Estimating Net Changes in Life-Cycle Emissions From Adoption of Emerging Civil Infrastructure Technologies

By Amponsah, Isaac Harrison, Kenneth W; Rizos, Dimitris C; Ziehl, Paul H

ABSTRACT There is a net emissions change when adopting new materials for use in civil infrastructure design. To evaluate the total net emissions change, one must consider changes in manufacture and associated life-cycle emissions, as well as changes in the quantity of material required. In addition, in principle one should also consider any differences in costs of the two designs because cost savings can be applied to other economic activities with associated environmental impacts. In this paper, a method is presented that combines these considerations to permit an evaluation of the net change in emissions when considering the adoption of emerging technologies/materials for civil infrastructure. The method factors in data on differences between a standard and new material for civil infrastructure, material requirements as specified in designs using both materials, and price information. The lifecycle assessment approach known as economic input-output life-cycle assessment (EIO-LCA) is utilized. A brief background on EIO-LCA is provided because its use is central to the method. The methodology is demonstrated with analysis of a switch from carbon steel to high- performance steel in military bridge design. The results are compared with a simplistic analysis that accounts for the weight reduction afforded by use of the high-performance steel but assuming no differences in manufacture.

(ProQuest: … denotes formulae omitted.)

INTRODUCTION

Civil construction is one of the largest users of energy, material resources, and water, as well as a formidable polluter. Approximately 5% of the total global industrial energy consumption, 5% of the total anthropogenic carbon dioxide (CO2) emissions, and significant emissions of sulfur dioxide (SO2) and oxides of nitrogen (NO^sub x^) are attributed to the production of approximately 1.45 billion Mg of global cement.1 Environmental impact assessment is increasingly considered in decisions related to civil infrastructure. 1,2 This paper presents a method for the environmental assessment of new technologies for civil infrastructure.

The method that is presented is based on economic input-output life-cycle assessment (EIO-LCA). Life-cycle assessment (LCA) is a common framework for environmental assessment as direct and indirect impacts are evaluated; standards exist for its implementation, e.g., the ISO 14001 system.3 EIO-LCA is one of two distinct approaches that have emerged; the other is SETAC-LCA (Society of Environmental Toxicology and Chemistry Life-Cycle Assessment). 4 EIO-LCA is built around economic input-output (EIO) modeling that allows analysis of the direct and indirect impacts of public decisions, for example, to assess changes in unemployment from shifts in government spending.5,6 The application of the EIO-LCA method has been facilitated with development of the web-based “EIOLCA model.”7

Previous applications of the EIO-LCA method to civil infrastructure were limited to standard infrastructure. In these applications, the civil infrastructure was “built” using the outputs of existing economic sectors included in the EIO national tables. For example, in an LCA of steel versus steel-reinforced concrete bridges, the bridges were built using the outputs of the economic sectors of iron and steel mill (Standard Industrial Classification [SIC] code 331111) and ready-mix concrete manufacturing (SIC code 327320).8 In another LCA study that compared asphalt and steel- reinforced concrete pavements, the pavements were built from the outputs of the iron and steel mill, ready-mix concrete manufacturing, and asphalt paving mixture and block manufacturing (SIC code 324121) sectors.9 The method that is developed in this paper focuses on the LCA of new materials (high-performance steels [HPS], high performance concrete, pervious concrete) that are not represented by the existing sectors.

There are several reasons why users of the EIO-LCA model may find it problematic to “build” new materials from existing sectors. First, the model user would need to identify all of the main sectors that contribute a significant burden of any type and then assign appropriate values. Because there are typically many sectors that contribute a significant amount of at least one type of environmental burden, this imposes a large information requirement. For example, an EIO-LCA model run of the steel sector reveals that if only the top six input sectors (on the basis of dollars) are included, 45% of volatile organic compounds would be missed. Second, it would be difficult to meet this information requirement in a way that ensures consistency with the information embodied in the input- output tables that underlies the EIO method. For example, in building HPS steel, it would be important to ensure consistency with the iron and steel sector. Third, without specific understanding of the input-output tables and knowledge and access to tools for linear algebra, errors such as double counting can be easily introduced. The developed method details how a user can incorporate external data to effectively modify the standard sector to reflect that of the new material. In doing so, the method lessens data requirements, enforces a measure of consistency, and prevents errors.

The method that is developed is applicable to the assessment of civil infrastructure (e.g., bridges, buildings, or pavements) and other examples of final demand (i.e., goods or sales to final consumers, which includes households and the government). To evaluate the impact of the use of new materials that are to be intermediate inputs to other producing sectors, a more involved approach is required, such as that of Joshi.10 Such an approach does require practitioner knowledge of linear algebra and associated computer implementations.

