Environ. Sci. Technol.

2006, 40, 1672-1680

Economic Analysis of Electronic Waste Recycling: Modeling the Cost and Revenue of a Materials Recovery Facility in California
HAI-YONG KANG* AND JULIE M. SCHOENUNG Department of Chemical Engineering and Materials Science, 3118 Bainer Hall, One Shields Avenue, University of California, Davis, California 95616

FIGURE 1. Process sequence at a representative materials recovery facility (MRF) for e-waste recycling. In fact, in the United States, electronic waste is the largest source of heavy metals in municipal solid waste (2). Many countries are now implementing or proposing legislation that would reduce the use of toxic materials in electronic products (3). For example, the European Union (EU) is in the process of implementing the Restriction of the Use of Certain Hazardous Substances (RoHS) Directive, which bans certain substances from future use in electrical and electronic equipment (4). Also, in the United States, some states have passed laws to encourage recycling of obsolete electronic products. Several years ago, the states of Massachusetts and California banned the disposal of CRT monitors in landfills (5, 6). Recently, the state of Minnesota has also proposed a bill to ban CRT-containing electronics from being disposed of in landfills (7, 8). These pieces of legislation seek to direct CRT glass, which contains lead (Pb), to proper recycling. Many other states are currently developing regulations for computer equipment disposal and recycling (7, 8). The state of California passed the California Electronic Waste Recycling Act (CEWRA) in 2003. According to this act, manufacturers and consumers, not local government, are responsible for the costs associated with recycling and disposal of e-waste. All manufacturers that sell their products in California must provide a proper and effective recycling system for recovered electronic devices. They must also submit a report that describes their efforts to reduce the use of hazardous materials and the estimated amount of recycled materials contained in their electronic devices (9). Also, this act imposes an extra charge to the consumer when a new LCD or CRT monitor is purchased. Used electronic devices also contain high-value materials such as gold, palladium, copper, and plastics. For example, 1 ton of electronic scrap from PCs contains more gold than that recovered from 17 tons of gold ore (2). Furthermore, several obsolete but functional electronic devices can be reused, if identified and sorted out by experts. For instance, some collected obsolete electronic devices still operate without any mechanical or functional problems. These resale or refurbished electronic devices are one of the major revenue sources for e-waste recyclers (10, 11). To cope with obsolete electronic equipment, several treatment options are used. Among these options, recycling and reuse of end-of-life (EOL) electronic products is the most feasible option to date (12). But electronic recycling has a short history, so there is not yet a broad and fixed infrastructure in place, and the size of the recycling companies are small. Most have fewer than 20 employees, which is small compared to other industries (13). In the electronic waste recycling industry, the materials recovery facility (MRF) plays a key role, both as part of private companies (manufacturers) and through municipalities. Figure 1 shows a process flow diagram for a representative e-waste recycling MRF. E-waste
10.1021/es0503783 CCC: $33.50 2006 American Chemical Society Published on Web 01/24/2006

The objectives of this study are to identify the various techniques used for treating electronic waste (e-waste) at material recovery facilities (MRFs) in the state of California and to investigate the costs and revenue drivers for these techniques. The economics of a representative e-waste MRF are evaluated by using technical cost modeling (TCM). MRFs are a critical element in the infrastructure being developed within the e-waste recycling industry. At an MRF, collected e-waste can become marketable output products including resalable systems/components and recyclable materials such as plastics, metals, and glass. TCM has two main constituents, inputs and outputs. Inputs are process-related and economic variables, which are directly specified in each model. Inputs can be divided into two parts: inputs for cost estimation and for revenue estimation. Outputs are the results of modeling and consist of costs and revenues, distributed by unit operation, cost element, and revenue source. The results of the present analysis indicate that the largest cost driver for the operation of the defined California e-waste MRF is the materials cost (37% of total cost), which includes the cost to outsource the recycling of the cathode ray tubes (CRTs) ($0.33/kg); the second largest cost driver is labor cost (28% of total cost without accounting for overhead). The other cost drivers are transportation, building, and equipment costs. The most costly unit operation is cathode ray tube glass recycling, and the next are sorting, collecting, and dismantling. The largest revenue source is the fee charged to the customer; metal recovery is the second largest revenue source.

Introduction
The development of the electronics industry has made our everyday life easier and more convenient. However, because the useful lifespan of consumer electronic devices is decreasing as a result of rapid changes in equipment features and capabilities, the waste stream created by these obsolete electronic devices is growing rapidly. The primary issue associated with these large amounts of obsolete electronic devices is the adverse effects they can have on human health and the environment, because various toxic materials such as lead, cadmium, and mercury are contained in e-waste. Personal computers (PCs), for instance, contain eight toxic metals that are categorized as hazardous by the Resource Conservation and Recovery Act (RCRA) (1).
1672
9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 40, NO. 5, 2006

