We study various implementations of a supernova (SN) feedback model and present the results of our `Osaka feedback model' using isolated galaxy simulations performed by the smoothed particle hydrodynamics code GADGET-3. Our model is a modified version of Stinson et al.'s work, and we newly add the momentum kick for SN feedback rather than only thermal feedback. We incorporate the physical properties of SN remnants from the results of Chevalier and McKee & Ostriker, such as the effective radius of the SN bubble and the remnant lifetime, in the form of Sedov-Taylor- like solutions with the effect of radiative cooling. Our model utilizes the local physical parameters such as the density and temperature of the interstellar medium rather than galactic or halo properties to determine the galactic wind velocity or mass- loading factor. The Osaka model succeeds in self-regulating star formation, and naturally produces galactic outflow with variable velocities depending on the local environment and available SN energy as a function of time. An important addition to our previous work by Aoyama et al. is the implementation of the CELIB chemistry library, which allows us to deal with the time- dependent input of energy and metal yields for Type Ia and II supernovae and asymptotic giant branch stars. As initial tests of our model, we apply it to isolated galaxy simulations, and examine various galactic properties and compare them with observational data, including metal abundances.