This paper explores an issue not investigated in the previous EIO- LCA studies. There typically will be differences in costs between alternative technologies for civil infrastructure that have the same function, for example, two bridges designed to the same standards with one built from a standard steel and the other from HPS. For proper comparison, such cost differences should be accounted for in an environmental assessment, because any savings from choosing the less costly alternative would be applied to other economic activities, with additional environmental burdens. The method that is presented addresses this issue.

The purpose of this paper is to develop a method for use by practitioners that can be used to evaluate the change in life-cycle burdens when switching from a standard material, which is well represented by an EIO sector, to a nonstandard, or emerging material that is not represented by an existing EIO sector. The evaluation of such a switch has not previously been explored. In the next section, a background on EIO and EIO-LCA is provided. We include this background to present a clear mathematical introduction for practitioners (that is not included in other EIO-LCA papers) and because the equations are used later in the development of the method. Following the development of the method is a demonstration involving a LCA of a switch from a carbon steel bridge (standard) to a high-performance (nonstandard) steel bridge. The results are compared with a simpler method that accounts for differences in material quantities for the carbon and HPS steels but ignores differences in their manufacturing inputs. The results are then discussed and conclusions are offered.

BACKGROUND ON EIO AND EIO-LCA

Economic Input-Output Modeling (EIO)

The EIO modeling approach was developed by Wassily Leontief, for which he received a Nobel Prize in Economics in 1973. Others have further explored the approach.11 EIO modeling can be used to assess the approximate effect on the column vector of sector output, x, from changes in final demand, represented by a column vector, f. EIO modeling is an approximation in that it assumes an economy at general equilibrium.5

An EIO table, or direct requirements matrix, D, describes the flow of goods and services between all of the individual sectors of an economy.5 It has traditionally been represented and expressed in monetary terms in a base year. The U.S. Department of Commerce Bureau of Economic Analysis regularly generates EIO tables for the national economy; the1997 table consisted of 491 = 491 sectors.12 A simplified three-sector version is shown in Figure 1.

Production of a steel bridge is used as an example in the illustration (Figure 1). The manufacture of steel requires as inputs many direct and indirect materials represented by several economic sectors, including for example iron ore mining (SIC code 212210), lime manufacturing (SIC code 327410), coal mining (SIC code 212100), ferroalloy and related product manufacturing (SIC code 331112), wholesale trade (SIC code 420000), and truck transportation (SIC code 484000). For the purposes of this illustration, only two direct inputs are considered: iron ore mining, designated sector 1, whose total output and final demand are x1 and f1, respectively, and power generation and supply (SIC code 221100), sector 2, with total output and final demand of x2 and f2; the total output and final demand of the iron and steel mill are x3 and f3. The sectoral interactions are represented by the direct requirements matrix D, which is shown in Table 1 for the illustrative example. Values for the illustration are actual values taken from the 1997 Bureau of Economic Analysis EIO data. The total sector output x can be broken down into that meeting final demand (f) and that which serves as an input to other sectors, represented as Dx. That is,

Ix = Dx + f (1)

where I is an identity matrix. Equation 1 can be rearranged to express x as a function of f:

x = (I – D)^sup -1^f (2)

The term (I – D)-1 is referred to as the Leontief inverse matrix. Equation 2 can be expressed also as a power-series10

x = f + Df + DDf + DDDf + . . . (3)

The power series expansion clearly indicates the indirect nature of the changes in economic output resulting from a change in final demand. Whereas Df refers to the direct inputs required for the production of f, DDf refers to the direct requirements for the production of Df; the higher order terms show still further indirect inputs for f.

In EIO analysis, the change in the output x from an incremental change in f is typically desired. Because matrix multiplication is associative and distributive with respect to addition:

Deltax = (I – D)-1 (f + Deltaf) – (I – D) – 1f (4)

simplifies to:

Deltax = (I – D)- 1Deltaf (5)

We can therefore evaluate with the same Leontief inverse changes in x from changes in f without knowledge of the baseline (f). For this reason, hereafter we drop the Delta, and intend for f and x to refer to Deltaf and Deltax, respectively. Consider, for example, that there is an exogenous (to the national economy) final demand of f1, f2, and f3, as represented in Figure 1, respectively, as 0, 0, and 1 (dollars output). Using the elements of D as in Table 1 and showing the elements of the matrices in eq 2, we have

… (6)

Evaluating eq 6, the total change in economic output (x1, x2, and x3) is determined to be 0.0447 and 0.0472 for the iron ore mining and power generation and supply sectors, respectively, and 0.130 (not counting the final demand itself of $1) for the iron and steel mills sector. It is noted here that the change in sector output used to meet the additional demand is greater than the direct inputs of 0.037, 0.026, and 0.113, respectively, by approximately 21, 45, and 15%.