recycling starts with the collection of e-waste and its subsequent transport to the MRF. Usually, for this transportation need, MRFs use either their own transportation equipment or rental transportation equipment available from local municipalities (i.e., waste management companies). These steps are then followed by a sorting process through which resale systems/components are isolated and nonfunctioning equipment is subsequently dismantled. The remaining materials are then shredded and separated to facilitate material recovery, as described below. The most common e-waste product is a personal computer system, which consists of a monitor (typically a CRT for waste units) and a central processing unit (CPU) (13). Thus, for the present case study, PCs have been chosen as the target product to calculate the cost and revenue for a representative e-waste MRF. Resalable equipment such as clean and functional monitors and CPUs can be sold to regional repair shops to generate revenue. The other equipment is then dismantled. After size reduction, some of the recovered materials (i.e., plastics and metals) are further separated; those with value are then sold to scrap dealers by auction. The CRT glass, however, is transported to a CRT glass recycling company for a specified recycling fee, which includes the transportation cost. The recovered output of the MRF process can be categorized into three parts: metals, plastics, and glass. Lead, copper, zinc, steel, and aluminum are representative metals used in electronics that can be separated and recovered for use in secondary markets (12). For plastics, high-impact polystyrene (HIPS), acrylonitrile butadiene styrene (ABS), polyphenylene oxide (PPO), and polycarbonate/acrylonitrile butadiene styrene (PC/ABS) are widely used in the electronic industry (14). Rios et al. (15) described an engineering approach for the effective separation of plastics at an e-waste MRF, and Basdere and Seliger (16) evaluated the feasibility of automation for end-of-life electronics disassembly. Much of the metal and plastic recovery is ultimately handled by organizations outside the MRF. The specific capabilities of a given MRF are unique. There are several MRFs for obsolete electronic products that are operated by original equipment manufacturers (OEMs), local organizations (nonprofit organizations, local government agencies), and private recyclers. Approximately 500 e-waste recycling companies are currently operating in the United States (13), and they are concentrated in the Midwest, New England, and the West; each region accounts for approximately a quarter of the recycling companies (12). Most often, an e-waste MRF is a small to mid-size recycling facility that collects obsolete electronic products from households and local communities through permanent collection programs. As a result, the electronic products that are collected are not homogeneous and, over time, exhibit large variations in size, physical/functional condition, and type of electronic product, especially in comparison to obsolete electronics collected from manufacturers and industrial users (13). Generally, homogeneous e-waste from large business computer users is more valuable than less homogeneous e-waste from households (17). These factors will influence the costs and revenues associated with a specific e-waste MRF. Several studies have been conducted to estimate the costs associated with the recycling of e-waste. For instance, Macauley and Palmer (18), and Boyce et al. (19) evaluated CRT monitor recycling. Lu and Stuart (20) evaluated the cost for industrial and residential returns via special collection events. Boon et al. (21) and Jung and Bartel (22) also considered the costs associated with an e-waste special collection event. Rios et al. (15) analyzed the difference in cost for bulk and separated plastic recycling. These studies are limited in scope, and, moreover, were conducted prior

FIGURE 2. Flow of costs and revenues in an MRF operation for e-waste. to state mandates banning landfill disposal and incineration of CRT monitors.

Scope
For the present study, we developed an input/output model to evaluate the economic drivers associated with a representative e-waste MRF operated in the state of California. Costs and revenues for each unit operation and for each item that causes cost or revenue, such as materials, labor, and system/component resale, were quantified. By estimating the costs and revenues for each unit operation, the model calculates the total processing costs and revenues in operating an MRF that handles e-waste. This methodology allows us to identify critical cost and revenue drivers. Sensitivity analysis was used to evaluate the sensitivity of the modeling results to key input values. Sensitivity of costs and revenue to variations in the total treatment amount and the relative amount of CRT monitors was evaluated. This study is focused on the situation in the state of California. The state of California has banned the disposal of CRT monitors to both landfills and incineration. Therefore, all CRTs must be recycled. This fact creates a unique feature in the California e-waste recycling industry. The MRF treatment ratio for CRTs and CPUs is 3:1 (23). For a previous study, which was not focused on California, the ratio was determined to be 2:1 (24). In addition, we assume the size of the e-waste MRF to be that which is typical today (20 employees, 2500-5000 tons/year treatment volume) (13, 24) for the treatment of heterogeneous waste collected at a permanent residential collection site. At this MRF, we assume the MRF is responsible for collecting the waste and transporting it to the MRF. We also assume that the steel, aluminum, copper, and zinc scrap are physically separated at the facility via magnetic and eddy current separation, respectively, before being sold to secondary suppliers; other metals are sold for further separation and refinement. No chemical extraction occurs at the MRF. For the plastics recovery, we assume that the large pieces, such as the ABS monitor housing, are separated by hand. Further plastics separation is conducted at specialty plastics recyclers, to whom the plastics scrap is sold.

Methodology
To set up our economic model for the e-waste MRF defined above, technical cost modeling (TCM) was utilized (25, 26). A TCM was used to estimate and analyze the costs and revenues associated with the MRF processes. Figure 2 shows the cost and revenue flows in an MRF operation. Costs, which consist of labor, building, equipment, energy, materials, and transportation, are shown as arrows out from the MRF box. The arrows pointing inward to the MRF represent revenue sources for the MRF. Revenues for an e-waste MRF consist of resale systems/components, materials recovery, and fees from customers.
VOL. 40, NO. 5, 2006 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9

1673

The model has two main constituents, inputs and outputs. Inputs are variables that are directly specified in the model. Outputs are the results of the modeling and consist of cost and revenue estimates. First, the basic unit operations are reviewed and process options within these steps are identified: collection (which includes transportation), sorting, testing, dismantling, size reduction, separation (which includes magnetic and eddy current separation), landfill, CRT recycling (which is outsourced), and sales are the unit operations considered for the present analysis. Cost elements are also defined: labor, energy, materials, transportation, equipment, and building. Second, for the selected unit operations, baseline costs are calculated and the key process parameters are evaluated. In this step we rely on published sources and on data from individuals and private companies. Third, with these process parameters, we establish the functional relationships that can be used to calculate the costs. Fourth, by adding up the key process-derived costs, we can get a cost for each unit operation. Finally, the total MRF operation cost is calculated by summing all the unit operation costs. Revenues for each unit operation and total MRF operating revenue can be estimated with the same logic as that used for the cost estimation. The revenue sources considered are fees charged to customers, materials recovery (metals and plastics), and resale of systems/components. Representative equations used for cost and revenue estimation are shown below.

