Abstract
The continued development of manufacturing industries along with increasing greenhouse gas emissions has become a critical concern that forces mankind to reduce global warming. Given the long atmospheric lifetimes of Perfluorocarbons (PFCs), it is especially important to reduce the emissions of PFCs which are commonly found in the semiconductor industry. In this study, we propose four master planning models that incorporate pollution taxes, subsidies, and/or progressive pollution taxes into capacity allocation to examine the problem of anthropogenic PFC emissions. The results show that master planning with subsidies and/or progressive taxes provides more flexibility to a foundry plant than that with flat taxes only. Setting emission limits and considering taxation for master planning is the first step toward the success of an environmental policy.
Keywords: Master planning,Pollution tax,Subsidy,Environmental,policy,Linear programming,Sustainable supply chain.
Received: 19 October 2016 / Revised: 11 November 2016/ Accepted:17 February 2017/ Published: 4 April 2017
Contribution/ Originality
This study proposes four master planning models for the semiconductor foundry plants by using ecological taxation methods. The paper’s primary contribution is finding that the master planning models with subsidies and/or progressive taxes provide more flexibility to a foundry plant than the model with flat taxes only.
1. INTRODUCTION
One of the biggest problems that the world is facing today is that of environmental pollution, which worsens every year and has caused irreparable damage to the earth. In the last 100 years, Earth's average surface temperature increased by about 0.8°C (1.4°F). Two thirds of the increase occurred over just the last three decades. Scientists were certain that it was caused mostly by the increasing concentrations of greenhouse gases produced by human activities such as deforestation and burning of fossil fuels. These phenomena were recognized by the National Academy of Science that consisted of all the major industrialized countries (National Academy of Science (NAS), 2007). To alleviate global warming, the United Nations, the European Union, and many countries have jointly enacted legislation or designed mechanisms such as the Kyoto Protocol (The United Nations Framework Convention on Climate Change (UNFCCC), 1997) and the Copenhagen Accord (UNFCCC, 2009) to curb the total amount of greenhouse gas emissions. Under the Kyoto Protocol, industrialized countries and those in transition to a market economy agreed to limit or reduce the emissions of six greenhouse gases, which are Carbon dioxide (CO2), Methane (CH4), Nitrous oxide (N2O), Hydrofluorocarbons (HFCs), Perfluorocarbons (PFCs), and Sulphur hexafluoride (SF6).
Among the most potent of non-CO2 greenhouse gases are the PFCs that have extraordinarily long atmospheric lifetimes of 10000 to more than 50000 years and high Global Warming Potentials (GWP). GWP is a relative measure of how much heat a greenhouse gas traps in the atmosphere. It is estimated that a molecule of Tetrafluoromethane (CF4), i.e., one of the PFCs, is as effective as 7390 molecules of CO2 for causing global warming (IPCC, 2013, Chap. 8). This means that if the same mass of CF4 and CO2 were introduced into the atmosphere, CF4 would trap 7390 times more heat than CO2 over the next 100 years. PFCs, such as Perfluoroethane (C2F6), Tetrafluoromethane (CF4), Fluoroform (CHF3), Sulphur hexafluoride (SF6), Octafluoropropane (C3F8), Octafluorocyclobutane (C4F8) and Nitrogen trifluoride (NF3), are commonly used in the wafer etching process, tungsten process and cleaning Chemical Vapor Deposition (CVD) tools of the semiconductor industry (see Table 1). Each PFC used in different processes has a different amount emitted to the air. Table 2 shows the GWP values of these PFCs over the next 100 years (Intergovernmental Panel on Climate Change (IPCC), 2013). Except for the natural emissions, CF4 emissions are mainly attributed to the manufacturing processes in the aluminum and semiconductor industries. Also, much of the C2F6 used in the semiconductor manufacturing was released to the atmosphere. By analyzing the stored air, Khalil et al. (2003) found that the emission concentrations of CF4 and C2F6 increased about two and ten times, respectively over the past four decades (from 1960 to 2000). Khalil et al. also found that the increasing trend of CF4 emissions has slowed in recent years, but the reduction in the emission rate in the aluminum industry is partially offset by the increasing use of CF4 in the semiconductor industry. To mitigate environmental damage from anthropogenic emissions, reducing PFC emissions is especially important given the growth of the semiconductor industry recently.
