Abstract <p>This paper presents an algorithm for the parametric identification of the heat transfer coefficient as a function of temperature in an evaporative heat exchanger operating under high vacuum conditions and subjected to a quasi-periodic specific heat flux. The problem is solved as a global minimization task by minimizing the mean squared error between the theoretical and experimental temperature fields at the locations of temperature sensors. To achieve this, the “direct” heat transfer problem is first solved in the chosen formulation with a given initial approximation of the heat transfer coefficient and its basis function, which accounts for its temperature dependence. The conjugate directions algorithm is selected as the minimization method, being the most accurate first-order gradient method with convergence. The second step involves finding the components of the gradient of the residual functional based on the solution of the adjoint heat transfer problem, obtained by differentiating the target formulation of the “direct” problem with respect to the parameterized heat transfer coefficient. The third step is to determine the descent step used in the proposed optimization method, based on the iterative regularization method. The systematic error in the measurements of the experimental temperature field is chosen as the stopping criterion for the iterative identification of the heat transfer coefficient.</p>