TABLE 1. Representative General Inputs Used in MRF Cost Modelinga

price of electricity (industry sector) ($/kWh) operating time (days/year) labor wage ($/h) price for building space ($/m2month) treatment volume (tons/year) CRT treatment amount (wt %) CPU and others treatment amount (wt %)

N(0.10, 0.02)b 240 N(9.0, 1) 3.55 2500 75 25

a Specific to California; see refs 23 and 27-29. b N(, ) indicates a normal distribution with mean and standard deviation .

TABLE 2. Representative Materials Composition Assumed for the Central Processing Units and Cathode Ray Tube Monitorsa
materialb lead copper steel aluminum zinc precious metals other materials HIPS ABS PPO glass
a

CPUs (wt %) 10.8 5.6 26.9 23.9 3.7 0.05 3 1.3 15.2 9.5

CRT monitors (wt %) 6.1 9.1

2.6 18 64.1

See refs 19 and 31-35. b HIPS, high-impact polystyrene; ABS, acrylonitrile butadiene styrene; PPO, polyphenylene oxide.

unit operation cost ) materials + energy + labor + transportation + equipment + building (1) total cost )

(cost at each unit operation)

(2)

TABLE 3. Representative Process Related Input Variables Used in Each Unit Operationa
MRF equipment cost recovery life (years) resale price for CPU (Pentium III) ($/unit) resale price for CRT monitor (17 in.) ($/unit) fee charged to customer for CRT monitor (17 in.) ($/unit) rebate for working computer ($/unit) resale price for board ($/kg) landfill tipping fee ($/kg) 7 N(32.5, 7.5)b N(30, 5) 10

revenue ) resale systems/components + recovered materials + fees to customers (3)

The materials cost includes rebates to the customer, waste disposal (landfill) fees, and CRT glass recycling costs. It should be noted that, from the perspective of the MRF, the CRT glass recycling cost can best be considered as simply a materials cost because it is a service provided by another company, even though the actual costs from the perspective of this other company would include transportation, labor, utilities, equipment, etc. The cost of energy consumption at the MRF is calculated on the basis of the information on energy costs and energy consumption per unit operation. The cost of labor is a function of the wages and the number of laborers associated with the unit operation. The transportation cost is estimated on the basis of the information on transportation distance to the MRF and to the landfill site. The equipment costs are determined by the number of machines required and the cost per machine. The building cost includes cost for building and land. Overhead costs are not considered.

N(3.5, 1.5) N(1, 0.5) N(0.1, 0.02)

a Specific to California; see refs 23, 36, and 37. b N(, ) indicates a normal distribution with mean and standard deviation .

systems/components resale values, plastics recovery values, and metals recovery values. The general input values for MRF cost modeling are listed in Table 1. These general input parameters are common to all of the unit operations but are specific to the state of California. Additional input data used to determine the costs and revenues are provided in Tables 2 and 3. Table 2 shows the typical materials composition assumed for the CPUs and CRT monitors. Table 3 shows the input variables used in each unit operation. Metals represent more than 73 wt % of the total CPU composition, and glass is the largest component of CRT monitors, representing more than 61 wt %. Recovered metals and plastics are sold by auction. The values of these fluctuate and are dependent on market conditions and recovered material quality (30). Systems and components can be sold only if they are clean and functional. Furthermore, at present, the main process chip for the CPU must be at least Pentium III clock speed, the random access memory (RAM) should be at least 128 Mb, and the recommended hard drive capacity must be larger than 8 Gb. Also, for CRTs, the display size should be 17 inch or larger and the plastic housing should show no signs of age. These are the current requirements. Clearly, these will vary with time, as secondary markets change.

Input Parameters
The role of an MRF is to take collected e-waste and, after several processes, resell the systems/components, sell the materials recovered from collected e-waste, and pay a fee for disposal of the remaining waste (see Figure 1). The input parameters for the e-waste MRF TCM are categorized into two parts: inputs for cost calculation and inputs for revenue calculation. The main input parameters needed to calculate costs for each unit operation include equipment costs, labor costs, material costs, and transportation costs. Input parameters for revenue estimation include the fees charged to customers,
1674
9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 40, NO. 5, 2006

FIGURE 3. Annual operating cost, distributed by cost element for an e-waste MRF. CRT 75 wt %; CPU 25 wt %; treatment amount 2500 tons/year.

FIGURE 4. Annual operating costs, distributed by unit operation, for an e-waste MRF. CRT 75 wt %; CPU 25 wt %; total treatment 2500 tons/year.

