An integrated energy–emergy approach to building form optimization: Use of EnergyPlus, emergy analysis and Taguchi-regression method

Accepted Manuscript An Integrated Energy-Emergy Approach to Building Form Optimization: Use of EnergyPlus, Emergy Analysis and Taguchi-Regression Method Hwang Yi, Ravi S. Srinivasan, William W. Braham PII:

S0360-1323(14)00334-5

DOI:

10.1016/j.buildenv.2014.10.013

Reference:

BAE 3858

To appear in:

Building and Environment

Received Date: 23 July 2014 Revised Date:

17 September 2014

Accepted Date: 13 October 2014

Please cite this article as: Yi H, Srinivasan RS, Braham WW, An Integrated Energy-Emergy Approach to Building Form Optimization: Use of EnergyPlus, Emergy Analysis and Taguchi-Regression Method, Building and Environment (2014), doi: 10.1016/j.buildenv.2014.10.013. This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

ACCEPTED MANUSCRIPT Building and Environment 2014

An Integrated Energy-Emergy Approach to Building Form Optimization: Use of EnergyPlus, Emergy Analysis and Taguchi-Regression Method

Abstract

RI PT

[email protected]/ +1 267 304 8323

EP

TE D

M AN U

SC

This research presents a new methodology of environmental building design with an integrated energy-emergy (spelled with an “m”) approach to study building form optimization in the schematic phases. In architecture, selection of sustainability assessment methods critically affects design goals, favoring or restricting choices designers can make. A building subsumes matter and energy to support human lives, but current building performance indicators are still hard to equate technical sides and human dominant sides of various scales in a synthetic metric. Moreover, in order to achieve global sustainability in a building, as a part of the whole built environment, it is necessary to integrate energy and environmental impacts at the highest scope of analysis. Emergy analysis coupled with building energy simulation can be suggested as a holistic indicator for architectural design process. To test the proposed method, a pilot study with a mid-size office building evaluates the consequences of early design decisions such as basic geometry, aspect ratio, window-wall ratio, construction types, etc. The integrated energy‒emergy approach to building form optimization consists of three modules namely, Building Energy Simulation (BES) module, Building EMergy Analysis (BEMA) module, and (iii) MetaModel Development (MMD) module. The BES module uses the EnergyPlus tool for whole building energy analysis, while the BEMA module employs analytical methods to estimate emergy quantities, and the MMD module employs the Taguchi method to develop a metamodel for faster and easier whole building emergy simulation. The metamodel developed using Taguchi-ANOVA method for building form optimization was validated with analytical test results to accelerate environmental design decision-making. This study demonstrates possibility of wider applications of emergy synthesis to building energy research and facilitates practical use of emergy simulation in the environmental design process.

AC C

Keywords: Energy simulation, emergy evaluation, building form optimization, Taguchi-ANOVA method

Nomenclature SST SSA

SS' SST MSS Ve θ AR WS1

Sum of squares Sum of the squares of the S/N variation induced by parameter A around overall mean Pure sum of squares Total sum of squares Mean sum of squares or variance Variance of error Building orientation Aspect ratio South-facing window-to-wall area ratio 1

ACCEPTED MANUSCRIPT East-facing window-to-wall area ratio North-facing window-to-wall area ratio West-facing window-to-wall area ratio Insulation level measured in U-value for South and North walls Insulation level measured in U-value for East and West walls Emergy change amount Change amount of energy use intensity Correction factor Degree of freedom

RI PT

WS2 WS3 WS4 M1 M2 ∆Em ∆EUI C.F. df

1. Background

AC C

EP

TE D

M AN U

SC

Nowadays the building production process is subjected to careful scrutiny based on growing environmental concerns. Architectural design processes towards “green” building, in which the domain of technology and aesthetics must be integrated, have also been advised by assessment methods for environmental performance and potential impacts. From the point of view of architects whose desire is to achieve optimal use of resources through a built form, definition and scope of environmental analysis strongly influence design results and transform architectural style at the end (e.g. large solar panels on a rooftop). So it is required to ensure that system boundaries and metrics of indicators are congruent to the full definition of building. A building is a thermodynamic engine maintaining thermal balance constant for human dwelling, and it consumes global resources in the form of matter [1]. Most of inputs are sourced beyond site and local contexts in the way of production or delivery. For example, primary sources of operational energy are imported in whole or in part, and building materials are manufactured with elements extracted off-site. Besides, quality of human labors and states of social/cultural systems outside buildings are another concerns in energetic exchanges. Those characteristics of building lead to identifying sustainable buildings in terms of ecology, which says, a sustainable built form has the least impact on natural cycle and it belongs to larger environmental contexts including history and society [2,3]. Therefore, in an ecological sense, we must say that heat flows in a building ultimately relate to thermal/resource balance of the largest environment system, namely, the earth. Monitoring the scope of mass/energy inflows to buildings is accordingly expected to be enlarged to the “maximum” extent to attain global sustainability [4,5]. Currently, there exist a number of elaborative techniques such as embodied energy analysis, Life Cycle Assessment (LCA), or exergy analysis as well as promoting standards (e.g. Building Research Environmental Assessment Method (BREEAM), Leadership in Energy and Environmental Design (LEED), Zero Energy Building (ZEB)) developed and applied in order to constrain resource consumption in increased energy efficiency. The green building standards (BREEAM, LEED) are comprehensive evaluation methods employing lists of indictors. Advanced efforts such as extended exergy analysis [6], LCA-ZEBs [7], renewable potential LCA [8], have attempted to acknowledge better different building typologies, materials, heat sources, and project boundaries, thereby encouraging building practitioners to adopt time and cost effective strategies. Despite significance of the existing indicators at multiple scales ‒ from their physical aspects (geometry, construction types, interior conditioning, lighting, etc.) to social mechanisms (user behavior, etc.) and at various perspectives (cost, source, emission, etc.) [9], they generally overlook to link the built environment effectively with loads of the larger geobiological system, including solar, wind, other renewable and non-renewable resources, and ecosystem services [10,11]. Embodied energy calculates energy demands from extraction of material and discounts environmental loads in 2

ACCEPTED MANUSCRIPT

RI PT

material formation process in nature. Although some current LCA-based tools, both sector and process-based, have attempted to describe a building as a conscious thermodynamic engine over its useful life [12,13], they still brush aside the thermo-physical makeup of a building that provides for human occupancy, and energy accounting of nature for raw material formation and human/social services. This thermo-physical composition informs the ways in which people choose to utilize building systems, often with direct implications for the amount of operational energy that building occupants utilize. Therefore, a holistic approach such as emergy that can analyze systems of different types and scales, and human communities, with a common indicator, is thus necessary to address the utmost sense of building sustainability and to provide synthetic information for facilitating energy benchmarking in a global context.

SC

1.1 Introduction to the emergy analysis

AC C

EP

TE D

M AN U

Emergy, pioneered by H.T. Odum, refers to past available energy use to measure mass/energy flows in and out of an open ecosystem in thermodynamic accounts. By definition, it is the available solar energy or solar embodied energy used up, directly and indirectly, to make a service or product [14]. Emergy study set off with a question of how to measure a dynamic living system in the field of ecology. In the early 1900s, a biophysicist Alfred Lotka found that an ecosystem tended to increase total energetic power, rather than efficiency, to achieve population growth, biomass production, and self-organization. This process of obtaining maximum power could be generalized to describe the development of ecosystems. Working with that principle, H.T. Odum proposed use of emergy to quantify this process in which an energetic flow i is simply computed by, (1)  =  •  where Em denotes solar emergy of energetic flow, Ei is energy or mass input, and Tri is transformity that refers to specific emergy value of the input. Transformity means “emergy input per unit of available energy output” [14]. Transformity is quality or value of energy, and it is usually obtained from previous studies, emergy database, or rarely, derived from the global baseline so that it becomes a sort of property of energy and matter of a system component simultaneously. Even though the maximum power conception is not yet fully demonstrated in characterization of building systems, it has demonstrated causality for energetic shifts discovered in ecology and terms of emergy can be directly applied to analysis of building energy dynamics [4,10]. The reason for this is because a building, much like any creature, thrives through energy exchanges. Buildings behave theoretically like a thermodynamic engine that requires a great deal of energy concentration and transformation for its birth (construction), life (operation and maintenance), and in some cases, death (deconstruction). Consequently, emergy analysis provides building environment research with the following three major strengths based on the emergy theory’s fundamental hypothesis that all energetic forms on the earth find their sources from solar, tidal, and deep heat.
Emergy is the most effective synthetic tool for linking environmental science to the goal of building for sustainability. Emergy identifies the earth with a gigantic open ecosystem and traces the origin of all energy generation to solar energy. So emergy evaluation integrates different metrics of resource use in a single measure of solar equivalent; solar emergy joule or sej. Among energy accounting related indices, few provides synthetic information in a single unit. Emergy can be an alternative to existing indicators used for switching and converting different metrics such as primary energy use or CO2 3

ACCEPTED MANUSCRIPT equivalent while allowing for the comparison of much broader environmental effects.

SC

RI PT


Emergy combines natural process and human-dominated system. A building is always situated within specific living contexts. Local climate, state of ambient settings, and social/cultural infrastructures are closely associated one another with building energy use. Thus, environmental assessment of buildings should not only be concerned with depletion of nonrenewable sources, but also man-made environment as well as natural resources [10]. Emergy evaluation aggregates energetic sources of various kinds, either in energy or matter, by assigning a universal common denominator (solar energy) for natural works and human services. So, accounting for building environmental impact in this way allows for all environmental pressures from different sectors (e.g. renewable sources, economy, cultural service, etc.). This method avoids troubles in the comparison of dissimilar energy, while allowing for the identification of the different manners of accumulation and dispersal of available energy to assess energy quality and hierarchy for all services and products of a system.

