The paper presents a thermodynamic optimization of 900MW power unit for ultra-supercritical parameters, modified according to AD700 concept. The aim of the study was to verify two optimisation methods, i.e., the finding the minimum of a constrained nonlinear multivariable function (fmincon) and the Nelder-Mead method with their own constrain functions. The analysis was carried out using IPSEpro software combined with MATLAB, where gross power generation efficiency was chosen as the objective function. In comparison with the Nelder-Mead method it was shown that using fmincon function gives reasonable results and a significant reduction of computational time. Unfortunately, with the increased number of decision parameters, the benefit measured by the increase in efficiency is becoming smaller. An important drawback of fmincon method is also a lack of repeatability by using different starting points. The obtained results led to the conclusion, that the Nelder-Mead method is a better tool for optimisation of thermal cycles with a high degree of complexity like the coal-fired power unit.