IMPACTS OF ARCHITECTURAL DETAILS ON THE WHOLE WALL THERMAL PERFORMANCE IN RESIDENTIAL BUILDINGS

 

 

J. Kosny and A.O. Desjarlais

Oak Ridge National Laboratory

Building Thermal Envelope Systems and Materials Group

 February, 1994

 

 

ABSTRACT

 

            Predicted heat losses through building walls are typically based on measurements of the wall system clear wall area using test methods such as ASTM C 236 or are calculated by one of the procedures recommended in the ASHRAE Handbook of Fundamentals that often is carried out for the clear wall area exclusively.  In this paper, the phrase "clear wall area" is defined as the part of the wall system that is free of thermal anomalies due to building envelope subsystems or thermally unaffected by intersections with other surfaces of the building envelope.  These experiments or calculations normally do not include the effects of building envelope subsystems such as corners, window and door openings, and structural joints with roofs, floors, ceilings, and other walls.  These details represent completely different constructions; it is apparent that the thermal properties measured or calculated for the clear wall area may not adequately represent the total wall system thermal performance.  Factors that would impact the ability of today's standard practice to accurately predict the total wall system thermal performance are the accuracy of the calculation methods, the amount of the total wall area that is clear wall, and the quantity and thermal performance of the various wall system details.

 

            Based on 3-D finite difference computer modeling, the thermal performance of several typical wall systems including various system details have been analyzed, and the overall wall system thermal performance for a typical single-story ranch house has been determined.  These data are compared to typical experimental and analytical techniques to ascertain their precision in predicting the overall wall system performance.

INTRODUCTION

 

            Most thermal calculations for building wall systems are based on the measured or calculated thermal performance of the clear wall area.  In this paper, the phrase "clear wall area" is defined as the part of the wall system that is free of thermal anomalies due to building envelope subsystems or thermally unaffected by intersections with other surfaces of the building envelope.

 

            Measurements of wall systems are typically carried out by apparatus such as the one described in ASTM C 236, Standard Test Method for "Steady-State Thermal Transmission Properties of Building Assemblies by Means of a Guarded Hot Box" [1].  A relatively large (approximately 8 x 8 ft or larger) cross-section of the clear wall area of the wall system is used to determine its thermal performance.  Thermal anomalies such as wood studs are typically included in the tested configuration.  The precision of this test method is reported to be approximately 8 percent [1].  Even if the test method was perfect, it is apparent that what is being measured only constitutes a portion of the wall system; details are rarely included as part of a series of measurements to ascertain the overall wall system thermal performance.

 

            The most widely used analytical techniques for estimating wall system thermal performance are described in Chapter 22 of the ASHRAE Handbook of Fundamentals [2].  The isothermal planes method allows the user to calculate the thermal performance of wall systems assuming that an isothermal surface exists whenever there is a change in the wall system geometry.  The error associated with this simplifying assumption is dependent on the wall system being analyzed.  For walls with metal framing, a modified parallel path calculation method is recommended [2].  This method involves two separate calculations.  The first calculation is used over a limited area containing the highly conductive element, while the second calculation models the remaining portion of the wall system. These two computations are then combined using the ASHRAE parallel path method [2].  As with the test methods, only the clear wall area is typically evaluated.

 

            These techniques for quantifying the thermal performance of wall systems appear to have obvious shortcomings.  Building envelope subsystems such as window and door frames, along with the additional structural support that these subsystems require, are ignored.  The impact of construction details such as wall corners, and floor and ceiling interfaces with the wall system are overlooked.  These simplifications can lead to errors in determining the energy efficiency of the building envelope. In addition, these techniques de-emphasize the importance of energy-efficient design of the wall details.  Since envelope system designers cannot claim performance benefits due to innovative detailing, the building community is less likely to concern itself with novel detailing concepts.

