Energy & Buildings 203 (2019) 109430
Contents lists available at ScienceDirect
Energy & Buildings journal homepage: www.elsevier.com/locate/enbuild
Whole building retrofit using vacuum insulation panels and energy performance analysis ✩ Kaushik Biswas a,∗, Tapan Patel b, Som Shrestha a, Douglas Smith c, Andre Desjarlais a a
Oak Ridge National Laboratory (ORNL), One Bethel Valley Road, Building 3147, P.O. Box 2008, M.S. - 6070, Oak Ridge, TN 37831, USA U.S. Army Engineer Research and Development Center (ERDC), Construction Engineering Research Laboratory (CERL), PO Box 9005, Champaign, IL 61826-9005, USA c NanoPore Incorporated, 2525 Alamo Ave. SE, Albuquerque, NM 87106, USA b
a r t i c l e
i n f o
Article history: Received 7 March 2019 Revised 29 July 2019 Accepted 11 September 2019 Available online 11 September 2019 Keywords: Vacuum insulation Modified atmosphere insulation Building retrofit Whole building energy analysis Exterior wall retrofit
a b s t r a c t Vacuum insulation panels (VIPs), due to their high thermal performance, provide an attractive alternative to traditional building insulation materials, especially as an option for retrofitting old, poorly insulated buildings. This article describes the complete retrofit of all exterior walls of a single-story building in a cold climate using VIPs. A recently-developed low-cost VIP, called modified atmosphere insulation (MAI), was used in this study. Two buildings of near-identical construction were studied, with one remaining unaltered and serving as the baseline while the other served as the retrofit building. The VIPs or MAI panels proved to be a feasible and durable option for retrofitting building envelopes. Thermal performance of both buildings was analyzed using in-situ temperature and heat flow sensors. Numerical models of the two buildings were created, benchmarked using experimental data and used for predictions of annual energy savings due to the addition of MAI panels to the exterior walls. The models predicted significant reduction in the annual heating energy consumption in the retrofitted building compared to the baseline building. © 2019 Elsevier B.V. All rights reserved.
1. Introduction Buildings consume about 40% of the total global energy, with a large fraction being used to maintain thermal comfort, and emit over 30% of the global carbon dioxide (CO2 ) [1]. Retrofitting ex-
Abbreviations: BLCC, Building Life Cycle Cost; CERL, Construction Engineering Research Laboratory; COP, Center-of-panel; CVRMSE, Coefficient of variance of root mean square error; ERDC, Engineer Research and Development Center; HFM, Heat flow meter; HFT, Heat flux transducer; HVAC, Heating, ventilation and airconditioning; IR, Infrared; MAI, Modified atmosphere insulation; NMBE, Normalized mean bias error; NE, Northeast; NIST, National Institute of Standards and Technology; NW, Northwest; ORNL, Oak Ridge National Laboratory; PIR, Polyisocyanurate; SE, Southeast; SIR, Savings-to-investment ratio; SW, Southwest; TARP, Thermal Analysis Research Program; VIP, Vacuum insulation panel. ✩ Notice: This manuscript has been authored by UT-Battelle, LLC, under Contract No. DE-AC05-00OR22725 with the U.S. Department of Energy. The United States Government retains and the publisher, by accepting the article for publication, acknowledges that the United States Government retains a non-exclusive, paid-up, irrevocable, world-wide license to publish or reproduce the published form of this manuscript, or allow others to do so, for United States Government purposes. DOE will provide public access to these results of federally sponsored research in accordance with the DOE Public Access Plan (http://energy.gov/ downloads/doe- public- access- plan). ∗ Corresponding author. E-mail address:
[email protected] (K. Biswas). https://doi.org/10.1016/j.enbuild.2019.109430 0378-7788/© 2019 Elsevier B.V. All rights reserved.
isting buildings presents a big opportunity for energy savings and reduction in emissions, due to the poor energy efficiency of major portions of the existing building stock, the long life time of buildings and the low turnover of the building stock [2,3]. There are numerous studies in the literature on building retrofits and evaluation of different retrofit strategies, including laboratory experiments, demonstrations in actual buildings, numerical simulations, or some combination thereof. Salvalai et al. [3] described the exterior façade retrofit of a multi-story, multi-family residential building using pre-fabricated insulation panels and observed a 36% reduction in energy consumption during winter of the first year after the retrofit compared to the average winter energy use during five years before the retrofit. Rodrigues et al. [4] evaluated a modular prefabricated external wall insulation system in the laboratory and used it to retrofit a residential building. While no details of the retrofit itself are provided, the authors provided descriptions of steady-state and dynamic simulations to evaluate the performance of the insulation system if applied to various building types. Shrestha et al. [5] monitored and performed whole-building simulations of an experimental army test hut retrofitted using typical foam and fiberglass insulation. Jones et al. [6] reported deep retrofits of five residential buildings, including added envelope insulation, energyefficient lighting, renewable energy generation, energy storage, etc.