Table-1. PFCs used in semiconductor manufacturing processes
| Manufacturing Process | Gas used in Process | |
| Etching | Metal | CF4, CHF3 |
| Polysilicon | SF6, NF3 | |
| Oxide | CHF3, NF3, C2F6, CF4, C4F8, C3F8 | |
| CVD | Silicon | C2F6, CF4 |
| Silicon Oxide | C3F8 | |
| Silicon Nitride | NF3 | |
| Tungsten | C2F6, NF3 | |
Table-2. GWP values of PFCs used in semiconductor manufacturing processes over the next 100 years (IPCC, 2013)
| Greenhouse Gas | Formula |
GWP (100 yr.) |
| Perfluoroethane | C2F6 |
12,200 |
| Perfluoromethane | CF4 |
7,390 |
| Fluoroform | CHF3 |
11,700 |
| Sulphur hexafluoride | SF6 |
22,800 |
| Perfluoropropane | C3F8 |
8,830 |
| Perfluorocyclobutane | C4F8 |
10,300 |
| Nitrogen trifluoride | NF3 |
17,200 |
If the market is left to operate freely, greenhouse gas emissions will be excessive, because there are insufficient incentives for companies to reduce emissions. Economists thus recommend that the polluters pay on carbon dioxide and other greenhouse gas emissions (e.g., (Pigou, 1920; Baumol, 1972)) while companies are likely to purchase more energy efficient equipment and the academia investigate low energy consumption technologies (Chang and Chang, 2006; Ou Yang et al., 2009). According to Gupta et al. (2006) the equipment cost in a new wafer fab exceeds 75% of the total factory capital costs. Energy efficient equipment and low energy consumption technologies usually cost more for a semiconductor company. On the other hand, sustainable supply chain design is one of the approaches that arises recently to embed economic, environmental, and societal concerns into the supply chain and tries to integrate and optimize firms’ operational decisions to reduce environment damage (Frota et al., 2008). This approach may reduce more greenhouse gas emissions with less cost than adopting energy efficient technologies (Benjaafar et al., 2013).
In recent years, the global green tax landscape is evolving rapidly and becoming more complex, as governments increasingly use taxation as a tool to achieve green policy goals and make firms operate more sustainably. Hundreds of pollution taxes are levied and subsidies are offered around the world (KPMG, 2014). In September 2012, Japanese government introduced a new tax to curb greenhouse gas emissions. France imposed a general tax on polluting activities (Taxe Générale sur les Activités Polluantes or TGAP) on a “pay as you pollute” basis. In the US, there are also a large number of sub-national state-based tax incentives related to pollution control and ecosystem protection. In Taiwan, air pollution fees are imposed according to the emission standard regulated by the government. Encouraging industries to reduce greenhouse gas emissions is widely seen as one of the government’s policy choices to mitigate the climate change. South Africa’s Section 121 Tax Allowance Incentive was designed to encourage the development of energy efficiency improvements. The UK offers a 100 percent first year allowance for specified energy saving plants and machinery. China provides financial subsidies to firms with energy saving evidence through central and provincial government agencies. The design of pollution taxes, subsidies, and economic considerations is central to an environment policy.
In addition, the long manufacturing lead time (around three months) and the high market volatility in the semiconductor industry make it difficult to capture the dynamics of this industry. With large capital investment, equipment idling may lead to high cost of wasted capacity. Hence, advanced planning methods play an important role in this industry. Master planning, which is one of the critical phases in the sustainable supply chain design, is to reserve available-to-promise capacity for customers and aims at the best match of supply and demand. The purpose of this research is to present four master planning models for the semiconductor foundry plants by using ecological taxation methods and evaluate both the economic and environmental performance of the four models in terms of profits, allocation quantities, and emission amounts.
The rest of this paper is organized as follows. In section 2, we will review the literature and models relevant to our study. In section 3, we will describe the modeling background and formulate the sustainable master planning models that restrict greenhouse gases emitted to the environment. In section 4, we present computational results to gain useful insights of the proposed models. Finally, section 5 concludes this paper and suggests topics for future research.
2. LITERATURE REVIEW
In the semiconductor foundry industry, fabrication is activated in response to an actual order, i.e., companies run make-to-order (MTO) operations. As inventory cannot be used to smooth the demand, capacity allocation is vital to the success of a foundry company. Chen et al. (2005) developed a capacity planning system that considered the capacity and capability of equipment for multiple semiconductor manufacturing plants. Ponsignon and Mӧnch (2012) applied heuristic approaches for solving master planning problems by taking demand and capacity constraints into account; Ponsignon and Mӧnch (2014) also devised two heuristic approaches, a genetic algorithm and a rule-based assignment procedure, for master planning in the semiconductor manufacturing and evaluated the performance of them while considering demand and execution uncertainty. Chiang and Hsu (2014) proposed a master planning model within an integrated order fulfillment system for a MTO foundry fab to maximize corporate profit. The above studies, however, concentrated only on capacity allocation and neglected the examination of environmental issues.