Results and Discussion

Operating Cost Analysis. Figure 3 shows the annual operating cost, distributed by cost element for an e-waste MRF in California. The largest cost driver is materials cost (which includes the cost to send CRT glass out for recycling, the customer rebate, and the landfill tipping fee for residual waste), at approximately 37% of the total operating cost. The second largest cost driver is labor cost, at approximately 28% of the total operating cost. It should be noted that this does not include overhead labor, which, if included, would result in labor costs being more significant. As seen from Figure 3, these two cost drivers are most critical and, combined, represent almost 65% of the total cost. Within the materials cost, the cost of sending the CRT glass out for recycling represents more than 80% of the total materials cost. In the state of California, collected CRT glass must be recycled (9). To fulfill this regulatory requirement, collected CRT glass is transported to glass smelters to be recycled, but these smelters are not located on the west coast of the United States (12). Thus, it is to be expected that the cost to recycle CRTs, which includes the cost of cross-country transportation to the glass recycler, will be high. The U.S. Environmental Protection Agency (EPA) reported that the general transportation cost for CRTs is approximately $2.253.41/mile for a full truckload in a 23-ton trailer, depending on transportation conditions (38). During MRF processing of e-waste, many unit operations such as sorting, testing, dismantling, and sales depend on

labor-intensive processes, which results in high labor cost. Transportation cost represents the costs for transportation of collected obsolete equipment to the MRF and the costs for transportation of residues to landfill sites after MRF processing. Energy cost is the smallest cost driver in the e-waste MRF. Figure 4 shows the distribution of annual operating cost by unit operation. CRT recycling cost represents the largest portion, approximately 30%, and then sorting, collection, and dismantling costs represent more than 10% each in the MRF operation. It is important to recall that this analysis is for the situation in the state of California, where CRTs must be recycled. If the same study is conducted for other states in which CRT glass recycling is optional, the percentage of the cost represented by the CRT glass recycling cost will be decreased. In the present case study, the cost to recycle CRTs has been calculated at $0.33/kg. But other studies, which were conducted outside California, show this cost to be $0.20-0.22/kg (39, 40). Other studies indicate that the most costly step in e-waste recycling is often collection (13, 41, 42). But in this study the most critical cost driver is the cost to have CRT glass recycled; secondary cost drivers include sorting, collection, dismantling, and separation. There are several reasons for these discrepancies. First, the former studies do not consider the entire process for e-waste recycling. They consider only special drop-off collection programs. Also, they do not model a conventional e-waste MRF operation, because they only
VOL. 40, NO. 5, 2006 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9

1675

Figure 6 shows the distribution in annual revenue by revenue source. The largest revenue source is the fee charged to customers for recycling, which represents approximately 60% of total revenue. This is consistent with the results of a survey conducted by the International Association of Electronics Recyclers (24). The second largest revenue source comes from recovered metals. Resale of systems/components generates the least revenue. Most of the equipment collected at an e-waste MRF is either nonfunctional or outdated. Only a limited amount (<2 wt %) of the machines collected are working properly, and only this equipment can be sold to regional repair shops. However, these resale items can yield more revenue than if they were demanufactured. It is of value to analyze the relationship between the revenue and the volume of treated material because this information can provide guidance on the economic viability of e-waste MRFs. Thus, we introduce the concept of a profitefficiency ratio (PER) to be used as the basis for describing this relationship: FIGURE 5. Distribution by cost element within each unit operation. CRT 75 wt %; CPU 25 wt %; total treatment 2500 tons/year. considered special drop-off events. At present, the most common e-waste collection program is not special drop-off events but permanent collection methods (13). Furthermore, these studies did not consider the costs of equipment for the collection, size reduction, and separation processes. Finally, these other studies were not focused on the state of California and, therefore, assumed it was legal to dispose of CRT glass in landfills or incinerators. Figure 5 shows the distribution by cost element within each unit operation. Labor cost is the largest portion in sorting, dismantling, and sales. Note that, in the dismantling unit operation, the labor cost represents approximately 90% of the cost. This result is an indication of the extent of manual labor needed within an e-waste MRF. Transportation cost represents most of the cost in the collection and landfilling steps. For all unit operations, the equipment cost recovery life is assumed to be 7 years (36). For the MRF, equipment costs are significant for the size reduction and separation steps. Materials costs contribute only to glass recycling (100%), customer rebate (100%), and landfill (18%). For the landfill step, the materials costs represent the cost to landfill the residual materials such as wood and laminated metal parts that cannot be recovered. Revenue Analysis. The resale of systems/components, recovered materials (plastics and metals), and fees charged to customers for disposal are the three main revenue sources for e-waste MRFs. The revenues from each of these sources depends on a number of variables such as secondary market value for resale systems/components, the efficiency of demanufacturing logistics, the working efficiency of employees, and the quality and/or quantity of the equipment that is collected.

profit-efficiency ratio (PER) ) percent of revenue/percent of treatment amount (4)

