A field framework for U-values and dynamic response in prefabricated rural dwellings: theory–measurement gaps and ranking reversal

• Field monitoring of three co-located rural prefabricated dwellings in autumn. • A sliding-window regression estimates in-situ wall thermal transmittance. • Dynamic envelope response is quantified using attenuation and time lag metrics. • Measured and calculated transmittance show deviations and ranking reversal. • Deviations are attributed to solar effects, thermal mass, and boundary conditions. Steady-state U-values from ISO 6946 can misrepresent envelope performance under free-running, non-steady conditions, risking biased evaluation and retrofit decisions for cost-sensitive rural housing. This study quantifies in-situ wall thermal transmittance and dynamic response under real autumn free-running operation and tests the applicability limits of U-value inversion using a co-located open-envelope reference unit. Heat flux and indoor/outdoor temperature and humidity were monitored in three prefabricated dwellings in Linyi, China: PRH-1 (external insulation), PRH-2 (sandwich wall), and PRH-3 (shell with an unsealed 2.0 m × 0.8 m opening). In-situ U-values were estimated via 8 h sliding-window regression of heat flux against indoor–outdoor temperature difference, using a ΔT > 10 °C eligibility criterion and convergence screening (R2 threshold and ΔU limit between adjacent windows), and compared with theoretical values. Theoretical U-values were 0.28 (PRH-1), 0.34 (PRH-2) and 1.67 W/(m2·K) (PRH-3). Measured U-values were 0.57 W/(m2·K) (95% CI: 0.55–0.59) for PRH-1 and 0.49 W/(m2·K) (0.47–0.51) for PRH-2, exceeding theory and reversing the calculated ordering. The reversal is attributed to predominantly daytime “outdoor-hot/indoor-cool” windows with strong solar forcing, wall-segment solar-loading differences, and PRH-2’s greater thermal mass buffering peak conductive flux. PRH-3 never developed a sustained ΔT because indoor air was strongly coupled to outdoors, so U-value inversion was not feasible. Dynamic indicators supported the ranking (PRH-2: f = 0.13, τ = 11.1 h; PRH-1: f = 0.15, τ = 6.1 h; PRH-3: f = 0.51, τ = 6.6 h), highlighting both the value and limits of long-period in-situ checks alongside steady-state calculations.