2
K. Biswas, T. Patel and S. Shrestha et al. / Energy & Buildings 203 (2019) 109430
Nomenclature CO2 Q k q R T
Carbon dioxide Solar irradiance (W/m2 ) Thermal conductivity (W/(mK)) Heat flux (W/m2 ) Thermal resistance (m2 •K/W) Temperature (°C or K)
Subscripts ext Exterior int Interior o Outside s Solar The simple paybacks based on savings and investment costs were in excess of 38 years, but the authors noted several other factors that should influence the decision-making as well as the potential for better savings-to-investment ratios (SIR) in future with increasing energy costs and greater adoption of retrofit measures. Farmer et al. [7] noted the uncertainties in the measured thermal performance of envelope retrofit strategies based on in situ measurements alone. The authors demonstrated a steady-state experimental methodology to evaluate the thermal performance of several combinations of retrofit measures applied to a full-scale solid-wall house that was located within an environmental chamber. Kosny et al. [8] evaluated multiple envelope retrofit strategies and noted the reduction in retrofit costs if thinner insulation materials with higher thermal performance could be used instead of current insulation materials. Due to their higher thermal resistance, vacuum insulation panels (VIPs) represent a more effective alternative to conventional building insulation materials [9]. Stateof-the-art, silica-based VIPs can achieve apparent thermal conductivities in the range of 0.0 021–0.0 043 W/(mK) [10] compared to fibrous or foam insulations that can achieve 0.02–0.03 W/(mK) or higher [11]. VIPs have been used to retrofit exterior building walls and facades of brick and concrete buildings and their long term performance was monitored using in situ temperature measurements [12,13]. Sveipe et al. [14] performed laboratory tests and simulations of exterior retrofits of timber walls with 20 mm and 30 mm VIPs with a focus on condensation risks due to addition of the VIPs. Ascione et al. [15] reported the dynamic testing of a multi-layered concrete wall with VIPs in a test building under multiple weather conditions and also performed simulations of a model office building to evaluate the relative benefits of the low thermal diffusivity of concrete and very low conductivity of VIPs under different weather conditions. Biswas et al. [16] demonstrated a new, lower-cost variant of VIPs called modified atmosphere insulation (MAI) panels in an additively manufactured research building, but the energy benefits of using MAI were not evaluated. Biswas et al. [17] reported the development of foam-MAI composite boards and the use of the composite boards to retrofit the roof of an occupied building in a cold climate; a limited number of temperature and heat flux sensors were installed on the roof, but their primary purpose was to monitor the long term performance of the composite board. Thus, while there are studies related to retrofits using VIPs, demonstrations of whole-building envelope retrofits with VIPs combined with systematic evaluations of their specific energy benefits are missing. This article presents the complete exterior wall retrofit of a single-story office building using MAI and energy-savings analysis using EnergyPlus (https://energyplus.net/). EnergyPlus is a whole building energy simulation program that has been used to study the benefits of applying various energy efficiency measures to building [18–20]. In this work, two identical buildings were studied
Fig. 1. Building 432B (retrofit building).
- the retrofit building with MAI added to the exterior wall and a baseline building that was left unaltered. The exterior walls of both the baseline and retrofit buildings were instrumented with temperature and heat flux sensors, which were monitored over a period of several months. EnergyPlus models of the two buildings were created and benchmarked using the temperature and heat flow data from the buildings. Finally, the benchmarked models were used to estimate the energy savings due to the addition of the MAI under standard operational and weather conditions. 2. Building description Two buildings with the same orientation and construction and located within 30 m of each other were utilized for this study. The buildings are located in Fort Drum, New York, which falls under climate zone 6A (cold-humid) [21]. The buildings were single storied with 186 m2 of floor space and built using wood-framed construction. The buildings were oriented with one corner pointing north. The external walls contained fiberglass cavity insulation with nominal thermal resistance of 3.4 m2 K/W and the roof insulation with nominal resistance of 7.0 m2 K/W. They also contained the typical interior and exterior sheathing layers, an air barrier film outside the exterior sheathing layer, and metal siding that was mechanically attached to the exterior sheathing. The exterior northeast wall of each building was adjacent to an arms vault and contained an approximately 0.3 m thick concrete wall inside the wood-framed exterior wall. The buildings were designated 431B and 432B; 431B was the baseline building and 432B was retrofitted using the MAI panels. Both buildings were used as classrooms for training purposes, but with different occupancy schedules. Fig. 1 shows a photograph of building 432B and Fig. 2 shows a plan of the test buildings. The buildings are oriented so that the bottom left corner of the schematic in Fig. 2 is pointing north. The building schematic includes the wall orientations, room locations and the approximate thermostat location. A major structural difference between the buildings was that the room indicated by dashed lines in Fig. 2 was only present in 432B; in building 431, it was an open space including the hallway. 3. Building retrofit and monitoring 3.1. MAI dimensions and layout MAI, a lower-cost VIP technology, was used for the retrofit. For this project MAI panels with a fumed silica core and a polymeric barrier film were utilized, similar to past work [17]. The advantage of polymeric films is that the edge-effects, i.e. heat flows around the edges of the panels, are significantly reduced compared to metallic or metallized films that are typically utilized in VIPs. MAI panels are similar to silica VIPs with respect to core and barrier film, but are manufactured in a different, simplified manner compared to traditional VIPs, with potential for significant cost
K. Biswas, T. Patel and S. Shrestha et al. / Energy & Buildings 203 (2019) 109430
3
Fig. 3. Variation in the area fraction of MAI panels vs. number of different panel sizes.
Fig. 2. Plan of the test buildings, with the room details and wall orientations (NE – northeast; SE – southeast; SW – southwest; NW – northwest). The approximate thermostat location is indicated by ‘X’. The room indicated by the dashed lines was only present in 432B and not in 431B. All dimensions are in meters.
reduction. MAI panels can achieve center-of-panel (COP) thermal conductivity of about 0.004 W/(mK), which is similar to regular silica core VIPs [10]. The typical VIP production process involves pressing fine powders of the core material into boards, which are cut to the desired size and dried; the latter is an energy-intensive process. Next, the core is evacuated and sealed within barrier films. MAI, on the other hand, is created by replacing the air in the fluidized powder core by a condensable vapor. After the core is sealed within barrier films, cooled and formed to final shapes, the condensation of the vapor creates the final vacuum. The MAI technology is protected by multiple patents, which are listed in Appendix A. An initial task was to determine the number of different MAI panel sizes as well as the total number of MAI panels needed to achieve adequate coverage of MAI on the surfaces of the exterior wall to be retrofitted, to increase the overall R-value to the extent possible. The overall cost of MAI when produced at commercial scales is expected to be dependent on the number of different sizes that need to be produced; the fewer the number of different sizes, the lower the cost. Hence, within this project, an attempt was made to minimize the number of MAI panel sizes while maintaining adequate MAI coverage. The gaps between the MAI panels were intended to be filled with rigid polyisocyanurate (PIR) foam insulation. For this task, drawings of the exterior wall surfaces were created to identify the locations of windows, doors, mechanical fas-
Table 1 MAI panel dimensions. MAI panel
X (cm)
Y (cm)
Number of panels
A B C D
50.7 54.1 48.7 52.3
71.1 71.1 64.8 64.8
87 83 26 135
teners for the metal siding as well as dimensions of the clear wall areas (i.e. areas without doors, windows, signs, etc.) where the MAI panels could be added. Using the measured wall dimensions, an analysis was done to calculate the percentage area that could be covered by MAI panels of various sizes. Fig. 3 shows the analysis results of area fraction of the MAI panels as a function of the number of different sizes of the MAI panels. The areas fractions were calculated using both clear wall and total wall areas of all external wall surfaces. A gradual decline in area fraction was observed when reducing the panels sizes from 10 to 4, but a sharp drop was observed below 4. Hence, a decision was made to produce MAI panels with the dimensions and numbers listed in Table 1. With the following selection, the overall area fraction of MAI panels on the four walls were: NE – 80%, SE – 78%, SW – 77%, NW – 72%. The thickness of all MAI panels was 2.54 cm. Fig. 4 shows a drawing of the SW exterior wall of the retrofit building showing features like doors, signs and location of the mechanical fasteners via the dash-dotted black lines. Also shown are the MAI panels designated as A, C and D. Assuming thermal conductivities of 0.004 W/(mK) and 0.026 W/(mK) for MAI and PIR, respectively, and using the parallel heat flow path calculations, the estimated average thermal resistance to be added to the exterior walls was 2.75 m2 K/W. Assuming 5% of the MAI panels would
Fig. 4. Schematic showing the distribution of the MAI panels on the SW exterior wall. All dimensions are in meters.