The EIO-LCA Method

The EIO-LCA method draws from the U.S. Environmental Protection Agency (EPA) and other databases to determine environmental burden coefficients for each of the EIO sectors. A column vector of environmental burden coefficients, e, is defined here as the change in environmental burden per unit output of each sector; for example, metric tons of an air pollutant per dollar economic output in the iron and steel mill sector. Horvath1 defines the diagonal matrix R, in which the elements along the diagonal are the elements of e.

A column vector of environmental burden by sector, b, is obtained by multiplying eqs 2 and 3 by R to obtain the inverse and power series expressions, respectively:

b = Rx = R(1 – D) – 1f (7)

b = R(f + Df + DDf + DDDf + . . .) (8)

The sum of the burden across sectors, sum(b), can be expressed directly in terms of e:

sum(b) = e^sup T^x = e^sup T^(I – D^sup -1^ 1f (9)

sum(b) = e^sup T^(f + Df + DDf + DDDf . . .) (10)

where e^sup T^ is the transpose of e.

Definition of Environmental Burden Coefficients

For later use, a unit environmental burden coefficient, ei, is defined as the amount sum(b) resulting from one dollar final demand in sector i, that is, f = (fi = 1, fj-i = 0). In Figure 1, the final demands corresponding to the evaluation of e3 are shown for the 3 = 3-sector economy. The amount ei differs from ei in that the indirect environmental burden is included. Substituting f = (fi = 1, fj=i = 0) into the power series of eq 10, ei is evaluated as

… (11)

This expression can be simplified. The product Df evaluates to di, the ith column vector of D. Substituting,

… (12)

Recognizing that the product of eT and f = (fi > 1, fj[not equal to]i = 0) evaluates to ei leads to:

… (13)

Comparing eq 13 to eq 10, it can be observed that ei is the sum of the direct burden associated with the production of the unit final demand itself plus the direct and indirect burden associated with the inputs to that production.

The EIO-LCA Model

The EIO-LCA model implements the EIO-LCA method. Hendrickson et al.13 describe the EIO-LCA model and its use. The matrices (I – D)- 1 and R in eq 7 have been determined and integrated into the web- based EIO-LCA model. The data sources include the 1997 industry- byindustry input-output matrix of the U.S. economy as developed by the Department of Commerce, electricity consumption from the U.S. 1998 Manufacturing Energy Consumption Survey, electricity consumption for mining sectors from the 1997 Economic Census, conventional pollutant emissions from fuel use from EPA, greenhouse gas emissions as calculated based on emissions factors from fuel use, and U.S. toxic release inventory (TRI) data derived from EPA’s 2000 Toxic Release Inventory.13

In its standard use, the model is set up to evaluate b and sum(b) given a change in final demand in a single sector i, that is, f = (fi > 0, fj[not equal to]i = 0). We define the function EIOLCA(i, fi) to refer to a single model run reporting sum(b), given f = (fi > 0, fj[not equal to]i = 0). The unit burden coefficient for a sector in the EIO-LCA model can therefore be evaluated with a single model run:

… (14)

(To accommodate apparent round-off errors, the coefficient is better computed as ei = EIOLCA(i, fi = 1M$)/106).

From coefficients ei, because matrix multiplication is associative and distributive with respect to addition, the burdens can simply be added. The burden from any f can be calculated as:

… (15)

ASSESSING THE CHANGE IN LIFE-CYCLE ENVIRONMENTAL BURDENS WITH A SWITCH TO AN EMERGING CIVIL TECHNOLOGY

A methodology is needed to estimate the difference in environmental burden if switching from use of a standard material to nonstandard material for use in civil infrastructure. The assessment must consider changes in material requirements and in material production. For example, in comparison to carbon steel (e.g., A36 steel), less HPS (e.g., HPS50) is required in the erection of a steel bridge because it has improved properties. Further, HPS50 has a different elemental composition and manufacturing process, requiring more ferroalloy and more energy but less iron ore.14

Two approaches are outlined here for the determination of the environmental burden with a switch from use of standard material to nonstandard material: a quantity approach that accounts for differences in material requirements but ignores differences in material production and an adjusted-coefficient approach that considers both differences in material requirements and in material production. The discussion below considers only one environmental burden parameter; the methodology can easily be generalized to several parameters. Throughout the remainder of the paper, we refer to the EIO sector representing the standard material with the subscript “std,” that is, i = std, and referring to the nonstandard sector with the subscript “non.”

