Development of Interactive Internet-Based Building ... - KU Leuven

Development of Interactive Internet-Based Building Envelope Materials Database for Energy Plus and other Whole-Building Energy Simulation Programs Jan Kosny PhD. Oak Ridge National Laboratory Oak Ridge, TN USA

INTRODUCTION In current residential and small commercial buildings, over 50% of the whole-building energy consumption is generated by heat transfer and air leakage through building envelope components. Thus it is essential to accurately represent full complexity of building envelope in energy analysis. A rapid development of energy-efficient building materials and systems means a single change in a building envelope or HVAC configuration will no longer significantly improve whole-building energy consumption in future buildings. Only an optimized combination of subsystems may cause notable changes in energy use. At the same time, a majority of building envelope designers and energy modelers understand only basic heat transfer principles and operates only in 1-D environment. Requesting 3-D transient heat transfer analysis for each envelope component seems unrealistic. That is why necessary computational tools and supporting data must be widely available and simple to apply. The first interactive envelope material database was introduced by Oak Ridge National Laboratory (ORNL) in 1994 (Kosny and Desjarlais 1994, Christian and Kosny 1996). At the beginning, it was oriented only toward whole-wall R-value calculations but later several new components like thermal mass, equivalent wall, and component air leakage have been added. The concept of Interactive Envelope Materials Database for Whole-Building Energy Simulation Programs was developed at ORNL to enable an accurate, fast, and simple parametric analysis of building energy consumption. This database uses several already existing subroutines, experimental results, and calculation techniques. The main purpose of this database is to serve energy modelers by enabling component analysis during whole-building energy simulations. However, it can also be utilized for performance comparisons between different envelope technologies. The user only selects the material configuration and sets dimensions. Later, the Internet program calculates whole-wall/roof/ or floor R-values, generates 3-D dynamic thermal characteristics, and calculates detailed air leakage for the selected building envelope system. Next, this information is converted into the format required by programs such as BLAST, DOE-2, or ENERGY PLUS. This paper summarizes the theoretical foundations for this new approach and presents some examples of component analysis for residential buildings. DESIGNING OF LOW ENERGY BUILDINGS AND THE NEED FOR PARAMETRIC ANALYSIS Since the 1970s, several zero-energy buildings have been constructed in different countries and in a wide variety of climatic conditions. The main lesson learned from these exercises was that while it is possible to design and construct a million-dollar zero-energy house, the real engineering challenge is to build such a house for a low-income family. A proper balance between the cost of “high-tech” materials and equipment and the reduction of whole-building energy consumption is critical for designing affordable low-energy buildings. The most effective way to optimize the building envelope is a parametric analysis of all

1

1

components. The following components and building characteristics are usually considered during this process: • • • • • • • •

geometry and orientation of the building whole-wall, whole-roof, and whole-floor R-values (all 3-D thermal bridges) dynamic thermal characteristics of all building envelope components application and configuration of thermal mass foundation type and foundation insulation size and location of glazing solar gain control systems inherent air leakage characteristics of main building envelope systems and interface connections • location of air ducts, etc.

To illustrate the importance of the parametric analysis during designing of low-energy buildings, Figure 1 shows potential energy savings calculated for four different configurations off massive walls of the same R-value. Massive walls are compared against R-3.5 wood-framed walls.

Energy savings M Wh/y

2.5 2

1.94

R-20 massive walls R-3.5m2K/W

1.97

1.41

1.5

1.25

1 0.5

Energy required by compared 143 m2 one-story rancher made of R-3.5 wood-framed walls: 19.60 MWh/y

0 CIC

CIC: Concrete 7.6-cm. Foam 10.2-cm. Concrete 7.6-cm.

Int.mass

ICI

Int.mass: Foam 10.2-cm. Concrete 15.2-cm.

Ext.mass

ICI: Foam 5.1-cm. Concrete 15.2-cm.. Foam 5.1-cm.