4
K. Biswas, T. Patel and S. Shrestha et al. / Energy & Buildings 203 (2019) 109430
Fig. 5. Schematic representing a plan view of the wall cross-section and locations temperature and heat flow sensors; sensor locations are not to scale.
to track the sun for measuring the direct normal solar irradiance (DNI) and diffuse horizontal radiation (DHR). However, the tracker operation failed, so the local DNI and HDR measurements required to create the weather file for EnergyPlus simulations could not be obtained. Therefore, solar irradiance data, including DNI and DHR, from a nearby commercial weather station located at the WheelerSack Airfield, New York were utilized. It is noted that utilizing offsite weather data does introduce additional uncertainties in the energy models [22]. All sensors were scanned at 15 s intervals and the hourly averages were stored in data files for monitoring and analysis. A heat flow meter was used to measure the thermal conductivity of MAI panels according to ASTM C518 [23] test method. The stated uncertainty of the heat flow meter is ±1%. 3.3. MAI installation
Fig. 6. Interior view of a wall sensors’ location; an ‘air’ thermistor (in a white pouch) and the gypsum board used for mounting the surface HFT are visible.
be damaged during installation or later, the added average thermal resistance was estimated to be 2.55 m2 K/W; the conductivity of damaged MAI is about 0.02 W/(mK) based on measurements. Based on later inspections using infrared imaging, the actual number of damaged MAI panels was about 3.3%. It is noted that these estimates are based on COP conductivity of MAI and do not include the edge effects. 3.2. Instrumentation and data acquisition The exterior walls of both buildings were instrumented with thermistors and heat flux transducers (HFT). A schematic representing the cross-section of the retrofitted wall as well as the relative locations of the thermistors and an HFT is shown in Fig. 5. The thermistors were installed on all exposed surfaces before and after the retrofit as well as mounted near the interior surface to measure the air temperature. The HFTs are 51 × 51 cm and 2.8 mm thick and were mounted on the interior wall surfaces using additional pieces of gypsum board. Fig. 6 shows an interior view of the “air” thermistor and the gypsum board used to attach the surface HFTs. Based on the manufacturers’ specifications, the thermistors have an accuracy of ±0.2 °C while the HFTs have an accuracy of ±5% and a sensitivity of 6.3 (W/m2 )/mV. To improve the sensitivity of the HFTs used with the higher-thermal resistance retrofitted walls and enable accurate measurement of the low heat flows, two HFTs combined in series were used in building 432B. HFTs produce a voltage signal in response to the local temperature gradient and resultant heat flow, and by adding two in series, a higher voltage signal is produced for the same temperature gradient and heat flow. Thus, the overall measurement sensitivity is increased. An onsite weather station was installed to gather local weather data needed for the whole-building energy simulations. Table 2 lists the instruments that were installed and corresponding parameters that were measured. In addition, a solar tracker was installed
The retrofit was performed during November 2016. The retrofit was done in sections and involved removal of the exterior metal siding, addition of the MAI panels using adhesives and reinstallation of the metal siding. PIR foam sheets were cut to size on site and fitted between the MAI panels. After some initial experimentation, the installation team observed that by first attaching the lighter PIR foam sheets to the air barrier film, the MAI panels could simply be fit into the gaps between the PIR strips without any adhesives. The foam strips would hold the MAI panels in place temporarily before the siding was reinstalled using mechanical fasteners at the same locations. The only architectural alterations needed were replacing the soffit vents at the intersection of the roof and external walls and replacing the original J-channels around doors and windows to accommodate the added thickness of the MAI panels. J-channels, as the name suggests, are J-shaped pieces of siding trim used around windows and doors to overlap and cover the siding. Fig. 7 shows photographs of the MAI installation. Adhesive construction tape was applied to seal the joints between the MAI panels and foam-MAI interfaces. In a related study, hygrothermal modeling of the retrofitted walls were performed and the model results indicated no additional moisture risks due to the exterior retrofit using MAI panels and PIR foam [24]. Infrared (IR) images of the facility were taken immediately after the installation and during February 2017. These images were used to determine whether any MAI panels were damaged. The IR imaging was performed during winter and clearly revealed the intact MAI panels as areas of lower apparent surface temperature compared to the foam areas; i.e. lower heat losses from the building interior through the MAI sections. One MAI panel was intentionally damaged and was identifiable in the IR images due to its contrast with the intact MAI panels. At some point after the retrofit, Ft. Drum staff installed building signs in both the baseline and retrofit buildings. Due to a lack of coordination, the signs were installed with screws and punctured five MAI panels in the process. The IR imaging during February revealed a total of eleven (11) failed MAI panels, including the one damaged intentionally and those due to the installation of the sign. Although it’s not certain why the additional five MAI panels failed, the team suspected the panels were damaged during the installation process and slowly allowed their internal pressure to rise over time. With a total of 331 MAI panels installed, the eleven damaged panels represent a 3.3% failure rate; if the ones damaged intentionally and due to lack of coordination are disregarded, the failure rate dropped to 1.5%. Thus, the MAI panels proved to be sufficiently durable during the envelope retrofit. Members of this research team have developed a foam-VIP composite insulation board in collaboration with another manufacturing partner and demonstrated its installation in an occupied building [17]. Use of composite insulation boards containing VIPs
K. Biswas, T. Patel and S. Shrestha et al. / Energy & Buildings 203 (2019) 109430
5
Table 2 Instrumentation to gather local weather data. Sensor/Instrument
Model and manufacturer
Measured parameter
Accuracy/Uncertainty
Weather transmitter
WXT520, Vaisala
Pyrheliometer Pyranometer Radiometer
DR02, Hukseflux SR20, Hukseflux NR01, Hukseflux
Wind speed (m/s) Wind direction (°) Precipitation (mm/h) Pressure (Pa) Temperature (°C) Relative humidity (%) Direct normal solar irradiance (DNI) (W/m2 ) Global horizontal irradiance (GHI) (W/m2 ) Infrared radiation (W/m2 ) Albedo
±3% at 10 m/s ±3° <5% ±50–100 Pa ±0.3 ±3–5% ±2% for daily sums ±3% for hourly totals
Fig. 7. Left - installation of MAI panels with foam strips in the gaps overlapping the fasteners for the exterior siding; Right – retrofitted wall with MAI before reinstalling the siding.