Some scholars addressed the impact of adding external control mechanisms such as carbon taxes or a cap-and-trade scheme on sustainable supply chain management practices. The concept of carbon taxes is to impose an explicit cost on each unit of greenhouse gas emissions to give firms an incentive to reduce pollution, while in a cap-and-trade system, emission permits are given out for free initially, but extra permits can be bought from other firms if needed. Avi-Yonah and Uhlmann (2009) argued that carbon taxes are a better solution to global warming than cap-and-trade due to its simplicity and capability to provide an immediate carbon pricing signal for firms producing greenhouse gases. It will also generate a bigger initial hit to the balance sheet and higher incentive to reduce greenhouse gas emissions (Hoeller and Markku, 1991). Oreskes (2011) suggested that a properly valued carbon tax might be a more effective mechanism than cap-and-trade to reduce greenhouse gas emissions, while Stram (2014) argued that the cap and trade system of European Union Emission Trading Scheme has virtually collapsed. A number of European countries, such as Denmark, Finland, the Netherlands, Norway, Sweden, Switzerland, and the UK, have implemented carbon taxes.
A carbon tax is a form of pollution taxes. Tseng and Hung (2014) proposed a supply chain model which showed that the higher the carbon tax level of CO2 emissions, the lower their amount. Pollution taxes are often grouped with two other economic policy instruments: pollution permits and subsidies. Johnson (2010) proposed a de-carbonization strategy for the electricity sector by imposing carbon fees and using the revenue exclusively to subsidize new, low-carbon generation sources. He et al. (2012) compared the effectiveness and efficiency of cap-and-trade and carbon tax policies in a generation expansion planning framework and provided insights on the pros and cons of these policies. Saysel and Hekimoğlu (2013) presented a dynamic simulation model which revealed that policy options such as investment subsidies and carbon taxes could help reduce CO2 emissions in Turkey; Li and Lin (2013) investigated the impacts of eight different carbon or energy tax policies on carbon emissions reduction in China. Absi et al. (2013) introduced carbon emission constraints in multi-sourcing lot-sizing problems, Dissou and Siddiqui (2014) provided a comprehensive analysis of the incidence of carbon taxes on inequality to discuss whether or not the carbon tax is progressive, and Yao et al. (2014) developed an energy-efficient subsidy program to ensure the effectiveness of energy consumption. Setting the right tax levels is critical to the success of an environmental policy. However, setting optimal tax and subsidy levels is not within the scope of this study. To facilitate a meaningful and fair comparison, carbon taxes and subsidies were designed to have an equivalent effectiveness on decision variables in He et al.’s model. We will also apply this equivalent effectiveness concept in our models. To the best of our knowledge, there are no articles in the literature that specifically devise master planning models with ecological sustainable concerns for the semiconductor industry. In this study, we propose four master planning models to address the issue of anthropogenic PFC emissions.
3. MODELS
Semiconductor foundry manufacturing consists of four main processes: wafer fabrication, circuit probing, assembly, and final test. Among the four, the wafer fabrication process is the most critical one with the longest flow time. It is a complex process, including wafer cleaning, oxidation, deposition, lithography, etching, diffusion, ion implantation, metallization, inspection, and measurement (see, e.g., Mӧnch et al. (2013)). It takes around three months to complete the whole fabrication. In this research, we study the wafer fabrication process only and propose four master planning models for allocating fabrication capacity to forecasted demand.
Wafer fabrication capacity is usually planned periodically in terms of technology codes (e.g., Chiang and Hsu (2014)) i.e., specifications of technology’s factory routing, average layer cycle time, throughout, and other manufacturing information, to generate projected production output targets; to streamline production as well as commit customers with exact delivery date and quantity, the production output targets are planned into daily slots (details of capacity planning are outside of the scope of this research), which become one of the two inputs of master planning (also called allocation planning). The other input of master planning is from demand planning (e.g., Chiang and Hsu (2014)). The demand forecast of a foundry company is usually based on regional sales data and requires sales managers to visit customers frequently. Also, when planning several months ahead, customers have difficulties in specifying exact product requirements to their foundry companies (which may include processing details for hundreds of wafer fabrication steps). Thus, customers usually provide only critical technology code information along with their demand forecast. The forecast data is combined and aggregated as monthly demand, which will be reviewed and revised internally by top managers according to marketing insights and strategy decisions; it will then be disaggregated to weekly demand based on historical trends and statistical forecasting, and divided proportionally into daily demand to match up with capacity’s time granularity. Next, capacity can be allocated for maximized profit in master planning, taking into account the environmental concerns, which were not examined in Chiang and Hsu (as mentioned above).
We assume that a semiconductor foundry plant with PFC emissions will be taxed by the government, who has also set the cap or limit of PFC emissions for a foundry plant on a monthly basis (it should not feasible and practical to set PFC emission permits on a daily basis. Thus, in addition to the daily time slot t in master planning, we use the index m to denote the monthly time bucket for the consideration of emission permits, and LMTm to stand for the cap of PFC emissions in month m and TAX to be the pollution tax per unit of PFC emissions.