 

            To address these uncertainties associated with the practice of evaluating only the clear wall area and the simplifications in the calculation methods, analytical experiments using a 3-D finite difference model have been performed on a variety of wall systems, their subsystems, and details.  Using a standard building elevation, these results have been combined to compute the amount of clear wall area and to determine the overall system thermal performance.  The overall system performance is compared to today's standard practices to determine the difference in performance due to their simplifying assumptions.

ANALYSIS METHODOLOGY

 

            A heat conduction computer code Heating 7.2, developed by Oak Ridge National Laboratory[1], was used for thermally analyzing clear wall areas, wall subsystems, and exterior wall intersections with other building elements.  Heating 7.2 can solve steady-state and/or transient heat conduction problems in one-, two-, or three-dimensional Cartesian, cylindrical, or spherical coordinates [3].  Multiple materials and time and temperature dependent thermal conductivity, density, and specific heat can be considered.  The boundary conditions, which may be surface-to-environment or surface-to-surface, can be specified temperatures or any combination of prescribed heat flux, forced convection, natural convection and radiation.  The boundary condition parameters can be time and/or temperature dependent.  The mesh spacing can be variable along each axis.  Heating 7.2 can solve transient problems by using of any one of several finite-difference schemes: Crank-Nicolson Implicit, Classical Implicit Procedure, Classical Explicit Procedure, or the Levy Explicit Method.  Two-dimensional modeling was used for most of the clear wall areas.  For wall details and for areas where the exterior wall intersects with other building elements, 3-D modeling was necessary.  The resultant temperature maps were used to calculate average heat fluxes, and then wall system R-values.

 

            The authors verified the accuracy of Heating 7.2's ability to predict wall system R-values by comparing Heating 7.2 simulation results with published test results for twenty-three masonry, wood frame, and metal stud walls.  The summary of the comparison between experimental and simulated R-values is presented in Table 1.  In Table 1, the data presented in the column entitled "Difference Between Results" were computed based on the following formula:

 

 

            Ten empty 2-core 12-in. concrete masonry units (CMUs), reported by Valore [4], Van Geem [5], and James [6] were simulated using Heating 7.2.  These data were selected for modeling because complete geometric descriptions and thermal properties of the components used to fabricate the wall systems were available.  The average difference between the simulated and tested R-Values for these ten wall systems was better than 4 percent.  This exercise was repeated for filled 2-core 12-in. CMUs; in this case, the average difference was less than 6 percent.

 

            A wood frame wall system reported by James [6] and four metal stud wall systems reported by Brown [7] and Strzepek [8] were similarly treated.  The average differences for the wood frame and metal stud wall systems were 2 and 5 percent, respectively.  Considering that the precision of the guarded hot box method is reported to be approximately 8 percent, the ability of Heating 7.2 to reproduce the experimental data is within the accuracy of the test method.

Table 1.

 

Accuracy of Heating 7.2 R-Value Calculations

 

 

The following 6 basic wall technologies were selected for detailed analysis:

            For each wall system, models of the clear wall area, corner, roof/wall intersection, floor/wall intersection, window header, windowsill, window edge, door header, and door edge were analyzed.  Geometries of these details were obtained from standard architectural drawings or system manufacturer's design guides [9, 10, 11, and 12].  A significant amount (up to twice the detail area) of clear wall area was included when modeling the subsystem or detail so that the interaction between the detail and the clear wall area was included in the computations and the area thermally affected by the subsystem or detail could be derived.  The temperatures and wind speeds used in all of the modeling runs were 70^oF and 0 MPH for the interior space and 20^oF and 15 MPH for the exterior environment.  Film coefficients were held constant for all configurations.

 

            The thermal resistance of each wall detail was calculated by dividing the average surface-to-surface temperature difference by the average heat flux for the area of the detail.  The arithmetic average temperature of each node on each surface of the wall system was used to determine the average surface temperature.  The method to determine this area is discussed in the following section.

