In this paper, we introduce a simulated annealing algorithm for single objective, trans-dimensional optimization problems. Trans-dimensional optimization refers to a class of problems where candidate solutions can have different number of variables. For such problems, the existing optimization methods need to be run for various models (i.e. problems with fixed number of variables) extensively, which is inefficient. The proposed optimization algorithm explores the model space and the corresponding variable space probabilistically, allocating more computational resources (function evaluations) to the promising models. The performance of the proposed algorithm is reported for a clustering problem and a warehouse optimization problem. The results of the proposed algorithm are compared with a conventional optimization algorithm with fixed number of variables (NSGA-II [3]) to highlight the benefits of the approach.