TE D

M AN U


Emergy evaluation is bookkeeping of resource consumption. Emergy can also be termed “energy memory” [4,14]. Evaluation of energy consumption dates back to the formative moments of a material. Other environmental accounting methods (e.g. exergy, embodied energy or Input-Output analysis, LCA) emphasize different temporal and spatial scales which can limit the analytical window. However, emergy is the most extensive of these accounting methods because it has considerations on a cosmological time-tracking scale. From a LCA perspective, emergy counts solar energy spent in geobiophysical material formation processes, which conventional LCA methods fail to address [11]. While exergy refers to the present state of available energy for use [15], emergy accounts for past available energy throughout history of a product or service for its formation. In spatial scale, emergy is the aggregate of energy demanded for direct/indirect supports, while embodied energy analysis lacks human efforts and supportive services. Even with more advanced exergy and embodied energy analysis being developed recently [16], this fundamental difference in conception should be remembered.

AC C

EP

Summarizing, the unification of different metrics into one universal value, the ability to account for environmental formation of matter, and the integration of considerations on indirect energy domains including renewable energy use and human services make emergy a unique indicator and demonstrate its potential to be a principle metric of environmental building design. The extension of the boundaries to include all energies used during formation and assessment with use of emergy analyses can illustrate the significance of more energies than just operational energy use in building [17]. However, despite advantages, as Hau and Bakshi [16] point out, emergy analysis still has several defects, namely: awkward generalization of transformity values, uncertainty in emergy synthesis process, definition of global energy baseline, and so on. And a handful of emergy studies on the built environment indicate that emergy analysis is new to building researchers. Nevertheless, some major efforts in this direction have taken aforementioned advantages and started to challenge traditional accounting models and position this method as a new building sustainability assessment with emergy. Emergy evaluations have been applied to whole building construction [4,18], to building assemblies [19, 20], to emergy-based eco-efficiency [21], to a new emergy-based approach of maximizing the use of renewable resources [22], as well as to the social agendas that a building contains [23]. 4

ACCEPTED MANUSCRIPT

1.2

Energy, Emergy and Building Design Decision-Making

M AN U

1.2.1 Trade-offs between building energy and emergy analysis

SC

RI PT

Introducing building to stewardship on global sustainability, this study primarily tackles with direct application of emergy evaluation as a practical principle at the early design phases. It is well identified that decisions on a building form – geometry, aspect ratio, window-wall ratio, construction types, etc. made through early stages are the most influential in environmental resource use [24‒26]. In design decision-making, operational energy reduction in particular has been marked as a main drive, and extensive research has been carried out to demonstrate form-creating potential of building energy simulation tools (e.g. EnergyPlus, DOE-2, etc) though the optimization process [27, 28]. Several studies exploring early design solutions now extend measures to adopt various indicators such as embodied energy [29], exergy [25], or LCA [24, 30]. Exergy and LCA research support practice with formulaic estimation as well as commercial tools based on relatively well-developed inventories, whereas the field of building emergy study is incipient. So, optimizing a building form in conjunction with emergy faces barriers concerned with the following distinct problems:

AC C

EP

TE D

Building spaces use two types of energies namely, (i) purchased energy use for maintaining occupants’ comfort, i.e., thermal comfort (heating and cooling), visual comfort (lighting), olfactory (ventilation) and (ii) other energies related to material and human labor for construction, maintenance, and demolition and/or deconstruction. Since selection of building materials that enclose the occupiable space considerably affects the indoor comfort, the trade-offs between building energy and emergy should be taken into account for any building evaluation. Pulselli et al. [19] recognized this issue and linked operational energy use with construction emergy using computational experiments on climate-conscious emergy variation of a building envelope. Although this study was one of the first research attempts that linked building energy and emergy evaluation, there were few limitations: (i) building operational energy use was calculated using a simple estimation with local degree day data which may not adequately represent the climate conditions of the site and, therefore, prone to error; (ii) the study considered wall insulation material only and did not consider other envelope materials such as glazing, finishing materials, etc., that considerably impact energy use; and (iii) the study used a useful life of an arbitrarily assumed value of 50 years, which may not relate to actual building design under investigation. Even though 50 or 60 years were acceptable useful life to some LCA researchers [30], as Shrestha et al. [32] noted, it has little grounds in history and may rarely exist in practice. In order to aggregate building’s annual Energy Use Intensity (EUI) and construction emergy, every factor should be divided by its useful life. Building materials have their own individual useful life which needs to be extended to reflect overall building service life representing the maintenance phase in real world scenarios. Effectively connecting building energy and construction emergy is crucial for better understanding and design decision-making. In order to provide robust parametric assessment of building systems and their energy use, more sophisticated energy simulation tools with capabilities to calculate dynamic thermal systems, etc., must be seamlessly integrated with conventional emergy approaches. Additionally, the diversity of building materials’ life span has to be highlighted and adequately implemented to remove potential errors. 1.2.2 Lack of emergy-based simulation for building performance 5

ACCEPTED MANUSCRIPT

M AN U

SC

RI PT

Improved focus on emergy approaches to buildings will, without doubt, strongly benefit the building industry by placing importance on description of renewable and nonrenewable resource consumption of a building in extensive scopes of temporal and spatial frame. However, the critical problem arises from the fact that building emergy evaluation has not always been employed for prediction, but as an afterthought or an extension to traditional analyses. Building emergy synthesis is meaningful in that it studies a building-related ecosystem of goods and services by taking the biophysical modeling approach. Needless to say, an anticipatory evaluation is important for building design decision-making since measuring expected impacts during the design process critically affects the actual building performance once built and operated. Among others, the correlation between alteration of building design components such as building form, fenestration, and materials, and their final target, i.e., total emergy, needs to be elucidated with the establishment of a proven analytical model. In other words, if a predictive model or formal procedure could be initiated, designers may successfully employ it as a decision-making tool to improve the building’s overall environmental performance. When it comes to building energy use, the history of energy accounting has been relatively long and design decision-making methods for the optimal energy use have been considerably substantiated [27, 28]. Meanwhile, although general emergy approaches have been introduced by Odum [14], his proponents continue to develop and expand its application in various fields including the built environment. Only a few studies identified a reasonable possibility of emergy-driven decision-making in the planning of a building [19, 20]. 1.2.3 Need of simplified approach to building emergy evaluation

AC C

EP

TE D

As is the case with database-based estimate, facilitating emergy prediction is hindered mainly due to time-consuming work and significant expertise. For building emergy analysis, a designer draws a system boundary and arranges the systems in a hierarchical order to complete a system diagram. Then, empirical energy use data and physical properties of direct material inputs are tabulated and represented with the diagram. Every building material has a specific emergy value, known as the Unit Emergy Value (UEV) or transformity, which represents the total energy dissipated in the production, delivery, and installation of that material. In principle, UEVs should be calculated on a case-by-case basis, however, in the present day, most of them are obtained from previous studies and literature. On this account, emergy evaluation always requires complex algebraic calculation including searching appropriate inventories for UEVs as well as material properties. Much of time expended in this search and identifying the relevance of existing UEVs to fit the model threatens the applicability of emergy methods to design practice, especially during the schematic stage, because architects and engineers have rough idea of planning, and are not skillful at emergy evaluation. So the larger question, now, is how to adapt emergy evaluation for easier and faster decision-making. Moreover, a fundamental problem is that it is too hard to formulate a validated analytical emergy model due to empirical search basis. In general, the development of a system model typically starts with an experimental inspection of the phenomena of a real system so that one finds regularity in it, and then description of the system is generalized into a form of empirical model for that case. If repetitive experiments with variable change indicate existence of a common rule, an analytical formulation characterizes the system performance. This process may accompany the first variable screening that eliminates uncritical factors. Development of a building energy model, for example, is formalized through such process as 6

ACCEPTED MANUSCRIPT

RI PT

it is based on the principle of mass and heat conservation. Once an analytical model is verified, it can be integrated or compiled into a simplified procedure. Examples can be found in a suite of building performance simulation tools. However, an analytical model may not be acceptable when the scale of the system boundary gets larger or if manipulation of cases under study demands a vast amount of iterations which may require massive computing power, for example, using parallel processing of high performance computing network. It follows further model development, i.e., a metamodel to enable less computational expense and time.

SC

For these reasons, the development of an energy-emergy integrated predictive model becomes a necessity for effective design decisions, thereby blending energy and material flows for design optimization. Advantages are briefly marked by: (i) a simple procedure of emergy simulation towards fast decision-making which enables to benchmark design alternatives; (ii) clarification of influential design variables underpinning emergy-based built forms. Section 2 describes our developed methodology and specific objectives for building emergy simulation at initial design phases.

M AN U

2. Integrated Energy-Emergy Approach to Building Form Optimization

TE D

The integrated energy-emergy approach to building form optimization consists of three modules namely, (i) Building Energy Simulation (BES) module, (ii) Building EMergy Analysis (BEMA) module, and (iii) MetaModel Development (MMD) module, figure 1. The BES module uses the EnergyPlus tool for whole building energy analysis, while the BEMA module employs analytical method to estimate emergy quantities and the MMD module employs Taguchi method to develop metamodel for faster and easier approximation. The integration of energy and emergy analyses, besides being novel and versatile for broader application, serves to answer two distinct inquiries: (1) Emergy theory effectively extends building energy metrics in the largest environmental context, facilitating the inclusion of building construction and other factors. Since emergy is a measure of solar embodied energy of construction materials and energy sources, the whole building emergy prediction needs to be integrated with operational energy analysis for overall performance checkup.

AC C

EP

(2) When designing an environmental building, building form and orientation, selection of material and mechanical system always come into conflict since it is hard to expect trade-off effects of operational energy use and embodied energy consumption. Emergy evaluation offers both quantitative (emergy flow) and qualitative (transformity) estimates of energy use of building design elements, thereby avoiding any conflicting claims from the designers’ perspective. Optimization of building forms using emergy helps distinguish subtle differences in direct comparison with energydriven optimized forms.

Figure. 1 Integrated Energy-Emergy Approach.

2.1 Building Energy Simulation Module The building energy simulation module aids in the estimation of operational energy use over its useful life. This measure becomes an input to emergy calculation Emuse, see Eq. (3). EnergyPlus is 7

ACCEPTED MANUSCRIPT used in this module to perform energy estimation, evaluating the energy demands of individual components. The information that can be input in an EnergyPlus Input Data File (Table 1). Since building form optimization through integrated energy-emergy assessment is the goal of this research, building geometry, window size, orientation, and building materials become model input variables, see table 3. Energy use data estimated within the BES module is exported to the BEMA module.