 

            The influence of subsystems on the overall wall thermal performance is different for every house because of the variety of architectural designs.  To normalize the calculations, a standard building elevation was used to combine the thermal resistances of the various details and to compute the overall wall system thermal resistance.  The standard elevation selected for this purpose is a single-story ranch style house that has been the subject of previous energy efficiency modeling studies [13].  A schematic of the house is shown in Figure 1.  The house has approximately 1500 ft² of living area, 1328 ft² of exterior (or elevation) wall area, 8 windows, and 2 doors (one door is a glass slider; its impact is included with the windows).  The elevation wall area includes 1146 ft² of opaque (or overall) wall area, 154 ft² of window area and 28 ft² of door area.

 

            The overall thermal resistance of the wall system, R_ws, was computed by combining in an area weighted method the thermal resistances of the subsystems, wall intersections, and clear wall area.

 

where A_cw is the clear wall area expressed as a percentage of the overall wall area, R_cw is the clear wall thermal resistance, A_i is the area of the ith detail expressed as a percentage of the overall wall area, R_i is the thermal resistance of the ith detail, and n is the number of details.  This technique is similar to the ASHRAE zone method [2] except that the zones of influence are computed instead of being prescribed.

 

            The amount of clear wall area was calculated by determining the zone of influence for each subsystem and subtracting that area from the total exterior wall area.  The zone of influence was determined by examining the isotherms produced by the modeling runs.  We defined the zone of influence as that area where the existence of the detail changed the slope of the isotherm by more than 5E.  This slope represents approximately a 1EF change in temperature per inch of length along the wall surface.  The area which depicted isotherms that were impacted by the presence of the subsystem, was defined as the zone of influence for that subsystem.

 

     The isothermal planes and parallel path methods [2] were used to calculate the clear wall area for the six wall systems that were simulated.  The purpose for these calculations was to determine the accuracy of the assumptions made when using the isothermal planes or parallel path methods and to compare the results to the overall wall system thermal resistance data.

 

 

WALL AREA DISTRIBUTION

 

            For each selected wall system, models of the clear wall area, corner, roof/wall intersection, floor/wall intersection, window header, window sill, window edge, door header, and door edge were created and analyzed.  As stated earlier, a significant amount of clear wall area was included in each model.  The purpose for including this clear wall area was to measure the influence of the detail on the clear wall area by determining the extent of the detail's effect on the isotherms.  As a first iteration, we allowed the geometry of the wall system to define how much wall area was modeled.  For instance, when we modeled a wall system containing studs 16 inches on center, we modeled from the middle of the cavity on one side of the detail to the middle of the cavity on the opposite side of the detail.

 

            We defined the detail's zone of influence by examining the isotherm map produced by the modeling runs.  As stated earlier, we defined the zone of influence as that area where the existence of the detail caused a change in the slope of the isotherm greater than 5E.  This slope represents approximately a 1EF change in temperature per inch of length along the wall surface.  An example of this procedure is depicted in Figure 2 (the isotherm map for the metal stud wall corner detail).  The effect of the thermal bridging from the corner detail impacts the isotherms for almost half of the modeled area.  When initially defining the modeling areas for the detail runs, several iterations in defining the modeling area were sometimes required until the entire zone of influence was included in the modeled area.

 

            This procedure was repeated for each detail and subsystem of the wall systems.  A summary of these results is presented in Figure 3.  Since the zones of influence for the individual details making up the door and window perimeter were relatively small for the standard elevation that was selected, they are combined to simplify the data presentation.  Figure 3 depicts the percentage of the total overall wall area that is thermally influenced by each detail.  The remainder of the wall area is defined as the clear wall area.

 

            Note that the data depicted in Figure 3 are percentages of the overall wall area and do not include the areas of openings such as windows and doors.  For the standard building elevation that is being simulated, the door and window areas represent 2 and 12 percent of the elevation area, respectively.  The terms "overall wall area" and "elevation area" are used in this document to differentiate between these two surface areas.  Our example house has an elevation area (perimeter times wall height) of 1328 ft² and an overall wall area (elevation area minus apertures) of 1146 ft².  Figure 4 portrays the clear wall area as a function of both the overall wall area and elevation area.