Fig. 8. Building model rendering.
is expected to significantly reduce the installation time and labor, both being critical considerations for retrofits. 4. Building energy model 4.1. Methodology and model details EnergyPlus, a whole building energy simulation program, was used for energy savings analysis. In this study, the building models were created using as-built dimensions and envelope details. Important parameters such as the exact location of HFTs, windows, and shading surfaces were verified using information gathered from the field. Sensor locations in the models matched the physical locations of the sensors in the baseline and retrofit buildings. A rendering of one of the building models, created using OpenStudio® SketchUp Plug-in [25], is shown in Fig. 8. The envelope covered by the sensors were modeled as separate envelope
sections to be able to calculate the appropriate area-averaged heat flows and temperatures for comparison with the measurements. In the retrofit building model, temperature-dependent thermal properties of MAI panels [17] were utilized using the Conduction Finite Difference heat balance algorithm of EnergyPlus. Areaweighted average thermal conductivity of insulation materials, accounting for the fractions of wall area covered by MAI panels and PIR foam, were used in the simulation. However, thermal conductivity at center of MAI panel was used on the small wall areas where the HFTs were located, which were modeled as independent envelope sections as indicated in Fig. 8. This was necessary as the HFTs were located near center of MAI panels and, based on data analysis, were not influenced by the adjacent PIR foam layers. It is noted that the area-weighted thermal conductivity calculations also used the COP conductivity of the MAI panels and did not include edge-effects. However, since a polymeric barrier film was utilized, the edge effects are expected to be relatively small. Due to the scale of the retrofit and variations in the geometries of the MAI panels and adjacent materials, it wasn’t feasible to precisely estimate and incorporate the edge effects of the MAI panels in the numerical analysis. Appendix B shows calculations of overall thermal resistance of a foam-MAI composite layer [17] using COP MAI conductivity and compares it to measured values. The calculations in Appendix B show that, with polymeric barrier films, the COP conductivity of MAI panels yields an excellent approximation of the overall thermal resistance. Exterior walls of the buildings were modeled using actual construction details. The models contained wood-framed walls with fiberglass cavity insulation of nominal thermal resistance of 3.4 m2 K/W. The roof was modeled as a low-slope roof with
6
K. Biswas, T. Patel and S. Shrestha et al. / Energy & Buildings 203 (2019) 109430 Table 3 Temperature-dependent thermal conductivity of MAI. Mean T (°C)
0
20
23.9
40
55
kMAI (W/(mK))
0.0039
0.0040
0.0041
0.0043
0.0047
7 m2 K/W thermal insulation. The models also contained the arms vault with 0.3 m thick concrete walls, which acted as a considerable thermal mass. Blower door tests of the two buildings were performed to measure the envelope airtightness, which is needed as input by the EnergyPlus models. Both buildings were served with identical heating, ventilation and air-conditioning (HVAC) equipment. Solar absorptance of exterior envelope surfaces was measured at various locations and the measured values ranged from 0.92 to 0.96 for the roof and 0.60 to 0.63 for walls. All these details were captured in the EnergyPlus models of the buildings. The simulations were performed using 20 timesteps per hour. The exterior and interior convection heat transfer was modeled using the ‘SimpleCombined’ and ‘Thermal Analysis Research Program’ (TARP) algorithms, respectively. Radiation heat transfer was modeled using the inputs of surface properties and building geometry; the latter is used by EnergyPlus to calculate view factors. Further details on heat transfer algorithms are available at https://energyplus.net/. 4.2. Model benchmarking and estimation of energy savings Thermal properties of materials used in the exterior walls were either measured directly using ASTM C518 [23] or were obtained from the ASHRAE Handbook of Fundamentals [26]. Temperaturedependent thermal conductivity (k) of the MAI panels [17], listed in Table 3, were used in the models. Two sets of models were developed, one for benchmarking against measured data and another for annual energy consumption estimations. For benchmarking, important parameters such as interior air temperature, HVAC schedule, and internal load were modeled using the collected data. Reliable data for some parameters such as occupancy and interior thermal mass (furniture, equipment, etc.) were not available, so those parameters were tuned in the models to reduce discrepancies between the EnergyPlus calculations and measurements. Weather data files were created using temperature, humidity, barometric pressure, wind direction, and wind speed that were measured at the building site and solar radiation data from Wheeler-Sack Airfield weather data. For annual energy savings predictions, parameters such as lighting, internal load, occupancy, HVAC operation, and thermostat setpoint schedule were used from a DOE commercial prototype building model of a small office [27]. Such generalized building parameters are more appropriate than the conditions that were specific to the baseline and retrofit buildings, since a common set of operational parameters are needed to appropriately compare the energy performance of the two buildings. All other parameters, such as the envelope details, were retained from the benchmarked models. For weather conditions, typical meteorological year (TMY) weather data [28] were utilized. TMY data represent average weather conditions over a period of several years, typically 30. TMY data were deemed more suitable for annual energy simulations rather than using actual weather data from a specific time period. 5. Results and discussion 5.1. Measured data Temperature, heat flux and weather data were recorded on an hourly basis. Fig. 9 shows sample measurements of temperatures
Fig. 9. Sample of measured transient temperatures across the southwest (SW) wall of the retrofit building and GHI during three winter days.