RI PT

Table 1. EnergyPlus inputs used for this study

2.2 Building Emergy Analysis Module

M AN U

SC

Total emergy intensity for building construction and maintenance, Embuilding, Eq. (1), can be determined as the summation of emergy inputs for each material or component, divisible into four distinct categories: (a) all embodied energy for building material production, Emmaterials; (b) onsite construction work including onsite renewable inputs, human labor, and transportation, Emconstruction; (c) building operation, Emuse, i.e, fuel use for space conditioning; and (d) demolition, recycling, and decomposition of materials, Emend-of-life. Materials and construction can be accounted to represent emergy for building manufacturing, Emmanufacturing i.e., embodied energy of all materials and construction services, Eq. (2). Since the emergy of onsite renewable inputs contributes less than 1% of the total [4], this study focuses on building operation and manufacturing. For this study, recycling is not included. The emergy of a construction assembly is calculated by multiplying individual material UEVs and quantities used in manufacturing, Eq. (2). Similarly, energy use for operation can be calculated taking different sources into account, i.e., heating, cooling, hot water, and lighting, Eq. (3).

TE D

   =   +   +   +  

! ! (1)

% &%'( ) = ∑  + ∑  5



,-. /. 0 . + 1 4 6, 8 = 1,2, … (2) 

EP

=  + 23 .23 ? 



AC C

 ;< =  ,=> 0 => + ( )  @, A, , , B = 1,2, … (3) =23 >23

where

23

is unit weight of an assembly, V is the volume, Tr is UEV, l is number of building

components, Hi is emergy of human labor and services for onsite construction work, m is number of conditioned space, n is number of interior lighting fixtures, and E is the energy demanded for each equipment and zone. Note that Emmanufacturing takes time variable (T) which represents expected life span of each building component. The annual emergy intensity (sej/yr) takes on integration of different time scales of emergy sources for comparison allowing for energy depreciation. 2.3 MetaModel Development Module 8

ACCEPTED MANUSCRIPT

SC

RI PT

Metamodels are well known in the field of simulation engineering as surrogate model, black-box model, or statistical emulator. And for building energy studies, several researches have implemented metamodel techniques and demonstrated validity as a fast and cheap calculation tool for building energy and thermal comfort [34,35]. The basic difference between an analytical model and a metamodel is the degree of freedom in formal expressions. Unlike the rigor of an analytical model, a metamodel is developed from observations of the input/output of a case. Moreover, the variables are significantly reduced to the domain of interest and the observation of expected responses is restricted to the researcher’s main concern. Despite the risk of off-target effects owing to the factor screening, a metamodel can be promising for disciplines that deal with ambiguous underlying structure or unknown internal organization behind exposed responses. Although a metamodel may obscure an exact process, it can adequately approximate a system’s outcome through factors of interest. The metamodel development begins with the observation of the real system, the building geometry and envelope, in this case. Practical application of the overall procedure to this study takes several important steps which are discussed in Appendix A.

M AN U

Design of Experiment and Taguchi method

TE D

Metamodels are typically developed with a standard regression model. That is, given the data, the qualitative characters of the factor-response relationship should be quantified to develop through statistical treatment. Fundamental to this method is the measurement of each factor-response pair identified by repetitive experiments. However, lack of a strategy to control experimental conditions and outcomes beforehand may yield undesirable results with numerous errors, and hence the need for a Design of Experiment (DoE) emerged. Practically, DoE is a technique of laying out multiple variable conditions, and applied to experiments with discrete and independent variables. DoE allows for an efficient and effectively reduced number of experiments. It is classified into three methods: full-factorial experiment, fractional (or partial)-factorial experiment, and one-at-a-time experiment. However, for n parameter to be tested, a straightforward full-factorial design requires 2n, i.e., if there are 15 parameters with two level variants, the experiment requires 215 = 32,768 repetition with a change of inputs each time. Taguchi [36] demonstrated that the number of runs can be significantly reduced based on a degrees-of-freedom approach as given below:

EP

D = 1 + 

GH

(E − 1) (4)

23

AC C

where N is total number of experiments required, NV is number of independent variables, and L is number of bound levels. Eq. (4) shows 32,768 experiments of above example is reduced to only 16 repetition. Taguchi’s breakthrough involved a new adaptation of the conventional DoE method, and was developed into a robust design philosophy referred to as the Taguchi method. Based on Eq. (4), the Taguchi method offers a pre-manipulated combinatorial design tables called orthogonal arrays that contains a finite set of variable combinations. The user can adopt the most adequate orthogonal array according to number of parameters and input levels of each. Table 2 lists the Taguchi procedure on the simplest experiment being carried with two variables and two discrete intervals such that x1: [p1, p2], x2: [q1, q2].

Table 2 Example of orthogonal array 9

ACCEPTED MANUSCRIPT

Assuming no interaction between x1 and x2, after obtaining all outcomes (y1,...,y4), a regression model can be explicitly formulated involving the independent variables as follows:

(RS T RU ) (RV T RW ) , N O(=W =V )

=

(RW T RU ) (RV T RS ) , O(>W >V )

K is the intercept, and P is an error. Eq. (5)

RI PT

where, L =

J = K + LM3 + NMO + P (5)

is a regression model of simulation. Presumably, this is at best approximation with the underlying assumption that simulation is certainly harder than just evaluating Eq. (5). The least square estimate of and

turns out to be linear transformations of the overall average responses. Note that Eq. (5)

3. Case Study Building

TE D

M AN U

SC

provides an alternative way to compute estimate of the effects using any statistical package, not simulation programs. The performance results (y1,...,y4) of each trial may not always be clear responses of the experiment factors. So, in this sense, signal-to-noise ratio (S/N ratio), rooted in Shannon’s information theory, can be a convenient yardstick to evaluate evenness of the signal transmission. See Appendix B for an explanation of S/N ratio applied in this study. Taguchi method can be used to measure sensitivity. Similar to other variance-based sensitivity methods, Taguchi method requires independency of number of parameters, but can explicate main interaction effects as well as main implications. Chelela et al. [37] validated applicability of Taguchi method to energy analysis. They selected 10 variables such as air-tightness, window types, sunshades, etc and put them into an orthogonal array to predict characteristics of building energy performance whose outputs lead to obtain a set of optimal parameter values. Appendix C discusses the TaguchiANOVA used for identifying the optimum response.

AC C

EP

3.1 Description For this study, the case being studied must be representative and be able to provide a justifiable baseline with practical utility. For this reason, we took advantage of the reference building types developed by the U.S. Department of Energy. Sixteen typical commercial building models, from small offices to large hotels, across 16 U.S. weather zones are available for public use [38]. Among the many building types, a medium office with a total floor area of 4,981.5 m2 was selected for this study. This is one of the most common commercial building types, which also exhibits very little formal variation. The baseline model consists of five separate thermal zones on each floor, but can be regrouped as a perimeter and an internal zone for each floor, figure 2. Perimeter zones surround the internal zone with a 4.57m width band. It is well known that the perimeter zone is more strongly connected to the climate and is the most sensitive area to changes in physical building configuration. Since the underlying interest of the study is to support decision-making in the early phase of design by clarifying the contribution of independent variables’, the analysis focused on design changes for the perimeter space, leaving the internal space unchanged.

Figure 2. Description of the test-case building; medium office, zone 4A Baltimore 10

ACCEPTED MANUSCRIPT

3.2 Design Scenarios and Variables

M AN U

SC

RI PT

For this study, a design development scenarios based on the reference building includes the building envelope and changes to the floor plan, i.e., length to width ratio. The fixed parameters are the floor area (A = 1660.5m2), height (H = 3.96m) of each floor. Early building designers face questions mainly on (i) orientation, (ii) building shape, (iii) structure, and (iv) building envelope [24]. The design variables whose change have major influence in energy use and directly affect an exterior form include the length and width of floor, window size (W), building orientation (θ), and wall insulation (M). Correlation of the plan length and width is regulated by the area, thus Area Ratio (AR) denotes length divided by its width. Note that every variable is independent the other, and the variables of window and insulation are compartmentalized into four and two items respectively. Final notations and variables are presented in table 3. To apply the L18 orthogonal array of Taguchi method for this case, each variable is divided into three different levels except for orientation. Angular measurement of orientation is on the azimuth basis, and set to be varied clockwise from zero to 90 degree. Window size variations consider each four elevation individually, and represented by windowwall ratio. They are marked with subscript S1, S2, S3, and S4 (Ws1 ~ Ws4). For wall material change, M1 represents insulation level for the South and North side walls and M2 for East and West. Underlined numbers in table 3 relate to the baseline model, and dominant values for the most common case of medium office.

TE D

Table 3. Design variable of each level

3.3 System Boundary for Emergy Evaluation

AC C

EP

Emergy evaluation starts by framing an observation boundary, also referred to as the analytical window, by drawing a system summary diagram. Since developing the building emergy model in this research relates to the prediction of emergy value changes prompted by variations in aggregation of energy and material flows, the boundary being analyzed has been narrowed to the perimeter spaces šof the test building, figure 3. This is because the energy use of spaces further from the building envelope is influenced by their internal states rather than changes to exterior conditions. Therefore, leaving all emergy-related specifications of internal space constant, we are able to identify the influence of variations of gross physical design elements such as building shape, window sizing, or façade materials, are on the whole building emergy use.

Figure 3. System boundary diagram for the test building: the storage of perimeter space structure is analyzed

4. Results and Discussion 4.1 Baseline Energy and Emergy Estimations

11

ACCEPTED MANUSCRIPT

RI PT

The emergy evaluation consists of two separated parts - emergy synthesis of building construction and simulation of operational energy for conditioning. Baseline results in the BES module, see figure 4, describe a baseline energy use intensity (EUI) of 0.58GJ/m2 (161.11kWh/m2). Office buildings are typically internal-load dominated, so interior equipment loads takes a large share of total energy use, 0.21GJ/m2 (58.33kWh/m2). Interior and exterior lighting rank second with an EUI of 0.17GJ/m2 (47.22kWh/m2). The loads directly connected to the design scenario are space conditioning (heating and cooling) and interior lighting energy. Together they use 0.30 GJ/m2 (82.94 kWh/ m2) which represent more than half of the total in relation to change in perimeter environmental conditions.