 

            For the wall systems analyzed, the clear wall area ranges from 67 to 87 percent of the overall wall area.  Four of the wall systems, namely the wood stud, metal stud, 2-core uninsulated CMU, and the EPS wall forms have clear wall areas of approximately 70 percent.  The stress-skin panel system wall area is the least impacted by wall system details; 87 percent of the overall wall area is thermally unaffected by the details.

 

            Based on the modeling runs, the roof/wall and floor/wall intersections have the largest impact on the overall wall area.  For the six wall systems analyzed, the roof/wall and floor/wall subsystems averaged 9 and 7 percent of the overall wall area, respectively.  The window perimeter also had an appreciable effect, averaging 5 percent of the overall wall area.  The effect of the window perimeter is probably underestimated for today's residential construction practice since the use of fenestration products has increased dramatically and the standard elevation's 12 percent of glazing area is probably too small.

 

            The wall system details that had the smallest impact on the overall wall area were the corners and door perimeter.  The average wall area attributed to these details was 3 and 1 percent, respectively.  Again, these results must be seen in the context of the selected floor plan.  A single-story residence with a rectangular configuration will diminish the effects of the corners while the typical residence certainly has more than two doors.

 

 

WALL SYSTEM THERMAL PERFORMANCE

 

            The previous section shows that a substantial portion of the overall wall area is ignored if procedures that are typically employed today are used to determine the wall system thermal performance.  In this section, we will quantify the differences in thermal performance that are likely to be obtained by ignoring the wall system details and subsystems. We will also estimate the impact of the wall details and subsystems thermal resistances on overall wall system thermal resistance.

 

            Based on the modeling, the thermal resistance of the clear wall area and each wall system detail was computed by dividing the average temperature difference by the average heat flux.  The overall wall system R-value was computed by combining the thermal resistances of the wall details, subsystems, wall intersections, and clear wall area in a parallel, area-weighted method as described in eq. (2).

 

            The simulation results for the clear wall and overall wall areas are summarized in Figure 5, along with the differences in these simulations.  With the exception of the uninsulated 2-core CMU and EPS wall form systems, the clear wall area thermal resistance is larger than the overall wall area.  In the cases of the metal stud wall system and the stress skin panel system, the clear wall areas are 24 and 12 percent higher (respectively) than the overall wall system R-Value.  These results suggest that significant improvements of the details in these wall systems are required.  For the uninsulated CMU system, the R-Value of the clear wall area is so low (1.56 hr ft² F/Btu) that poor detailing may actually increase the R-Value of the overall wall area.  For the EPS wall forms, thermally well-designed details may also improve the R-Value of the overall wall area.

 

              Our validation of the computer code suggests that competent experimentation yields results that are comparable to our simulations.  If experiments were performed on the clear wall area of these systems and the results were similar to our modeling predictions, then the discussion in the previous paragraph would apply to the experimental results.  The differences noted in our clear wall and overall wall simulations would be applicable to the differences in the experimental results and the thermal performance of the overall wall area.

 

            To observe the effects of the wall detail thermal resistances on overall wall thermal resistance, we introduced a new parameter, the detail influence factor (DIF).  DIF combines the effects of the detail's zone of influence and thermal resistance so that the full impact of the individual details can be quantified and compared.  DIF is a measure of the amount of heat loss that can be attributed to each component of the overall wall system.  DIF is defined by the following formula:

 

 

 

 

            Figure 6 shows the distribution of heat losses from the overall wall area.  It is apparent that the amount of energy lost through the clear wall area can be appreciably different than the overall wall area for the wall system technologies that were simulated. The heat losses from the clear wall area range from 29 percent for the 2-core CMU wall system to 63 percent for the stress-skin panel wall system.  Wall/roof and wall/floor intersections account for 6 to 25 percent of overall wall heat losses.  Opening perimeters (the combination of door and window perimeters) are responsible for 9 to 19 percent of overall wall heat losses.  Results from these calculations demonstrate how important the thermal performance of wall details and subsystems are.  When designing wall systems and determining their thermal performance, appreciable errors can be introduced when the details and subsystems are not included in the analysis.