across the cross-section of the southwest (SW) wall of the retrofit building and the GHI over three winter days (January 18–20, 2017). The locations of the measured temperatures were the exterior surface of the siding (‘Siding ext.’), the exterior and interior surfaces of the MAI panels (‘MAI ext.’ and ‘MAI int.’, respectively), the interior sheathing surface (‘Int. sheathing’) and room air adjacent to the interior wall surface. During the daytime hours, the temporal profiles of the exterior surface temperature and GHI are similar. As expected, large temperature differences were observed across the MAI layer and across the original wall cross-section, i.e. between the interior sheathing surface and interior MAI surface. Also noticeable in the plot are the changes in the room air temperature adjacent to the SW wall due to the thermostat settings. The air temperatures remained at about 23 °C during period from 8:00 AM till 7:00 PM, and then started dropping overnight till about 7:00 AM the following morning. This indicates thermostat setbacks outside typical operational hours. Similar observations were made in the baseline building. Fig. 10 compares the room air and exterior siding temperatures measured across all four exterior walls of the retrofit (R) and baseline (B) buildings. While the measured exterior siding temperatures were very close between the baseline and retrofit buildings, significant differences were observed in the room air temperatures. It is noted that the air temperatures near the northwest and northeast walls were consistently higher in the retrofit building. It is also noted that the diurnal fluctuations in the air temperatures near the northeast (NE) wall were lower compared to the other walls, due to the presence of the thermally-massive concrete section of the NE wall. The measured air temperatures near the southwest (SW) and southeast (SE) provide some insights into the benefits of retrofitting with MAI panels. The average air temperatures during the nominal operational hours (8 AM – 7 PM) were observed to be similar between the two buildings, but during the evening and overnight hours the air temperature dropped to discernibly lower values in the baseline building. Thus, the MAI panels were effective in reducing the temperature drop during the thermostat setback hours. Table 4 compares the average (‘Avg.’) air temperatures during the operational hours and the minimum (‘Min.’) air temperatures during the subsequent evening and overnight hours. For clarity, the corresponding date and time periods used for calculating the average and minimum temperatures are listed in Table 4. The impact of the MAI panels added to the retrofit building are clearly observed in the overnight drops in the air temperature near the SE and SW walls of the retrofit vs. baseline buildings; the overnight temperature drops were 1.5–2.5 °C lower in the retrofit building.
K. Biswas, T. Patel and S. Shrestha et al. / Energy & Buildings 203 (2019) 109430
7
Fig. 10. Comparison of exterior surface temperatures and room air temperatures adjacent to the interior wall surfaces in the baseline (B) and retrofit (R) buildings. Table 4 Comparison of room air temperatures in the retrofit and baseline buildings. Retrofit building
Avg. (Jan 18, 8 AM Min. (Jan 18, 7 PM Difference Avg. (Jan 19, 8 AM Min. (Jan 19, 7 PM Difference Avg. (Jan 20, 8 AM Min. (Jan 20, 7 PM Difference
– 7 PM) – Jan 19, 6 AM) – 7 PM) – Jan 20, 6 AM) – 7 PM) – Jan 21, 6 AM)
Baseline building
NE
SE
SW
NW
NE
SE
SW
NW
21.5 19.0 2.5 21.6 19.2 2.5 21.7 19.3 2.4
22.7 18.3 4.4 23.1 17.8 5.3 22.7 18.0 4.7
22.9 18.2 4.7 23.1 17.8 5.3 22.8 18.1 4.7
26.8 18.7 8.1 26.4 18.7 7.7 26.6 18.9 7.7
18.7 16.3 2.5 19.2 16.6 2.6 19.0 16.5 2.4
21.3 15.1 6.2 22.0 15.5 6.5 21.7 15.6 6.1
21.8 14.8 7.0 22.8 15.4 7.4 22.7 15.4 7.3
21.5 15.2 6.3 22.3 15.7 6.6 21.9 15.7 6.2
Conversely, near the NW wall the temperature drops in the retrofit building were higher; it is noted that the daytime (operational period) air temperatures near the NW wall were about 4 °C higher than near the SW and SE walls, which can partly explain the larger overnight drops. The reasons for the higher daytime temperatures near the NW wall are unknown. The thermally-massive NE wall behaved similarly in both the retrofit and baseline buildings, with no discernible impact of the added MAI panels. Fig. 11 compares the measured heat fluxes between the retrofit and baseline buildings. There are no readily discernible trends to indicate the overall improvement in the thermal performance of retrofit building. The heat flux magnitudes through the NE and SE walls are very close between the retrofit and baseline buildings. At the SW wall, the peak heat loses are greater in the baseline building and vice-versa in the retrofit building. It is noted that the heat fluxes are based on single-point measurements and may not be considered representative of the entire walls in the two buildings. Energy modeling has been utilized to estimate the overall thermal performance of the walls and their impact on energy consumption.
5.2. Model benchmarking results EnergyPlus models were created to match the building geometry and construction, and benchmarked by comparing the hourly calculations of the temperatures and heat fluxes with the measurements from the baseline and retrofit buildings. Figs. 12 and 13 compare the EnergyPlus-calculated and measured heat flows from different walls of the retrofit building during two different time periods in January and April 2017. While the calculations follow the temporal trends of the measurements, different levels of agreements were observed at different times and for different wall orientations. The calculated peak heat losses were observed to be in excellent agreement with the measurements, except for the thermally-massive NE wall. Greater discrepancies were observed between the hours of 12 PM and midnight during each 24-hour period. Fig. 14 compares the measured and calculated temperatures at the interior surfaces of the walls of the retrofit buildings during April 2017. The measurements are from thermistors located next to the HFTs on the walls. Similar to the heat flow results, the cal-
8
K. Biswas, T. Patel and S. Shrestha et al. / Energy & Buildings 203 (2019) 109430
Fig. 11. Comparison of heat flows through the walls of the baseline (B) and retrofit (R) buildings.
Fig. 12. Comparison of measured and calculated heat flows through the walls of the retrofit building during January 11–15, 2017.
K. Biswas, T. Patel and S. Shrestha et al. / Energy & Buildings 203 (2019) 109430
9
Fig. 13. Comparison of measured and calculated heat flows through the walls of the retrofit building during April 16–20, 2017.
culated temperatures were in excellent agreement with measurements during the overnight hours, while discrepancies of 1–1.5 °C were observed during the peak temperature periods. Calculations and measurements from the baseline building were also compared and showed similar trends as the retrofit building. The exact thermal characteristics of the NE concrete wall were not known making it difficult to model the NE wall accurately. Two statistical metrics typically used to assess how the simulation results (Calc.) compare with measured data (Meas.) are normalized mean bias error (NMBE) and coefficient of variance of root mean square error (CVRMSE). However, these metrics are not suitable for parameters like heat flow that fluctuate between positive and negative values and have values that are often very close to zero, i.e. their overall average would approach zero. Therefore, the daily maximum and minimum values of the measured and calculated heat fluxes from January 1, 2017 to April 30, 2017 were selected for assessment using NMBE and CVRMSE. The maximum values typically correspond to the peak daytime heat gains into the building and the minimum values correspond to peak nighttime heat losses. NMBE and CVRMSE are calculated as follows [5].