Ext.mass: Concrete 15.2-cm. Foam 10.2-cm. .

Figure 1. Energy savings comparisons for R-3.5 massive walls computed for Boulder, CO.

As shown on Figure 1, a simple change in configuration of the same wall materials brought about a 5% savings in total energy consumption. Adding 5-cm. thick layer of exterior EPS sheathing can produce similar energy effect. This example demonstrates that sometimes it might be wise to optimize a configuration of building envelope materials before making recommendations for a costly addition of thermal insulation. ACCURACY PROBLEMS IN DYNAMIC ENERGY MODELING

2

2

Most whole-building energy simulation programs require 1-D descriptions of envelope components. Unfortunately, proper analysis of complex thermal envelope systems sometimes requires application of advanced 3-D transient heat transfer analytical tools. This situation creates many accuracy problems in whole-building energy modeling and complicates the lives of many building envelope designers and energy modelers. It might also generate uncertainties in sizing HVAC equipment because of inaccuracies in building load calculations. To reduce the cost of the process and minimize the potential for inaccuracies, a method of developing architectural component descriptions in simulation programs has to be as simple as selecting the specific material configuration, setting dimensions, and determining building orientation. Inaccuracies in Approximating Thermal Bridges Thermal bridges created by highly conductive structural materials can significantly reduce local R-values and change the transient response for building envelope details. A simple thermal modeling exercise illustrates the differences between heat flow calculated using a simplified 1-D model and using more-complicated (closer to reality) models. Three walls with 20% framing factor were simulated. Three different framing materials were assumed for thermal modeling: wood, 0.116 Wm/K; concrete, 1.40 Wm/K; and steel, 46.20 Wm/K. Expanded Polystyrene (EPS) foam (0.035 Wm/K) served as a wall insulation. Figure 2 shows that differences in R-value estimations depend on the thermal conductivity ratio between framing and insulation materials and the number of framing material inserts. For the traditional method of describing a wall in whole-building modeling input files (Case I), errors in R-value calculations may exceed 44% for steel framing and 27% for concrete framing while being less than 2% for wood framing. 20% framing material

80% thermal insulation 1.3 cm. 8.9 cm. 1.3 cm. gypsum board

Difference in R-value calculations are computed using the case with a single framing material insert as a base for comparisons. Errors are computed against cases with two and four framing material inserts.

1.3 cm. 8.9 cm. 1.3 cm. gypsum board 1.3 cm. 8.9 cm. 1.3 cm. gypsum board

Framing material Wood Concrete Steel

Ratio between thermal cond. Insul./Fram. 3 40 1330

R-value differences Base case 1 insert 2 inserts 4 inserts 1.4 % 1.8 % 17.8 % 27.5 % 28.1 % 44.4 %

Figure 2. Results of comparisons between R-value estimation for three walls of the same framing factors.

The second example shows how hand calculations can easily generate errors in R-value estimations for residential attic. Assume that a building (8.5 × 16.8 m) has a pitched roof with rafters installed at 40.6 cm o.c., and a high point in the ridge of 1.6 m. Nominal roof insulation is R-8.8 m2AW/K (thermal conductivity - 0.0417 W/mAK,. thickness - 36.8 cm.). The ceiling is hung to the wood joists (25.4 × 1.2 cm) installed at 40.6 cm o.c. and finished with 1.2-cm layer of