Table 4 shows the summary of the profit-efficiency ratios for each revenue source and each revenue element (which are subsets of the revenue sources), as calculated from eq 4. The CRT recycling fee charged to consumers generates the most revenue; however, it has a low profit-efficiency ratio of 0.8 (59.9% 75.3% ) 0.8). In contrast, the profit-efficiency ratio for resale systems/components is highest, with a value of 2.35. But this revenue source represents only 5% of the total revenue, because its absolute treatment amount is small. The number of systems/components that can be resold depends on the age and mechanical/functional condition of the collected equipment. Generally, the resale price of CPUs (Pentium III) and monitors (at least 17 in.) is more than $25/unit (23, 43) and depends on the functional conditions of the CPU and the actual size of the monitors. This is one of the most profitable items for an e-waste MRF in terms of profit-efficiency ratio. Also, resale is a preferable end-of-life strategy. Metals recovery has the second highest profit-efficiency ratio, 0.9, as it represents 31.1 wt % of the total treatment amount and 27.9% of the total revenue. Among the revenue elements for this revenue source, precious metals have the highest profitefficiency ratio, 41.1, but because the absolute amount of these metals is small, this revenue element represents less than 0.5% of total revenue. Plastic recovery shows the lowest profit-efficiency ratio at 0.4. Figure 7 provides a comparison of the distribution in the relative amount of the different metals recovered from e-waste with the distribution in revenue generated by these metals. Printed wiring boards (PWBs) contain precious metals, which have high economic value. In general, onethird of the precious metal found in e-waste is gold (1).

FIGURE 6. Distribution in annual revenue by revenue source. CRT 75 wt %; CPU 25 wt %; total treatment 2500 tons/year.
1676
9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 40, NO. 5, 2006

TABLE 4. Calculated Profit-Efficiency Ratios for Each Revenue Source and Revenue Element in the Defined E-waste MRFa
revenue source fees charged to customer resale components/systems PER by source 0.8 2.3 revenue elementb CRT monitor computer breakage CPU CRT monitor board HIPS ABS PPO PC/ABS Pb Cu Al steel (Fe) Zn precious metals (Au, Ag, Pt) PER by element 0.8 0.1 1.4 2.5 2.4 0.4 0.3 1.1 1.2 0.4 2.1 1.5 0.07 1.1 41.1

plastics recovery

0.4

metals recovery

0.9

a CRT 75 wt %; CPU 25 wt %; total treatment 2500 tons/year. b HIPS, high-impact polystyrene; ABS, acrylonitrile butadiene styrene; PPO, polyphenylene oxide; PC/ABS, polycarbonate/acrylonitrile butadiene styrene.

FIGURE 7. Distribution in metals recovered from e-waste and in the revenues they generate. CRT 75 wt %; CPU 25 wt %; total treatment 2500 tons/year. Precious metals can be reclaimed profitably through precious metals recovery processes (12). Recovered precious metals from electronic scrap is one of the most profitable aspects of the recycling industry, with gold valued at more than $44/kg, even though the absolute quantity is very small. Other metals such as aluminum and copper, with secondary market values of more than $1/kg (18), also yield high profit-efficiency ratios, but in the case of steel, the profit-efficiency ratio is only 0.07; that is, the treatment amount of steel is 14.5 wt % but revenue is only 1.1%. Likewise, Figure 8 provides a comparison of the distribution in the relative amount of the different plastics recovered from e-waste with the distribution in revenue generated by these plastics. Recovered plastics show the lowest profitefficiency ratio for an e-waste MRF. These results can be determined from the selling price per kilogram for each item. Generally, market prices for secondary recycled plastics are lower than $1/kg. The largest portion of the plastics consists of HIPS and ABS, the secondary market prices for which are less than $0.70/kg (44). Different from metal recovery, plastic recovery is more complicated (10). There are many types of plastics used in the electronic industry compared to metals, and recovery processes for plastics are still under development. Also, markets for recovered plastics are not mature. The market price of the materials recovered from e-waste is generally dependent on factors beyond the control of the

FIGURE 8. Distribution in plastics recovered from e-waste and in the revenues they generate. CRT 75 wt %; CPU 25 wt %; total treatment 2500 tons/year. MRF, such as the price of virgin materials, the demand for recovered materials, and the economic situation. But, to maximize profit, MRFs need to understand the variables that affect the yield of recovered materials, such as which collection method is best. In general, e-waste collected from commercial or large organizations provides more economic value than e-waste collected from residential areas. Also, within the materials recovered, the highest values to recyclers are for those that have high precious metal content in their circuitry. The most profitable items for MRFs are complete and working systems. Sensitivity Analysis. The purpose of the sensitivity analysis is to demonstrate the sensitivity of the model output to uncertainty in the values of the model input data. This analysis provides for the investigation of the effects of variations in one or more inputs. Typically, one or two variables are systematically changed. Other key inputs are stated with the figure. Figures 9 and 10 show the changes in cost and revenue (percent basis), respectively, when the weight percent of CRTs relative to total e-waste is changed and when the total treatment amount is changed. Figure 9A indicates the changes in cost for each unit operation when CRT monitor treatment amounts are changed. Except for the fraction represented by the CRT glass recycling cost, all the other unit operation costs have a tendency to decrease slightly in percent contribution with increasing CRT monitor treatment amount. Only the fraction of the cost for CRT glass recycling increases,
VOL. 40, NO. 5, 2006 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9

