This presentation discusses how to conduct optimal design of an indoor environment based on specific design objectives by controlling the thermo-fluid boundary conditions, such as air supply location, size, and parameters. This study used a CFD-based adjoint method to identify the othermo-fluid boundary conditions. Through defining the air distribution in a certain area (design domain) as a design objective in an indoor space, the adjoint method can identify the air supply location, size, and parameters. The adjoint method can only achieve local design optimal but the computing costs did not depend on the number of design variables.