Quantity Approach

This approach considers the difference in the quantity of standard and nonstandard material required for the infrastructure but ignores differences in material production. The underlying assumption of this approach is that the sector in the EIO-LCA model that represents the standard material is also adequate for assessing the environmental burdens associated with the nonstandard material.

In determining the environmental burden, first a functional unit for the infrastructure is defined. In the demonstration section, a steel bridge is the functional unit. Other examples of functional units are “a parking lot with appropriate stormwater runoff controls” and “1-km two-lane road designed to meet applicable standards.” Next, the quantity of the standard and nonstandard material required for production of the functional unit, Qstd and Qnon, respectively, are determined based on designs of the functional unit. The environmental burden associated with use of the standard material, Bstd, is evaluated as:

… (16)

The product pstdQstd is the cost of the standard material required, where pstd is the price for the standard material.

The environmental burden associated with use of the nonstandard material as calculated with the quantity approach, Bnonq , is equivalent to eq 16 with the exception that Qnon replaces Qstd:

… (17)

The resulting difference in environmental burden, DeltaBq, is:

… (18)

With the quantity approach, the percentage change in environmental burden is simply the percentage change in the material requirement.

Adjusted-Coefficient Approach

The adjusted-coefficient approach efficiently addresses differences in both the amount of material in the functional unit and difference in material production. Addressing the differences between the burdens involves four steps (Figure 2). The first two steps are equivalent to those of the quantity-based approach: define a functional unit and assess the quantities of the materials, Qnon and Qstd, required for the functional unit. The third step is to develop enon from estd. In developing enon, data are gathered (step 3a) on the inputs required to make one unit of production (e.g., 1 t of steel) and on the prices of the standard and nonstandard materials, pstd and pnon, respectively (in commensurate units, e.g., $/ton steel). Such data may include material composition, energy usage, and other resource consumption. From the gathered data, for each input sector i, the fractional change, ki, in the amount of input from the standard to the nonstandard material is determined (step 3b); refer to eq 24 for the mathematical definition. For example, if data indicate that the iron content in nonstandard steel is 20% lower than that of standard steel, then the fractional change for the iron ore mining sector is -0.20. For many input sectors, the material input quantities will remain unchanged and the adjustment will not have to be made (i.e., ki = 0). We defer to a later section the handling of the case in which the fractional change is undefined, that is, where an input is required for the production of the nonstandard material but is not required for production of the standard material. From the gathered information and the ki’s, enon is then determined (step 3c). The development of enon is given in the appendix. Its development involves several steps. First, the direct requirements matrix D is enlarged to include a new sector for the nonstandard material, as is shown in Table 2. The bottom row includes only zeros because civil infrastructure is final demand, that is, the nonstandard sector is not an input to any other sector. The appendix details the matrix manipulations to show that, given the factors ki, enon can be simulated using the original matrix D:

… (19)

where set G consists of those sectors requiring adjustment, i.e, for which ki is not zero. The term di,std is the ith row of the column of the direct requirements matrix D that represents the inputs to the standard sector: di,std dollars of sector i output are directly required for one dollar’s worth of output from the standard sector.

Equation 19 is central to the adjusted-coefficient method, specifying how the burden coefficient for the nonstandard sector can be obtained. The life-cycle burden per dollar spent in the nonstandard sector, enon, is effectively a correction to estd. If one sees past the price terms, which only have the effect of ensuring that the units are correct, enon can be seen to be equated with estd plus two corrections: a correction for the emissions associated with production of the final demand itself-subtraction of estd and addition of enon-and a correction for the life-cycle emissions associated with the differences in the inputs to manufacture, which is represented by the summation term in eq 19.

To implement eq 19, one standard EIO-LCA model run is required to obtain estd. Additional runs are required to obtain ei for each sector requiring adjustment. The values of di,std can be obtained by multiplying the “Direct Economic %” column by the “Total Economic” column (or obtained directly from the input-output tables themselves). (Equivalently, a single EIO-LCA run can be conducted using the “Custom Product Builder” in the EIOLCA model, adding as “ingredients” the sectors requiring adjustment and entering for each the product eikidi,std).

With enon determined, the final step (step 4) is to then compute the life-cycle environmental burden for the nonstandard material. The life-cycle environmental burden associated with use of the nonstandard material, Bnon adj , is determined as:

… (20)

The difference in environmental burden DeltaBadj between the standard and the nonstandard technologies is:

… (21)

If ki is Undefined

It may be the case that an input is needed for the nonstandard material that was not needed for the standard material. In such a case, di,std would be zero and ki would therefore be undefined (as a fractional change from zero is undefined). In such a case, the amount of the input per unit nonstandard material, xi, and the price of each input, pi, would be required. The dollars input per dollar nonstandard material output would be given by xi pi/pnon. This leads to the addition of a term to eq 19:

… (22)

where the set H refers to those sectors for which there is input required for the nonstandard material production that was not required for the standard material. For such sectors, price information is needed.

