In smoke pipe boilers, the thermal efficiency of the boiler depends on the smoke pipe diameter, smoke pipe length and the heat transfer between the smoke pipe and the outlet chimney. If the heat in the smoke pipes is effectively transported through the pipes, the heat distribution on the surfaces is balanced and the thermal efficiency of the boiler increases. In this study, the improvement of heat transfer in a solid fuel boiler with 125,000 kcal / h heat capacity with a diameter of 42 mm, chimney diameter of 230 mm and water inlet and outlet diameters of 65 mm was investigated by using 4 different types of strip turbulators. Experiments were carried out with turbulators placed in all the smoke pipes in the boiler. Firstly, experiments were carried out without placing a turbulator inside. In the second step, by placing turbulators in the smoke pipes, experiments were made for each type and heat transfer was calculated. In the experiments, the flow rate of the fan was changed with the help of damper and the reynolds number was calculated between 18000 and 28000. Turbulator experiments for heat transfer improvement have increased by at least %15 and at most %41 compared to turbulator free experiments. For the heat transfer increase values obtained because of calculations, predictive models were obtained using machine learning algorithms SVM (support vector machine) and decision tree (M5P model tree). The resulting models have been analyzed for error analysis and have been shown to successfully predict heat transfer increase values.