GIS-based Planning Support System for Transportation and Industrial Location Analyses:

A Case Study of the Cokemaking Sector in Province, China by Yan Chen

Bachelor of Architecture, 1998 Tsinghua University, China Submitted to the Department of Urban Studies and Planning and the Center for Real Estate in partial fulfillment of the requirements for the degrees of Master in City Planning and Master of Science in Real Estate Development ROTCH

at the MASSAC HUSTTSNSTITUTE TECHNOLOGY MASSACHUSETTS INSTITUTE OF TECHNOLOGY OF 1 9 2003 February 2003 MA @2003 Yan Chen. All Rights Reserved LIEBRARIES

The author hereby grants to MIT permission to reproduce and to distribute publicly paper and electronic copies of this thesis document in whole or in part.

Author Department oft rban Studies and Planning Center for Real Estate January, 2003

Certified by Professor Karen R. Polenske Professor of Regional Political Economy and Planning Thesis Supervisor

Accepted b Professor Dennis Frenchman Chair, MCP Committee Department of Urban Studies and Planning

Accepted b Prdfessor William C. Wheaton Chairman, Interdepartmental Degree in Real Estate Development GIS-based Planning Support System for Transportation and Industrial Location Analyses: A Case Study of the Cokemaking Sector in Shanxi Province, China

by Yan Chen

Submitted to the Department of Urban Studies and Planning and the Center for Real Estate in partial fulfillment of the requirements for the degrees of Master in City Planning and Master of Science in Real Estate Development.

ABSTRACT

I created a Shanxi Province GIS -based Planning Support System (SPGPSS) for transportation and industrial plant location studies of the cokemaking sector in Shanxi Province.

By integrating database, map viewer, scripts, and professional models in the GIS environment, on the provincial level, I designed the SPGPSS to have capabilities of optimizing plant locations, transport routes and modes under the different scenarios and computing the corresponding cost, energy consumption, and pollution emissions in the transportation process. Policy makers and industrial organizations can utilize the SPGPSS to value the economic and environmental impacts from different policy possibilities and assist their planning decisions on location rearrangements and structural changes. On the plant level, a plant manager can use the SPGPSS to conduct spatial analyses and multi-plan valuations for an individual plant in the planning of transport routes and new plant location.

By the applications of SPGPSS, I tested my hypothesis that combining cokemaking plants into several large-capacity plants or industrial parks is preferable to having them distributed throughout the area. From the perspective of total cost, the large-capacity plants and industrial parks instead of the distributed small-capacity plants would reduce the total cost both from the transportation and cokemaking process. From the perspective of total energy consumption and pollution emissions, however, the large-capacity plants and industrial parks would increase the total energy consumption and pollution emissions. Thus, my hypothesis is only partially proven.

Thesis Supervisor: Karen R. Polenske ' Title: Professor of Regional Political Economy and Planning

Thesis Reader: David M. Geltner Title: Professor of Real Estate Finance ACKNOWLEDGMENTS

I would like to thank my research and thesis supervisor, Professor Karen R.

Polenske, for her continual help, support, and encouragement through out the entire time I have worked with her. Her kindness and devotedness will always remain in my heart. This thesis would not have gone anywhere without her strong support and insightful comments. I also thank Professor David Geltner for reviewing my thesis and giving me valuable comments and suggestions.

My most sincere appreciation is extended to Professor Steven Kraines and Mr.

Takeyoshi Akatsuka inTokyo University. They started their pioneering transportation studies in Shanxi Province and gave me lots of enlightenment and suggestions for my work in this area. I also thank all other team members in the

AGS MRP Group, especially Professor Fang Jinghua from University of

Technology for helping me on the coke-oven technologies.

This research was sponsored by AGS grants (No. 005151-042 and 008282-008) and NSF grant (No. 006487-001). I thank AGS and NSF for making this research possible.

Finally I want to thank my parents and my husband for their love and support all the time. TABLE OF CONTENTS

T IT L E ...... 1

A B S T R A C T ...... 2

A C K O W LED G E M ENT ...... 3

TA B LE O F C O NTENTS ...... 4

LIST O F FIG U R ES ...... 6

LIST O F TA B LE S ...... 8

A B B R EV IA T IO N S ...... 10

I IN TR O D U CTIO N ...... 11

2 HYPOTHESIS and OBJECTIVE ...... 18 2.1 Hypothesis ...... 18 2.2 Objectives...... 21

3 LITERATURE REVIEW ...... 23 3.1 GIS-based Planning Support Systems...... 23 3.2 Transportation NETFLOW Model ------...... 33 3.3 Process Flow Model...... 35 3.4 Industrial Location Theories...... 36

4 METHODOLOGY and PROJECT DESIGN ...... - 39 4.1 SPGPSS Components and Organization...... 39 4.1.1 SPGPSS Components...... 40 4.1.2 SPGPSS Advantages...... 47 4.1.3 SPGPSS Organization...... 49 4.2 Analysis of Alternatives ...... 50 4.3 Case Study...... 54 5 IM PLEM ENTATION ...... 55 5.1 Data Collection and Database Creation...... 55 5.2 GIS Modeling, Programming and Processing...... 61

6 APPLICATIONS...... 64 6.1 Analysis of Alternatives at Provincial Level...... 64 6.1.1 Minimize Total Cost...... 65 6.1.2 Minimize Total Pollution Emissions and Energy Consumption 72 6.1.3 Coke Oven Technology Impact Analysis...... 74 6.1.4 Highway Construction & Speed Improvement Impact Analysis 76 6.1.5 Industrial-Park Location-Choice Analysis...... 78 6.2 Plant Case Studies...... 79 6.2.1 Choose transport routes and modes...... 79 6.2.2 Selection of a new location...... 81 6.2.3 Choose coke-oven technology...... 83

7 CONCLUSIONSS ...... 84

APPENDICES ...... 86

Appendix A: Transportation Cost, Energy Consumption, and NOx Emissions for Each Transportation Link inthe GIS System...... 86

Appendix B: Application Results...... 88

BIBLIOGRAPHY...... 95 LIST OF FIGURES

Figure 1.1: Country-share of World Coke Production in 2000...... 12

Figure 1.2: Location of Shanxi Province in China...... 13

Figure 1.3: Cokemaking Supply Chain in Shanxi Province...... 15

Figure 2.1: Taiyuan Yingxian Non-recovery Coke Ovens...... 20

Figure 2.2: Qingxu Meijing Large-machinery Coke Ovens...... 20

Figure 3.1: GIS-based Urban Transportation Planning System in Portland 25 Metro, Oregon......