1677

FIGURE 9. Sensitivity of cost to changes in total treatment amount and CRT monitor treatment amount. (A) Cost changes by unit operation. Total treatment amount is 2500 tons/year. (B) Fixed and variable cost change. CRT 75 wt %; CPU 25 wt %. glass recycling cost. Thus, more CRTs result in more fees. Also, as the amount of CRT monitor treatment is increased, the revenue from metal recovery is decreased. Metals are mainly recovered from CPUs (see Table 2). As the amount of CRTs increases, by definition, the amount of CPUs decreases. Furthermore, as the amount of metals recovered decreases, revenue from recovered metals also decreases. The other two revenue sources, resale systems/components and plastics recycling, show only a slight change with CRT monitor treatment amount. Figure 11 shows the change in both normalized cost and normalized revenue when the total e-waste treatment amounts are changed. The costs and revenues are normalized to the baseline values determined for a total e-waste treatment amount of 2500 tons/year. Figure 11A shows that the normalized annual operating cost (cost per year) exhibits different sensitivity behavior when compared to the cost on a per-unit-mass basis (cost per kilogram). When the total treatment amount is less than 2500 tons/year, normalized cost per unit mass (cost per kilogram) increases sharply and has a value of 3.6 when total treatment amount is 500 tons/ year. The cost per unit mass shows a typical pattern for economies of scale; that is, the costs are distributed over more units, and hence the total cost per unit decreases (45). The normalized annual operating cost (cost per year) increases linearly with increases in total treatment amount. The strong effect of total treatment amount change on annual revenues (revenue per year) is shown in Figure 11B. However, normalized revenue per unit mass (revenue per kilogram) is not affected by total treatment amount, because revenue depends on the sales price of recovered materials, which is held constant for the analysis in Figure 11B. The results in Figure 11B demonstrate that changes in total treatment amount strongly affect the total annual revenue but do not affect the revenue per unit mass. As total treatment amount increases, both normalized annual operating cost (cost per year) and normalized annual revenue (revenue per year) increase, but the extent of increase in normalized revenue is greater than that for the normalized cost. A technical cost model was developed to evaluate the costs and revenues of an e-waste MRF operating in the state of California and to determine the key cost and revenue drivers. Cost/revenue drivers were identified by cost/revenue and by unit operation. For the case study evaluated, the cost elements of materials and labor are the primary cost drivers. The cost drivers by unit operation showed that the cost to have CRT glass recycled is the primary cost driver. Dismantling, collection, sorting, and separation are secondary cost drivers. In the e-waste MRF, most of the costs are variable costs.

FIGURE 10. Sensitivity of revenue to changes in CRT monitor treatment amount. Total treatment amount is 2500 tons/year. which is a consequence of the material composition of the monitors. Monitors are composed largely of glass (see Table 2), and this fact causes higher cost with increasing CRT monitor treatment amount. Regardless of the actual CRT monitor treatment amount, for amounts between 60% and 90%, the cost of the CRT glass recycling is the primary cost driver. Figure 9B shows the changes in the distribution between fixed and variable costs according to changes in the total e-waste treatment amounts. From Figure 9B it is seen that the fixed cost fraction of the total cost decreases with an increase in the total treatment amount, but the variable cost fraction increases with total treatment amount. When total treatment amount increases, the CRT treatment amount increases proportionally. This increased CRT treatment amount increases the variable cost portion, which includes the CRT glass recycling cost. Furthermore, because CRT glass recycling cost is the predominant cost driver in both the materials cost and the entire MRF operating cost, that is, almost 30% in total cost (see Figures 3 and 4), an increase in the CRT treatment amount causes an increase in the CRT recycling cost and, consequently, the variable cost fraction of the total cost. However, the absolute value of the fixed cost does not depend on changes in the total treatment amount, because the fixed cost is a one-time investment. As a result, the relative fraction of the fixed cost to the total cost decreases with an increase in the amount of total treatment volume. Figure 10 shows the changes in revenue for each revenue source when the CRT monitor treatment amount is changed. Revenue from fees charged to customer increases, but revenue from metals recycling decreases with increasing CRT monitor treatment amount. This phenomenon is expected when we consider the composition of CRT monitors (see Table 2). More than 60 wt % of the monitor is glass; other materials are copper (9.1 wt %), steel (11.4 wt %), and ABS (18.2 wt %). Even though there is a cost for CRT glass recycling ($0.33/kg), the fees charged to customer to accept CRTs at an e-waste MRF ($10/unit, average) are greater than the CRT
1678
9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 40, NO. 5, 2006

