Nuclear reactors exhibit excess reactivity at start-up to ensure continuous operation over the length of the fuel cycle. For sodium fast reactors, this excess reactivity should cover burn-up reactivity loss, operation margin and uncertainty margin. However, contrary to light water reactors, control rods are the only available mean of reactivity control, boron dilution in sodium being not possible. Therefore, at the beginning of cycle, one part of control rods should be inserted into the core to balance this excess reactivity. The control rods are then withdrawn slowly during the cycle to compensate for burn-up reactivity loss. The malfunction of a control rod mechanism would lead to a so-called control rod withdrawal (CRW) accident that is considered as a typical event for unprotected transient over-power. This event could lead to the local melting of fuel assemblies and even to the global melting of the core. As a consequence, this accident must be evaluated at the core design stage to ensure good margins. This paper proposes to use the newly deterministic code APOLLO3 to optimize the model of the con-trol rods, the transient calculation code MAT4DYN to calculate in the core response to a CRW, and the GERMINAL code to study the fuel pin thermal-mechanical behavior during incidental conditions. This methodology is applied in a small sodium fast reactor that has an important reactivity loss and thus a high excess reactivity at start-up. The space for the implementation is limited by its mechanical motors especially for small reactors. To achieve the objectives defined for Generation-IV reactors, the CRW accident becomes the limiting factor for small modular fast reactors by comparing with other requirements such as maximum fuel burn-up. Three different options are proposed and studied to obtain core designs with a favorable behavior in case of CRW accident. The first solution is to reduce calculation uncertainty, but this is long process. The second solution is to enhance Doppler constant. The last solution is the application of new system, such as burnable poisons, to compensate for reactivity loss. This paper investigates the required ability of such potential systems and its impact on the core design.