According to Kilger and Schneeweiss (2005) master planning is to reserve “committed ATP” capacity for customers in the medium term (e.g., from 6th to 17th month), i.e., it is usually performed monthly. We suppose that the customer demand forecast is known and deterministic, and capacity is stringent and difficult to be expanded quickly. Since on-time delivery is often one of the manufacturing objectives for foundry companies (Taiwan Semiconductor Manufacturing Company (TSMC), 2013) we do not consider in master planning the inventory holding and backorder costs, which are usually handled when orders are promised by firms as they arrive (for details of how order promising follows from master planning, please see Chiang and Hsu (2014). We will formulate four different master planning models with linear programming. The first model with flat pollution taxes only is denoted as MPT, the second model with flat taxes and subsidies as MPTS, the third model with flat and progressive pollution taxes as MPTP, and the fourth model with flat and progressive pollution taxes and subsidies as MPTSP.
Model 1: MPT
We consider a finite planning horizon that spans over 12 months. Suppose each demand includes only one customer c. We let Dcfgmt denote the customer c’s demand forecast that can be satisfied with technology code g at factory f in time period t of month m, and QTYcfgmti be the required quantity of order i from customer c satisfied with technology code g at factory f in time period t of month m. Demand is to be served with capacity CAPfgmt that is supplied at factory f with technology code g in time period t of month m. The goal is to find the optimized committed ATP profile “ALPcfgmt” that maximizes forecasted corporate profit (minus pollution taxes). The per wafer forecasted margin Mcfgmt can be obtained by subtracting the per wafer manufacturing cost from the per wafer forecasted selling price for demand Dcfgmt. Note that the same technology code g may have a different forecasted margin for a different customer c, at a different factory f, in a different time period t of a different month m for a wafer foundry company. In a foundry plant, the expensive implanter equipment often represents the bottleneck operation and thus its capacity needs to be taken into account in particular in the model. The bottleneck machine consumption (in hours) per wafer and total bottleneck machine consumption limit are denoted by CMWcfgmt and CMfgmt, respectively.
We use EMfgsh to represent per wafer emission amount of chemical h at stage s with technology code g at factory f, and TEMfg to be the sum of EMfgsh with technology code g at factory f, i.e.,
We assume that unsatisfied orders are lost and unmet forecasts are ignored for the rest of the horizon. Table 3 shows the notation of the basic master planning model (MPT).
Table-3. Notation of the master planning model
Formally, MPT is expressed as follows:
Objective function (2) shows that profit of a foundry company is deducted by the pollution taxes paid to the government. Constraints (3)-(6) are the same as expressions (2)-(5) of the allocation planning stage in Chiang and Hsu (2014) except that the monthly time buckets are introduced here. Note that Rcfg is used to denote whether or not customer c’s products with technology code g can be manufactured at factory f. A customer’s product was qualified in a specific plant with requested technology code at the product engineering phase. It’s quite cost and time consuming to change the manufacturing plant or technology code. Thus, a customer places an order only after the product was qualified already. Constraint (3) ensures that the ATP capacity committed to customers be less than or equal to the available capacity. Constraint (4) specifies that the committed ATP quantity cannot exceed the customer demand forecast, while constraint (5) ensures that it should at least satisfy the order quantity placed. Since master planning is to reserve capacity for forecasted demand in the medium term (e.g., from 6th to 17th month) when only a small number of orders arrive at this stage,is usually true. Constraint (6) requires that the total consumption of bottleneck machine hours should be less than or equal to the maximum bottleneck capacity. Constraint (7) states that the sum of daily PFC emission amounts cannot be larger than the monthly emission cap.
Model 2: MPTS
Economic incentives to reduce environmental damage can be used through subsidies as well. A subsidy is a form of financial compensation for polluters to reduce PFC emissions. Thus, in model 2 (MPTS), we assume that in addition to the pollution taxes on PFC emissions, subsidies will be given to foundry plants with unused emissions. We use MBm to denote the remaining emission permits in month m, as given by expression (10) below,and SI as the subsidy per unit of unused emission permits. The total compensation can be obtained from multiplying
Constraint (10), instead of (7), is used to ensure that the consumed emission permits plus unused permits be equal to the total permits.
Model 3: MPTP
A progressive tax is a tax applied when the taxable base increases above a certain level. It is frequently used in personal income taxes. We incorporate it along with flat taxes into the master planning model, which is named as MPTP. We suppose that the producer pays a higher tax rate, denoted by TAXh, for his greenhouse gas emissions when the emission amount exceeds LMTm. The tax rate is TAX when the emission amount is less than or equal to LMTm, and will climb to TAXh when the amount is greater than LMTm. MPTP is expressed by
implying that there are flat taxes only for month m in (11).
