Colloquium announcement

Faculty of Engineering Technology

Department Thermal Engineering
Master programme Sustainable Energy Technology

As part of his / her masterassignment

Cosmin Iordache

will hold a speech entitled:

Adjoint-based optimization study of a heat exchanger: single and multi-objective shape optimization



An adjoint gradient-based optimization method is used for optimizing a pin-fin heat exchanger mounted in a domestic boiler. The advantage of this method is that an improved shape of the pins can be reached with respect to the operational conditions at the same cost as the CFD simulation. The optimum shape is computed within the boundary that encloses the cross-section of the pins. A modification of this boundary is realized via the adjoint optimization method and a new shape is obtained. This process of shape optimization and its functionality is investigated by optimizing a simplified model of a pin-fin heat exchanger  which represents the main goal of this thesis. This is done via a CFD numerical simulation tool ANSYS Fluent. The study is carried on by optimizing heat transfer rate and pressure drop. The optimization of these objectives is done by analyzing its changes due to a modification in the shape. The changes in flow field will be analyzed as well to understand its effects on the variation of the objectives.

This study used a two dimensional simplified  model for the optimization of each objective. The results show that an extremely accurate converged flow solution as well as using constant thermophysical properties improve the predictability of the optimized objectives. However, the predictability is decreasing as the degree of shape change increases. The predictability is also affected by the choice of discretization scheme used in the flow solver. It was concluded that the pressure drop is more sensitive to changes in shape than the heat transfer rate.

Further study investigates the multi-objective optimization. Different combinations of pressure drop and heat transfer rates have been used as a multi-objective optimization. Results show that optimization of both objectives is possible and that a higher optimization rate is achieved for a certain combination of the objectives. Also, due to large changes in shape after a high number of optimization cycles ( >30) an improvement in mesh quality is required. Moreover, the results obtained from  the two dimensional domain allow for a successful approximation of the three dimensional domain.

The optimization of the three dimensional geometry results in improved heat transfer rate and pressure drop. An influence of the area chosen as an enclosed boundary for optimizing the pins gives different results. A larger area provides more degree of freedom for the movement of the shape boundary, leading to a higher optimization of the objectives. Finally, the end geometry is obtained for the optimized solutions.