Figure 3.2: Inter-Oceanic Corridor Passing Through the Republic of Bolivia 26

Figure 3.3: The A IDA IR System ...... 28

Figure 3.4: NOx Industrial Emissions Plumes on a Fresh Northeasterly W indy Day in Geneva...... 29 Figure 3.5: NOx Emissions Due to Traffic in Geneva Region...... 30

Figure 3.6: Different Land-Rent Gradients...... 36

Figure 4.1: Structure of GIS-based Planning Support System (GPSS)...... 40

Figure 4.2: View of Coal Transportation in Shanxi Province, China...... 43

Figure 4.3: View of Coke Transportation in Shanxi Province, China...... 43

Figure 4.4: ArcView's Automatic Links between Map and Database...... 44

Figure 4.5: Comparison between the SPGPSS and System Developed by Kraines and Akatsuka (K&A System...... 47

Figure 4.6: System-Flow Chart of GPSS...... 49

Figure 6.1: Total Cost of Non-recovery Coke-Oven Technology...... 66

Figure 6.2: Total Cost of Large-machinery Coke-oven Technology...... 67 Figure 6.3: Plant-Cost Composition of Three Coke-oven Technology Optio ns ...... --

Figure 6.4: Coke Transportation Choices of X Cokemaking Company by Railway and by Highway......

Figure 6.5: Locational Choice of X Cokemaking Plant......

Figures in the Appendices:

Figure B.1: 2000 Plant-Min Scenario for Total Cost Minimization...... 91

Figure B.2: 2000 Transport-Min Scenario for Total PM Emission Minim ization...... 92

Figure B.3: Highway System in Shanxi Province...... 93

Figure B.4: Different Scenarios of Industrial-Park Locations in the PM Emission Minimization...... 94 LIST OF TABLES

Table 5.1: Coal Production, Coke Production Capacity, and Coke Consumption in Shanxi Province, 1990 and 2000...... 57

Table 6.1: Total Cost of Non-recovery Coke-Oven Technology...... 66

Table 6.2: Total Cost of Large-machinery Coke-oven Technology...... 67

Table 6.3: Plant-Cost Composition of Three Coke-oven Technology Options...... 74 Table 6.4: Comparisons Before and After the New Highway Construction and Road-speed Improvements...... 77

Tables in the Appendices:

Table A. 1: Transportation-cost Coefficients for Diesel T rucks...... 86

Table A.2: Transportation Cost Coefficients for Diesel and Electric Trains...... 86

Table A.3: Transportation Energy Consumption Coefficients for Diesel Trucks and Trains...... 87

Table A.4: Transportation Energy Consumption Coefficients of Electric Trains...... 87

Table A.5: Transportation NOx Emission Coefficients for Diesel trucks, Diesel Trains and Electric Trains...... 87

Table B.1: Road Transportation vs. Rail Transportation (2000 Base Scenario)...... 88

Table B.2: Coal Transportation vs. Coke Transportation (2000 Base Scenarios)...... 88

Table B.3: Particulate Emissions from Transportation and Cokemaking Plants...... 88

Table B.4: SOx Emissions from Transportation and Cokemaking Plants, 2000...... 89

Table B.5: Transportation Energy Consumption, 2000...... 89 Table B.6: Comparison of Three Cokemaking Industrial Park Scenarios...... 89

Table B.7: Coke Transportation of X Cokemaking Pla nt...... 89

Table B.8: Locational Choice of X Cokemaking Plant: Old Location Scenario...... 90

Table B.9: Locational Choice of X Cokemaking Plant: New Location Scenario...... 90

Table B.10: Plant-Cost Comparison of Different Coke-Oven Technologies.. 90 ABBREVIATIONS

AGS Alliance Global Sustainability

EPB Environmental Protection Bureau

ESRI Environmental Systems Research Institute

GIS Geographic Information System

GPS Geo-referenced Positioning System

GPSS GIS-based Planning Support System

GUI Graphic User Interface

IOPM Input-Output Process Model

MRP Multiregional Planning

NOx Nitrogen Oxide(s)

Plant-Min Plant-Minimization

PM Particulate Matter

RMB Renminbi (Chinese Currency)

SOEs State-Owned Enterprises

SOx Sulfur Oxide(s) SPGPSS Shanxi Province GIS -based Planning Support System Transport-Min Transport-Minimization

TVEs Town and Village Enterprises Chapter 1 INTRODUCTION

With the advances in information technology and database management,

Geographic Information System (GIS) technology has been widely used by analysts for planning decision-making. GIS makes it possible to store and visualize graphic maps together with linked attribute data. Planners can use GIS as a database management system to perform spatial data storing and quantitative analyses based on visual maps. GIS also provides a platform for planners to work with other application software and programming languages.

Nowadays, government, academia, and business analysts use GIS for various applications, including making thematic maps, building their own analyses and query models, and developing sophisticated planning support systems.

A GIS-based Planning Support System (GPSS), such as the one I developed in this study, is a computer-based system that uses GIS technologies to assist decision makers and policy analysts in the planning process by conducting spatial information processing and studies. Such systems provide simulations of planning alternatives in a GIS environment and utilize the GIS spatial tools or other professional models linked with the system to conduct analyses for specific purposes. Planners and decision makers can refer to these simulation results and choose the best alternative. Planners have widely used

GPSS in the assessment of public policies and strategies in various areas.

Examples include the Geneva AIDAIR project, a GPSS, to help decision makers assessing the impact of urban air-quality management in the Geneva region, Switzerland (CUEH, 1998). The German GAF mbH Company developed and implemented a GPSS to support multi-model transportation studies, focusing on the inter-oceanic corridors in Bolivia (GAF mbH Company, 1998). Those systems successfully simulate the environmental and transportation initiatives and evaluate the possible impacts from those policy and project implementations.

Based on the GIS technologies, GPSS can play a very useful and crucial role in the planning decision-making process.

Coke is a very important energy and commercial commodity in the international trade markets. While the cokemaking production capacities in

Europe and the United States have been dwindling in the past decade, China has been greatly increasing its production capacity and is becoming the largest coke production and consumption country in the world. In 2000, China produced approximately 30 percent of the total coke production in the world (Figure 1.1, Coke Outlook 2000 Conference), while its coke exports made up approximately

78 percent of the total exports in the world (Hua, 2001).