N NMBE =
CV RMSE =
i=1
Meas.i − Calc.i
× 100;
(N − 1 ) × Meas. 2 1 N i=1 (Meas.i − Calc.i ) N−1 Meas.
× 100
(1)
In Eq. (1), ‘N’ is the number of observation points. The acceptable range of the statistical metrics for hourly results is ±10% for NMBE and ±30% for CVRMSE [5]. Table 5 lists the NMBE and CVRMSE of the daily minimum heat fluxes from the different walls of the retrofit and baseline buildings. Most of the NMBE and
CVRMSE values are within acceptable ranges except the NE wall of the retrofit building, NMBE of the SE wall of the retrofit building and NW wall of the baseline building. NMBE and CVRMSE of the daily maximum (i.e. peak daytime) heat fluxes were too far outside the acceptable limits and have not been listed here. The most probable reason for the discrepancies between the measured and calculated heat flows, especially during daytime hours, is the use of solar irradiance data from an offsite location due to the failure of the onsite solar tracker. Fig. 15 compares the solar irradiance on a horizontal surface, or GHI, measured using the onsite pyranometer and from the offsite weather station (‘Wea. Stn.’), and the differences are clearly observable. Further, assumptions were made about the wall construction in the model such as defect-free construction, uniform and homogeneous insulation in the cavities, etc., which could not be verified. 5.3. Annual energy savings Following the benchmarking of the models against measured data, these models were utilized for annual energy simulation and evaluation of energy savings due to the MAI retrofit. TMY weather data were used for the annual simulations. Lighting, internal load, occupancy, HVAC operation, and thermostat setpoint schedule were used from the DOE commercial prototype building model of a small office[27]. The heating and cooling set points were assumed to be 21.1 °C and 23.9 °C during operational hours and changed to 15.6 °C and 29.4 °C during non-operational hours. The main focus is on the reductions in cooling and heating energy consumption in the retrofit building. The model assumed cooling using electricity and heating via natural gas, the same as the real buildings.
10
K. Biswas, T. Patel and S. Shrestha et al. / Energy & Buildings 203 (2019) 109430
Fig. 14. Comparison of measured and calculated temperatures at the interior wall surfaces of the retrofit building during April 16–21, 2017. Table 5 NMBE and CV-RMSE of daily minimum heat fluxes. Retrofit building
NMBE CVRMSE
Baseline building
NE
SE
SW
NW
NE
SE
SW
NW
33.2 −40.8
24.4 −26.9
3.0 −6.9
2.5 −5.4
9.8 −20.2
−7.4 −10.1
−2.0 −7.4
−30.5 −37.2
Fig. 15. Comparison of solar incidence on a horizontal surface measured on-site and from an off-site weather station.
Fig. 16 compares the calculated monthly cooling electricity and natural gas consumption between the baseline and retrofit buildings. Since Ft. Drum lies in a heating dominated climate zone, overall gas consumption was much higher than cooling electricity use. It is noted that, during some months, the calculated electricity consumption of the retrofit building was higher than the baseline building, which is counter-intuitive. One probable reason for this is
that the retrofit building captured and retained the heat from internal loads better than the baseline facility. As expected, the calculated natural gas usage was consistently lower for the retrofit facility. It is further noted that the both the baseline and retrofit buildings studied here were relatively newly-constructed buildings and contained cavity insulation with nominal thermal resistance of
K. Biswas, T. Patel and S. Shrestha et al. / Energy & Buildings 203 (2019) 109430
11
Fig. 16. Comparison of calculated monthly cooling electricity and natural gas consumption in the baseline and retrofit buildings based on annual energy simulations. Table 6 Calculated annual consumption of electricity for cooling and natural gas for heating.
Electricity (kWh) Natural gas (kWh)
Baseline
Retrofit
% Difference
Baseline∗
Retrofit∗
% Difference
1060.7 14,414.8
1028.4 12,616.9
−3.1 −12.5
1458.2 26,165.0
1090.9 14,032.4
−25.2 −46.4
Table 7 Material, installation and energy costs assumed for the preliminary cost analysis. Item
Unit cost
Unit
MAI PIR Installation Electricity Natural gas
$30.90 $6.70 $11,220 $0.073 $3.83
m2 m2 kWh GJ
ongoing efforts are expected to yield more accurate estimates of investment costs and payback periods of building retrofits. In addition, team members are continuing to perform research on MAI and VIP technologies to improve thermal resistance and long-term performance by investigating different core materials, barrier films and sealing methods as well as to reduce cost by investigating new processing methods. Based on the outcomes of the ongoing research, the updated cost analysis results may be quite different. 6. Summary and conclusions
3.4 m2 K/W. Older buildings needing retrofit are not nearly as well insulated. Thus, to further evaluate benefits of retrofitting an old building with MAI, additional simulations were performed by assuming no cavity insulation. Table 6 compares the calculated annual cooling and heating energy use for both scenarios, with and without cavity insulation. ‘Baseline∗ ’ and ‘Retrofit∗ ’ represent cases without any cavity insulation and, as expected, much higher savings are predicted for this scenario. In reality, by retrofitting an old building of similar construction and under similar operating conditions using MAI panels, the percentage savings can be expected to be somewhere in between the cases listed in Table 6. A preliminary cost analysis was performed using the Building Life Cycle Cost (BLCC) program from National Institute of Standards and Technology (NIST) [29]. Table 7 lists the costs assumed for this analysis. The cost of MAI is a projection assuming large-scale production. For the current retrofit, 117.2 m2 and 38.6 m2 of MAI and PIR insulation were used. The installation costs assumed $60/hour and estimated the cumulative hours needed for installation. Finally, a 16% overhead was added to the total costs. Using the costs listed in Table 7 and assuming a building with no pre-existing insulation in the exterior walls, a simple payback of 108 years was calculated. Biswas et al. [17] reported the retrofit of a low-slope roof using foam-VIP composites, which could be installed similarly to regular insulation boards. If similar foam-VIP or foam-MAI composites could be used, the installation time is expected to be small fraction of the time taken to install individual MAI panels. Thus, another cost analysis was performed based on material costs alone and it resulted in a simple payback of 28 years. Members of this research team are currently engaged in projects on further demonstrations of MAI for building retrofits as well as evaluating the long-term performance of MAI panels. These
