Computational fluid dynamics (CFD) simulations of red blood cells (RBCs) in flow require robust validation of the behaviour of individual cells before progressing to multi-cell simulations. The optical tweezers case is often used for validation as robust experimental data is available and simulations can be performed both with and without modelling the internal cytosol and surrounding plasma fluid. The cell loading approach in optical tweezers simulations is a point of interest. A transient graphics processing unit (GPU)-based lattice Boltzmann-immersed boundary (LB-IB) CFD solver coupled with a spring-network RBC model was used for examining loading approaches. The model allows for simulating RBC behaviour both with and without the fluid solver, with a membrane viscosity model implemented for the standalone RBC solver (which was used for this work). The number of nodes on the RBC model surface was 642. Three loading approaches were studied, all involving the application of 10 and then 20 pN loads each over 0.2 s to specified nodes at either side of the RBC: instantaneous loading at the start of the simulation with an RBC settling time, gradual loading using divisions of the original load at every time step without any settling time, and a hybrid approach involving gradual loading over 0.1 s with 0.1 s of settling time. The results were of interest to determine if the gradual loading approach could be successfully applied and if the results were more or less accurate than the previously validated instantaneous loading approach. Gradual loading without settling resulted in a reduction of 0.82% for the axial diameter and an increase of 0.47% for the transverse diameter for the 20 pN load when compared to the instantaneous loading results. This showed that gradually loading the RBC can recover accurate results using a more physically intuitive loading approach. Gradual loading over a shorter period with settling afterwards gave results equivalent to the instantaneous loading simulations. Overall, assessing the loading approach and allowed settling time is deemed important for physical consistency, and the loading approach can have an impact on simulation results.