Figure 1.1: Country-share of World Coke Production in 2000

Country-share of Coke Production in 2000

Others 14%

Poland30

S. Aftc

German

Australia 6%

Russia U.S. 6% India 6%

Source: Coke Outlook 2000 Conference Shanxi Province is located in the north of China (Figure 1.2) and possesses abundant coal resources. Currently, Shanxi Province produces the most coke in China, representing about 40 percent of the total coke production in China and 16 percent of the total coke production in the world (Polenske and

McMichael, 2002). In 2000, Shanxi Province exported 6,440,000 tonnes of coke to the international coke market (Shanxi Statistical Yearbook, 2001). Shanxi also supplies a great amount of coke to domestic users, including iron-steel plants, electric-power plants, chemical plants, etc. Shanxi Province has become the biggest coke production and export region in the world.

Figure 1.2: Location of Shanxi Province in China

Shanxt Provnce

Source: http://www.chinatour.com/map/a.htm In the cokemaking sector, the supply chain runs from coalmines to the coal transportation to cokemaking plants, and then from cokemaking plants to the coke transportation to coke consumers (Figure 1.3). The cokemaking plant is the core component of this supply chain, and an analyst can view it as the connecting point for coal transportation and coke transportation. Coal and coke transportation play a very important economic and environmental role in the supply chain of the cokemaking sector in Shanxi Province. The transportation cost accounts for approximately one-third of the total production cost (AGS MRP

Field Trip Interview, 2001). The choice of transport routes and modes, the location of the cokemaking plant, and also the access to coalmines and coke consumers all affect the transportation cost. As we have documented in the different part of the team research in which I am participating, the cokemaking industry is a high-pollution and energy-intensive industry (Polenske and

McMichael, 2002). Researchers usually focus on the pollution and energy consumption from cokemaking plants, but the high pollution and energy consumption from diesel trucks in the transportation process aroused our attention on our field trips to Shanxi Province. We realize that the pollution and energy consumption from coal and coke transportation cannot be ignored. When we consider the future industrial planning of the cokemaking sector in Shanxi province, we should measure and minimize the overall cost and pollution both from plants and the transportation process. Currently, the Chinese national environmental regulations require closing the low-efficient and high-pollution small-capacity cokemaking plants and replacing these plants by some large-capacity plants or enlarging already existing plants. After the #367 official directive issued by the State Economic and

Trade Commission and State Environmental Protection Agency in 1997, the

Shanxi local government started to close the small-capacity plants using indigenous and modified indigenous coke ovens and replacing them by the large- machinery cokemaking plants whose annual capacity is usually larger than

200,000 tonnes. In the future provincial industrial planning, the Shanxi government plans to build several cokemaking industrial parks in the province to consolidate its cokemaking industry.

Figure 1.3: Cokemaking Supply Chain in Shanxi Province

Cokemaking Supply Chain

Stage I Stage 2 Stage 3 Stage 4 Stage 6 Coal Mines Coal Transportation Cokemaking Plants Coke Transportation Coke Consumers

" Coal mining """"""" iea a a technology Vehicle Oven Vehicle Domesti " Coal quality technology technology technology.

Suppliers Distributors Manufacturers Distributors Consumers Information Flow Source: AGS MRP Team, 2002 Those regulations not only have changed the plant capacity now being

used in Shanxi Province, but also affected the adoption of cokemaking technologies. The national environmental regulations prohibited the plants from

continuing to use the indigenous and modified-indigenous coke-oven

technologies and recommended that plant managers should adopt the large-

machinery coke-oven technology with by-product recovery. Currently, large-

machinery coke ovens are the ovens whose height is equal or bigger than 4.3

meters (AGS MRP Team Field Trip Interview, 2002). In China, the machinery ovens in the range of 2.8-3.2 meters height are called middle-machinery coke ovens and the ones shorter than 2.5 meters are called small-machinery coke ovens (Fang, 2002). In the United States, all of these ovens are called slot ovens.

The local Shanxi Environmental Protection Bureau (EPB) also recommended the non-recovery cokemaking technology, which represents the least-polluting coke-oven technology and virtually eliminates all hazardous air pollutants. This technology has better environmental performance than that of other coking methods, because it operates under negative pressure within the ovens (Li and Shen, 1995). The Chinese types of non-recovery coke ovens usually use cold-coal loading and cold-coke unloading, and the smoke released from these processes is less than the non-recovery coke-oven types used in the

United States (Fang, 2002).

Different coke-oven technologies have different investment costs and land costs. On the one hand, the use of non-recovery technology has lowered operational and investment costs but raised land costs due to the intensive land use. On the other hand, the use of large-machinery coke ovens usually has raised operational and investment costs but lowered land costs due to its compact machinery production.

As a result of these policy initiatives, coke managers and policy makers in

Shanxi Province need to integrate factors both from the transportation and the cokemaking plant (capacity and technology considerations), and optimize the location and transport routes for large-capacity plants and industrial parks to reduce the overall costs and pollutants from the cokemaking and transportation processes. In the planning decision-making process, GPSS plays a very important role. A planner can do simulations under different alternatives and assist the managers and policy makers to choose the optimized transportation and industrial plant respective routes and locations. Also, the planner can develop and use such GPSS systems in the transportation and industriaocation analyses of other sectors in other regions. The GPSS can generate specific spatial and quantitative analyses with visual maps for the studied region in the areas of transportation, location choice, and environmental pollution control. Chapter 2 HYPOTHESIS and OBJECTIVE

How to create and use the GIS -based planning support system to resolve the real-world problems is an interesting research topic. I choose the cokemaking sector in Shanxi Province, China as the study focus and create a Shanxi Province GIS-based Planning Support System (SPGPSS) to resolve its transportation planning and industrialocation issues.

2.1 Hypothesis

For the transportation and industrial-location study of the cokemaking sector in Shanxi Province, my hypothesis is that the large-capacity cokemaking plants and industrial parks instead of the distributed small-capacity plants will reduce the total cost, energy consumption, and pollution emissions both from transportation and the cokemaking process in Shanxi Province, China.

The total cost includes the transportation cost and the plant cost. The transportation cost is usually one-third of the production cost (AGS MRP Field

Trip Interview, 2001). I consider four kinds of major transportation activities in the cokemaking sector inShanxi Province: (1) transporting coal by road, (2) transporting coal by railway, (3) transporting coke by road, (4) transporting coke by railway. The plant cost consists of three parts. The first part is the operational cost, which is the cost for the daily operation and maintenance, as measured by the quantities of all the inputs used in cokemaking and the prices of each input.