3

3

gypsum board with thermal conductivity of 0.16 W/mAK. In this case, some modelers may make the following assumption in their input files: Attic Materials: 1. Gypsum board 2. Insulation Total R-value

thickness 1.2 cm thickness 36.8 cm R-8.88 m2AW/K

thermal conductivity 0.16 W/mAK thermal conductivity 0.0417W/mAK

However, the roof has a triangular shape. Therefore, on 22% of the attic area, the declining roof surface reduces the thickness of the attic insulation. Consequently, the average insulation thickness is not 36.8 cm, but 30 cm. Moreover, wood joists penetrate the insulation at 40.6 cm o.c. Based on all these facts, effective attic R-value is reduced by about 30%. DOE 2.1E simulation performed on this house for Atlanta climate yielded about 4% differences in energy consumption. Note that in that case, when configuration of the attic was relatively simple, it took about half an hour to perform all computations. It is unlikely that similar calculations would be made manually for more complex structures. Steel framing is considered more difficult to analyze. Let’s assume the building used in the previous example has 2 × 4 steel stud walls insulated with fiberglass batts. Steel studs are installed at 40.6 cm o.c. On the exterior, the wall is finished with a 1.2-cm layer of plywood and wood siding. Some energy modelers probably will make the following assumptions: Exterior walls materials: 1. Gypsum board thickness 1.2 cm 2. Insulation thickness 8.9 cm 3. Steel studs web depth 8.9 cm 4. Plywood thickness 1.2 cm 5. Wood siding

thermal conductivity thermal conductivity thermal conductivity thermal conductivity R- 0.17 m2AW/K

Wall heat losses

Wall area

0.51

0.67

0.01 0.10 0.06

0.16 W/mAK 0.041 W/mAK 45 W/mAK 0.115 W/mAK

0.01

0.05

0.16 0.14

0.07

0.13

0.10

clear wall corner

roof/wall wall/ceil

wind. perimeter door perimeter

Figure 3. Distributions of the elevation area and wall heat losses for a conventional 8.9-cm steel stud wall.

Clear wall R-value calculated using the Modified Zone Method (ASHRAE 2001a) is R-1.16 m2AW/K. As depicted on Figure 3, the clear wall represents only 67% of the whole opaque area of the elevation for a considered building. Because wall details generate about 50% of the total heat transfer, the whole-wall R-value is much closer to reality than the clear wall R-value. In our

4

4

example, the whole-wall R-value is R-0.94 m2AW/K (about 18% less than the clear wall R-value). Two DOE 2.1E simulations were performed for Atlanta climate, with both clear-wall and wholewall R-value inputs for a steel stud wall. These computations yielded a 4.5% difference in predicted total energy consumption. In both simulation exercises, miscalculations of energy consumption were below 5%. However, similar miscalculations for other components could easily result in a total combined error exceeding 10%. Very often, material configurations are more complex than those in the above examples. They require application of computer simulation tools for R-value estimations. It is likely that the modeler would prefer to use approximate values instead of performing complex and timeconsuming calculations. Thus it is important to offer a simple Internet-based tool for this type of calculation. Inaccuracies in 1-D Simplifications of Complex Building Envelope Assemblies An insulated concrete form (ICF) wall was analyzed, as shown on Figure 4. Inside this wall, there are 3-D networks of vertical and horizontal channels that are filled with concrete during construction of the wall. Response factors, heat capacity, and R-value were computed using validated finite difference model. They enabled generation of a series of 1-D equivalent walls. Equivalent wall has a simple multilayer structure. Its dynamic thermal behavior is identical to that of the actual wall (Kossecka and Kosny 1997). Physical properties of equivalent wall can be used in whole-building energy simulation programs.

Figure 4. Insulated concrete form wall made of EPS shells.

A simple 1-D model was developed for the ICF wall as well. Because computer programs such as DOE-2, BLAST, or Energy Plus can perform only 1-D thermal analysis, it is likely that most whole-building modelers would make a similar 1-D simplification. R-value calculated for the ICF wall using simple 1-D model was 38% higher from R-value calculated using detailed 3-D simulation. At the same time, R-values for the ICF wall and for the equivalent walls were equal.

