Multiphase compressible flow problems are widespread in aviation, aerospace,
transportation, military, and industrial fields, for instance, in underwater
explosion bubble dynamics, fuel injection for hypersonic vehicles, liquid
sloshing in propellant tanks, and supercavitating underwater vehicles. This
paper proposes an improved THINC (Tangent of Hyperbola for Interface Capturing)
method for multiphase flow simulations, based on a selective reconstruction
strategy for the dominant material. The core of the strategy is to apply the
THINC reconstruction exclusively to the material with the largest volume
fraction within a multiphase mixed cell, which numerically governs the local
interface evolution. The volume fractions of non-dominant materials are then
obtained through a proportional distribution that inherently ensures the
summation (Σαk = 1) and boundedness (0 ≤
αk> ≤ 1) constraints are met without explicit
corrections. This approach reduces the number of THINC reconstructions for each
time step in a multiphase mixed cell from Nm (the
number of materials) to one, significantly simplifying the algorithm and
lowering computational cost. It thereby avoids the error accumulation and
complex renormalization procedures associated with conventional schemes that
reconstruct all materials. While strictly maintaining volume fraction
conservation, the proposed method preserves interface sharpness through the
underlying THINC framework. The method is implemented in a diffuse-interface,
multiphase Eulerian framework and validated with a series of challenging
benchmarks, including shock-helium bubble interaction, triple-point problem, gas
impact, and the more complex modified gas impact. Numerical results show that,
compared with conventional multiphase THINC approaches that reconstruct every
material, the proposed scheme can reduce CPU time by about 40.0% without
compromising the accuracy of key physical quantities.