It can be separated into materials cost, labor cost, and maintenance cost (Chen, 2000). The second part is the investment cost, which is the cost for the design and purchase of facilities and equipment. The third part is the land cost, which is the cost for the purchase of land and the improvement of ground infrastructure.

To estimate the land cost, I use the opportunity cost of land used for farming.

The plant cost is directly related to the plant capacity and the coke-oven technology. Obviously, the larger the plant capacity, the more the plant cost. To compare plants of different capacities and technologies, I use the unit plant cost

(total plant cost divided by total production capacity) as the measure. After the implementation of the 1997 #367 national environmental regulation,' there are two major recommended coke-oven technologies, large-machinery coke-oven technology and non-recovery coke-oven technology. The plants using the large- machinery technology usually have a larger capacity than the ones using the non-recovery technology. From the interviews with coke managers on our field trips, I found that the large-machinery cokemaking plants usually have higher unit operational cost and unit investment cost, but lower unit land cost than the non- recovery cokemaking plants. Because almost all the plants are located in the countryside and the unit price for the use of land is almost the same in this region, the amount of land used determines the land cost.2 Because of the technological requirements, the land use of non-recovery coke ovens is much more intensive than large-machinery coke ovens with the same capacity (Figures

Refer to the details of the 1997 #367 national environmental regulation in Chapter 1. 2 The land ownership and property rights in China are very different from those in the United States, but a detailed explanation is beyond the scope of my current study. Here, I assume that the plants own the use-right of the land, and the price of the use of land is equal to the opportunity cost of land used for farming. 2.1 and 2.2). That is why the unit land cost of non-recovery cokemaking plants is much higher than large-machinery cokemaking plants.

Figure 2.1: Taiyuan Yingxian Non-recovery Coke Ovens

Source: Author, 2002 Summer Field Trip in Shanxi Province, China

Figure 2.2: Qingxu Meijing Large-machinery Coke Ovens

Source: Author, 2002 Summer Field Trip in Shanxi Province, China The location does not affect the land cost much, but it does affect the transportation cost. The plants close to coalmines, coke consumers, or with easy access to railways or highways, have lower transportation cost than those farther away. On the one hand, the scattered small-capacity plants are very location- flexible in that they can locate near suppliers or consumers and choose the best transportation routes. On the other hand, the large-capacity plants do not have such advantages. There is therefore an important tradeoff between the transportation cost and the plant cost. In the case that several distributed small plants have the same capacity as one big plant, the scattered small plants have lower total transportation cost due to the contiguity to suppliers and customers and higher plant cost due to more initial investment. The large-capacity plant has higher transportation costs because it cannot locate near all its suppliers and consumers, but it has lower plant costs due to the economies of scale that accrue as the plant becomes larger.

2.2 Obiectives

By creating a SPGPSS for the transportation and industrial-location choice study of the selected areas and integrating a transportation model, a plant model, and a process-flow model and a friendly Graphic User Interface (GUI) within a

GIS environment, I am able to use the SPGPSS to assist real-time decision- making.

The following are major issues I explore for coke managers and policy makers in Shanxi Province: * What are the locations for specific cokemaking plants and industrial parks that

minimize the total transport cost, energy consumption, or pollution emissions?

* What are the transport routes and modes for specific cokemaking plants and

industrial parks that minimize the total transport cost, energy consumption, or

pollution emissions?

* What are the economic or environmental impacts on the total cost, energy

consumption, and pollution emissions of the plants of using different coke-

oven technologies?

* What are the benefits from the location rearrangements and industrial-

structure changes as measured by the reduction of the total cost, energy

consumption, or pollution emissions both from the transportation and

cokemaking process?

* What are the benefits from the recent transportation infrastructure

improvements as measured by the reduction of the total cost, energy

consumption, or pollution emissions both from transportation and cokemaking process?

I am able to test my hypothesis by applying the SPGPSS. The SPGPSS also provides the plant managers and industrial-park organizations with the optimized plant locations and transport routes and modes in the province or in the surveyed region, and shows the benefits from these location rearrangements and structure changes. Chapter 3 LITERATURE REVIEW

To create a GIS -based planning support system and conduct transportation and industrial-location analyses in the case of the cokemaking industry in Shanxi Province, I did a literature review of the critical aspects of the

GIS-based planning support systems, transportation NETFLOW model, process- flow model as well as the industrial-location theories.

3.1 GIS-based Planning Support Systems

The GIS-based Planning Support System (GPSS) has been widely used in the planning decision-making process in various areas (Gittings et al., 1993).

GPSS has the capability to improve the efficiency and quality of planning and program development. There are lots of innovative GPSSes using GIS technologies to solve real-world problems and assist planners and the public making decisions in the project planning and assessments. In the following part,

I review several relevant GPSSes in the areas of transportation and environmental studies. Those systems are all applied in the real-world planning process.

* GIS-based Urban Transportation Planning Studies in Portland Metro, Oregon (U.S. DOT, 1998) With the region's growing population, Portland Metro needed to reduce reliance on the car and vehicle miles of travel per capita during the next decade.

It has become an important issue to improve accessibility to employment, education and non-work activities while traffic congestion is expected to get worse during the plan period. To analyze these issues, Portland Metro decided to create a GIS-based urban-transportation planning system and to look at a range of transportation-system alternatives, including motor-vehicle alternatives with varying levels of investment in roadway improvements and transit and pedestrian alternatives with varying levels of investment intransit and pedestrian access to transit. To evaluate these alternatives, Portland Metro applied a kind of GPSS with travel-demand forecasting model that predicts how each alternative would affect transit ridership, traffic congestion, access to jobs, movement of goods and many other factors. The analyst uses the GPSS to collate and manage the data needed for the transportation model and also to display the model outputs, such as predicted employment densities and pedestrian environment factor. Their GPSS utilizes ArcView GIS as a major GIS environment. With the geocoding data of the location of households and activity centers, they were able to have the GPSS perform accurate spatial analyses of trip- generation and trip-distribution factors. Compared with the traditional modeling techniques that may be a poor approximation to reakworld conditions, GIS is able to measure very accurately the distance or travel time and to produce an

average value for a group of points by using actual network distance from each site. Thus, the use of geocoded locations in the GPSS improved the accuracy of the model data and the modeling process. They also used the GPSS to conduct the demographic and employment characteristics analyses, mixed land-use

24 measures, analyses of pedestrian accessibility to transit and pedestrian environment factors, and mapping and displays of model outputs.

