The article considers the creation of a digital twin model of a capital construction facility at the stage of design and survey works (DSW). The aim is to develop a grounded model and algorithm for building a digital twin at this early stage and to assess its advantages and limitations. The methods include analysis of Russian and international publications, systematization of approaches, generalization of implementation practice, and decomposition of the facility life cycle into stages relevant to building the model. A step-by-step methodology is proposed: collection and normalization of initial data, creation of a base BIM model, integration of process information (4D/5D modeling), connection of sensor and monitoring systems to provide two-way linkage with the physical asset, verification and validation procedures, and transition to operational use. It is shown that using a digital twin in DSW improves design accuracy, reduces clashes, optimizes material, time and labor resources, and extends preventive risk management through forecasting wear and deviations from design parameters. Empirical data on the share of organizations that have implemented or plan to implement digital twins confirms growing demand, while highlighting barriers such as high costs, lack of regulatory and methodological support, data integration complexity and information security requirements. The conclusion is that the proposed approach forms a basis for full life-cycle management and can be scaled further as standards and AI/Big Data analytics evolve.