Model 4: MPTSP
It is possible for government to include in an environmental policy all of the three policy instruments considered in this paper: flat pollution taxes, subsidies, and progressive pollution taxes. The resulting master planning model for a foundry plant is named as MPTSP, which is expressed by
4. RESULTS
We will provide a case study of a semiconductor company to examine the performance of the proposed four master planning models. The experiment design and corresponding data are described in section 4.1. Computational results of the four models are stated in section 4.2. The models can be solved to optimality by any LP software or optimization solver. For this research, we used the modeling and optimizing language Lingo 13.0 and the simple database Microsoft Access 2010. Also, all the computations were executed on a personal computer with an Intel Core i7-2600 2.80 GHz processor and 8 GB RAM, operated by the Microsoft Windows 7 professional system. All the four models were solved within six seconds. Discussion of the results is provided in section 4.3.
4.1. Experiment Design
The environment where the company operates is a classical MTO operation. The data was collected from it with some modifications (see also Chiang and Hsu (2014)). We consider three customers, three technology codes at a factory, and a planning horizon of twelve months. Each month is assumed to have 30 working days. Tables 4 and 5 list the experimental data of capacity and demand forecast, respectively used in the four models for a horizon from 6th to 17th months.
Table-4. Experimental data of capacity
| Month | 6Th |
7Th |
8Th |
9Th |
10Th |
11Th |
12Th |
13Th |
14Th |
15Th |
16Th |
17Th |
Total |
| Technology 1 | 40950 |
40950 |
40950 |
40950 |
40950 |
40950 |
42950 |
42950 |
42950 |
43550 |
43550 |
43550 |
505200 |
| Technology 1 | 46800 |
46800 |
46800 |
46800 |
46800 |
46800 |
46800 |
46800 |
46800 |
46800 |
46800 |
46800 |
561600 |
| Technology 1 | 46200 |
46200 |
46200 |
46200 |
46200 |
46200 |
46200 |
46200 |
46200 |
46200 |
46200 |
46200 |
554400 |
| Total | 133950 |
133950 |
133950 |
133950 |
133950 |
133950 |
135950 |
135950 |
135950 |
136550 |
136550 |
136550 |
1621200 |
Table-5. Experimental data of demand forecast
| Month | 6Th |
7Th |
8Th |
9Th |
10Th |
11Th |
12Th |
13Th |
14Th |
15Th |
16Th |
17Th |
Total |
| Customer 1 | 50400 |
49300 |
50400 |
50650 |
50400 |
50000 |
50400 |
50650 |
51800 |
50400 |
50500 |
50800 |
605700 |
| Technology 1 | 14400 |
13500 |
14400 |
14400 |
14400 |
14400 |
14400 |
14650 |
14400 |
14800 |
14400 |
14400 |
172350 |
| Technology 2 | 20100 |
20100 |
20100 |
20100 |
20100 |
20100 |
20100 |
20100 |
20100 |
20100 |
20100 |
20100 |
241200 |
| Technology 3 | 15900 |
15700 |
15900 |
16150 |
15900 |
15700 |
15900 |
15900 |
16300 |
16500 |
16000 |
16300 |
192150 |
| Customer 2 | 52200 |
52200 |
52100 |
52200 |
52200 |
51900 |
53100 |
52400 |
53050 |
53100 |
53100 |
53100 |
630650 |
| Technology 1 | 15150 |
15150 |
15250 |
15150 |
15150 |
14850 |
15450 |
15250 |
16050 |
15650 |
15450 |
15150 |
183700 |
| Technology 2 | 22800 |
22800 |
22600 |
22800 |
22800 |
22800 |
22800 |
22900 |
22800 |
23100 |
23400 |
23700 |
275300 |
| Technology 3 | 14250 |
14250 |
14250 |
14250 |
14250 |
14250 |
14850 |
14250 |
14200 |
14350 |
14250 |
14250 |
171650 |
| Customer 3 | 38400 |
38500 |
38700 |
38750 |
38600 |
37900 |
38600 |
38500 |
38800 |
38400 |
39200 |
38750 |
463100 |
| Technology 1 | 10800 |
10900 |
10950 |
11000 |
11000 |
10300 |
11000 |
10900 |
11100 |
10800 |
11600 |
10800 |
131150 |
| Technology 2 | 12300 |
12300 |
12300 |
12450 |
12300 |
12300 |
12300 |
12300 |
12300 |
12300 |
12300 |
12300 |
147750 |
| Technology 3 | 15300 |
15300 |
15450 |
15300 |
15300 |
15300 |
15300 |
15300 |
15400 |
15300 |
15300 |
15650 |
184200 |
| Total | 141000 |
140000 |
141200 |
141600 |
141200 |
139800 |
142100 |
141550 |
142650 |
142900 |
142800 |
142650 |
1699450 |
Table-6. Experimental data of tax and subsidy values
| Month | 6Th |
7Th |
8Th |
9Th |
10Th |