This article describes the use of low-cost VIPs, called modified atmosphere insulation or MAI, for retrofitting the exterior walls of a one-story office building located in Ft. Drum, New York. Two buildings of near-identical construction were studied; one building was left unaltered and served as the baseline and the other was retrofitted with MAI. Thermal performance of the baseline and retrofitted walls was analyzed using in-situ temperature and heat flow sensors. The measured data were used to benchmark EnergyPlus models, which were subsequently used to estimate the annual energy savings due to the addition of MAI panels to the exterior walls. An important outcome of this project was that VIPs/MAI panels proved to be a feasible and durable option for retrofitting building envelopes, with a failure rate of 1.5% during installation. The walls retrofitted with MAI panels were more effective in regulating the indoor temperatures and reducing the temperature drops during the thermostat setback hours during nights and weekends. EnergyPlus models predicted annual reduction of 12.5% in the natural gas consumption for heating in the retrofitted building compared to the baseline building. It is important to note that both the baseline and retrofit buildings contained well-insulated exterior wall cavities. For an older building with no wall insulation, the heating energy savings were estimated to be 46%. Preliminary cost analyses indicated payback periods of about 28 years based on projected cost of the MAI panels used for the current retrofit study. Members of this research team and, the VIP industry in general, are continuing to perform research, development and demonstration work related to MAI and VIPs. The research work includes evaluation of long-term performance, improving performance by investigating new core and film materials, and reducing cost by altering production methods. Additional demonstrations in real
12
K. Biswas, T. Patel and S. Shrestha et al. / Energy & Buildings 203 (2019) 109430
buildings and verification of adequate long-term performance are needed to increase the adoption of VIPs in buildings.
Table B1 Measured thermal resistance of foam-MAI composite boards and individual components. k (W/(mK))
Declaration of Competing Interest One of the co-authors, Dr. Douglas Smith, is the President of NanoPore Inc, which is a manufacturer of VIPs. Beyond that, the authors declare that there are no conflicts of interest.
PIR (outer layer) 0.026 MAI (center layer) 0.0049 PIR (center layer) 0.026 HD PIR (outer layer) 0.029 Composite board (RTotal ) MAI-PIR (center layer) (RMAI-PIR )
Thickness (mm)
R (m2 •K/W)
12.7 25.4 25.4 12.7
0.49 5.18 0.98 0.44 4.46 3.53
Acknowledgments Funding for this project was provided by the Environmental Security Technology Certification Program (ESTCP) under project EW201512. ORNL authors are supported by the Building Technologies Office of the U. S. Department of Energy under Contract No. DEAC05-00OR22725 with UT-Battelle, LLC. We thank Mr. Philip Childs, Mr. Anthony Gehl, Mr. Jerald Atchley and Mr. Bradley Brown of ORNL for calibration, set up and troubleshooting of all sensors and data acquisition systems as well as Mr. Rohit Jogineedi, a research intern at ORNL, for creating computer aided drawings. We also thank Mr. Lake Lattimore, Ms. Megan Kreiger, and Dr. Andrew Nelson of ERDC-CERL and Mr. Ron Esparza of NanoPore for their support in project execution and the installation of MAI panels. We are grateful to Mr. Steve Rowley, Energy Manager at Ft. Drum, for his efforts to coordinate between the project team and Ft. Drum staff. We thank the Ft. Drum safety office for allowing us to retrofit one of their facilities and the Ft. Drum carpentry shop for their support during the installation of the sensors in the buildings and MAI installation process. Supplementary materials Supplementary material associated with this article can be found, in the online version, at doi:10.1016/j.enbuild.2019.109430. Appendix A. MAI-related patents •
•
•
•
Douglas M. Smith, US Patent 9,133,973. Method of using thermal insulation products with non-planar objects, September 2015. Douglas M. Smith, US Patent 9,598,857. Thermal insulation products for insulating buildings and other enclosed environments, March 2017. Douglas M. Smith, US Patent 9,726,438. Production of thermal insulation products, August 2017. Douglas M. Smith, US Patent 9,849,405. Thermal insulation products and production of thermal insulation products, December 2017.
Appendix B. Approximation of overall thermal resistance using COP conductivities of MAI panels Biswas et al. [17] measured the overall thermal resistance of foam-MAI composites using a guarded hot box, following ASTM 1363 (Standard Test Method for Thermal Performance of Building Materials and Envelope Assemblies by Means of a Hot Box Apparatus). The composite consisted on an array of 25.4 mm thick MAI panels sandwiched by 12.7 mm of regular PIR and 12.7 cm highdensity (HD) PIR. There were gaps between the MAI panels that were filled with regular PIR. The area fraction of MAI in the center layer was 89.8%. The overall thermal resistance of the composite board was measured according to ASTM C1363. Table B1 lists the measured conductivity (k), thickness and resistance (R) of the different layers and total measured resistance of a composite board. The conductivities listed in Table B1 were measured via ASTM C518 using materials similar to those used in the foam-MAI composite
board. The second-to-bottom row of Table B1 shows the total measured resistance of the composite board (RTotal ), which includes the impacts of the edge effects of the MAI panels. The bottom row shows the resistance of the center MAI-PIR layer (RMAI-PIR ), which was obtained by subtracting the measured resistance of the sandwiching regular and HD PIR layers from RTotal . Next, RMAI-PIR was estimated via an area-weighted parallel path calculation using COP MAI resistance and area fraction of MAI as follows:
1 AMAI (1 − AMAI ) = + RMAI−PIR RCOP, MAI RPIR ‘AMAI ’ is the area fraction of the MAI panels in the center layer, which is 0.898. RCOP,MAI and RPIR in the center layer are 5.18 and 0.98 m2 K/W, respectively, from Table B1. Thus, using only COP thermal conductivity/resistance of the MAI panels, the value of RMAI-PIR is 3.60 m2 K/W, which is within 2% of the experimentally determined RMAI-PIR of 3.53 m2 K/W (which includes the edge effects). It is noted that the edge effects will be different in the retrofit building at Ft. Drum compared to the values listed in Table B1 due to differences in the overall configurations. References [1] L. Yang, H. Yan, J.C. Lam, Thermal comfort and building energy consumption implications–a review, Appl. Energy 115 (2014) 164–173. [2] E. Di Giuseppe, M. Iannaccone, M. Telloni, M. D’Orazio, C. Di Perna, Probabilistic life cycle costing of existing buildings retrofit interventions towards nZE target: methodology and application example, Energy Build. 144 (2017) 416–432. [3] G. Salvalai, M.M. Sesana, G. Iannaccone, Deep renovation of multi-storey multi-owner existing residential buildings: a pilot case study in Italy, Energy Build. 148 (2017) 23–36. [4] L. Rodrigues, J. White, M. Gillott, E. Braham, A. Ishaque, Theoretical and experimental thermal performance assessment of an innovative external wall insulation system for social housing retrofit, Energy Build. 162 (2018) 77–90. [5] S. Shrestha, A. Pagan-Vazquez, D. Chu, M. Kreiger, K. Biswas, Evaluation of energy efficiency of U.S. army hard shelters, Buildings XIII: Thermal Performance of Exterior Envelopes of Whole Buildings, 2016. [6] P. Jones, X. Li, E. Perisoglou, J. Patterson, Five energy retrofit houses in South Wales, Energy Build. 154 (2017) 335–342. [7] D. Farmer, C. Gorse, W. Swan, R. Fitton, M. Brooke-Peat, D. Miles-Shenton, D. Johnston, Measuring thermal performance in steady-state conditions at each stage of a full fabric retrofit to a solid wall dwelling, Energy Build. 156 (2017) 404–414. [8] J. Kosny, A. Fallahi, N. Shukla, Cold Climate Building Enclosure Solutions, National Renewable Energy Lab.(NREL), Golden, COUnited States, 2013. [9] M. Alam, H. Singh, M.C. Limbachiya, Vacuum Insulation Panels (VIPs) for building construction industry - a review of the contemporary developments and future directions, Appl. Energy 88 (11) (2011) 3592–3602. [10] S.S. Alotaibi, S. Riffat, Vacuum insulated panels for sustainable buildings: a review of research and applications, Int. J. Energy Res. 38 (1) (2014) 1–19. [11] B.P. Jelle, Traditional, state-of-the-art and future thermal building insulation materials and solutions - Properties, requirements and possibilities, Energy Build. 43 (10) (2011) 2549–2563. [12] P. Mukhopadhyaya, D. MacLean, J. Korn, D. van Reenen, S. Molleti, Building application and thermal performance of vacuum insulation panels (VIPs) in Canadian subarctic climate, Energy Build. 85 (2014) 672–680. [13] P. Johansson, B. Adl-Zarrabi, A.S. Kalagasidis, Evaluation of 5 years’ performance of VIPs in a retrofitted building façade, Energy Build. 130 (2016) 488–494. [14] E. Sveipe, B.P. Jelle, E. Wegger, S. Uvsløkk, S. Grynning, J.V. Thue, B. Time, A. Gustavsen, Improving thermal insulation of timber frame walls by retrofitting with vacuum insulation panels–experimental and theoretical investigations, J. Build. Phys. 35 (2) (2011) 168–188.
K. Biswas, T. Patel and S. Shrestha et al. / Energy & Buildings 203 (2019) 109430 [15] F. Ascione, R.F. De Masi, R.M. Mastrullo, S. Ruggiero, G.P. Vanoli, Experimental investigation and numerical evaluation of adoption of multi-layered wall with vacuum insulation panel for typical Mediterranean climate, Energy Build. 152 (2017) 108–123. [16] K. Biswas, J. Rose, L. Eikevik, M. Guerguis, P. Enquist, B. Lee, L. Love, J. Green, R. Jackson, Additive manufacturing integrated energy-enabling innovative solutions for buildings of the future, J. Sol. Energy Eng. Trans. ASME 139 (1) (2017). [17] K. Biswas, A. Desjarlais, D. Smith, J. Letts, J. Yao, T. Jiang, Development and thermal performance verification of composite insulation boards containing foam-encapsulated vacuum insulation panels, Appl. Energy 228 (2018) 1159–1172. [18] S. Shrestha, W. Miller, T. Stovall, A. Desjarlais, K. Childs, W. Porter, M. Bhandari, S. Coley, Modeling PCM-enhanced insulation system and benchmarking EnergyPlus against controlled field data, in: Proceedings of Building Simulation, 2011. [19] J. Ferdyn-Grygierek, K. Grygierek, HVAC control methods for drastically improved hygrothermal museum microclimates in warm season, Build. Environ. 149 (2019) 90–99. [20] G. Martinopoulos, A. Serasidou, P. Antoniadou, A.M. Papadopoulos, Building integrated shading and building applied photovoltaic system assessment in the energy performance and thermal comfort of office buildings, Sustainability 10 (12) (2018) 4670.
13
[21] A. ASHRAE, ANSI/ASHRAE/IES Standard 90.1-2016, Energy Standard for Buildings Except Low-Rise Residential Buildings, 2016. [22] M. Bhandari, S. Shrestha, J. New, Evaluation of weather datasets for building energy simulation, Energy Build. 49 (2012) 109–118. [23] ASTM C518-17, Standard Test Method for Steady-State Thermal Transmission Properties by Means of the Heat Flow Meter Apparatus, ASTM International, 2017. [24] K. Biswas, A. Desjarlais, T. Jiang, T. Patel, A. Nelson, D. Smith, Demonstration of the hygrothermal performance of a next-generation insulation material in a cold climate, Advances in Hygrothermal Performance of Building Envelopes: Materials, Systems and Simulations, ASTM International, 2017. [25] OpenStudio SketchUp Plug-in. Available from: https://nrel.github.io/ OpenStudio- user- documentation/. [26] ASHRAE Handbook - Fundamentals. 2013; Available from: https://www.ashrae. org/technical-resources/ashrae-handbook. [27] Commercial Prototype Building Models. Available from: https://www. energycodes.gov/development/commercial/prototype_models. [28] Wilcox, S. and W. Marion. Users manual for TMY3 data sets, NREL/TP-58143156. Available from: https://www.nrel.gov/docs/fy08osti/43156.pdf. [29] Building Life Cycle Cost Programs. Available from: https://www.nist.gov/ services-resources/software/building-life-cycle-cost-programs.