Measured Baseline of a Double-Wide Manufactured Home Testbed: Hygrothermal and Energy Performance

Abstract

A 1990 double-wide manufactured home (MH) dimensioned 26 ft × 40 ft with an area of 1,040 ft2 and total volume of 8,320 ft3 was installed and instrumented on Oak Ridge National Laboratory’s (ORNL) main campus in Oak Ridge, Tennessee to establish baseline (“as-is”) performance for envelope, airtightness, and space-conditioning energy use. The unoccupied test facility was instrumented with more than 200 sensors to quantify hygrothermal behavior across the four wall orientations and major assemblies (walls, floor, and ceiling/roof), including temperature (thermistors/thermocouples), relative humidity, and heat-flux measurements. Boundary conditions were recorded using an on-site weather station (air temperature, relative humidity, pressure, precipitation, solar radiation, wind speed and direction). Electrical consumption of space-conditioning equipment was monitored with circuit-level power metering, and additional instrumentation supported condensate measurement and mean radiant temperature characterization. Field-measured, monthly integrated thermal resistance (R-value) was calculated from December 2025 through May 2026 using surface temperatures and in-situ heat-flux data. Wall thermal performance was stable and consistent with calculated values: measured integrated R-values ranged from 18.5 to 19.5 hr·ft²·°F/Btu versus a calculated value of ~19.2. In contrast, the floor assembly underperformed substantially with measured integrated R-values ranging from 10.1 to 12.8 hr·ft²·°F/Btu compared to a calculated value of ~20.4. This reduction is consistent with observed degradation of the belly cavity system (sagging/failed membrane and displaced fiberglass insulation), which likely increased convective bypass and reduced effective insulation performance. Ceiling integrated R-values were most reliable during winter (12.0–12.7 hr·ft²·°F/Btu from December–February), aligning with the calculated ceiling value (10.8 ± 2) given measured variability in loose-fill thickness; spring results became highly variable as temperature differentials weakened due to solar effects. Sheathing temperature and relative humidity (RH monitoring showed broadly similar behavior across all wall orientations, with periods above 80% RH occurring primarily during colder conditions and generally declining as temperatures increased, indicating seasonal drying potential and no clear, persistent mold-risk signature in the monitored wall locations. The floor sheathing exhibited wider RH swings, consistent with the compromised belly condition. Roof OSB sheathing experienced pronounced solar-driven temperature peaks (maximum recorded ~160.5°F in spring), with modest RH differences between slopes, underscoring the importance of summer overheating considerations for any future above-sheathing insulation retrofits. Airtightness testing using multipoint depressurization blower door measurements found the home to be extremely leaky in its baseline condition at 18.1 ACH50. Sealing the vents and central HVAC distribution system reduced leakage to 14.8 ACH50 (an ~18% reduction), demonstrating that duct/distribution pathways contribute materially to whole-house leakage. Two off-the-shelf tape-based “DIY” duct-sealing approaches produced smaller reductions (to 17.3 and 16.8 ACH50), indicating limited whole-house benefit from simple register/duct sealing measures alone. Space heating during winter baseline operation was provided by three variable-speed window space conditioner (WSC) units (two from Manufacturer A and one from Manufacturer B). Based on monitored inlet/outlet air temperatures and power measurements for the most instrumented unit, average WSC winter performance (December 2025 and January 2026) ranged from coefficient of performance (COP) ~2.5 to 3.08, with uncertainty driven primarily by airflow measurement limitations on one unit. Using conservative assumptions for building-level comparison, WSC operation was estimated to reduce heating energy use by roughly 55% and peak demand by roughly 40% relative to distributed electric resistance (ER) heating providing equivalent delivered heat. Planned and unplanned power interruption tests further highlighted peak-demand implications during recovery. After an unplanned ~4-hour outage (average outdoor ~22°F), three WSCs restored indoor conditions to setpoint in under 2 hours with a measured peak of 2,224 W. In a planned 9-hour shutdown at ~20°F using four 1,500 W resistance heaters, recovery required ~3 hours with a 5,424 W peak. A planned 9-hour outage using three WSCs recovered in ~3 hours with a 4,440 W peak. A mixed strategy (two WSCs plus two 1,500 W resistance heaters) recovered in ~2.5 hours at ~30°F with a 4,236 W peak and reduced recovery energy use by 46% relative to resistance-only heating. Collectively, these baseline results quantify key envelope deficiencies (notably the floor), document very high infiltration, and demonstrate the potential of WSC heating and cooling systems to reduce energy use and peak electrical demand in older, leaky MHs.