11Th |
12Th |
13Th |
14Th |
15Th |
16Th |
17Th |
Total |
| Customer 1 | 50400 |
49300 |
50400 |
50650 |
50400 |
50000 |
50400 |
50650 |
51800 |
50400 |
50500 |
50800 |
605700 |
| Technology 1 | 14400 |
13500 |
14400 |
14400 |
14400 |
14400 |
14400 |
14650 |
14400 |
14800 |
14400 |
14400 |
172350 |
| Technology 2 | 20100 |
20100 |
20100 |
20100 |
20100 |
20100 |
20100 |
20100 |
20100 |
20100 |
20100 |
20100 |
241200 |
| Technology 3 | 15900 |
15700 |
15900 |
16150 |
15900 |
15700 |
15900 |
15900 |
16300 |
16500 |
16000 |
16300 |
192150 |
| Customer 2 | 52200 |
52200 |
52100 |
52200 |
52200 |
51900 |
53100 |
52400 |
53050 |
53100 |
53100 |
53100 |
630650 |
| Technology 1 | 15150 |
15150 |
15250 |
15150 |
15150 |
14850 |
15450 |
15250 |
16050 |
15650 |
15450 |
15150 |
183700 |
| Technology 2 | 22800 |
22800 |
22600 |
22800 |
22800 |
22800 |
22800 |
22900 |
22800 |
23100 |
23400 |
23700 |
275300 |
| Technology 3 | 14250 |
14250 |
14250 |
14250 |
14250 |
14250 |
14850 |
14250 |
14200 |
14350 |
14250 |
14250 |
171650 |
| Customer 3 | 38400 |
38500 |
38700 |
38750 |
38600 |
37900 |
38600 |
38500 |
38800 |
38400 |
39200 |
38750 |
463100 |
| Technology 1 | 10800 |
10900 |
10950 |
11000 |
11000 |
10300 |
11000 |
10900 |
11100 |
10800 |
11600 |
10800 |
131150 |
| Technology 2 | 12300 |
12300 |
12300 |
12450 |
12300 |
12300 |
12300 |
12300 |
12300 |
12300 |
12300 |
12300 |
147750 |
| Technology 3 | 15300 |
15300 |
15450 |
15300 |
15300 |
15300 |
15300 |
15300 |
15400 |
15300 |
15300 |
15650 |
184200 |
| Total | 141000 |
140000 |
141200 |
141600 |
141200 |
139800 |
142100 |
141550 |
142650 |
142900 |
142800 |
142650 |
1699450 |
We are interested in investigating the economic and environmental performance of the four models with respect to different tax rate and subsidy settings. We will use the same tax and subsidy values in each experiment, as explained in section 2. Table 6 lists different TAX, SI, and TAXh values used, where TAXh is greater than TAX by 100 in each experiment. Following the 2013 TSMC corporate social report (TSMC, 2014) we set the emission amount per wafer between 0.156 to 0.302 tons in our experiments.
4.2. Computational Results
We present the results of computational experiments by running the four models. Figures 1, 2, and 3 show the results in terms of profit, allocation quantity, and emission amount, which are given in thousands of dollars, pieces, and tons, respectively.
Figure-1. Comparison of Profits
Figure-2. Comparison of Allocation Quantities
Figure-3. Comparison of Emission Amounts
From Fig. 1, we observe that profit and TAX are negatively correlated for MPT and MPTP, which makes sense intuitively. This is also true of MPTS and MPTSP when TAX is small. When TAX and SI are low, MPTP and MPTSP are likely to earn more profits by keeping on allocating capacity and emitting PFCs beyond the limit, as seen from Fig. 2 and 3 (Fig. 3 is similar to Fig. 2, for a foundry plant will yield more PFC emissions if it allocates more capacity to customers). On the other hand, when TAX and SI are high, MPTSP seems to earn more profits as well by allocating less capacity which results in fewer emissions. This shows the advantage of including progressive taxes and subsidies in master planning.
In the experiments, when TAX and SI are less than or equal to 1500, MPTP and MPTSP allocate the same capacity and earn the same profits. As TAX and SI are greater than 1500, it does not pay off to allocate capacity to the wafers with low margins; hence, MPTS and MPTSP perform the same as they allocate the same capacity to customers and progressive taxes do not come into play. When TAX and SI are greater than or equal to 3500, MPTS and MPTSP do not allocate capacity and thus receive subsidies from the government. Different levels of the pollution taxes and subsidies can allow the government or a decision maker to consider “what-if” analysis so as to determine the extent of the environmental “footprint” of an industry (the semiconductor industry in our models).
