Experimental investigation on heat transfer to supercritical CO2 in a microtube up to 30 MPa for application in the NET Power cycle

• Heat transfer in supercritical CO 2 microflows up to 30 MPa was investigated. • An increase in mass flux enhances heat transfer rates at all pressures tested. • Higher heat flux enhances supercritical CO 2 heat transfer rates for 15–30 MPa. • Buoyancy effects diminish with increasing pressure and are more marked at 10 MPa. • The new heat transfer model captures the data points with a 6.74 % of error. Microtube heat exchangers represent a high-performance alternative to conventional printed circuit designs for the thermal recuperator of the innovative oxy-combustion NET Power cycle, offering potential improvements in both system efficiency and compactness. To support this technology transition, this study presents an experimental investigation of heat transfer in CO 2 at supercritical pressures up to 30 MPa. Experiments were conducted using a 1700 mm long, 0.88 mm inner diameter, uniformly heated horizontal microtube designed to replicate the operating conditions of a microtube heat exchanger. An experimental setup was built to measure local heat transfer coefficients of CO 2 , with a parametric analysis performed to evaluate the influence of mass flux, heat flux, inlet temperature, buoyancy, and flow acceleration. Tests were conducted at pressures of 10, 15, 20, 25 and 30 MPa. Results show that the heat transfer improves with increasing mass flux. At 10 MPa, the heat transfer coefficient exhibits a peak near the pseudo-critical temperature, followed by a deterioration and subsequent recovery. With increasing thermal input, the peak is attenuated, while heat transfer performance improves at higher pressures. Raising inlet temperatures enhances heat transfer in the thermal inflow region, reduces the peak value at 10 MPa, and causes the heat transfer coefficients to converge across different pressures. Buoyancy effects are most pronounced at 10 MPa and become weaker as pressure increases. Moreover, a new deep neural network model was developed to predict heat transfer coefficients, demonstrating an average deviation of 6.34 %. The present study substantially expands the existing experimental database, provides new physical interpretations of key phenomena, and translates these findings into a predictive tool applicable to engineering design.