In a Pressurized Water Reactor (PWR), the rod ejection is a design basis accident for uncontrolled evolution of the nuclear reaction.In case of failure of a rod mechanism, the rod ejection is caused by the pressure differential between the primary loop (155 bar) and the confinement-s enclosure (atmospheric pressure).It leads to a local power transient and a fast fuel temperature increase.The power transient is limited by the reactivity feedbacks before the automatic reactor shutdown.The CABRI experimental pulsed reactor is funded by the French Nuclear Safety and Radioprotection Institute (IRSN) and is operated by CEA at the Cadarache research center.It is designed to study fuel rods behavior under Reactivity Initiated Accident (RIA) conditions.The tested fuel rod is placed at the center of the CABRI core, inside a pressurized water loop reproducing PWR conditions.CABRI is a pool type reactor, made of 1487 UO$_2$ fuel rods and controlled by 6 Hafnium control rods.A specific device allows the fast depressurization of $^3$He contained in 4 transient rods to reproduce control rods ejection conditions.Based on a BEPU approach, we developed a tool, named SPARTE, for CABRI power transients calculation.This tool is based on point kinetics, simplified thermal-hydraulics and thermal-mechanics.It computes the global behavior of the core by the calculation of a mean fuel rod. It includes models of reactivity insertion specific to the CABRI transient rods system, variable kinetics parameters and variable Doppler coefficient.This code is validated on the basis of 66 CABRI start-up power transients realized during the first quarter of 2017. One goal of the SPARTE code is to be used for the prediction of future CABRI power transients.This paper focuses on methods for optimizing a specific CABRI power transient (FWHM $\simeq$ 30 ms, Deposited energy $\simeq$ 130 $MJ$) using the target characteristics of the pulse. The selection of a method may help the experimentalists and the operation team to minimize the number of white- power transients to perform before the final test with the fuel sample. The optimization can lead to different results, that can be ranked according to their projected uncertainties. Different optimization methods are tested and compared in this paper. The Subplex method based on reiterations of the Nelder-Mead algorithm (simplex method) was selected for its high precision. Indeed, the CABRI power transients are not completely reproducible and present some uncertainties linked to the test parameters. This article focuses on the uncertainties propagation in order to identify and select the parameters that minimize the output uncertainties. The results are very satisfactory and lead to several optimized scenarios that will be tested during the next qualification test campaign.