5

5

Simple DOE-2.1E modeling was performed on a previously used ranch house to illustrate how inaccuracy in 1-D descriptions can affect simulation of cooling and heating energies. The following three wall descriptions were developed: (I) (II) (III)

1-D Equivalent wall generated for the EPS form wall using finite difference program. Simple 1-D multilayer description of the material configuration created to match approximate wall dimensions. Simple 1-D multilayer description of the material configuration created to match approximate wall dimensions and R-value. Modelers would use this method if they know the exact wall R-value.

DOE-2.1E input files were developed using all three 1-D approximations. Equivalent wall case was used as a base for comparisons. For cases II and III, differences in predicted energy consumption for six US climates were between 3 and 19% (Kosny and Kossecka 2000). COMPONENT AIR LEAKAGE DATA IN WHOLE BUILDING ENERGY SIMULATIONS

Analyzing component air leakage during whole-building energy simulations can be complex and time-consuming. It has been investigated by Dickerhoff et al. (1982), Harrje and Born (1982), and Colliver et al. (1992). They attributed, on an average, 35% of air leakage to exterior walls, 18% to ceiling details, and 15% to details such as windows and doors (ASHRAE 2001a). These proportions may vary depending on factors such as materials used, building geometry, and complexity of architectural details. Unfortunately, data presented by ASHRAE Handbook of Fundamentals 2001 (ASHRAE 2001a, Table 1, Chapter 26) represents only a limited number of building envelope technologies and sealing techniques. At the same time, the type of building envelope determines whole-building air leakage characteristics. A house built of structural insulated panels (SIPs) has different air leakage characteristics from the house made of ICFs or steel framing. Moreover, most building components demonstrate different air leakage characteristics when matched with different building envelope technologies. Therefore, ideally, each building envelope system would need its own list of component air leakage values (characteristic for each standard components and sealing techniques). This data should be based on experimental results accomplished using a uniform and widely accepted testing procedure. A two-story residential building of floor area 282 m2 was used to illustrate the effect of component air leakage generated by wall/floor detail on whole building energy consumption. “Best Estimate”, air leakage values (ASHRAE 2001a) were used for all building components except wall/floor detail. For wall/floor detail, an ASHRAE-recommended maximum value for unsealed detail (5.45-cm2/lnm.) was compared with air leakage value measured by ORNL for proprietary sealant (0.57-cm2/lnm.) (Kosny 2003). Later, Effective Air-leakage Area (ELA) was computed for each building configuration. Data presented in Figure 5 shows that even a single wall component may significantly influence results of the whole building energy simulations. That is why, application of experimentally verified component air leakage characteristics can be critical for accuracy of whole building energy simulations.

6

6

Difference 43%

0.12

2

ELA [m ]

0.1 0.08 0.06 0.04 0.02 0 ASHRAE best estimate

Difference 13.5%

Gas consumption [Therm]

Electric energy [K Wh]

1100

Proprietary sealant

1050 1000 950 900 850 ASHRAE best estimate

Proprietary sealant

1600

Difference 17.5%

1500 1400 1300 1200 1100 1000 ASHRAE best estimate

Proprietary sealant

Figure 5. Results of ELA calculations and electric and gas consumption predictions based on whole building energy simulations for Minneapolis climate data.

MATERIALS DATABASE FOR WHOLE-BUILDING ENERGY SIMULATION PROGRAMS In time, when rapid development of energy-efficient technologies made possible significant reductions in whole building energy consumption, parametric analysis is a key advantage of using whole-building energy simulation tools. Because it is unrealistic to expect serious improvement in whole-building energy consumption from a change in a single building component, designers of energy-efficient buildings have to treat a building as a collection of subsystems generating insignificant energy effects if they are analyzed separately. That is why, a concept of Interactive Envelope Materials Database for Whole-Building Energy Simulation Programs was developed at ORNL to reinforce an accurate, fast, and simple parametric analysis of building energy consumption. In 1994, ORNL introduced a whole-wall R-value procedure (Kosny and Desjarlais 1994). It was based on hot-box test results and 3-D heat conduction simulations. Whole-wall R-value combines thermal performance of the clear wall area with typical envelope interface details, including wall/wall (corners), wall /roof, wall/floor, wall/door, and wall/window connections. Results from these detailed simulations are combined into a single whole-wall R-value and compared with simplified “center-of-cavity” and “clear-wall” R-values. The Whole Wall Thermal Performance Calculator (Christian and Kosny 1996) is available on the Internet at http://www.ornl.gov/roofs+walls/calculators. In 1996, ORNL developed the equivalent wall concept (Kossecka and Kosny 1996), which transforms complex 3-D thermal characteristics of building envelope components into simple