 

            For an "ideal" wall system, the local thermal resistances created by the wall system details should be at least as good as the clear wall area.  Heat losses through details and subsystems should be proportional only to the wall area distribution.  In the case of most existing wall technologies, only the clear wall area has been developed to optimize its thermal resistance.  The development of wall details is important to control wall system thermal performance.

 

 

COMPARISON TO TODAY'S METHODS OF COMPUTING AND MEASURING WALL SYSTEM THERMAL PERFORMANCE

 

            The ASHRAE Handbook of Fundamentals lists several procedures for computing the thermal performance of wall systems.  With the exception of the metal stud wall system, the isothermal planes method is recommended for all the wall systems analyzed in this study.  For the metal stud wall system, the handbook recommends the use of the parallel path method.  These calculation techniques were applied to the clear wall areas for each of the wall systems that we modeled.  A comparison of the calculated and simulated clear wall R-Values is shown in Figure 7.

 

            The difference between simulated and calculated (using ASHRAE methods) thermal resistances for clear wall area is a measure of the effects of the simplifying assumptions that must be made when using the ASHRAE methods.  Very good agreement was obtained for the wood frame, Larsen truss, and stress-skin panel wall systems; a difference of less than 3 percent was noted between the simulated and calculated thermal resistances for each of these systems.  In these three cases, the simulations predicted slightly higher thermal resistances.  For the remaining three wall systems, the metal stud, 2-core uninsulated CMU, and EPS-form wall systems, the agreement between the simulated and calculated R-Values was -8, -15, and -23 percent, respectively.

 

            Since wall details are typically not considered, the data produced from the calculations of clear wall area performed in accordance with the ASHRAE isothermal planes and parallel path methods must be applied to the overall wall system.  This is also true when describing wall systems with experimental results.  Therefore, a comparison of the calculated or measured clear wall area thermal resistance to the simulated overall wall area thermal resistance will yield an estimate of the error associated with using the ASHRAE calculations or laboratory experiments for computing overall wall system heat losses.  This comparison is shown in Figure 8.

 

            In general, the simulated overall wall and the calculated clear wall thermal resistances are in relatively good agreement; all of the wall systems except metal stud and the EPS forms have less than a 7 percent difference.  In the case of the metal stud and EPS forms, the calculation method overstates the wall system's thermal performance by 33 and 20 percent, respectively.  It is interesting to note that all of the calculated results predict an equal or higher thermal resistance, and on average, that overstatement is approximately 7 percent.

 

            Although the data presented in Figure 8 suggests that the calculated clear wall R-values and simulated overall wall R-values are in relatively good agreement, care must be taken in extending these findings to other systems and other details.  The references used in modeling the details depicted recommended constructions; actual construction practices may vary significantly from the cases we have analyzed.

 

 

CONCLUSIONS

 

            Using a finite difference computer code, six wall systems with their typical details have been modeled. The modeling has been used to compute the amount of area that is thermally impacted by the details, and the effect of the details on the overall thermal performance of the wall system.  Calculated values of thermal performance have also been produced, representing the procedures that are typically employed for assessing wall system thermal performance.  These results have been examined and compared.  The following series of conclusions have been developed; these conclusions may be useful in the design and performance characterization of wall systems.

 

 1.        The finite difference code that we used for the simulations was compared to a series of critically reviewed experimental results.  The average difference between the simulations and experiments was less than 4%.  This difference is less than the stated uncertainty of the test method.

 

 2.        Building wall systems are a combination of clear wall area, wall details, and subsystems, and cannot be accurately modeled simply by studying the clear wall area.  For the wall systems reported in this study, as much as 25% of the overall wall area is different in construction and thermal performance than the clear wall area.  We have used a fairly straightforward building elevation for this modeling; the wall area distribution of most other elevations will probably have a smaller percentage of clear wall area.