Figure 4. Annual energy use breakdown of the reference building (baseline)

AC C

EP

TE D

M AN U

SC

For this study, the emergy synthesis for construction in the BEMA module includes only the building assemblies which are potentially sensitive to gross design change. To be specific, those which belong to internal spaces such as internal structure, indoor equipment, and fixtures are not included. Results of the emergy analysis are listed in table 4. Major material specifications follow as described in the template file. However, the DOE reference building does not contain materials that are not directly connected to heat flow, so the omitted materials are supplemented by assumptions from typical construction. For instance, the reference file defines the envelope structure as lightweight steel framed, with no information about finishes or spacing of structural members. Hence, the quantity and composition of materials (exterior finish-sheathing-insulation-structure-interior finish), thickness of sheathing board (15.9mm plywood), steel stud (2×6 C-Channel), and footing structure are assumed based on typical construction. As discussed in Section 2.2, life spans of individual assemblies play an important role in emergy intensity. Individual assemblies’ expected life is extended appropriately such that they coincide with the overall building useful life. Apart from natural depreciation, O’Connor [39] conducted a survey of 227 building demolitions in North American cities and reported that steel frame buildings only last 36.5 years. Thus, for the materials such insulation and steel studs whose durability is longer than the averaged actual building service life, 36.5 has been applied as their expected life spans. References for material properties and UEVs are noted in Table 4. The final emergy intensity of the baseline is estimated to be 1.08 ×1013 sej/m2 yr; total emergy per year is estimated to be 5.38 ×1016 sej/yr.

Table 4. Baseline emergy calculation for construction (Emconstruction)

4.2 Experimental Results 4.2.1 BES Results

There is always a trade-off between space heating and cooling. So, the energy intensity performance for heating and cooling are compared for each run. Simulation inputs except for design variables (Table 3) follows the baseline model (see Table 1). Table 5 shows the Taguchi array for the variables. A Taguchi L18 orthogonal array has been adopted for 8 parameters. This layout shows that only 18 runs are required to cover any case of combination. Three discrete numbers (-1, 0, 1) denote 12

ACCEPTED MANUSCRIPT

RI PT

levels of design quality. As the number increases from -1 (level 1) to 1 (level 3), fenestration area becomes larger and wall insulation is intensified. Run #3 shows increase in construction emergy and the greatest increase in electricity use (125×106 J/m2 yr), but the lowest in gas (18.87×106 J/m2 yr). We may assume that the largest area of windows (applied level 3 for all directions) and the lengthwise form (AR = 0.3) strongly affect higher demand for cooling energy, and lower for heating relatively. However, as shown in the figure 3, even for heating, electricity is the main energy source unlike residential buildings. Therefore, when cooling and heating are contrary to each other, total energy use mostly follows changes in the amount of electricity use. Emegy for these results (Emconditioning) is estimated by multiplication of UEVs: 1.70 ×105 sej/J for electricity and 4.35×104 sej/J for natural gas [45].

SC

Table 5. Orthogonal array (L18) for 8 parameters with three levels and energy simulation results

4.2.2 BEMA Results

M AN U

Referenced from table 5, calculation of construction emergy (Emconstruction) is performed in the BEMA module and compared for each response (y1). Run #3 shows the greatest increase in construction emergy (35.2×1011 sej/m2 yr), and run #4 shows the greatest decrease (-5.12×1011 sej/m2 yr). Variations of total emergy (y2, ∆Embuilding) are obtained adding the emergy of operational energy after executing energy simulations for each run (Table 6).

TE D

Table 6. Parameter values and BEMA results

From the results, variation of responses on each design variable can be analyzed. Figure 5 compares operational energy (∆EUI) and construction emergy (∆Emconstruction) according to their

AC C

EP

respective factors and levels. Mean and S/N ratio of the performance results are plotted over each factor. In quality control design, S/N is calculated for repetitions of each run and becomes an objective value of the design because low variability of outputs guarantees high quality products. However, in this case, S/N is utilized to represent distribution of observed data over each factor, since this study is interested in the response value itself. By definition, the highest S/N ratio can be approximated to a uniform distribution. For the EUI plot, see figure 5a, S/N ranges from -154.1 to 143.2, and for the Emconstruction, S/N varies from to -260.7 to -247.9. S/N variations are largely consistent with output data for both plots. As response values become greater, S/N also decreases, meaning the likelihood of uneven distribution of outputs. However, notable exceptions are found in Ws1, M1, and M2 for EUI, as well as θ and Ws1 for Emconstruction. For example, when the performance output of Ws1 setting changed considerably from level 1 to 2, S/N did not change significantly. That means, those factors are likely to have a lot of noise, and discrepancy between mean and S/N trends implies irregularity in relationship of factor level-response variations.

Figure 5. Response values plotted against different factor levels (S/N: right axis)

13

ACCEPTED MANUSCRIPT 4.2.3 Design Optimization

Fig. 6

M AN U

SC

RI PT

Optimal design solution can be realized by implementing results from the Section 4.2.1 and 4.2.2. Considering the independence of variables, the final optimization is obtained from explicit combination of sub-optimization of each factor. Therefore, for operational energy use, optimal combination of design variables is θ(2) + AR(3) + Ws1(1) + Ws2(1) + Ws3(1) + Ws4(1) + M1(1) + M2(2) (numbers in parenthesis refer to factor levels), which ensures the most reduced energy consumption for building operation. For emergy, optimal combination is θ(1) + AR(2) + Ws1(1) + Ws2(1) + Ws3(1) + Ws4(1) + M1(1) + M2(1). The optimal configuration is illustrated in figure 6. As discussed, operational energy use takes a large amount of total emergy, and selected variable levels for form optimization are quite similar to each other. However, the aspect ratio has a critical influence on the construction emergy value, so the alternative with the least surface area shows minimum emergy value. The difference does not seems too significant, but cannot be ignored. Note that there are certain properties that contributed different behaviors of energy and emergy. Accordingly, it is necessary to identify how much influential each factor is on the total outputs.

Optimized building form: (a) EUI-optimal model

(b) Emergy-optimal model

4.3 Characterization of Contribution of Each Design Factor to Energy and Emergy

AC C

EP

TE D

Table 7 lists the result of ANOVA for EUI responses. SST is 1.80×1016. The greatest SS is 4.71×1015 for AR, and the least is 2.14×1014 for M2. From calculation of pure sum of squares (SS'), AR ranks first in percentage contribution (19.59%), followed by Ws3 (12.88%). However, error is large (55.89%) and the side wall insulation (M1 and M2) shows negative contribution. For this case, Taguchi demonstrated it is justifiable to rule out insignificant factors until df of the error term comes to approximately half the total df of an experiment [36]. To be specific, if percent contribution of each parameter that is less than one of the error, negative, or too small to be ignored, it can be eliminated by adding it up to error, which is referred to as “pooling.” M1and M2 turned out to be unimportant, and came to be pooled. Table 8 shows the pooled result. Again, AR has the largest ratio of 23.12%, and ratio of others are also increased from around 1% to 4% by pooling effect.

Table 7. Calculation of contribution level from ANOVA (∆EUI)

Eventually, the contribution to total building emergy (∆Embuilding) is calculated. Table 9 presents the results pooled for building orientation (θ) and insulation (M1 and M2) that have less regard to the total outcome. Similar to the EUI result, AR is reported to be the most effective factor (20.67%). Ws3, Ws4, and Ws1 are other major factors with 10.97%, 7.71%, and 6.81% respectively.

Table 8. Final result: manipulation with pooling (∆EUI)

14

ACCEPTED MANUSCRIPT Table 9. Final result of ANOVA (∆Embuilding)

Figure 7. Contribution breakdown of design variables

M AN U

SC

RI PT

ANOVA is tested for construction emergy change as well. Effects of the design factors acts differently on outcomes. Figure 7 compares the contribution levels of design factors. Building size change (AR) accounts for the largest portion in common for all three outcomes (∆EUI, ∆Emconstruction, and ∆Embuilding) but is distinctively high (66%) in construction emergy. Orientation has impact for energy use, but bears no reference to others. On the other hand, it is worth noting that highperformance insulation material has little effect on the energy reduction. Taking it into consideration that energy consumption of residential buildings largely depends on quality of wall insulation, this result may look unreasonable; however, it is understandable because medium office consumes relatively much of its energy due to internal demands and lighting, and cooling control is more important than heating. Moreover, large window area contributes to more heat flows through fenestration than side walls. ∆EUI and ∆Emconstruction have different distribution, but distribution of ∆Embuilding became quite similar to ∆EUI owing to which emergy for operational energy consumption offsets emergy for building manufacturing. Nevertheless, contributiveness between ∆EUI and ∆Embuilding takes on clear difference. 4.4 Regression Analysis: Development of BES Metamodel

EP

TE D

The development of the metamodel is accomplished through regression analysis of postprocessing data obtained from Taguchi method. Regression analysis is one of the oldest but most popular methodologies to clarify system input/output relationships. It is the integration of mathematical and statistical techniques, and used for the construction of empirical metamodel as well as for design-decisions to optimize system output. Employing normalized variation of factors, see table 5, experimental results can be migrated into a polynomial function. The order of the function is determined by the complexity of interaction and statistical t-test on coefficients of function, see table 10. Second-order polynomial regression analysis leads to the following formula with normalized

AC C

variables (z) such that J = 80.57\3O + 53.01 \O \? , where J = Δab , and \3 , \O , \? ∈ { −1, 0, 1}. However, this model cannot be accepted because the p-value is 0.124 which is larger than limit of significance level (0.05). On the other hand, adjustment to the first-order regression gives the p-value 0.00841 which is much less than 0.05, hence a significant relationship between the variables in the following linear regression model: J = 10.13 − 37.46\3 + 29.75\O + 34.53\? + 31.56\5 (10)

The coefficient of z2 is turns out to be insignificant (p > 0.05), and eliminating z2 gives, J = 10.13 − 37.46\3 + 34.53\? + 31.56\5 (11) Introducing rescaled variable for each design factors, above equation is transformed as follows, 15

ACCEPTED MANUSCRIPT MO − 0.32 M? − 0.32 M3 − 0.93 l + 34.53 k l + 31.56 k l (12) J = 10.13 − 37.46 k 0.6 0.19 0.19 Finally, Eq. (12) can be rewritten as,

RI PT

J = −62.43M3 + 181.74MO + 166.11M? − 43.13 (13)

where 0.3 ≤ x1 ≤ 1.5, 0.13 ≤ x2 ≤ 0.5, 0.13 ≤ x3 ≤ 0.5, y = ∆Embuilding, x1 = AR, x2 = Ws3, and x3 = Ws4.