4.3. Discussion
It appears from the above results that the more policy options or instruments a master planning model is given, the better the profit obtained. For example, MPTP outperforms MPT if TAX/SI is small, and MPTS outperforms MPT if TAX/SI is large; MPTS also yields a smaller amount of PFC emissions than MPT if TAX/SI is large. Moreover, MPTSP performs the best for a foundry plant, since it allows the use of all three policy options: flat pollution taxes, subsidies, and progressive pollution taxes. When TAX/SI is small, it allocates capacity beyond the emission limit, thus paying progressive taxes to the government; on the other hand, when TAX/SI is large, it does not allocate enough capacity to customers, thus receiving some subsidies from the government.
We can interpret the results in another way. We know from linear programming that relaxing a constraint allows one to consider more choices, resulting in better objective function values. In our models, relaxing the PFCs emission constraint allows a foundry plant to use more policy options given by the government, resulting in improved profits. Hence, MTPSP can be regarded as the best proposed master planning model.
5. CONCLUSIONS AND POLICY IMPLICATIONS
In this research, we study a greenhouse gas emission problem that is relevant for master planning in the semiconductor industry. We consider taxation for capacity allocation to resolve the trade-off between economic growth and environmental protection. Specifically, we incorporate flat taxes, subsidies, and/or progressive taxes into master planning. We demonstrate that the planning model with subsidies and/or progressive taxes provides more flexibility than the model with flat taxes only.
The main contribution of this study is on considering four ecological taxation methods and evaluating both the economic and environmental performance of the corresponding four models. From the experiments, we find that setting tax rates and subsidy values is key to the models studied. If they are set low, firms would prefer to allocate more capacity to customers, thus perhaps paying progressive taxes to the government, and the environment hurts. On the contrary, if they are set higher than necessary to reduce emissions, firms would allocate less capacity, thus receiving some subsidies from the government, and the economy grows slowly. Designing optimized taxation and subsidy policies seems to be an attractive avenue of future research. However, we believe that the current study lays the foundation for such future research.
Funding: The authors would like to thank the National Science Council of Taiwan for financial support. The Grant No. is NSC 102-2410-H-009-040. |
Competing Interests: The authors declare that they have no competing interests. |
Contributors/Acknowledgement: Both authors contributed equally to the conception and design of the study. |
REFERENCES
Absi, N., S. Dauzere-Péres, S. Kedad-Sidhoum, B. Penz and C. Rapine, 2013. Lot sizing with carbon emission constraints. European Journal of Operational Research, 227(1): 55-61. View at Google Scholar
Avi-Yonah, R.S. and D.M. Uhlmann, 2009. Combating global climate change: Why a carbon tax is a better response to global warming than cap and trade. Stanford Environmental Law Journal, 28(3): 9-28. View at Google Scholar
Baumol, W.J., 1972. On taxation and the control of externalities. American Economic Review, 62(3): 307-322. View at Google ScholarBenjaafar, S., Y. Li and M. Daskin, 2013. Carbon footprint and the management of supply chains: Insights from simple models. IEEE Transactions on Automation Science and Engineering, 10(1): 99-116. View at Google Scholar | View at Publisher
Chang, M.B. and J.S. Chang, 2006. Abatement of PFCs from semiconductor manufacturing processes by nonthermal plasma technologies: A critical review. Industrial & Engineering Chemistry Research, 45(12): 4101-4109. View at Google Scholar | View at Publisher
Chen, J.C., C.W. Chen, J.C. Lin and H. Rau, 2005. Capacity planning with capability for multiple semiconductor manufacturing fabs. Computers & Industrial Engineering, 48(4): 709-732. View at Google Scholar | View at Publisher
Chiang, C. and H.L. Hsu, 2014. An order fulfillment model with periodic review mechanism in semiconductor foundry plants. IEEE Transactions on Semiconductor Manufacturing, 27(4): 489-500. View at Google Scholar | View at Publisher
Dissou, Y. and M.S. Siddiqui, 2014. Can carbon taxes be progressive? Energy Economics, 42: 88-100. View at Google Scholar | View at Publisher
Frota, N.J.Q., J.M. Bloemhof-Ruwaard, J.A.E.E. Van Nunen and E. Van Heck, 2008. Designing and evaluating sustainable logistics networks. International Journal of Production Economics, 111(2): 195-208. View at Google Scholar | View at Publisher
Gupta, J.N., R. Ruiz, J.W. Fowler and S.J. Mason, 2006. Operational planning and control of semiconductor wafer production. Production Planning & Control, 17(7): 639-647. View at Google Scholar | View at Publisher
He, Y., L. Wang and J. Wang, 2012. Cap-and-trade vs. Carbon taxes: A quantitative comparison from a generation expansion planning perspective. Computers & Industrial Engineering, 63(3): 708-716. View at Google Scholar | View at Publisher
Hoeller, P. and W. Markku, 1991. Energy prices, taxes and carbon dioxide emissions. No. 106. Paris: OECD.
