We present a new class of component-wise numerical schemes that are in the family of relaxation formulations, originally introduced by [S. Jin and Z. P. Xin, Comm. Pure Appl. Math., 48(1995), pp. 235-277]. The relaxation framework enables the construction of schemes that are free of nonlinear Riemann solvers and are independent of the underlying eigenstructure of the problem. The constant relaxation schemes proposed by Jin & Xin can however introduce strong numerical diffusion, especially when the maximum characteristic speeds are high compared to the average speeds in the domain. We propose a general class of variable relaxation formulations for multidimensional systems of conservation laws which utilizes estimates of local maximum and minimum speeds to arrive at more accurate relaxation schemes, irrespective of the contrast in maximum and average characteristic speeds. First and second order variable relaxation methods are presented for general nonlinear systems in one and two spatial dimensions, along with monotonicity and TVD (Total Variation Diminishing) properties for the 1D schemes. The effectiveness of the schemes is demonstrated on a test suite that includes Burgers' equation, the weakly hyperbolic Engquist-Runborg problem, as well as the weakly hyperbolic gas injection displacements that are governed by strong nonlinear coupling thus making them highly sensitive to numerical diffusion. In the latter examples the second order Jin-Xin scheme fails to capture the fronts reasonably, when both the first and second order variable relaxed schemes produce the displacement profiles sharply.