SC

Table 10. Result of regression analysis

M AN U

One of the greatest strengths of a metamodel is that it transforms discrete experimental results into a model built over a continuous variation of variables so that practitioners can estimate outcomes at any point of the variation. Figure 8 displays error analysis between analytical measurement and expected value from the metamodel. It compares the 18 observations of L18 array (Table 5) with the metamodels developed in Eq. (10) and (11). Residual standard error of Eq. (10) is 0.496, and F-value is 5.452. Thus, it is confident that prediction from the model fits analytical test values. As known from the failure of the second-order regression, we reconfirm that design variables are interdependent.

5. Conclusions

TE D

Fig. 8 Analytical measurement vs. expected value from the developed metamodel

AC C

EP

This study addresses the question of emergy-based form optimization for early building design process, and developed an integrated energy-emergy approach to explore feasibility of a building emergy simulation model. The results of this study showed optimization of building emergy with Taguchi method being employed and established a metamodel for further utilization of emergy analysis in decision-making for advanced design studies. The proposed methodology outlined an approach integrating building energy simulation, emergy analysis, and experimental design methods. And the case study has highlighted the nature of emergy simulation in quantitative terms. The Taguchi method was applicable since design decisions pertain to changes in forms of building assemblies which are independent one another in general cases. The final model, Eq. (13), generated by Taguchiregression analysis approximates predictive behaviors of the building emergy changes, and the bottom-up approach makes this data-driven construction computationally cheap to evaluate. The developed model does not deviate significantly from the analytical outputs as shown in figure 8. Emergy is a new environmental measure that incorporates all the upstream work and resources expended in construction and expresses them in a single unit. In spite of its holistic potential as a new design discipline against a current design industry preoccupied in operating-energy efficiency as well as limited definition of embodied energy, emergy has been less studied, and considered as an afterthought of building design. In this context, this paper makes clear contribution to two definite 16

ACCEPTED MANUSCRIPT domains of environmental building study: (i) architectural design and (ii) advanced emergy research as follows:

RI PT


A critical problem in introduction of emergy evaluation at early design stages is its timedemanding calculation and limited available data. So the establishment of fast emergy analysis procedure and the development of an easy-to-handle model was the essential task of this study, providing the logic for a wider application of emergy analysis in practice. The proposed approach was clearly proved, and the final metamodel development provides an overview of building emergy simulation without a large numbers of iterative calculations, thereby easily adaptable to practice.

M AN U

SC


This study demonstrates the synthetic value of emergy metrics by integrating dynamic energy analysis and emergy evaluation. Examined difference of the final forms and influences of variables between energy and emergy-driven optimization shows how a set of design components can be effectively allocated in association with a deep sense of global resource depletion grounded in the ecological point view of sustainability.
The case study also yields a benchmark model that is useful to future building emergy researchers. There are few of cases presented for building emergy study. Rigorous emergy evaluation of the DOE reference building provides an established point of reference. Buildings in similar size and shape can be estimated by referring to the result of this research.

TE D

To the best of our knowledge, this is the first attempt to model building emergy evaluation in formulaic expression. Thus, despite achievements and contributions stated above, our research is not without limitations, and there remain further considerations as follows,

AC C

EP


The main limitation for this study is the difficulty in general application of the empirical findings obtained from the case study. Though eliminating representativeness of the test building, the quantitative methodology used is meant to develop a pilot simulation model, and it reveals new theoretical relationships of design variable and emergy values that have previously been unexplored. However, it may neglect typological issues as our case study model is bounded to a particular building type. More case studies are required to complete development of a model applicable to different building types and to apply our proposed model to real circumstances for wider practical implications.
Because our main focus is to capture dependence of design decisions directly related to formmaking, various kinds of building components cannot be included. Our findings are constrained by limited number of variables which may not include other influential factors such as glazing types. In case, emergy of additional variables can be attached to our model with much less laborious work. However, further research is necessary to gain a better understanding of the effects of the various building materials and assemblies with more general design considerations.
Life-cyclic approaches deserve to be considered for our future works. Emergy prediction on demolition and emissions may affect decision-making. Emergy of indoor human activities and mechanical equipments can also be followed to navigate design development. Enhancement of emergy prediction and precision of results with this question would lead to far more interesting parametric studies suggesting various ways of construction of building assemblies. 17

ACCEPTED MANUSCRIPT Appendix A. Metamodel development

M AN U

SC

RI PT

The steps involved in the development of a metamodel, specific for this project are listed below. Step 1: Observation of system being studied. Building emergy flows are hard to be captured through phenomenological observation directly, the system boundary of the analytical model developed in section 3.1 carries out a role of a surrogate system. Step 2: Suite of calculation (Eq. (1) ~ Eq. (3)) works for tabulated-form calculation of baseline emergy. Step 3: Proposed design analysis is performed through the real system again encapsulated by analytical model of Step2. Step 4: Step 3 is directed to Meta modeling process. A problem being analyzed (re-design) is delimitated with design interest of practitioners. Step 5: Building design variables (e.g. space height, window size, material change, etc.) are also defined. Step 6: Meta modeling technique (Design of Experiment) is employed for simulation. Step 7: Evaluation of the system under study is interlocked with analytical model of Step 3. Step 8: Emergy performance results obtained from the previous step are formulated into a mathematical model.

Figure 9. Scheme of Meta modeling procedure. Number of variables are reduced as mimicry level moves toward higher abstract level such that p > q > r

TE D

Appendix B. Explanation of Signal-to-Noise Ratio (S/N) applied in this study. By definition, quality of telecommunication is characterized by how clear discernible signals are distributed on the way of message delivery when comparing to measurement of incorrect bit (noise). Accordingly, S/N ratio (S/N) is formulated as:

EP

rs (m6tBsu) m/D = 10log ( ) (6) rs (Dv6wx)

AC C

where var(Signal) is variation of signal sent from a source, and var(Noise) is variation of noise in the signal. Similarly, in quality control experiments, the output of each run is assessed by S/N ratio which is the sum of reciprocal number of variance. It determines the most robust set of operating conditions from variation (Roy, 1990). For smaller-the-better performance characteristics (lesser value benefits the overall system work such as amount of energy consumption), since the target value approximates to zero, the generic form of S/N ratio for undesirable characteristics becomes, 

1 m/D = −10uvt (  JO ) (7) B 23

where n is number of observations. S/N is basically a form of information that characterizes uncertainty of transmission. That is, greatest information implies maximum evenness of observed data. Negative sign of Eq. (7) indicates that larger S/N ratio ensures higher reliability of a test. Along with 18

ACCEPTED MANUSCRIPT S/N ratio test, the analysis of variance should be conducted to identify relative influence of individual factors. Appendix C. Explanation of Taguchi-ANOVA applied in this study.



({ − {| )O

23 

= 

= 

{O − B{|O

23 

{O

23

O − (}. ~. ) (8) B

SC

yz = 

RI PT

Suppose that we obtain n number of individual performance observation Yi, and the total performance average represented by Y0. m is the number of trials at the same level of a factor (m < n). First, total sum of squares of observations is calculated as follows,

O

where = , ∑23 { 0 

y = 

.23 

= 

= 

‚‚‚‚ . ,{ € − {| 0

.23 

M AN U

For each factor such as A, B, ..., the sum of squares is,

.23

O

O . ‚‚‚‚ {€ − 2  O > ,∑ƒ23 {€.ƒ 0

.



.23 O

−

. ‚‚‚‚ {€ {| + 



.23

. {|O

(}. ~. ) (9) B

TE D

Where, l = number of treatment groups of a factor, q = number of observations within a single treatment group, ∑.23 . = B , L ∈ {„, …, … }, 8 ∈ {1, 2, … , u} , † ∈ {1, 2, … , A} .Taguchi method followed by ANOVA is promising for quantification of percentage contribution of explanatory factors, which consequently leads to identification of major variables to produce the optimum response.