Intergovernmental Panel on Climate Change (IPCC), 2013. Fifth assessment report: Changes in atmospheric constituents and in radiative forcing. Retrieved from http://www.ipcc.ch/report/ar5/wg1/ [Accessed 2015/01/07].
IPCC, 2013. Fifth assessment report: Changes in atmospheric constituents and in radiative forcing. Retrieved from http://www.ipcc.ch/report/ar5/wg1/ [Accessed 2015/01/07].
Johnson, K., 2010. A decarbonization strategy for the electricity sector: New-source subsidies. Energy Policy, 38(5): 2499-2507. View at Google Scholar | View at Publisher
Khalil, M.A.K., R.A. Rasmussen, J.A. Culbertson, J.M. Prins, E.P. Grimsrud and M.J. Shearer, 2003. Atmospheric perfluorocarbons. Environmental Science & Technology 37(19): 4358-4361. View at Google Scholar | View at Publisher
Kilger, C. and L. Schneeweiss, 2005. Demand fulfillment and ATP. In: Stadtler, H., Kilger, C. (Eds.), Supply chain management and advanced panning. Berlin: Springer. Ch. 9.
KPMG, 2014. The KPMG green tax index 2013. Retrieved from http://www.kpmg.com/Global/en/IssuesAndInsights/ArticlesPublications/green-tax/Documents/kpmg-green-tax-index-2013.pdf [Accessed 2015/01/27].
Li, A. and B. Lin, 2013. Comparing climate policies to reduce carbon emissions in China. Energy Policy, 60: 667-674. View at Google Scholar | View at Publisher
Mӧnch, L., J.W. Fowler and S.J. Mason, 2013. Production planning and control for semiconductor wafer fabrication facilities: Modeling, analysis, and systems. Springer.
National Academy of Science (NAS), 2007. Joint science academies’ statement: Global response to climate change. Retrieved from http://nationalacademies.org/onpi/06072005.pdf [Accessed 2015/01/07].
Oreskes, N., 2011. Metaphors of warfare and the lessons of history: Time to revisit a carbon tax? Climatic Change, 104(2): 223-230. View at Google Scholar | View at Publisher
Ou Yang, C.F., S.H. Kam, C.H. Liu, J. Tzou and J.L. Wang, 2009. Assessment of removal efficiency of perfluorocompounds (PFCs) in a semiconductor fabrication plant by gas chromatography. Chemosphere, 76(9): 1273-1277. View at Google Scholar | View at Publisher
Pigou, A.C., 1920. The economics of welfare. London: McMillan & Co.
Ponsignon, T. and L. Mӧnch, 2012. Heuristic approaches for master planning in semiconductor manufacturing. Computers & Operations Research, 39(3): 479-491. View at Google Scholar | View at Publisher
Ponsignon, T. and L. Mӧnch, 2014. Simulation-based performance assessment of master planning approaches in semiconductor manufacturing. Omega-International Journal of Management Science, 46: 21-35. View at Google Scholar | View at Publisher
Saysel, A.K. and M. Hekimoğlu, 2013. Exploring the options for carbon dioxide mitigation in Turkish electric power industry: System dynamics approach. Energy Policy, 60: 675-686. View at Google Scholar | View at Publisher
Stram, B.N., 2014. A new strategic plan for a carbon tax. Energy Policy, 73: 519-523.
Taiwan Semiconductor Manufacturing Company (TSMC), 2013. Accurate delivery. Retrieved from http://www.tsmc.com/english/dedicatedFoundry/manufacturing/efficiency.htm [Accessed 2015/01/07].
The United Nations Framework Convention on Climate Change (UNFCCC), 1997. Kyoto protocol. Retrieved from http://unfccc.int/kyoto_protocol/items/3145.php [Accessed 2015/01/07].
Tseng, S.C. and S.W. Hung, 2014. A strategic decision-making model considering the social costs of carbon dioxide emissions for sustainable supply chain management. Journal of Environmental Management, 133: 315-322.
TSMC, 2014. TSMC 2013 corporate social responsibility report. Retrieved from http://www.tsmc.com/download/csr/2014_tsmc_csr/chinese/files/c_all.pdf [Accessed 2015/01/07].
UNFCCC, 2009. Copenhagen accord. Retrieved from http://unfccc.int/resource/docs/2009/cop15/eng/11a01.pdf [Accessed 2015/01/07].
Yao, X.L., Y. Liu and X. Yan, 2014. A quantile approach to assess the effectiveness of the subsidy policy for energy-efficient home appliances: Evidence from Rizhao, China. Energy Policy, 73: 512-518. View at Google Scholar | View at Publisher
Views and opinions expressed in this article are the views and opinions of the author(s), International Journal of Management and Sustainability shall not be responsible or answerable for any loss, damage or liability etc. caused in relation to/arising out of the use of the content. |