Figure 3.1: GIS-based Urban Transportation Planning System in Portland Metro, Oregon - Households, Employment Sites, and Activity Centers

Source: U.S. DOT, 1998. The GPSS increasingly and continually assisted the planning activities in the improvement of the travel-modeling program at Portland Metro, Oregon. The system enabled planners to enhance travel models with more quantitative databases and collect and organize consistent information for analysis and validation. They benefited from the technical support provided by this comprehensive regional GPSS that allowed them to advance their travel- modeling efforts. This GPSS applied in Portland Metro demonstrated to planners how GIS could be used to support urban transportation planning. GIS databases provided an effective system for data management, and the GIS contained a number of tools for spatial analysis and data display that added value to the modeling process. In the future, GIS is likely to play an increasing role in transportation planning and become an important technology to shape the system framework and model development.

GIS-Based Planning Decision Support System for Transportation Planning and Infrastructure Management, Bolivia (GAF mbH Company, 1998)

The governments of Argentina, Bolivia, Brazil, Chile, Paraguay, and Peru worked together on projects of physical integration, so that the objectives of this project were to develop and implement a GIS -based planning system to support multi-modal transportation studies. The main objectives of the project were to develop the Bolivia highway transportation master plan and the transport alternatives of South-American inter-oceanic corridors, which start from central west Brazil and Bolivia and are destined to the Pacific Basin market.

Figure 3.2: Inter-Oceanic Corridor Passing Through the Republic of Bolivia

Source: GAF mbH Company, 1998 The project also included a highway inventory system using a satellite- based Geo-referenced Positioning System (GPS) technology. The planners conducted alternative analyses of potential volumes of traffic between the study area (Central West Brazil and Bolivia) and Pacific Basin markets through Pacific coast ports. They used the system as a platform and framework to update existing exports to the Pacific Basin and corresponding transport infrastructure needs from ports, rail and highways, while eventually strengthening sub-regional cooperation among the participating countries.

In this process, GIS technologies enhanced the study by making available further analytical tools, and providing planning and management capabilities. By using a GIS-based support system, an analyst can use more efficient and cost- saving methods than with any alternative approach to update the transportation information in the studies of outlets to the Pacific from the inner regions of

Central South American. Also, it is easier for the analyst to bring new infrastructure and technological developments and changes of productive capacity into the decision environment. This GPSS provided analysts GIS-type capabilities for the transportation sectors and a comprehensive spatial database, which they used primarily for project planning and infrastructure management. GIS also provided the means for analysts to visualize the spatial aspects of the problem being studied and a number of additional analytical tools. The analysts also can employ GIS -related technologies and planning capability in other studies for this region. * GIS-based Planning Support System for Human Ecology and Environmental Studies (CUEH, University of Geneva, 1998)

Figure 3.3: The AIDAIR System

GIS-DataBase 'Economic Acivity

Ene19 Trnspot teh. iTraffic

I Topography Polluttinndispersonn

Source: CUEH, University of Geneva, 1998. AIDAIR is a GIS -based decision-support system for air-pollution management in an urban environment. It is built around the AIRWARE, a product developed by the Austrian company Environmental Software and Services. Inthe AIDAIR-GENEVA project, policy makers in the Geneva region used AIDAIR system in their assessment of public policies and strategies concerning urban air-quality management.

This system is composed of three interconnected modules: EGIS (Energy

GIS), TAP (Traffic and Air Pollution) and APPH (Air Pollution and Public Health).

The system structure and the relations among the three modules are represented in Figure 3.3. EGIS is based on a technical-economic model of energy and technology choices. This module displays the results from the model by the GIS maps and integrates the energy technology choices directly in the APPH module, which simulates atmospheric pollutants emissions and dispersion. Figure 3.4 shows a simulation of Nitrogen Oxide(s) emissions from industrial sources in the

Geneva region.

Figure 3.4: NOx Industrial Emissions Plumes on a Fresh Northeasterly Windy Day in Geneva

Source: CUEH, University of Geneva, 1998.

The TAP module (Traffic and Air Pollution) is based on a traffic equilibrium model inthe AIDAIR system, which computes the air-pollution emissions due to traffic. From the traffic loads computed by the traffic equilibrium model, the TAP

module estimated and displayed the emissions of pollutants from traffic activities

(Figure 3.5). Analysts can use the APPH module to model the atmospheric-

pollution-dispersion and to apply epidemiological studies on air-pollution effects

to the population of the canton of Geneva. Concerning impacts of air pollution on public health, the AIDAIR system takes suspended particulates (PM10) as pollution indicators. As these PM10 emissions are not measured in Geneva, researchers inferred them from the NOx emissions. The epidemiological analysts tried to show a relationship between disease prevalence and different air- pollution levels. Figure 3.5: NOx Emissions Due to Traffic in Geneva Region & ~ ~M ~eINENEEPA ~ ~ m P LE TRARC

Source: CUEH, University of Geneva, 1998.

* Summary

From my reviews of the GIS -based planning decision-support systems, there are several common elements in these systems besides accurate spatial analysis from GIS tools:

1) GPSS enhances the efficiency and quality of alternative analyses in the

planning and program development. It is both time-saving and cost-saving to

conduct altemative analyses by GIS technologies in the motor-vehicle choice analysis inthe Portland Metro urban-planning studies and traffic-volume

analysis of inter-oceanic corridors in the Bolivia case.

2) To conduct specific studies, GPSS links professional models in the system, such as a travel-demand-forecasting model in the Portland Metro case, an

energy-technology-choice model, pollution-dispersion model and health-

impact model in the AIDAIR system. Those professional models expand the

GIS functions and make GPSS applicable for solving real-world problems.

3) GPSS contains a comprehensive database shared by different modules and

models in the system. In the Bolivia case, the sharing of the GPSS database

with the updated information of exports and corresponding infrastructure

needs undoubtedly strengthened sub-regional cooperation among the

participating countries. In the AIDAIR system, the model integration is based

on the comprehensive database for the whole Geneva region, which includes

statistical socio-economical data, traffic data, emission data, and

meteorological data.

4) GPSS effectively maps and displays model outputs and makes results more

observed than before by using GIS technologies. The AIDAIR system

visualizes the model results and links them with other demographic data for further analysis. The overlay of NOx emissions with the population

distribution clearly indicates visually the exposure of the Geneva population to

health risks.

