Abstract: Determining ultimate bearing capacity problem is of great importance in design and construction of the subgrade above void. The limit damage state of subgrade is numerically simulated by using upper bound finite element method. Firstly, the basic principles and calculation process of the upper bound finite element method are briefly introduced. Secondly, reasonable calculation assumption is proposed; and then a numerical model that can take various factors into consideration is established. Thirdly, the upper-limit solutions of ultimate bearing capacity under various conditions are calculated by using upper bound finite element method; and the bearing capacity coefficients Nc, Nq, N? are computed by using the formula of ultimate bearing capacity, as Tergaghi suggested; and then their influencing factors are analyzed. Finally, based on the large amount of energy dissipation and velocity field pattern analysis, three typical failure modes are presented and their influencing factors are analyzed. It is shown that Nc and Nq increase with the increasing of internal friction angle of ? , and when ? is larger, N? is positive number, and increases with the increasing of ? ; Both Nc and Nq increase with the increasing of the ratio of viod roof thickness and footing width H/B; and when ? is smaller, N? decreases with the increase of H/B. When ? is larger, N? increases with the increasing of H/B; and Nc、Nq and N? both decrease with the increase of D/B. The subgrade above void has three typical failure mechanisms: punching failure mode, punching shear press failure mode and Prandtl failure mode.

Key words: void, upper bound solution, finite elements, bearing capacity coefficient, failure mechanism