We observed three "composite" type SNRs, W44, CTB1 and HB-3, with ASCA. The "compoisite" type SNRs have shell in radio band and center-filled morphology in X-ray. We observed thermal radiation of these SNRs in X-ray. About W44, the temperature is 0.50+0.03-0.02 keV, interstellar absorption is (2.33+0.22-0.16) × 1022 cm-2, and its metal abundances are 0.63+0.14-0.10 on Mg, 0.77+0.18-0.11 on Si, 0.45+0.23-0.17 on S, and 0.16+0.18-0.15 (relative to solar value) on Fe. About CTB1 the temperarure is about 0.9 keV, and about HB-3 the temperature is about 0.4 keV. We perform spatially resolved X-ray spectroscopy of three "composite" SNRs. About W44, the temperature and interstellar absorption are consistent with constant along to the radius from the center of the SNR. The metal abundances of the center are about 3 times of those in the 10 arcmin off the center on Mg, Si, and S. The metal aboundance gradients about Ne, Ar, and Fe are not well determined and are consistent with both the constant and 3 times higher at the center than those in the 10 arcmin off the cneter. CTB1 and HB-3 also show higher metal abundances of Mg in some region around center than in the rest regions. We discussed the formation mechanism of "composite" SNRs, especially comparison of two models; White and Long model(1991) and the radiative shock model discussed by Harrus et al (1997). Both models can explain the morphology and the center high metal abundances. About radiation shock model, we perform the hydrodynamic simuration of Type II SNR using VH-1 code. The front shock of about 4000 years after explosion is cooled due to radiation, and its temperature is about 104 K and the shell no more radiate X-ray. Then the center-filled morphology is showed by the simulation. Our new calucuration shows that the radius of the contact discontinuity is about half of the radius of the front shock after 2000 years. The radiation shock model can explain the center metal rich gradient due to the existence of the ejecta from SN explosion. On the other hand, the WL model can explain it due to the evaporation of the metal rich dust. Therefore the existence of the inert gas of Ne and Ar at the center of the SNR will give the answer of the validity of these models.