The SPGPSS I developed for transportation and industrial-location analyses of the cokemaking sector in Shanxi Province has the above characteristics, but it is distinct in the some aspects. Usually, GPSS needs to have a large database with multi-area information to support its integrated modeling work. In the AIDAIR-Geneva project, the comprehensive and shared database made it possible for the analysts to integrate different professional models in the system and conduct advanced analyses between pollution emissions and health risks. In the SPGPSS, I need considerable data for typical transportation and environmental studies, which are not available due to the limitation of the data collection in China. I therefore designed the SPGPSS for two levels of research with different data requirements.

I did the first-level research work on the provincial level. Based on the basic information of the transportation network and cokemaking plants, my colleagues and I calculated the transport cost, energy consumption, and pollution emissions for each transportation link using the formulae and parameters we collected during the interviews and research, which to some extent made up for the lack of original detailed data. By connecting with the NETFLOW optimization model, an analyst only needs to know the transport cost and capacity of each link in order to have the SPGPSS conduct network optimization and generate the best transport routes and flows to minimize total transport cost. The analyst can also use the same method to optimize the transport routes and flows for energy consumption and pollution emissions. I ran the SPGPSS to conduct alternative analyses and compared the results to examine the impacts from the changes in prerequisites. Taking advantage of the efficiency and quality of SPGPSS in the alternative analyses, I focused more on the relative changes between alternatives than on the absolute results of each alternative and explored the reasons behind those changes.

The second-level research is on the plant level. Our AGS MRP research group did specific and detailed surveys in some Town and Villages Enterprises

(TVEs) and State-Owned Enterprises (SOEs) in the cokemaking industry in

Shanxi Province. I located those plants in the GIS transportation network by the addresses in the surveys. With the detailed information from the surveys, I used spatial tools imbedded in Arcview GIS to do spatial analyses and multi-plan valuations for an individual plant in the planning of transport routes and new plant location. My approach is to rely on the GIS spatial tools to conduct the comparison between the different transportation and location plans for cokemaking plants and to let coke managers themselves choose the best plan.

3.2 Transportation NETFLOW Model

I applied the freeware "NETFLOW" (Kennington and Helgason, 1980) to the optimization of transport routes and flows in the SPGPSS. NETFLOW is a mathematical optimization model to solve the cost-minimization problem of transporting a given quantity of material from a single-supply node to a single- demand node across a network of links and nodes. Each link has an associated cost and capacity. The program minimizes the total cost of transport across the network, subject to the capacities of each link. An analyst also can use it to minimize the total cost, energy consumption or pollution emission across the transport network. Kraines and Akatsuka in Tokyo University first applied this model to the transportation study of cokemaking sector in Shanxi Province. They constructed coal and coke transportation networks expressing the entire transportation infrastructure in Shanxi Province, using the cost-converted GIS-link data as described by Akatsuka (Akatsuka, 2001). Then, they connected the coal network and coke network at each cokemaking plant node with a link having the plant's production capacity as the link capacity and zero link cost. To satisfy the requirement of the NETFLOW program, they also created the single-supply node, single-consumption node, the links between single-supply node and the nodes for each coalmine, and the links between single-consumption node and the nodes for each coke consumer. They set these links with capacity equal to the coalmine production capacity or the coke-user consumption requirement and cost equal to zero. After setting the capacity and cost for each link, they ran the

NETFLOW program to do the transportation-flow and cost-minimization calculation and got the minimized total cost of transport across the network and the optimized flow route.

It is a very successful example of how to apply a mathematical optimization model to the real-world problem solving. They used the Excel spreadsheets as a database and Java-developed Graphic User Interface (GUI) to visualize the optimal routes and plant scales. But with a separate database and GUI, users have difficulty of updating the information and operating the program. Also the Java-developed GUI does not show as accurate spatial information as professional GIS software. For advanced spatial analysis, the SPGPSS has major advantages with many specific-designed GIS spatial- analysis tools.

3.3 Process-Flow Model

I connected a modified version of the input-output process model (IOPM) developed by Lin and Polenske with the GIS environment. Based on the survey data, the cokemaking process-flow model can provide a great deal of information for each facility within a plant and the interrelationships among the facilities that comprise this type of coal-using enterprise. In terms of costs, mass flows, and environmental emissions, I used the IOPM to evaluate alternative cokemaking technologies and distinguish the differences between them. In the process-flow balance sheet, the rows contain four key sections: main products, purchase inputs, byproducts and wastes, and primary inputs representing financial costs.

The columns pertain to the seven facilities in the plant: (1) steam coal washer, (2) metallurgical coal washer, (3) producer-gas generator, (4) steam coal-fired electric power plant and (5-7) three types of ovens used at the hypothetical plant.

This balance sheet provides information on the investment, labor, and maintenance costs plus costs of purchased coal and utilities costs. It also estimates revenues obtained from the sales of coke, chemical by-products, and generated steam and electricity (Polenske and McMichael, 2002). 3.4 Industrial-Location Theories

In their book "Urban Economics and Real Estate Markets", DiPasquale and Wheaton (1996) illustrated that the pattern of spatial separation and location of different land uses in the urban area is due to their different land-rent gradients

(Figure 3.6). The industrial uses have a relatively flat land-rent gradient and require lower land rent than the residential and commercial uses, which resulted in the industrial firms largely locating in suburban, even rural, areas. To some extent, the flat land-rent gradient for industrial use is due to the extensive land use of industrial firms and the lower land rent they are able to afford than the other uses. Due to the spatially diffuse character of truck and rail transportation, industrial plants decentralize, because they are less willing than others to compete with denser land users with higher rent, such as residential, commercial, and retail users. Figure 3.6: Different Land-Rent Gradients Land Rent

Industries

CBD Residenc

------

Distance Source: DiPasquale and Wheaton, 1996. From the twentieth century, DiPasquale and Wheaton (1996) indicate that the changes in both production and storage methods greatly increased the amount of land used per unit of output by industrial firms. The integrated horizontal assembly lines and modern inventory technology both have high land requirements and increased the amount of land needed for the industrial use.

Thus, they argue that the production and technology requirements have driven the industrial use, consuming more land than the other users. Even in the same industry, the industrial firms with different production technologies have the different requirements of land-use intensity, which eventually causes those firms to have different land-rent gradients and choose the different locations.