FIGURE 11. Normalized change in cost and revenue when total e-waste treatment amount is changed. (CRT monitor 75 wt %; CPU 25 wt %). (A) Normalized cost per year and per unit mass. (B) Normalized revenue per year and per unit mass. For the revenue drivers, the fee charged to customers is the primary revenue source, and metals recovery ranks second. The contribution of the other two revenue sources, resale of systems/components and plastics recovery, is very small. To maximize the revenue and ensure the robust economic viability of an e-waste MRF, the quality of the collected equipment must be high and the equipment collected must contain materials that have high market value. Revenue sources with higher profit-efficiency ratio (PER) are an example. Furthermore, customers need to be willing to pay the fees charged by the MRF. These fees are the result of legislation, as is the ban on landfill disposal. It is beyond the scope of the present study to investigate the effectiveness of this legislation, but the incentives for compliance (or penalties for noncompliance) will clearly impact the effectiveness, and ultimately the economic viability of e-waste MRFs. To decrease the labor cost, which is a primary cost driver, it is recommended that, for unit operations where labor cost is high, automated processes be considered. However, automation of processes is currently a challenge because of the variations in product design and the randomness factor in the type and size of products collected. The quality and quantity of the collected equipment is also dependent on consumer support and education. One might also think that labor, as well as energy, cost could be reduced by moving the MRFs outside of the state of California to locations with lower wages and utility costs. It is not clear, however, that the mandates requiring recycling are enforceable outside of the state. This issue merits further study. Tonglet et al. (46) reported that consumer attitudes are a major determinant for the recycling behavior of consumers, and these attitudes are influenced by having appropriate opportunities, facilities, and knowledge to recycle. Other researches have shown that recycling behaviors depend on age, gender, income, education, and ideology of consumers (10, 47, 48). Research done by the U.S. EPA shows that about 80% of consumers are willing to pay a fee of less than $5 for the recycling of obsolete consumer electronic devices (10). The response of consumers to landfill bans is complex and not fully understood. Moreover, it can be very difficult to forecast actual consumer behavior. In addition, there are numerous issues that lead to market uncertainty relative to the future economic sustainability of the electronic recycling industry. These factors include potential changes in legislation/regulations, recycling technology, markets for recovered materials and/or components, and the steady supply of obsolete equipment from customers (13, 24). In-depth analysis of this market uncertainty and consumer behavior, although beyond the scope of this paper, merits further investigation. Because of the seriousness of e-waste, many states have started to regulate e-waste. These regulations are another factor that will affect the e-waste recycling industry. Compliance with these regulations is essential to the survival of e-waste MRFs. The state of California passed CEWRA in 2003. This act will have an additional impact on the e-waste recycling industry. After this act is fully implemented, further study will be needed.

Acknowledgments
We thank Paul Y. Gao of California Electronic Asset Recovery in Sacramento, CA, and Sarah Kittle of the Division of Integrated Waste Management in Yolo County, CA, for valuable insight and technical discussions.

Literature Cited
(1) U.S. Geological Survey. Obsolete computers; gold mine or hightech trash? Resource recovery from recycling. USGS Fact Sheet FS-060-01, 2001. (2) The Basel Action Network and Silicon Valley Toxic Coalition, 2002. Exporting harm: the high-tech trashing of Asia. http:// www.svtc.org/cleancc/pubs/technotrash.htm (accessed August 2005). (3) Schoenung, J. M.; Ogunseitan, O. A.; Saphores, J. D. M.; Shapiro, A. A. Adopting lead-free electronics: policy differences and knowledge gaps. J. Ind. Ecol. 2005, 8 (4), 59-85. (4) European Council. Directive 2002/95/EC of the European Parliament and of the Council of 27 January 2003 on the restriction of the use of certain hazardous substances in electrical and electronic equipment. Off. J. Eur. Union 2003, L 37/19-L 37/23. (5) California Department of Toxic Substance Control, 2001. Managing waste cathode ray tubes: fact sheet. http:// www.dtsc.ca.gov/HazardousWaste/CRTs/HWM_FS_CRTEmergencyRegs.pdf (accessed August 2005). (6) U.S. Environmental Protection Agency. Hazardous waste. http:// www.epa.epaoswer/hazwaste/recycle/recycling/rules.htm (accessed August 2004). (7) U.S. Environmental Protection Agency. Product stewardship, electronics: state action. http://www.epa.gov/epaoswer/nonhw/reduce/epr/products/ele-state.htm (accessed August 2005).
VOL. 40, NO. 5, 2006 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9

