Integrated Deterministic Probabilistic Safety Assessment (IDPSA) methodology is a powerful tool for identification of failure domains when both stochastic events and physical time dependent processes are important. Computational efficiency of deterministic models is one of the limiting factors for detailed exploration of the event space. Pool type designs of Generation IV heavy liquid metal cooled reactors introduce importance of capturing intricate 3D flow phenomena in safety analysis. Specifically mixing and stratification in 3D elements can affect efficiency of passive safety systems based on natural circulation. Conventional 1D System Thermal Hydraulics (STH) codes are incapable of predicting such complex 3D phenomena. Computational Fluid Dynamics (CFD) codes are too computationally expensive to be used for simulation of the whole reactor primary coolant system. One proposed solution is code coupling where all 1D components are simulated with STH and 3D components with CFD codes. However, modeling with coupled codes is still too time consuming to be used directly in IDPSA methodologies, which require thousands of simulations. The goal of this work is to develop a computationally efficient surrogate model (SM) which captures key physics of complex thermal hydraulic phenomena in the 3D elements and can be coupled with 1D STH codes instead of CFD. TALL-3D is a lead-bismuth eutectic thermal hydraulic loop which incorporates both 1D and 3D elements. Coupled STH-CFD simulations of TALL-3D typical transients (such as transition from forced to natural circulation) are used to calibrate the surrogate model parameters. Details of current implementation and limitations of the surrogate modeling are discussed in the paper in detail.