With the suburban decentralization of households, many firms are decentralizing to be closer to their workforce, which consequently lets them pay lower wages and receive more profits (DiPasquale and Wheaton, 1996).

Employment is one of the factors for decentralization of the industrial firms, but I argue that employment is not the determining reason for the decentralization of industrial firms in developing countries, such as China. As we know, China has a large population and plentiful labor, so that the workforce is not a driving factor for the firms to decide where they locate to reduce the cost. Comparatively, the transportation cost is a more important factor determining the industrial locations.

Industrial plants prefer to locate near the raw-material resources, suppliers, customers, and major transportation routes to reduce their transportation cost.

They locate and decentralize with the attraction of the easy access to suppliers/customers and major transportation network. Thus, in the conditions that labor cost is not as important as transportation cost, the industrial-location choice is not employment-driven, but transportation-driven.

Industrial firms have a tendency to locate by merging and by reaping agglomeration economies. Rees and Stafford (1986) indicate that firms might derive economic advantages from locating in larger, more central, clusters.

There is a fundamental tension between economies of scales and the impact of distance. Larger economies of scale result in larger, fewer, more widely separated plants. Smaller plants located in a finer, more dispersed spatial network would have low friction of distance-related (transportation) costs, but higher investment cost in total. Larger plants may be internally more efficient, and, in the aggregate, easier to manage; but transportation costs are higher, single investments are larger, flexibility is reduced, and the risks of a poor locational choice are greater. The trick is to balance these opportunity forces correctly (Rees and Stafford, 1986).

How to reach the optimization point of the decentralization and aggregation is an interesting question. With the use of the GIS technologies and network optimization algorithm, I conducted transportation and industrial-location analyses to get the balancing point between decentralization and aggregation in the case study of the coal and coke transportation of the cokemaking sector in

Shanxi Province, China. Chapter 4 METHODOLOGY AND PROJECT DESIGN

I have designed a GIS -based Planning Support System (GPSS) with the technologies of Geographic Information System (GIS) to assist decision--making and policy analysis. It is now commonplace for business, government, and academia to use GIS for many diverse applications. I apply the Shanxi Province

GPSS (SPGPSS) to transportation and location choices faced by the coke managers and local government officials in Shanxi Province, China.

4.1 SPGPSS Components and Organization

GIS is a computer-based system capable of holding and using data describing places on the earth's surface. "GIS doesn't hold maps or pictures - it holds a database. The database concept is central to a GIS and is the main difference between a GIS and drafting or computer mapping systems, which can only produce good graphic output" (ESRI Inc., 1997). GIS gives users a powerful analytical tool to combine both graphic maps and linked attribute data, such as population, property value, and pollution emissions, into one integrated system.

It also allows users to build their own analyses and query models connected with the GIS system, and it automatically generates final results for alternatives. In my study, GIS plays a crucial role by providing a database, a map viewer, and analytical tools. I use the GIS package ArcView GIS developed by

Environmental Systems Research Institute Inc. (ESRI Inc.) as the major platform for my research. 4.1.1 SPGPSS Components

The SPGPSS has four major components: (1) database, (2) map viewer,

(3) scripts, and (4) professional models. Although I use it for a study of the cokemaking sector in China, I designed it to be applicable to other sectors and other countries. Each component has its special functions and is independent of other components, but they are closely connected and the basic structure of this

SPGPSS is shown in Figure 4.1. Figure 4.1: Structure of SPGPSS SPGPSS in ArcView GIS

Source: Author

(1) Database -Core of SPGPSS

Spatial data are at the heart of every ArcView GIS application. Spatial data are geographic data that store the geometric location of particular features, along with attribute information describing what these features represent, also

known as digital-map or digital-cartographic data (ESRI lnc, 1997). The spatial data have three forms in the SPGPSS. (1) A point data set contains data of

"node" features, such as cities, towns, villages, and cokemaking plants, as well as their attribute data, such as city population and plant-production rates. (2) A line data set contains the data of "links", representing primarily the roads and railways of the transportation network. (3) An area data set mainly stores data of regions, such as industrial districts, residential areas, and even cities, towns, and villages. Except for the spatial data, the non-spatial attribute data set stores information with numerical and characteristic values, like numbers and qualitative evaluations. Spatial data and non-spatial data sets can contain the information users need to run spatial models and make the specific quantitative and qualitative analyses. I prepared and integrated the data sets to form the major database for the SPGPSS for Shanxi Province cokemaking sector, which contains economic and technological data of mines, cokemaking plants, and consumers in the transportation network.

In this SPGPSS, most of the spatial data are stored in the form of ArcView shapefiles. The shapefile format has five files with specific file extensions, of which the most useful two files are the .shp file - the shape file that stores the feature geometry, and the .dbf file - the dBASE file that stores the attribute information of features. Users can add a shape file (.shp) into a map viewer as a theme, which represents all the features of a particular feature class in the data source. They can then display the related dBASE file as a feature table in

ArcView. They can export the dBASE file into Excel and other database- processing software and also import the dBASE file from Excel to Arcview by the function of adding tables. The dBASE file is one kind of attribute data, which can include almost any data set, whether or not it contains geographic data. Users can display some tabular tables in a map viewer directly, and they can join others that provide additional attributes to the existing spatial data. ArcView can support the data from database servers, such as Oracle, dBASE III and IV files,

INFO tables, and text files with fields separated by tabs or commas.

(2) Map Viewer-Graphic User Interface (GUI) of SPGPSS The ArcView GIS can have different spatial data stored in the different views, which provides users with a separate layer to display and query a collection of user-defined themes. By displaying or hiding those views, users can have the different combinations of the spatial data for analyses. Below, I provide two views with different theme combinations, showing the different parts of the supply chain in the cokemaking sector in Shanxi Province. In the view of coal transportation (from coalmines to cokemaking plants), I include themes of coalmines, cokemaking plants, railways, roads categorized by the service levels, cities and towns, and the border of Shanxi Province (Figure 4.2). In the view of coke transportation (Figure 4.3), I turn on the theme of coke consumers and turn off the theme of coalmines, and then the view shows the coke-transportation routes from the cokemaking plants to the coke consumers. Figure 4.2: View of Coal Transportation in Shanxi Province, China

I Source: Author Figure 4.3: View of Coke Transportation in Shanxi Province, China

Source: Author By using Arcview tools, users can move or zoom the views in and out conveniently. When users select an object inthe view, they can select and highlight the data stored in the related feature table, as shown in Figure 4.4. The automatic link between the map in the view and the data in the feature table is a great advantage in spatial analyses, because users do not need to switch back and forth between the different programs, and they can show all the changes and updates in the feature table in the corresponding map. Users can use the GIS environment to combine spatial data as well as corresponding non-spatial data.