Accounting for Cost Differences

Earlier EIO-LCA studies8,9 reported the life-cycle burdens associated with technology alternatives with different costs. A factor considered here that was not considered in those studies is the consequence of the differences in costs on the overall analysis of the environmental burden of a switch from use of the standard material to the nonstandard material. If there are savings with use of the nonstandard material, the savings would be applied to some other economic activity, resulting in likely additional environmental burden. Alternatively, should use of the nonstandard material cost more, there would be a displaced environmental burden. Leontief, in an economic inputoutput modeling study on the reallocation of the defense budget to civilian sectors after World War II, considers a similar offset analysis.5 The change in emissions from a switch from the standard to the nonstandard technology can be determined as:

DeltaB^sup adj +^ = DeltaB^sup adj^ + LCA+p^sub non^Q^sub non^ – p^sub std^Q^sub std^) (23)

where the function LCA() represents the environmental burden resulting from the use of the savings or, if the new technology is more costly, the displaced burden. In the demonstration of the developed method, which is presented next, this issue is examined. In some cases, the alternative use of the savings will be under control of the decision-maker. In other cases, it may not be known specifically, and in such cases, using a long-run argument, it may be reasonable to assume that savings are applied to a general sector. For example, the cost savings could be assumed to apply to a general civil infrastructure sector such as highway, street, bridge, and tunnel construction (SIC code 230230).

DEMONSTRATION OF THE METHOD

The purpose of presenting the demonstration is (1) to illustrate the application of the adjusted-coefficient method, and (2) to investigate its worth through comparison to the simpler quantity- based approach. An additional goal is to investigate the significance of the application of any cost savings. An EIO-LCA analysis of a switch from using A36 steel to HPS50 steel for a military steel bridge was conducted to demonstrate the developed method. A36 was formerly the conventional steel used in military steel bridge design; HPS50 steel is the conventional steel currently used. Among the environmental burdens considered were emissions of conventional pollutants, greenhouse gases, and energy consumption.

Aside from quantity differences, A36 and HPS50 steels have different material composition. In addition, the two steels require different heat treatment in their production. Production of HPS50 grades of steel may require a quenching and tempering heat treatment process to achieve the desired microstructure of the steel.14 Because of the differences in material input and energy, it is therefore expected that the manufacturing of the two different types of steel will have different life-cycle environmental burdens.

There are several caveats to the analysis. To focus on the differences between the quantity-based approach and adjusted- coefficient approach, the boundary of analysis was restricted to the life-cycle assessment of the steel required in the bridge design. The analysis therefore excluded the fabrication, transportation, and erection of the steel members. Consideration of these additional factors would alter the results. An additional and important caveat is that only the cross-sectional geometry was altered in the design. The design was not optimized to take advantage of the higher strength of the HPS50 steel. The reported life-cycle environmental burdens for the HPS50 bridge were anticipated to exceed those of a design that is optimized for HPS50 steel.

The adjusted-coefficient method as outlined in Figure 2 was applied to this problem. The functional unit selected was “a 40-ft span military nonstandard fixed bridge” (step 1; “nonstandard” here refers not to the type of steel but instead to the military use of the bridge). The military bridge design is a simply supported single span utilizing a laminated wood deck supported by laterally braced steel stringers; the wood deck is the same for the conventional and HPS bridges and is therefore excluded from the analysis. The weight of steel required for the 40-ft span was then determined (step 2). The design included five girders (for both the standard and HPS50 steel bridges). The weight per linear foot (lb/ft) for the A36 and HPS steels that was necessary to meet the bridge design standards (moment capacity while meeting deflection limits) was 116 and 94 lb/ ft, respectively. All analysis and design tasks were performed according to FM 3-34.343, the Military Nonstandard Fixed Bridging document15 and AASHTO LRFD Bridge Design Specifications.16

Data, including price data (pnon and pstd), the chemical composition of the steels (Table 3), and information on the energy inputs was collected (step 3a). The year 2002 prices, pnon and pstd, for manufactured steel were $0.39/lb and $0.46/lb, respectively.14 The higher unit price of the HPS50 steel members is due to the further heat treatment required to improve the crystalline microstructure of the steel.

Using the chemical composition data given in Table 3, the fractional changes, ki, were determined (step 3b). The values are reported in Table 4. The sectors for which the direct input amounts were altered include: iron ore mining, carbon and graphite manufacturing, coal mining, ferroalloy and related product manufacturing, and copper, nickel, lead, and zinc mining. Although lime serves as a direct input to the steel making process, it was assumed that the quantity of lime consumed per ton of the A36 steel and the HPS50 steel produced is unchanged (i.e., k = 0).

