Flat heat sinks are devices for efficient heat dissipation and are important components in manufacturing. For example, in precision glass molding, it is utilized as a cooling plate contributing to improve the quality of the product by reducing residual stress and shape deviation, during the gradual cooling phase. In this study, we attempted to achieve precise temperature design of the cooling plate using variable lattice optimization. Fins allow for controlling the flow direction and achieving efficient heat transfer. Therefore, elliptical cylindrical fins were adopted as lattice unit cells, increasing the design freedom of the flow field. To account for the asymmetric flow field within the unit cell, we introduced an asymmetric tensor for the coefficients of the effective physical properties in the Darcy–Forchheimer equation. The proposed method's effectiveness and validity are discussed through numerical calculations with effective material properties, reproduced detailed shapes, and experimental verification. The optimization aiming to minimize the average temperature yielded a numerical simulation temperature decrease of 24.6 % and an experimental temperature decrease of 14.1 % compared to the benchmark model.