Abstract
Steam-assisted gravity drainage (SAGD) remains one of the most energy-intensive technologies for heavy-oil recovery because production response is controlled not only by injected heat but also by spatial heat delivery, wellbore losses, viscosity reduction and steam chamber geometry. This paper develops a physics-guided digital twin for SAGD with distributed thermocable assistance and a bounded residual machine learning correction layer. The framework combines a heat-delivery model, temperature-dependent oil mobility, scenario analysis and decision-oriented visualization within a reproducible computational experiment. A reference operating envelope was formulated for heavy-oil reservoirs, including depth, horizontal well length, permeability, porosity, oil viscosity, steam temperature, injection rate, thermocable power and cable coverage. The analysis includes sensitivity testing, cumulative-production/SOR dynamics and Pareto-type operating-window mapping. In the reference computational scenario, which is treated as an illustrative screening case rather than as field-history validation, the thermocable-assisted hybrid configuration changed the model-calculated eight-year cumulative oil from 452.5 × 103 m3 to 615.2 × 103 m3 and the mean SOR from 3.17 to 2.72 t/t relative to the conventional SAGD physics-core case. The largest sensitivities were associated with steam rate, steam temperature, initial viscosity and permeability. Within the declared operating envelope, the results support the use of the framework as a pre-field screening tool and indicate that thermocable assistance should be interpreted primarily as spatial heat distribution control rather than as a field-validated production-improvement guarantee.