Because of limited data and because energy usage for the manufacturing of the steel is plant-specific, the fractional changes for the energy required for the manufacturing of the steel were left as variables in the demonstration. Fractional changes for the energy- related sectors-power generation and supply, coal mining (coke), and natural gas distribution-are represented, respectively, as kpgs, kcm, and kng. Importantly, the EIO-LCA environmental databases and sectoral inputs for the iron and steel mill sector are much more reflective of carbon steels, which in 1990 comprised 92.7% of the steel market by tonnage, whereas alloy steels comprised 5.5% of the steel market.17 Consequently, the iron and steel mill sector is assumed to represent A36 steel, because it is a carbon steel. The HPS50 steel is considered the nonstandard material (even though it is a standard steel for bridges) for which enon needs to be developed. It would be expected that emerging technologies/ materials and special application materials like HPS50 would constitute a small share of the overall market represented by the most closely associated EIO sector. If this were not the case, the more detailed analysis involving the disaggregation of sectors outlined by Joshi10 would be required.

After obtaining ki, enon was determined (step 3c) using eq 19. Finally, the environmental burden for the HPS50 steel was estimated using eqs 17 and 20 for the quantity-based and adjusted-coefficient approaches, respectively (step 4). The environmental burden associated with A36 steel was evaluated using eq 16. The change in environmental burden was then computed using the quantity-based and the adjusted-coefficient approaches with application of eqs 18 and 21, respectively. The significance of the application of the cost differences was then explored with reference to eq 23.

RESULTS AND DISCUSSION

The weight and the weight savings, cost and cost savings, and the change in environmental burden from using HPS50 in place of A36 steel in the example bridge designs are shown in Tables 5 and 6. Use of HPS50 steel lessens the cross-sectional area of the steel members that is required, which leads to a weight reduction of 19%.

Although 19% less steel is required, there is a higher unit cost for the HPS50 steel. The overall steel cost reduction per bridge is 4.42% (Table 5). Further cost savings would be expected for an optimized design for the HPS steel, and when incorporating the savings in transportation and erection costs-functions largely of weight-and savings in maintenance costs, as HPS50 is a weathering steel.

Estimates of the environmental burdens associated with switching from use of A36 to HPS50 steels are given for the two approaches in Tables 5 and 6. With the quantity-based approach, the emissions and energy usage are reduced by 19%, in line with the weight reduction (Table 5).

The adjusted-coefficient approach results in a more complicated assessment. The various environmental burdens are given in Table 6 as linear functions of four parameters: the ratio enon/estd, and the fractional increases-kpgs, kcm, and kng-for the three energy- related sectors, power generation and supply, coal mining, and natural gas distribution, respectively. Insight into the impact of a switch to HPS can be drawn for each environmental burden parameter by examining the intercept and coefficients. For example, if examining the function for SO2 given in the top row of Table 6, if the SO2 emissions per dollar output at the iron and steel mill were equivalent for the A36 and HPS50 steels and no additional energy input (per ton of steel) were required, then there would be a decrease in SO2 emissions for each bridge of 0.00527 metric tons (t) (= -0.01928 0.01401). In this case, the SO2 emissions would be approximately 34% greater than that estimated by the quantity-based approach. As there is a reduction, a switch to HPS50 would be favorable for SO2 emissions. However, if the energy inputs for the quenching and tempering heat treatment process results in a 100% increase in the energy inputs (for all three energy sources), then there would be slight increase in SO2 emissions of 0.00786 t (= – 0.01928 0.01401 0.01229 0.00063 0.00021). The quantitybased approach in this case would have the wrong sign under such an assumption.

Unlike the quantity-based approach in which all burden parameters were reduced in proportion to the quantity reduction, with the adjusted-coefficient approach a switch may be favorable for some environmental burden parameters but unfavorable for others. For example, consider as before that enon/estd = 1 and assuming a 100% increase in the energy inputs. If ignoring the cost savings, as in Table 5, an increase in GWP of 1.286 t (= -9.45892 + 7.04208 + 2.40849 + 1.1141 + 0.17997) CO2 equivalent is estimated. In such cases, an overall environmental assessment of a switch would have to consider some weighting (formal or informal) of the burden parameters.