 

 3.        Differences between simulated R-values and ASHRAE isothermal planes and modified parallel path method R-value calculations of the clear wall area are a measure of the effect of the simplifications required to perform the calculations.  For the six wall systems that we studied, the calculations overstated the thermal resistance of three wall systems, the metal studs, the 2-core CMU and the EPS forms, by as much as 23 percent.  For the remaining three wall systems, the calculations conservatively estimated the clear wall thermal resistances.

 

 4.        Calculated clear wall R-Values should clearly not be used to characterize the overall thermal resistance of metal stud and EPS form wall systems.  Although the calculated clear wall R-values and simulated overall wall R-values were in fairly good agreement for the remaining four wall systems that we modeled (differences of less 7 percent), it is clear that there is potential for significant differences.  A more elaborate elevation with energy-inefficient detailing could have an appreciable effect on the overall wall system thermal performance.  A more extensive review of wall details and elevations is required.

 

 5.        Experimental procedures that are employed to characterize wall systems typically only evaluate the performance of the clear wall area.  Assuming the simulated clear wall area thermal resistance results are representative of the quality of experimental data, the previous conclusion is also applicable to the experimental results.

 

 6.        We have determined that only 29 - 63 percent of overall wall heat losses are due to the clear wall area.  The remaining heat losses are attributed to the portions of the wall systems that are typically not handled correctly when characterizing the thermal performance of wall systems.  Additional emphasis on developing energy-efficient details could lead to significant heat loss reductions for building wall systems of the future.

 

 

REFERENCES

1.	ASTM C 236-89, Standard Test Method for "Steady-State Thermal Performance of Building Assemblies by Means of Guarded Hot Box," Vol 04.06, pp. 53-63.
2.	ASHRAE, ASHRAE Handbook of Fundamentals, ASHRAE, 1993.
3.	K.W. Childs, "HEATING 7.2 User's Manual," Oak Ridge National Laboratory Report ORNL/TM-12262, February 1993.
4.	R.C. Valore, "Thermophysical Properties of Masonry and it's Constituents - Part II: Thermal Transmittance of Masonry," International Masonry Institute, Washington, 1988.
5.	M.G. Van Geem, "Thermal Transmittance of Concrete Block Walls with Core Insulation", Journal of Thermal Insulation, Vol. 9, January 1986.
6.	T.B. James, Final Draft of "Manual of Heat Transmission Coefficients for Building Components," Dept. of Mechanical Engineering, University of Massachusetts, Amherst, MA, Nov. 1990.
7.	W.C. Brown and D.G. Stephenson, "Guarded Hot Box Measurements of the Dynamic Heat Transmission Characteristics of Seven Wall Specimens: Part II," ASHRAE Transactions, vol. 99, Pt.1, 1993.
8.	W.R. Strzepek, "Thermal Resistances of Metal Frame Wall Constructions Incorporating Various Combinations of Insulating Materials" Insulation Materials, Testing and Applications, ASTM/STP 1030, 1990.
9.	J.R. Hoke Jr., "Architectural Graphic Standards," The American Institute of Architects, John Wiley & Sons, ISBN 0-471-81148-3.
10.	AISI, "Low-Rise Residential Construction Details - TECHNICAL DATA," Publication RG-934, American Iron and Steel Institute, June 1993.
11.	Marino Industries Corporation, "NAHB Resource Conservation House - Construction Drawings," January, 1992.
12.	National Concrete Masonry Association, "Design and Construction of Plain and Reinforced Concrete Masonry Basement and Foundation Walls. NCMA - TR68-A, 1975.
13.	Huang, Y.J., Ritschard, J., Bull, S., Byrne, I., Turiel, D., Wilson, C., Sui, H., and Foley, D., "Methodology and Assumptions for Evaluating Heating and Cooling Energy Requirements in New Single-Family Residential Buildings, Technical Support Document for the PEAR Microcomputer Program," Lawrence Berkeley Laboratory Report No. LBL-19128. Berkeley, CA, 1987