7

7

one-dimensional equivalents. A potential application of the equivalent wall theory in wholebuilding energy simulations was analyzed by ASHRAE project TRP-1145 (ASHRAE 2001b). Three novel testing procedures were introduced by ORNL to collect experimental data on component air leakage (Kosny 2003). These procedures enabled separate air leakage analysis for building envelope details, such as window and door perimeter, wall/ foundation intersection, wall/ceiling intersection, and wall/roof connection. At the beginning, a series of tests were performed on conventional wood-frame technology. Intersections incorporating a concrete basement wall, floor, above-grade wall, and wall/window interface were investigated. Several types of air-sealing methods were analyzed during these experiments. Interactive Envelope Materials Database for Whole-Building Energy Simulation Programs utilizes all the theoretical concepts and experimental procedures described above. It consists of four computational modules: 1. Building geometry calculator 2. Whole-wall, roof, ceiling thermal calculator 3. Air leakage calculator 4. Input file generator The building geometry calculator remembers all geometry data for the building (e.g., building dimensions, number of corners, windows and doors, shape of roof, size and distribution of structural members, etc.…). It enables calculations of elevation area distribution for major building envelope components. The whole-wall, roof, ceiling, thermal calculator consists of five independent sections: 1. Hot-box test results database 2. Clear-wall, roof, and floor R-value database and detail R-value database 3. Whole-wall, roof, floor R-value calculator 4. Experimental dynamic thermal characteristics database 5. Dynamic characteristics calculator All historic hot box results for hundreds of wall, roof, and floor material configurations will be available in the ORNL hot box test result database. The R-value calculator will be based on the Whole Wall Thermal Performance Calculator (Christian and Kosny 1996). Its calculation capability will extend to roof and floor structures. Using already existing thousands of results of detailed 3-D heat transfer simulations for clear walls, wall details, and roof and floor details, it will process them into whole-wall, whole-roof, or whole-floor R-values. Dynamic hot box results and dynamic thermal characteristics for complex building envelope systems will be accessible as well. The dynamic thermal characteristic calculator will be based on the Equivalent Wall Program (Kossecka and Kosny 1997). It will generate a series of response factors, structure factors, and equivalent wall for a given materials configuration. It will also reconfigure dynamic thermal characteristics to incorporate the effects of building envelope details using computational procedures developed by ASHRAE research project TRP 1145 (ASHRAE 2001b). The air leakage calculator will utilize experimental results on component air leakage. It will process detailed information linking available component air leakage experimental data with the type of building envelope, complexity of architectural components, type and number of windows and doors, etc. This calculator will simplify the process of the whole-building air leakage analysis and minimize the possibility of errors and miscalculations. At the end of the process, the input file generator will combine all data developed by all calculation modules and develop an envelope-related part of the input file for whole-building energy simulation programs.