The 4.42%, or $392, steel cost savings per bridge would be applied to another activity that may result in additional environmental burden. There are many possible scenarios that could be considered for the allocation of the savings. To illustrate the potential importance of the allocation of the cost savings, three scenarios were considered: (1) the savings are returned to an infrastructure-related budget, (2) the military uses the savings for hospital services, or (3) the military allocates the savings to research and development. The sectors of highway, street, bridge, and tunnel construction (SIC code 230230), hospitals (SIC code 622), and scientific research and development services (SIC code 5417), were used to represent these three scenarios.

In Table 7, the results of applying the savings to each of these sectors are shown. The emissions associated with the cost savings are highly dependent on the sector. Focusing on global warming potential (GWP), the burden associated with application of the savings to highway, street, bridge, and tunnel construction is 7.2 and 9.8 times the burden if applied to hospitals and scientific research and development services, respectively. The burden associated with the allocation of savings can make a considerable difference in the assessment of the switch. If the cost savings were applied to the highway, street, bridge, and tunnel construction sector, an increase of 2.368 t (= 1.286 1.082) CO2 equivalent would be estimated (using eq 31) under the earlier assumptions of enon/ estd = 1 and a 100% increase in energy inputs. Accounting for the application of the cost savings would therefore result in an increase of 84% over the prior estimate of 1.286. If, however, the cost savings were applied to scientific research and development services, the increase in estimated metric tons CO2 equivalent would be 8.6%.

CONCLUSIONS

A method based on use of the EIO-LCA model (“adjustedcoefficient method”) was developed to allow for an environmental life-cycle assessment of new materials for which there are no corresponding sectors in the EIO-LCA model. The method is limited to civil infrastructure and other examples of “final demand,” demands that directly meet household and government needs.

A demonstration was conducted to show how, starting with data collection, the adjusted-coefficient method could be applied in an environmental assessment of emerging or application-specific materials for civil infrastructure (those materials representing a small fraction of an EIO sector). The demonstration involved an environmental assessment of a change from conventional (A36) steel to the lighter and stronger HPS50 steel in a standard military bridge design. For clarity of the demonstration, the boundaries of the analysis were restricted to the supply chain for steel manufacture and did not incorporate consideration of the fabrication or transportation of the steel, or erection of the bridge.

An approach ignoring differences in manufacturing inputs (a “quantity-based approach”) would suggest that the smaller steel requirement would lead to an equal percent reduction in each of the environmental burden parameters. In contrast, the adjusted- coefficient approach that was developed and that accounts for differences in the inputs to manufacture suggests that the changes in environmental burden parameters should not be expected to be in accordance with the reduction in the steel requirement and will not be uniform across the range of environmental burden parameters.

In addition, it was demonstrated that if there are cost savings in use of the new material, the application of the cost savings should be considered in the analysis, because there are associated environmental burdens.

ACKNOWLEDGMENTS

This work was partially supported by the Advanced Technology Institute Inc. through the Vanadium Technology Program and the U.S. Army.

IMPLICATIONS

New materials for civil infrastructure are continually being developed and proposed for use, e.g., high-performance steel. The typically superior properties (e.g., stronger, lighter) often translate to fewer material requirements in the engineering design. However, as the inputs (e.g., energy, raw materials) to their manufacture are different than the standard materials, the manufacturing and associated lifecycle emissions may be higher and therefore need to be estimated. The method presented in this paper aims to allow decision makers to estimate life-cycle environmental burdens while considering both changes in quantity and differences in manufacture.

REFERENCES

1. Horvath, A. Construction Materials and the Environment. Annu. Rev. Environ. Res. 2004, 29, 181-204.

2. Leadership in Energy and Environmental Design (November, 2002): Green Building Rating System for New Construction and Major Renovations (LEED-NC), Version 2.1.

3. Martin, R. ISO Guidance Manual; Technical Report; NCEDR/98- 06; National Center for Environmental Decision-Making Research: Knoxville, TN, 1998.

4. Horvath, A. Ph.D. Dissertation, Carnegie Mellon University, Pittsburgh, PA, 1997.5. Leontief, W. Input-Output Economics; Oxford University: New York, 1996. 6. Baumol, W.J.; Wolff, E.N. A Key Role for Input-Output Analysis in Policy Design; Regional Sci. Urban Econ. 1994, 24, 93-113.

7. Economic Input-Output Life-Cycle Assessment Model; available at http://www.eiolca.net (accessed March 6, 2006).

8. Horvath, A.; Hendrickson, C.T. Steel vs Steel-Reinforced Concrete Bridge: Environmental Assessment; ASCE J. Infrastruct. Systems; 1998, 4, 111-117.

9. Horvath, A.; Hendrickson, C.T. A Comparison of the Environmental Implication of Asphalt and Steel-Reinforced Concrete Pavement; Transportation Research Record No. 1626, Transportation Research Board, Washington, DC, 1998; pp 105-113.

