Bu çalışmada, türbülanslı boru akışındaki taşınım yolu ile gerçekleşen ısı geçişi zamana bağlı sayısal olarak çeşitli yöntemlerin kullanımı ile modellenip, incelenmiştir. Bu maksatla, muhtelif türbülans modelleri ile cidar yakını modellemesi yaklaşımları denenerek, bu modellerin performansı etüd edilmiştir. Çalışmaya ilk olarak daimi akış koşulu altındaki analizler ile başlanmıştır. Daimi hal için, literatürde yer alan ampirik ifadelerden türetilen Nusselt sayıları hesaplamalardan elde edilen Nusselt sayıları ile karşılaştırılarak, hesap değerleri doğrulanmaya çalışılmıştır. Buna göre, denenen muhtelif iki denklemli türbülans modellerinin birbirlerine çok benzer neticeler ürettikleri, dolayısıyla bu modellerin daimi hal analizlerindeki performansları arasında önemli farkların bulunmadığı gözlemlenmiştir. Zamana bağlı hal analizlerinde ise, daha çok, iki denklemli türbülans modellerinin cidar fonksiyonlarıyla birlikte kullanımına odaklanılmıştır. Bunun yanı sıra, cidar fonksiyonlarının kullanılmadığı iki tabaka bölgeli model de bazı haller için kullanılarak elde edilen çözümlerin iyileştirilebilme olasılıkları incelenmiştir. Basamak şeklinde, darbe şeklinde ve sinüzoidal şekilde akış debisinin değiştiği muhtelif zamana bağlı akışlarda çeşitli sayıda Reynolds sayıları ele alınmıştır. Yürütülen hesaplamalardan elde edilen sonuçlar deneyler ile kıyaslanarak sunulmuş, ve elde edilen hesap neticelerinin ölçüm değerlerini öngörebilme kabiliyeti sorgulanmıştır. Burada, hız ve ısı akısı gibi parametrelerin zamana bağlı değişimlerinin hem nitelik hem de nicelik olarak, özellikle yüksek Reynolds sayılarında, makul bir biçimde öngörülebildikleri tespit edilmiştir. Ancak diğer taraftan, cidardaki ısı akısı cevabındaki gecikme gibi bazı hususların hesaplamalar tarafından yeterince yakalanamadığı da gözlemlenmiştir.

Unsteady turbulent flows and the heat transfer which occur in pipe flows take place in a broad range of engineering devices. Investigation of such phenomena and understanding the underlying mechanisms would be very beneficial for the design of such practical devices, i.g. preswirl systems in gas turbine cooling and inlet manifolds in automotive engines. Analyzing the fluid flow problems by numerical simulations has gained such maturity within the last decade that they are utilized as an important design and analysis tool in a wide and ever increasing range of applications. Likewise, convective heat transfer is also analyzed using computational methods. Therefore, validation of numerical methods is a crucial matter for the modeling of such phenomena accurately. Validation of computational procedures has been carried out by some authors, but they were rather interested in solving steady-state problems. Therefore, there is a lack of validation studies of numerical predictions for unsteady convective heat transfer problems. This is the scope of the present investigation. In this study, unsteady turbulent pipe flow which includes convective heat transfer has been investigated computationally. The numerical predictions have been validated by considering the recent experiments of Barker and Williams. This numerical study has been carried out using the general purpose code Fluent as basis, which utilizes finite volume method to discretize the governing equations, and a pressure correction formulation to handle the pressure-velocity coupling. Ensemble averaged continuity, Navier-Strokes, and energy transport equations have been solved for the incompressible, unsteady, 2D, axisymmetrical, turbulent pipe flow. Various two-equation turbulence models have been tested and examined in different computations in order to validate the models, since they are often used in industrial applications. Those are, namely, the standard k-ε, RNG k-ε, and Realizable k-ε turbulence models. A one-equation model has also been used in steady-state analysis, for comparison. As the modeling of near wall region is very important and have influence on convective heat transfer, different models have been employed. Those are, namely, some formulations including the standard or nonequilibrium wall-functions, and two-layer-zonal methods. Although the two-layer methods which resolve the near wall region with fine cells can principally produce better predictions the wall functions approach has been mainly used in this investigation because the two-layer methods demand high computational costs, and their use is still restricted to practical applications. For steady-state analysis, the predictions have been validated by empirical correlations. The predictions agree quite well with the empirical values where no substantial differences were observed in performance of two-equation models considered. On the other hand, the two-layer methods did not offer a significant advantage over wall-function approach. The transient analysis has been carried out with step-like, pulse-like and sinusoidal perturbations of the flow rate at different Reynolds numbers, by mainly using, the standard k-ε turbulence model with the standard wall-functions. The computational results have been compared with the experiments. It has been observed that in step-like changes the amount of wall heat flux could be predicted well, but the time delay of the heat flux response to the change of flow rate could not be predicted. In pulselike changes the peak values and the variation of the wall heat flux with respect to time could be predicted well, especially at higher Reynolds numbers. However, the time delay of wall heat flux response to the change of flow rate has been fairly under-predicted. In sinusoidal perturbations, ensemble averaged measurements of wall heat flux could be predicted reasonably well. However, the deviation between the peak values of the computed wall heat flux and the measurements became greater with increasing relative amplitude of the flow rates. Computations could not predict sufficiently the form of the measured ensemble averaged heat flux curve with respect to time, which deviate from the sinusoidal shape, especially at highest relative amplitude of the flow rate. This might be due to low Reynolds number effects around temporally minimum flow rates during oscillations. It has been shown that the mean computed wall heat flux was not influenced by sinusoidal perturbations, and this finding was also in agreement with the experiments very well. Two-layer-zonal methods did not improve the observed time delay between the wall heat flux and the centerline velocity for the considered cases.