8

8

CONCLUSION Parametric analysis will be a primary method of optimizing buildings’ energy performance in the future. Eventual implementation of the full range of physical characteristics of building envelope components into whole-building energy simulation programs requires development of an advanced interactive materials database. Such a database would enable a modeler unfamiliar with advanced heat transfer analysis to develop simple and accurate descriptions of envelope systems in a form readable by most simulation programs. To reduce the cost of the designing process and minimize the potential for inaccuracies, developing architectural component descriptions in whole-building energy simulation programs has to be as simple as selecting the specific material configuration, setting dimensions, and orientation. Computational examples presented in this paper demonstrated that by simply adding insulation to the building does not necessary mean huge savings in energy consumption. Very often, proper analysis of configuration of building envelope can be more efficient and less expensive. Analysis of building envelope assemblies containing thermal bridges requires very often application 3-D simulation tools. It is very common that application of simplified (not accurate) 1-D description created for a single building envelope detail may generate relatively insignificant errors in energy consumption predictions. For a single building simulation, this kind of “small errors” can be found in many point of the input file (thermal bridges in walls, attic, or floor, improper dynamic models of complex assemblies, etc..). In some cases these inaccuracies (combined together) may generate error of significantly larger magnitude than 5%. Complex massive structures and steel-framed assemblies are most sensitive for these inaccuracies, where errors in predicted energy consumption may reach 20%. Some building components may notably influence whole building air leakage characteristic. That is why, it is crucial to use experimentally validated air leakage values for components used in whole building air leakage calculations necessary during energy simulations. Several sophisticated whole-building energy simulation programs have been developed, but they cannot be fully utilized because of the lack of a proper materials database (especially for new non-wood technologies). This hinders the use of novel energy-efficient technologies as well. That is why, a development of an interactive materials database is a critical step in introducing a new generation of whole-building energy simulation tools. To serve this need, ORNL has introduced the Interactive Internet-Based Envelope Material Database for Whole-Building Energy Simulation Programs, which links experimental data on thermal and air-tightness characteristics of building envelopes with advanced analytical methods available for thermal and energy analysis.

REFERENCES ASHRAE. 2001a. 1989 ASHRAE Handbook: Fundamentals. Atlanta: American Society of Heating, Refrigerating and Air-Conditioning Engineers, Inc. ASHRAE. 2001b. “Modeling Two-and-Three Dimensional Heat Transfer Through Composite Wall and Roof Assemblies in Transient Energy Simulation Programs.” ASHRAE Project 1145-TRP. March 2001. Christian, J. E., and J. KoÑny. 1996. “Thermal Performance and Wall Ratings.” ASHRAE Journal, March 1996,

9

9

Colliver, D.G., W. E. Murphy, and W. Sun. 1992. “Evaluation of the Techniques for the Measurement of Air Leakage of Building Components,” ASHRAE Project RP-438, University of Kentucky, Lexington. Dickerhoff, D. J., D. T. Grimsrud, and R. D. Lipschutz. 1982. “Component Leakage Testing in Residential Buildings.” Proceedings of the ACEEE 1982 Summer Study, Santa Cruz, Calif. Washington, D.C.: American Council for an Energy-Efficient Economy. Harrje, D.T. and G. J. Born. 1982 “Cataloguing Air Leakage Components in House.” Proceedings of the ACEEE 1982 Summer Study, Santa Cruz, Calif. Washington, D.C.: American Council for an Energy-Efficient Economy. KoÑny, J. and A. O. Desjarlais. 1994. “Influence of Architectural Details on the Overall Thermal Performance of Residential Wall Systems.” Journal of Thermal Insulation and Building Envelopes, Vol. 18, July 1994. KoÑny, J., and E. Kossecka. 2000 “Computer Modeling of Complex Wall Assemblies—Some Accuracy Problems.” Presented at International Building Physics Conference, Eindhoven, The Netherlands. KoÑny, J., 2003 “Testing Air Sealing Techniques.” Home Energy, January/February 2003. Kossecka E., and J. KoÑny. 1996. “Relations Between Structural and Dynamic Thermal Characteristics of Building Walls. Presented at the Conseil International du Batiment Symposium, Vienna, Austria, August 1996. Kossecka E., and J. KoÑny. 1997. “Equivalent Wall as a Dynamic Model of Complex Thermal Structure.” Journal of Thermal Insulation and Building Envelopes, Vol. 20, January 1997.

10

10