1679

(8) Silicon Valley Toxics Coalition, 2004. Poison PCs and toxic TVs. http://www.svtc.org/cleancc/pubs/poisonpc.htm (accessed August 2005). (9) California Electronics Waste Recycling Act, 2003. (10) Electronic reuse and recycling infrastructure development in Massachusetts; EPA-901-R-00-002; U.S. Environmental Protection Agency: Washington, DC, 2000. (11) Dunn, B. Logging into e-cycling: will cyber-trash ever compute? Am. Met. Market. 2003, 111 (19), 3-4. (12) Kang, H.-Y.; Schoenung, J. M. Electronic waste recycling: a review of U.S. infrastructure and technology options. Resour. Conserv. Recycl. 2005, 45 (4), 368-400. (13) International Association of Electronics Recyclers. IAER electronics recycling industry report, 2003. (14) American Plastics Council. Plastics from residential electronics recycling, 2000. (15) Rios, P.; Stuart, J. A.; Grant, E. Plastics disassembly versus bulk recycling: engineering design for end-of-life electronics resource recovery. Environ. Sci. Technol. 2003, 37, 5463-5470. (16) Basdere, B.; Seliger, G. Disassembly factories for electrical and electronic products to recover resources in product and material cycles. Environ. Sci. Technol. 2003, 37, 5354-5362. (17) Grenchus, E.; Keene, R.; Nobs, C. Comparison and value of returned consumer and industrial information technology equipment. Proc. IEEE Int. Symp. Electron. Environ. 2000, 324329. (18) Macauley, M.; Palmer, K. Dealing with electronic waste: modeling the costs and environmental benefit of computer monitor disposal. J. Environ. Manage. 2003, 68, 13-22. (19) Boyce, J.; Lichtenvort, K.; Johansson, G. Monitor disassembly and CRT recycling as an example for green cost management system. Proceedings of Going Green CARE Innovation 2002 or http://www.greeen.it (accessed August 2005). (20) Lu, Q.; Stuart, J. A. An investigation of bulk recycling planning for an electronics recycling facility receiving industrial returns versus residential returns. IEEE Trans. Electron. Packag. Manuf. 2003, 26 (4), 320-327. (21) Boon, J. E.; Isaacs, J. A.; Gupta, S. M. Economics of PC recycling. Environmentally Conscious Manufacturing Conference, Boston, MA, 2000. (22) Jung, L. B.; Bartel, T. An industry approach to consumer recycling: the San Jose project. Proc. IEEE Int. Symp. Electron. Environ. 1998, 36-41. (23) California Electronic Asset Recovery Inc. http://www. cearinc.com (accessed August 2005). (24) Northeast Recycling Council. Inc. (NERC). Used electronics market study survey analysis, August 2003. http://www.nerc.org/ adobe/Used_Electronics_Market_study8-22-03.pdf (accessed August 2005). (25) Rosato, D. V.; Rosato, D. V. Technical cost modeling. In Blow modeling handbook; Busch, J. V., Field, F. R., III, Eds.; Hanser Publications: New York, 1989. (26) Schoenung, J. M. Process cost modeling - a summary. In Smart Processing of Materials Symposium; 27th International SAMPE Technical Conference, Albuquerque, NM, 1995. (27) California Energy Commission. http://www.energy.ca.gov/ electricity/current_electricity_rates.html (accessed August 2005). (28) U.S. Department of Labor. Employment Standards Administration Wage and Hour Division. http://www.dol.gov/esa/ minwage/america.htm (accessed August 2005). (29) Jackson Properties. Inc. http://www.jacksonprop.com (accessed August 2005). (30) Scrap Metals Index. Recyclers world online market prices, 2004. http:// www.recycle.net/computer/index.html (accessed September 2004).

(31) Materials for the Future Foundation. The monitor of electronics recycling issue: CRT smelting, Issue 2, 2002. (32) Minnesota Office of Environmental Assistance. Recycling used electronics: report on Minnesotas demonstration project, 2001. (33) Desktop computer display: a life-cycle assessment, Vol. 2; EPA 744-R-01-004b; U.S. Environmental Protection Agency: Washington, DC, 2001. (34) Huisman, J.; Stevels, A. N.; Stobbs, I. Eco-efficiency considerations on the end-of-life of consumer electronics products. IEEE Trans. Electron. Packag. Manuf. 2004, 27 (1), 9-25. (35) Smith, D.; Small, M.; Dodds, R.; Amagai, S.; Strong, T. Computer monitor recycling: a case study. Eng. Sci. Educ. J. 1996, 5 (4), 159-164. (36) Peters, M. S.; Timmerhause, K. D.; West, R. E. Plant Design and Economics for Chemical Engineers, 5th ed.: McGraw-Hill: New York, 2003. (37) Global Futures Foundation and U.S. Environmental Protection Agency. Computers, e-waste, and product stewardship: is California ready for the challenge? Report for USEPA Region IX, 2001. (38) Economic analysis of cathode ray tub management, notice of proposed rulemaking. Office of Solid Waste, U.S. Environmental Protection Agency: Washington, DC, February 2002. (39) University of Massachusetts Technical Report 7: Scrap electronic processing; Office of Waste Management, University of Massachusetts: Boston, MA, 1998. (40) Demanufacturing of Electronic Equipment for Reuse and Recycling (DEER2) Technology and Demonstration Center. Final Information Report Subtask GL-04; Electronic demanufacturing exploration (CRT Glass), 2002. (41) Lonn, S. A.; Stuart, J. A.; Losada, A. How collection method and e-commerce impact product arrival rate to electronic return, reuse, and recycling center. Proc. IEEE Int. Symp. Electron. Environ. 2002, 228-233. (42) Hainault, T.; Smith, D. S.; Cauchi, D. J. Minnesotas multistakeholder approach to managing electronic products at endof-life. Proc. IEEE Int. Symp. Electron. Environ. 2000, 310-317. (43) Goldsmith Group Inc. Equipment resale. http://www. goldsmithgroup.com/recycling_services/equipmentresale.htm (accessed August 2005). (44) Plastics Recycling. Recyclers world online market prices, 2004. http://www.recycle.net/Plastic/index.html (accessed October 2004). (45) Evans, J. R.; Anderson, D. R.; Sweeney, D. J.; Williams, T. A. Applied production and operation management, 2nd ed.; West Publishing Co.: 1987. (46) Tonglet, M.; Phillips, P. S.; Bates, M. P. Determining the drivers for householder pro-environmental behavior: waste minimization compared to recycling. Resour. Conserv. Recycl. 2004, 42, 27-48. (47) Saphores, J.-D. M.; Nixon, H.; Ogunseitan, O. A.; Schoenung, J. M.; Shapiro, A. A. Household willingness to recycling electronic waste. Environ. Behav. 2005 (in press). (48) Darby, L.; Obara, L. Household recycling behavior and attitudes toward the disposal of small electrical and electronic equipment. Resour. Conserv. Recycl. 2005, 44, 17-35.

Received for review February 23, 2005. Revised manuscript received December 16, 2005. Accepted December 20, 2005. ES0503783

1680

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 40, NO. 5, 2006