞ Her mühendislik yapısının gerçekleştirilmesi, projelendirme adımı ile başlar. Projelendirmenin çekirdeğini ise modellemenin oluşturduğu bilinmektedir. Bu aşamada, yapıda kullanılması planlanan malzemenin fiziksel özelliklerinin yanısıra, etkiyen dış yüklerin yapı elemanları arasındaki paylaşımının ve buna benzer her türlü durumun ideal şartlar içinde yer aldığı veya oluştuğu kabul edilmektedir. Böylece ele alman yapının hesapları ve boyutlandırılması, bu kabul gözönünde tutularak yürütülür ve yapının etkiyen yükler altında kendisinden beklenen görevi yerine getirecek şekilde tamamlanır. Modelleme aşamasında yapılan ideal kabullerin, ele alman yapının veya malzemenin doğasındaki belirsizlikleri tanımlamada yeterli olamayacağı aşikardır. Nitekim deneylerden elde edilen ölçümlerde hiçbir zaman, yapılan modellemenin tamamına uygun beklenen çözümler elde edilememektedir. Öte yandan, projecinin ve gerçekleştiricinin de bazı hatalar yapabileceği gerçeğini gözden uzak tutmamak gerekmektedir. Ayrıca, kullanılan yapı elemanlarının kusursuz şekilde düzgün eksenli olması, mafsallı birleşimlerin mafsallı, rijit bağlantıların ise gerçekten rijit olarak çalışmasını sağlamak da mümkün olamamaktadır. Tüm bu ve buna benzer nedenlerden ötürü yapı, inşaası yeni bitirildiğinde dahi yük taşıma kapasitesi açısından projelendirme esnasında tasarlanan yapıdan farklı bir durumda olacaktır. Arada oluş an bu fark, yük taş ima kapasitesinin artış ı doğrultusunda ortaya çıkabileceği gibi, bazen tamamen tersi bir duruma da raslamak mümkün olabilmektedir. Örneğin, 1971 yılında meydana gelen San Fernando Depremi esnasında California Teknoloji Enstitüsü ne ait Millikan Kütüphanesi yapısından elde edilmiş datalara dayanarak, Udwadia ve Jerath [1] ve de Lemura ve Jennings [2] sözkonusu yapıda, %50 den %70 e kadar değişebilen bir rijitlik azalmasını tesbit etmişlerdir. Ayrıca bazen, özellikle inşaat mühendisliğini yakından ilgilendiren yapılarda, daha önceden tüm ömrü boyunca yapıya etkimesi ihtimali düşünülen veya varsayılan dış yüklerin çok üzerinde dış kuvvetlerin de, ele alman yapıya
Tez (Yüksek Lisans) -- İstanbul Teknik Üniversitesi, Fen Bilimleri Enstitüsü, 1993
Anahtar kelimeler
Parametreler, Yapı sistemleri, Parameters, Yapı sistemleri