This paper proposes a robust topology optimisation (RTO) methodology for minimizing both the expected compliance and the variance given arbitrarily distributed uncertain loading conditions.However, such a stochastic loading accompanies a substantial computational expense and this can quickly become computationally prohibitive for high-fidelity high-dimensional topology optimisation.A common methodology for accounting for the stochastic elements in RTO is to obtain a set of deterministic problems that represent the original non-deterministic problem.According to the means of separation, the existing approaches can be classified into two categories.