EP

Appendix D. Calculation of percentage contribution To obtain SS for ∆Embuilding (yi is an individual output, and subscripts mean index number of trials), average values of ∆Embuilding for each factor are calculated as below,

AC C

‚‚‚‚‚ (‡)

= (y1 + y2 + ... + y9) / 9 = 2.08×1012,

‚‚‚‚‚ (ˆ)

= (y10 + y12 + ... + y18) / 9 = -5.72×1010,

12 ‚‚‚‚‚‚‚ AR (‡) = (y1 + y2 + y3 + y10 + y11 + y12)/6 = 6.08×10 12 ‚‚‚‚‚‚‚ AR (ˆ) = (y4 + y5 + y6 + y13 + y14 + y15)/6 = -1.64×10 12 ‚‚‚‚‚‚‚ AR (‹) = (y7 + y8 + y9 + y16 + y17 + y18)/6 = -1.41×10

‚‚‚‚‚‚‚‚‚ Ws1(‡) = (y1 + y4 + y7 + y10 + y13 + y16)/6 = -2.39×1012 ‚‚‚‚‚‚‚‚‚ Ws1(ˆ) = (y2 + y5 + y8 + y11 + y14 + y17)/6 = 1.86×1012 ‚‚‚‚‚‚‚‚‚ Ws1(‹) = (y3 + y6 + y9 + y12 + y15 + y18)/6 = 3.56×1012 Similarly, 11 ‚‚‚‚‚‚‚‚‚ 12 ‚‚‚‚‚‚‚‚‚ ‚‚‚‚‚‚‚‚‚ Ws2(‡) = -9.01×1011, Ws2 (ˆ) = 3.25×10 , Ws2(‹) = 3.61×10 , 19

ACCEPTED MANUSCRIPT 11 ‚‚‚‚‚‚‚‚‚ 12 ‚‚‚‚‚‚‚‚‚ ‚‚‚‚‚‚‚‚‚ Ws3(‡) = -2.06×1012, Ws3 (ˆ) = 2.47×10 , Ws3(‹) = 4.85×10 , 11 ‚‚‚‚‚‚‚‚‚ 12 ‚‚‚‚‚‚‚‚‚ ‚‚‚‚‚‚‚‚‚ Ws4(‡) = -1.99×1012, Ws4 (ˆ) = 7.06×10 , Ws4(‹) = 4.32×10 , 11 ‚‚‚‚‚‚‚‚ 12 ‚‚‚‚‚‚‚‚ 12 ‚‚‚‚‚‚‚‚ M1 (‡) = -1.05×10 , M1(ˆ) = 1.51×10 , M1(‹) = 1.63×10 , 12 ‚‚‚‚‚‚‚‚ 11 ‚‚‚‚‚‚‚‚ 12 ‚‚‚‚‚‚‚‚ M2 (‡) = 1.68×10 , M2(ˆ) = 3.66×10 , M2(‹) = 1.02×10 ,

Degree of freedom(df) is computed by,

RI PT

dftotal = 18 ‒ 1 = 17, df = 2 ‒ 1 = 1, dfAR = 2, dfWs1 = dfWs2 = dfWs3 = dfWs4 = 2, dfM1 = dfM1 = 2, dferror = 17 – 1 – 14 = 2.

Consequently, percentage contribution of a i th factor can be obtained such that, SSi' = SSi - dfi × Ve

SC

Pi = SSi' /SST × 100,

M AN U

Where Pi is percentage quantification of the i th factor’s contribution, SSi' is pure sum of square of the i th factor, SST is total sum of square, df is the degree of freedom, and Ve is variance of error.

References

AC C

EP

TE D

[1] L. Fernández-Galiano, Fire and Memory on Architecture and Energy, MIT Press, Cambridge, MA, 2000. [2] A.F. Tzikopoulos, M.C. Karatza, J.A. Paravantis, Modelling energy efficiency of bioclimatic buildings, Energy and Buildings 37 (2005) 529–544. [3] J. Godfaurd, D. Clements-Croome, G. Jeronimidis, Sustainable building solutions: a review of lessons from the natural world, Building and Environment 40 (2004) 319–328. [4] R.M. Pulselli, E. Simoncini, F.M. Pulselli, S. Bastianoni, Emergy analysis of building manufacturing, maintenance and use: Em-building indices to evaluate housing sustainability, Energy and buildings 39 (2007) 620–628. [5] M.T. Brown, S. Ulgiati, Emergy evaluations and environmental loading of electricity production systems, Journal of Cleaner Production 10 (2002) 321‒334. [6] Sciubba, E., Beyond thermoeconomics? The concept of extended exergy accounting and its application to the analysis and design of thermal systems. Exergy Int. J. 1 (2) (2001), 68–84. [7] M. Cellura, F. Guarino, S. Longo, M. Mistretta, Energy life-cycle approach in Net zero energy building balance: Operation and embodied energy of an Italian case study, Energy and Buildings 72 (2014) 371‒381. [8] G.A. Blengini, Life cycle of buildings, demolition and recycling potential: A case study in Turin, Italy, Building and Environment 44 (2009) 319–330. [9] A.J. Marszal, P. Heiselberg, J.S. Bourrelle, E. Musall, K. Voss, I.Sartori, A. Napolitano, Zero Energy Building-A review of definitions and calculation methodologies, Energy and Buildings 43 (2011) 971‒979. [10] W.W. Braham, Architecture, style, and power: the work of civilization, in: W.W. Braham, D. Willis (Eds.), Architecture and Energy Performance and Style, Routledge, 2013, 9‒24. [11] V. Buranakarn, Evaluation of recycling and reuse of building materials using the emergy analysis method, Ph.D. Dissertation (1998) University of Florida. [12] USEPA, Draft guidance on the development, evaluation, and application of regulatory environmental models, US EPA, Washington, DC, 2003. [13] C. Bayer, , M. Gamble, R. Gentry, S. Joshi, AIA Guide to Building Life Cycle Assessment in Practice, The American Institute of Architects, U.S., 2010. [14] Odum HT. Environmental accounting: emergy and environmental decision making. New York, NJ: Wiley; 20

ACCEPTED MANUSCRIPT

AC C

EP

TE D

M AN U

SC

RI PT

1996. [15] S. Bastianoni, N. Marchettini, Emergy/exergy ratio as a measure of the level of organization of systems, Ecological Modelling 99 (1997) 33–40. [16] J.L. Hau, B.R. Bakshi, Promise and problems of emergy analysis, Ecological modeling 178 (2004) 215– 225. [17] R.S. Srinivasan, W. Ingwersen, C. Trucco, R. Ries, D. Campbell, Comparison of energy-based indicators used in life cycle assessment tools for buildings. Building and Environment 79 (2014) 138‒157. [18] F.M. Meillaud, J.-B. Gay, M.T. Brown, Evaluation of a building using the emergy method, Solar Energy 79 (2005) 204–212. [19] R.M. Pulselli, E. Simoncini, N. Marchettini, Energy and emergy based cost-benefit evaluation of building envelopes relative to geographical location and climate, Building and Environment 44 (2009) 920–928. [20] R.S. Srinivasan, W.W. Braham, D.E. Campbell, C.D. Curcija, Building envelope optimization using emergy analysis in environmental building design. Proceedings of the 12th International Building Performance Simulation Association Conference (2011) 358‒365. [21] D. Li, J. Zhu, E.C.M. Hui, B.Y.P. Leung, Q. Li, An emergy analysis-based methodology for eco-efficiency evaluation of building manufacturing, Ecological Indicators 11 (2011) 1419–1425. [22] R.S. Srinivasan, W.W. Braham, D.E. Campbell, and D.C. Curcija, Re(de)fining Net Zero Energy: Renewable Emergy Balance of Environmental Building Design, Building and Environment 47(1) (2012) 300-315. [23] W.W. Braham, Environmental Building Design: Forms of Emergy, Proceedings from the Seventh Biennial Emergy Analysis Research Conference (2013)141‒146. [24] H.Gervasio, P. Santos, R. Martins, L. Simoes da Silva, A macro-component approach for the assessment of building sustainability in early stages of design. [25] W. Wang, R. Zmeureanu, H. Rivard, Applying multi-objective genetic algorithms in green building design optimization, Building and Environment 40 (2005), 1512‒1525. [26] D. Hawkes, J. McDonald, and K. Steemers, The Selective Environment: An Approach to Environmentally Responsive Architecture, Spon Press, London, 2002. [27] Yi, H. (In press), Automated Generation of Optimized Building envelope: simulation based multi objective process using evolutionary algorithm, International Journal of Sustainable Building Technology and Urban Development 5 (2014). [28] N. Bouchlaghem, Optimizing the design of building envelopes for thermal performance, Automation in Construction 10 (1) (2000) 101–112. [29] Sattary, A and Thorpe, D (2012) Optimizing embodied energy of building construction through bioclimatic principles In: Smith, S.D (Ed) Procs 28th Annual ARCOM Conference, 3-5 September 2012, Edinburgh, UK, Association of Researchers in Construction Management, 1401-1411. [30] J. Basbagill, F. Flager, M. Lepech, M. Fischer, Application of life-cycle assessment to early stage building design for reduced embodied environmental impacts, Building and Environment 60 (2013) 81‒92. [31] C. Thormark, A low energy building in a life-cycle-its embodied energy, energy need for operation and recycling potential, Building and Environment 37 (2002) 429‒435. [32] S.S. Shrestha, K. Biswas, A.O. Desjarlais, A protocol for lifetime energy and environmental impact assessment of building insulation materials, Environmental Impact Assessment Review 46 (2014) 25‒31. [33] M.S. Taskhiri, R.R. Tan, A.S.F. Chiu, Emergy-based fuzzy optimization approach for water reuse in an ecoindustrial park, Resources, Conservation and Recycling 55(2011) 730–737. [34] A. Rabl, Parameter Estimation in Buildings: Methods for Dynamic Analysis of measured Energy Use, Journal of Solar Energy Engineering 110 (1988) 52–66. [35] S. Wang, X. Xu, Simplified building model for transient thermal performance estimation using GA-based parameter identification, International Journal of Thermal Science, 45 (2006) 419–432. [36] R. Roy, A primer on the Taguchi method, Van Nostrand Reinhold, New York, 1990. [37] F. Chelela, A. Husaunndee, P. Riederer, C. Inard, A statistical method to improve the energy efficiency of an office building. Proceedings of International Building Performance Simulation Association Conference 21

ACCEPTED MANUSCRIPT

AC C

EP

TE D

M AN U

SC

RI PT

(2007) 1756‒1764. [38] DOE, 2012. Files are available at: http://energy.gov/eere/buildings/commercial-reference-buildings [39] O’Connor, J., Survey on actual service lives for North American buildings, Woodframe Housing Durability and Disaster Issues conference, Las Vegas, October 2004. [40] Pulselli RM, Simoncini E, Ridolfi R, Bastianoni S. Specific emergy of cement and concrete: an energybased appraisal of building materials and their transport. Ecological Indicators 2007;8:647–56. [41] Brown, M.T., Buranakarn, V., Emergy indices and ratios for sustainable material cycles and recycle options, Resources, Conservation and Recycling 38 (1) (2003) 1–22. [42] EPA, San Luis Basin Sustainability Metrics Project: A methodology for evaluating regional sustainability, 2010 (Available at:http://nepis.epa.gov) [43] Odum HT, Odum EC, King R, Richardson R. Ecology and economy: emergy analysis and public policy in Texas. Energy system in Texas and the United States, policy research project report number 78. Texas: The Board of Regents, University of Texas; 1987. [44] M.Y.H. Bangash, T. Bangash, Elements of Spatial Structures: Analysis and Design, Thomas Telford Publishing, London, UK, 2003. [45] S. Bastianoni, D.E. Campbell, R. Ridolfi, F.M. Pulselli, The solar transformity of petroleum fuels, Ecological modelling 220 (2009) 40–50.

