Simulation Methods of Glass Composition and Properties: A Short Review

Seven methods of modeling glass properties were briefly reviewed, i.e., covering additive method, phase diagram method, Priven method, topological theory, molecular dynamics simulation, machine learning, and mathematical statistical modeling (composition-property, structure-property, and composition-structure-property). The principle, theoretical basis, procedure of each method and their application were highlighted. The additive method demonstrates its application modeling multiple glass properties. The phase diagram approach is suitable for binary, ternary and quaternary systems of silicate, borate and borosilicate glass systems. The Priven method combines glass structure, thermodynamic equations and computer simulation; topological theory is used to simulate several properties of simple oxide and sulfide glass. The molecular dynamics simulation provides insight of molecular structures of glass of various compositions. The machine learning method is utilized based on a large database from the literature to predict properties of complex glass systems. The statistical modeling methodology is applied to bridge mathematical interrelationships of composition (C) – structure (S) – property (P) of multicomponent systems of silicate, borosilicate, and phosphate glasses. The C–S–P statistical modeling approach shows the improvement in accuracy and precision rather than the conventional C–P statistical modeling in the design of new glasses.