10. Joshi, S.V. Product Environmental Life-Cycle Assessment Using Input-Output Techniques; J. Indust. Ecol. 2000, 3, 95-120.

11. Miller, R.A.; Blair, P.D. Input-Output Analysis: Foundation and Extension; Prentice Hall: Englewood Cliffs, NJ, 1985.

12. Bureau of Economic Analysis Economic Input-Output Data; available at http://www.bea.gov (accessed March 15, 2006).

13. Hendrickson, C.T.; Lave L.B.; Matthews, H.S. Environmental Life Cycle Assessment of Goods and Services: an Input-Output Approach. Resources for the Future: Washington, DC, 2006.

14. Lwin, M.; Myint M. High Performance Steel Designers’ Guide, 2nd ed.; Federal Highway Administration; Western Resource Center; U.S. Department of Transportation: San Francisco, CA, 2002.

15. Military Nonstandard Fixed Bridging; Field Manual Number FM 3- 34.343 (SS FM 5-446); Headquarters, Department of the Army: Washington, DC, 2002.

16. American Association of State and Highway Transportation Officials-Load and Resistance Factor Design Bridge Design Specifications, Customary U.S. Units, 4th ed.; 2007.

17. Houck, G.W. U.S. Geological Survey Minerals Yearbook, Vol. I, Metals & Minerals. Iron and Steel; 1994; available at http:// minerals.usgs.gov/minerals/pubs/commodity/iron_&_steel/index.html (accessed October 17, 2007).

Isaac Amponsah, Kenneth W. Harrison, Dimitris C. Rizos, and Paul H. Ziehl

Department of Civil and Environmental Engineering, University of South Carolina, Columbia, SC

About the Authors

Isaac Amponsah is a Ph.D. candidate and research assistant in the environmental program area of the Department of Civil and Environmental Engineering at the University of South Carolina. Kenneth W. Harrison is an assistant professor in the environmental program area in the Department of Civil and Environmental Engineering at the University of South Carolina. Dimitris C. Rizos is an associate professor in the structural program area in the Department of Civil and Environmental Engineering at the University of South Carolina. Paul H. Ziehl is an assistant professor in the structural program area in the Department of Civil and Environmental Engineering at the University of South Carolina. Please address correspondence to: Kenneth W. Harrison, 300 Main Street, University of South Carolina, Columbia, SC 29208; phone: 1-803-777-1917; fax: 1- 803-777-067; e-mail: harriskw@engr.sc.edu.

APPENDIX

Development of enon

To develop enon, we consider the introduction of a new EIO sector for the nonstandard material. A new direct requirements matrix, D , is created that includes an additional row and column to accommodate the nonstandard sector. D is composed of the matrix D, a column vector dnon of the same size as dstd, and 0, a row vector of zeros. Using matrix notation, D = [Ddnon;0]; it is shown in Table 2 for the illustrative 3 = 3-sector economy. The bottom row of D is 0, as the nonstandard sector is not an input to other producing sectors but satisfies final demand only. If the nonstandard material were an input to other producing sectors, the bottom row of D would not be 0 and the approach presented by Joshi10 would be required.

Price information for the standard and nonstandard materials, pstd and pnon, respectively, can be used to relate k, dnon, and dstd. The units of di, non are sector i output, in dollars, per nonstandard output, also in dollars. Multiplication of di, non by the price of the nonstandard material, pnon, converts the units to sector i output per unit of nonstandard material. Similarly, the units of di, std can be converted using the price pstd; the prices should be in units commensurate with those of the nonstandard material, for example, both in units of $/ton steel. The difference in the sector i input required per unit output is pnondi, non – pstddi, std. The fractional change ki is:

… (24)

Next, eq 24 is solved for di, non:

… (25)

or in matrix form:

… (26)

where the operator = is to denote element-by-element multiplication.

Determination of the Environmental Burden Coefficient enon

To develop enon, eq 13 is used with i = non and with the now larger direct requirements matrix D :

… (27)

[dnon;0] is zero and the bottom row of D is 0, the expression simplifies to terms involving the original matrix D, and column vectors e and dnon:

… (28)

Then, substituting the expression for dnon (eq 26) and pulling out the price ratio:

… (29)

This can further be broken down into two power series:

… (30)

With reference to eq 13 with i = std, the first power series above can be replaced by estd – estd:

… (31)

The remaining power series can be evaluated with model runs conducted only for those sectors represented by the set G for which ki is not zero:

… (32)

Copyright Air and Waste Management Association Jan 2008

(c) 2008 Journal of the Air & Waste Management Association. Provided by ProQuest Information and Learning. All rights Reserved.