22

ACCEPTED MANUSCRIPT

List of Tables Table 1. EnergyPlus inputs used for this study.

RI PT

Table 2. Example of orthogonal array Table 3. Design variable of each level. Table 4. Baseline emergy calculation for construction (Emconstruction). Table 5. Orthogonal array (L18) for 8 parameters with three levels and energy simulation results. Table 6. Parameter values and BEMA results. Table 7. Calculation of contribution level from ANOVA (∆EUI). Table 8. Final result: manipulation with pooling (∆EUI). Table 9. Final result of ANOVA (∆Embuilding).

M AN U

SC

Table 10. Result of regression analysis.

List of Figures

AC C

EP

TE D

Figure 1 Integrated Energy-Emergy Approach. Figure 2 Description of the test-case building; medium office, zone 4A Baltimore. Figure 3 System boundary diagram for the test building: the storage of perimeter space structure is analyzed. Figure 4 Annual energy use breakdown of the reference building (baseline) Figure 5 Response values plotted against different factor levels (S/N: right axis). Figure 6 Optimized building form: (a) EUI-optimal model (b) Emergy-optimal model Figure 7 Contribution breakdown of design variables Figure 8 Analytical measurement vs. expected value from the developed metamodel. Figure 9 Scheme of Meta modeling procedure. Number of variables are reduced as mimicry level moves toward higher abstract level such that p > q > r.

23

RI PT

ACCEPTED MANUSCRIPT

M AN U

SC

Table 1. EnergyPlus inputs used for this study. List Units Item Site: location USA MD-Baltimore Shadow calculation frequency 7 2 Zone floor area per person m /person 18.58 Lights W/m2 10.76 Zone infiltration m3/s-m2 0.000302 Cooling/Heating supply air tem 12.8/50 °C p. 0.3 Minimum system airflow rati

AC C

EP

TE D

o Fan (efficiency) Air terminal

1

Variable volume, two speed DX (0.5915) Single duct, VAV, reheat

M AN U

SC

RI PT

ACCEPTED MANUSCRIPT

Table 2. Example of orthogonal array

TE D

AC C

EP

# 1 2 3 4

L(22) Orthogonal array Independent Variables x1 x2 p1 q1 p1 q2 p2 q1 p2 q2

2

Performance y1 y2 y3 y4

SC

RI PT

ACCEPTED MANUSCRIPT

Table 3. Design variable of each level. AR

Level 1

0

0.3

0.13

0.13

0.13

0.13

0.91

0.91

Level 2

90

1.0

0.33

0.33

0.33

0.33

0.40

0.40

Level 3

WS1

1.5

0.50

Fenestrated area fraction WS2 WS3 WS4

Wall Insulation (W/m2K)(1) M1 M2

Orientation (θ)

(2)

M AN U

#

0.50

(2)

0.50

(2)

0.50

(2)

0.15

(3)

0.15(3)

AC C

EP

TE D

Note: underlined values are defaults for the base case. (1) Insulation board thickness: 54mm, (2) Height of ceiling plenum space is 1.22m, and maximum available height of glazing is 2.74m (69% of the side wall area). Considering opaque parts such as window frames in fenestration, 50% is maximum for glazing in practical sense. (3) U-value of exterior walls < 0.15W/m2K for passive house, θ = Azimuth, AR = aspect ratio (plan width/depth), WS1 = South, WS2 = East, WS3 = North, WS4 = West, M1 = insulation level for South and North, M2 = insulation level for East and West.

Table 4. Baseline emergy calculation for construction (Emconstruction). 3

ACCEPTED MANUSCRIPT

Element

Material

Density 3

2

kg/m Footing and foundation wall structure

waterproofing

Weight [54] kg/m

Area

Volume

2

3

m

249.6

74.88

m

concrete

2179.0

N/A

steel(2)

7840.0

N/A

bituminous material

N/A

7.32

249.6

1073.5

N/A

1325.03

600.0

8.79

1325.03

30.0

N/A

1325.03

600.0

8.79

96.0

N/A

7850.0

29.29

7.6

19.54

1073.5

N/A

1325.03

N/A

652.63

N/A

4.04 N/A

Steel stud cavity wall and fenestration

exterior plywood

vapor barrier

0.1mm PVC

non-insulating sheathing

15.9 mm plywood

Insulation

cavity insulation

mineral fiber

Structure

metal stud

2×6 C-Channel

Interior

interior sheathing

12.7mm gypsum board (double layer)

Exterior Sheathing

finish Window

glass frame(4)

acrylic paint(3) 24mm double glazing (6mm glass) Aluminum

2739.2

2×4 C-Channel 12.7mm gypsum board (double layer)

7850.0 7.6

acrylic paint(3)

1073.5

Human work

1325.03

1325.03

N/A

37.10

Ref.

E(life)(1) [55] yr

Emergy sej/yr

1.81 ×1012

sej/kg

[40]

100

8.09 ×1015

3.2 ×104

6.94 ×1012

sej/kg

[41]

100

6.03 ×1015

1.8 ×103

9.86 ×1012

sej/kg

[41]

10

1.80 ×1015

3.3 ×102

2.55 ×1013

sej/kg

[11]

15

5.64 ×1014

1.2 ×104

2.40 ×1012

sej/kg

[14]

60

7.66 ×1014

4.0

9.86 ×1012

sej/kg

[41]

10

3.91 ×1012

1.2 ×104

2.40 ×1012

sej/kg

[14]

60

7.66 ×1014

3.6 ×103

3.09 ×1012

sej/kg

[42]

lifetime

3.02 ×1014

1325.03

N/A

3.9 ×104

6.94 ×1012

sej/kg

[41]

lifetime

7.38 ×1015

1325.03

N/A

2.6 ×104

2.15 ×1012

sej/kg

[42]

60

1.53 ×1015

0.31

3.3 ×102

2.55 ×1013

sej/kg

[11]

15

5.64 ×1014

7.83

2.0 ×104

1.42 ×1012

sej/kg

[43]

17.5

1.64 ×1015

N/A

N/A

1.3 ×104

2.13 ×1013

sej/kg

[41]

17.5

1.59 ×1016

19.53

1068.06

N/A

2.1 ×104

6.94 ×1012

sej/kg

[41]

lifetime

3.97 ×1015

39.08

1068.06

N/A

4.2 ×104

2.15 ×1012

sej/kg

[42]

60

2.46 ×1015

5.3 ×102

2.55 ×1013

sej/kg

[11]

15

9.09 ×1014

N/A

AC C

Internal walls interior wall

2579.0

N/A

0.13

unit

1.6 ×105

SC

siding

0.31

UEV

kg

M AN U

acrylic paint(3)

TE D

finish

EP

Exterior finish

Raw data

RI PT

Assembly

N/A

1068.06

0.50

1.08 ×1015

2.03% of total manufacturing emergy [15]

Total 5.38 ×1016 Note: (1) Average commercial building service life [39], 36.5 years, is applied to elements that last longer than the full lifespan of the building. (2) Volume of steel is estimated to be 5.4% of the total concrete volume. (3) Typically, 1m3 of paint covers 4287.6 m2 for double layer. (4) Weight of frame is estimated by subtraction of glass from total window weight which is 51.0 kg/m2 on average. Actual value ranges from 40.8 to 61.2 (kg/m2) [44]. (5) Not Applicable(N/A) refers to no direct regard to the emergy calculation.

4

RI PT

ACCEPTED MANUSCRIPT

Table 5. Orthogonal array (L18) for 8 parameters with three levels and energy simulation results. Insulation(W/m2K) M1 M2

#

θ

AR

1

-1

-1

-1

-1

-1

-1

-1

-1

2

-1

-1

0

0

0

0

0

0

3

-1

-1

1

1

1

1

1

1

4

-1

0

-1

-1

0

0

1

1

5

-1

0

0

0

1

1

-1

-1

6

-1

0

1

1

-1

-1

0

0

7 8

-1 -1

1 1

-1 0

0 1

-1 0

1 -1

0 1

1 -1

9

-1

1

1

-1

1

0

-1

0

10

0

-1

-1

1

1

0

0

11

0

-1

0

-1

-1

1

1

12

0

-1

1

0

0

-1

13

0

0

-1

0

1

-1

14

0

0

0

1

-1

0

15

0

0

1

-1

0

1

16

0

1

-1

1

0

17

0

1

0

-1

18

0

1

1

0

∆EUI* (×106 J/m2 yr) Electricity Natural gas -5.33

0.94

50.69

Total -4.39 38.29

-18.87

106.13

-42.36

4.91

-37.45

29.55

-6.28

23.27

-14.78

-1.89

-16.67

-35.79 -13.43

2.86 0.74

-32.93 -12.69

-7.54

2.87

-4.67

-1

15.69

1.73

17.42

0

-10.31

8.24

-2.07

-1

1

-21.20

1.65

-19.55

1

0

-26.01

-0.38

-26.39

-1

1

-14.46

0.65

-13.81

0

-1

17.53

-4.95

12.58

TE D

M AN U

SC

-12.40

125.00

1

-1

0

0.80

-1.37

-0.57

1

-1

0

1

4.69

-8.06

-3.37

-1

0

1

-1

-0.10

-4.07

-4.17

EP

WS1

Window fraction WS2 WS3 WS4

AC C

Note: *On site measurement, multiplication of UEV for each source converts the onsite energy to emergy.

5

RI PT

ACCEPTED MANUSCRIPT

Table 6. Parameter values and BEMA results. #

θ

AR WS1

Window fraction WS2 WS3

WS4

Insulation M1 M2

∆Em (×1011 sej/m2 yr) Construction(y1) Total(y2)

0

0.3

0.13

0.13

0.13

0.13

0.91

0.91

0.48

-8.17

2

0

0.3

0.33

0.33

0.33

0.33

0.40

0.40

18.60

99.34

3

0

0.3

0.50

0.50

0.50

0.50

0.15

0.15

35.20

239.53

4

0

1.0

0.13

0.13

0.33

0.33

0.15

0.15

-5.12

-75.00

5

0

1.0

0.33

0.33

0.50

0.50

0.91

0.91

3.61

51.12

6

0

1.0

0.50

0.50

0.13

0.13

0.40

0.40

-2.52

-28.47

7

0

1.5

0.13

0.33

0.13

0.50

0.40

0.15

-4.11

-63.71

8

0

1.5

0.33

0.50

0.33

0.13

0.15

0.91

1.93

-20.58

9

0

1.5

0.50

0.13

0.50

0.33

0.91

0.40

4.96

-6.61

10

90

0.3

0.13

0.50

0.50

0.33

0.40

0.91

22.90

50.36

11

90

0.3

0.33

0.13

0.13

0.50

0.15

0.40

16.60

2.67

12

90

0.3

0.50

0.33

0.33

0.13

0.91

0.15

16.60

-18.71

13

90

1.0

0.13

0.33

0.50

0.13

0.15

0.40

-4.01

-48.39

14

90

1.0

0.33

0.50

0.13

0.33

0.91

0.15

-1.29

-25.59

15

90

1.0

0.50

0.13

0.33

0.50

0.40

0.91

0.47

28.12

16

90

1.5

0.13

0.50

0.33

0.50

0.91

0.40

0.88

1.64

17

90

1.5

0.33

0.13

0.50

0.13

0.40

0.15

0.46

4.93

18

90

1.5

0.50

0.33

0.13

0.33

0.15

0.91

1.77

-0.17

M AN U

TE D

EP

AC C

SC

1

6

M AN U

SC

RI PT

ACCEPTED MANUSCRIPT

Table 7. Calculation of contribution level from ANOVA (∆EUI). Variance 6.43E+14 2.35E+15 1.27E+15 8.01E+14 1.75E+15 1.67E+15 1.23E+14 1.07E+14 5.91E+14

TE D

Df 1 2 2 2 2 2 2 2 2 17

EP

SS 6.43E+14 4.71E+15 2.55E+15 1.60E+15 3.50E+15 3.34E+15 2.46E+14 2.14E+14 1.18E+15 1.80E+16

AC C

Source θ AR Ws1 Ws2 Ws3 Ws4 M1 M2 ε (Error) Total

7

F-value 1.09 3.98 2.15 1.35 2.96 2.83 0.21 0.18 1.00

SS' 5.19E+13 3.52E+15 1.37E+15 4.20E+14 2.32E+15 2.16E+15 -9.36E+14 -9.68E+14 1.01E+16

Contribution (%) 0.29 19.59 7.59 2.33 12.88 12.01 -5.20 -5.38 55.89 100

RI PT

ACCEPTED MANUSCRIPT

df 1 2 2 2 2 2

Variance 6.43E+14 2.35E+15 1.27E+15 8.01E+14 1.75E+15 1.67E+15

1.64E+15 1.80E+16

6 17

2.74E+14

F-value 2.35 8.59 4.65 2.93 6.39 6.10 pooled pooled 1.00

AC C

EP

TE D

M AN U

SS 6.43E+14 4.71E+15 2.55E+15 1.60E+15 3.50E+15 3.34E+15

SC

Table 8. Final result: manipulation with pooling (∆EUI). Source θ AR Ws1 Ws2 Ws3 Ws4 M1 M2 ε (Error) Total

8

SS' 3.69E+14 4.16E+15 2.00E+15 1.05E+15 2.95E+15 2.79E+15

Contribution (%) 2.05 23.12 11.12 5.86 16.41 15.54

4.66E+15

25.89 100

SC

RI PT

ACCEPTED MANUSCRIPT

Table 9. Final result of ANOVA (∆Embuilding). df

Variance

2.32E+26 1.13E+26

2 2

1.16E+26 5.64E+25

1.48E+26 1.20E+26

2 2

7.42E+25 6.02E+25

2.44E+26 8.57E+26

9 17

F-value pooled 1.63 0.80

TE D

M AN U

SS

EP

2.72E+25

AC C

Source θ AR Ws1 Ws2 Ws3 Ws4 M1 M2 Error(ε) Total

9

1.05 0.85 pooled pooled 1.00

SS'

Contribution (%)

1.77E+26 5.84E+25

20.67 6.81

9.41E+25 6.61E+25

10.97 7.71

4.62E+26

53.84 100

SC

RI PT

ACCEPTED MANUSCRIPT

Table 10. Result of regression analysis. 10.12833 -37.46 29.74667 34.53167 31.56333

11.69854 14.32773 14.32773 14.32773 14.32773

EP

TE D

M AN U

Std. Error

AC C

Intercept AR Ws1 Ws3 Ws4 p-value

Coeff. estimate

10

t-value

Pr(>|t|)

0.86578 -2.61451 2.07616 2.41013 2.20295

0.402308 0.021408 0.058277 0.031482 0.046246 0.00841

ACCEPTED MANUSCRIPT Building information /Design variables

Building Energy Simulation

RI PT

Simulation model Energyplus

Tabulated form calculation

TE D

Inventory (UEVs, Material life cycle)

M AN U

Building Emergy Analysis

SC

Whole building simulation

MetaModel Development

AC C

EP

Design alternatives

Taguchi method

Regression analysis

Building form optimization

Building emergy metamodel

Figure 1 Integrated Energy-Emergy Approach.

RI PT

ACCEPTED MANUSCRIPT

internal space fenestration

SC

24.13m

perimeter space

AC C

49.91m Length

EP

TE D

M AN U

40.76m

11.88m

33.27m Width

Figure 2 Description of the test-case building; medium office, zone 4A Baltimore.

ACCEPTED MANUSCRIPT

SC

Building Construction & Operation

Core space Structure

M AN U

Sun

Service Labor

RI PT

Purchased Energy

Materials

Perimeter space Structure

Wind

Interior equipment

Conditioning

Rain

fans pumps

Light

analysis boundary

EP

Ambient Heat

cooling

TE D

heating

AC C

building boundary

Figure 3 System boundary diagram for the test building: the storage of perimeter space structure is analyzed.

AC C

EP

TE D

M AN U

SC

RI PT

ACCEPTED MANUSCRIPT

Figure 4 Annual energy use breakdown of the reference building (baseline)

ACCEPTED MANUSCRIPT

(a) ∆EUI (J/m2 yr) 3.00E+07

RI PT

-160.0

2.00E+07

-155.0

1.00E+07

-150.0 -145.0

-1.00E+07 -2.00E+07 1 2 3

Orientation

AR

1 2 3

1 2 3

1 2 3

1 2 3

1 2 3

Ws1

Ws2

Ws3

Ws4

M1

M AN U

1 2

SC

0.00E+00

Mean S/N

-140.0 -135.0

1 2 3 M2

(b) ∆Emconstruction(sej/m2 yr)

-250.0

TE D

2.00E+12 1.50E+12 1.00E+12

-240.0 -235.0

0.00E+00 -5.00E+11 Orientation

-230.0 -225.0

1 2 3

1 2 3

1 2 3

1 2 3

1 2 3

1 2 3

AR

Ws1

Ws2

Ws3

Ws4

M1

AC C

1 2

Mean S/N

EP

5.00E+11

-245.0

1 2 3 M2

Figure 5 Response values plotted against different factor levels (S/N: right axis).

(b)

M AN U

SC

(a)

RI PT

ACCEPTED MANUSCRIPT

40.75m 40.75m

EP AC C

33.27m

TE D

49.91m

Zone of interest (perimeter)

Window

Level 1 insulation

Level 2 insulation

Figure 6 Optimized building form: (a) EUI-optimal model (b) Emergy-optimal model

RI PT

ACCEPTED MANUSCRIPT

SC

70%

M AN U

60%

50%

40%

Emconstruction Embuilding

TE D

30%

EUI

20%

0% AR

Ws1

AC C

θ

EP

10%

Ws2

Ws3

Ws4

M1

M2

Figure 7 Contribution breakdown of design variables

SC

RI PT

ACCEPTED MANUSCRIPT

-

Eq. (10) Eq. (11)

TE D

-

2

3

4

5

6

7

EP

1

AC C

∆Em (×1013 sej/m2 yr)

Analytical model

M AN U

-

8 9 10 11 12 Number of observation

13

14

15

16

17

18

Figure 8 Analytical measurement vs. expected value from the developed metamodel.

ACCEPTED MANUSCRIPT

Meta modeling process

6

Design of Experiment

X1 X2 Xn X1 X2

Physical/Empirical Model

Y

Simulation Model or Analytical Model

Y*

M AN U

Parameter selection

SC

Problem specification

5

2

Xp

Black box evaluations

Proposed Design Analysis

EP

X1 X2

Baseline Analysis 8

Meta Model

Xq

AC C

7

f(X1 ,..., Xn ) = Y ,

TE D

4

3

RI PT

1

Real System

q
Figure 9 Scheme of Meta modeling procedure. Number of variables are reduced as mimicry level moves toward higher abstract level such that p > q > r.

Y**

ACCEPTED MANUSCRIPT Building and Environment 2014

An Integrated Energy-Emergy Approach to Building Form Optimization: Use of EnergyPlus, Emergy Analysis and Taguchi-Regression Method

Highlights

RI PT

Corresponding author [email protected]/ +1 267 304 8323

We model a whole building emergy simulation.




The emergy-driven optimal building form is different from the energy-driven one.




Geometry aspect ratio shows the highest responsibility to global sustainability.




The effectiveness of a metamodel to predict building emergy is tested.

AC C

EP

TE D

M AN U

SC