Such a combination overcomes the traditional bottleneck between maps and related tabular data. The map viewer works as a Graphic User Interface (GUI) for the SPGPSS. By adding additional buttons and menus for specific users, I customize and design a user-friendly GUI.

Figure 4.4: ArcView's Automatic Links between Map and Database

Source: Author (3) SCRIPTS - BRIDGES OF SPGPSS

Script is a piece of code written by an object-oriented language named

Avenue, which is the programming language embedded inArcView GIS. Scripts can control how and when to send requests to function objects in ArcView.

Through Avenue scripts, users can expand the GPSS by connecting the ArcView

GIS with professional models and by customizing ArcView's look and functionality. In the SPGPSS, the major functions of scripts are to retrieve the needed data from the tables in the ArcView GIS, sort or format those data, write the data into the professional models outside the ArcView GIS and execute those models, read the results back from the models to the ArcView GIS, and update the old tables.

Script works as a bridge connecting the professional models with the

ArcView GIS and integrates the different components into one system. The customized buttons on the toolbar and the menus of map viewer can link those scripts. Users only need to click those buttons or menus to start those scripts to run the professional models inthe background and then get the results back to the SPGPSS automatically. These designs are especially beneficial for the analysis of alternatives. By changing the data in the maps or tables for different alternatives, users can run models efficiently and get results quickly for comparison and analysis. (4) Professional Models-Branches of SPGPSS

In the SPGPSS, I connect the system with two professional models: the transportation NETFLOW model (Kraines et al., 2001) and the process-flow model (Polenske and McMichael, 2002). I introduced these two models in the literature review.

These two models can share a common database in the system and retrieve the data needed for their requirements by the scripts. Users can conveniently start the models using a GUI, making the models run in the background, and get the results back to the SPGPSS as soon as the models finish their internal computations. The professional models also can be other models in other areas, such as emission models for environment studies or logistics models for supply-chain management. The type of model that can be connected depends on the database design of SPGPSS and the connection interface of the model with the system. Some models have very specific data requirements, for example, the emissions model for transportation studies usually need emission-factor data, which relate to the data of vehicle type, speed, and traffic volume. To implement such models, users must incorporate such a dataset into the SPGPSS database and resolve the connection requirements for each model. 4.1.2 SPGPSS Advantages

As noted in Section 3.2, Kraines and Akatsuka successfully modeled the transportation activities of the cokemaking sector in Shanxi Province. They used

Excel spreadsheets as a database and Java-developed Graphic User Interface

(GUI) to visualize the optimal routes and plant scales. Compared with the system developed by Kraines and Akatsuka (K&A System), the SPGPSS has the following advantages for spatial-information processing and alternative analyses as an integrated system with many powerful components (Figure 4.5). Figure 4.5: Comparison between the SPGPSS and System Developed by Kraines and Akatsuka (K&A System) SPGPSS

K&A System

Source: Author 1) Integration: The SPGPSS integrates the database, GUI, and professional

models in the GIS ArcView Environment. These components are

automatically linked with each other in the system. Users can do all the

operations in one program, GIS ArcView, and do not need to switch between

the different programs, such as Mapinfo, Excel and Java-developed GUI as in

the K&A System. Thus users can avoid mistakes in the operational process.

2) Interaction: Users are able to interact with the system in a real-time mode.

Users do not need to know the complicated structure inside the SPGPSS,

which works as a black box, but they just insert different alternatives from a

friendly GUI and obtain the results quickly.

3) Accuracy: The GIS-based processing maintains data integrity and accuracy.

GIS has a great advantage for use in spatial-information processing and

analyses because of the automatic links between maps and tables. The

ArcView GIS program also provides many spatial-analysis tools, such as the

identify function and the measure function. Although the K&A System uses

another GIS software, MapInfo, contains the GIS data, but it only provides the

distance and some other data to Excel for computation, and does not conduct

spatial analysis and visualize the optimal results in the program by the GIS

technologies. Thus, for advanced spatial analysis, the SPGPSS, which has

many specific-designed GIS spatial-analysis functions, has major advantages

over the K&A System. 4.1.3 SPGPSS Organization

Figure 4.6: System-Flow Chart of SPGPSS

Source: Author The SPGPSS is organized as shown in Figure 4.6. Users can start from the ArcView Map Viewer (GUI). An interactive GUI is an important component of the SPGPSS. It includes GIS special features, such as maps, tables, legends, and common components including toolbar, menu, and application tips. The GUI automatically interacts with related databases by the special link function of ArcView GIS. By the tools on the menu bar and buttons of the GUI, users can select and update the data on maps and tables. For example, users can insert a

new plant onto the map, change production data of the cokemaking plant in the table, or update the capacity of a specific road or railway. The inserted

information will immediately update the feature tables in the database. In this

SPGPSS, I use scripts to connect the professional models. The scripts are

linked with the buttons on the toolbar. After users click the button, the script will write an input data file from the system database, and tell the professional model to read the input data and execute itself to do internal computations. Then, the professional model gives computation results by writing an output file. The models stay in the background of the system and do the computation work inside a black box. The output file returns to the system database and updates the related tables automatically by another script. Simultaneously, the maps in the

GUI spatially show the updated information.

4.2 Analysis of Alternatives

By applying this SPGPSS for the different alternatives (i.e., different inputs in the SPGPSS database), I found that it is effective and efficient to use the alternative analyses for this research. I compare the minimized total costs, energy consumption, or emissions from transportation and cokemaking plants

under different alternatives in the year 2000; explore the underlying reasons

behind those differences; evaluate several coke-oven technology options for the plants in terms of plant costs and emissions; measure the impacts of new highway construction on the coal and coke transportation by the reduction of

costs, energy consumption, and emissions; and finally recommend the optimized

plant location, capacity allocation, transportation routes and modes by comparing

the results from these alternatives. Kraines and Akatsuka first developed this

method of alternative analysis and used the Base Scenario, Transportation-

Minimization Scenario, and Plant-Minimization Scenario based on the 1990 data

in their previous transportation studies in Shanxi Province (Kraines et al., 2000). I keep using these scenarios to make our research consistent. Based on the updated 2000 data of coalmines, cokemaking plants, and coke consumers, I conduct further analyses of the following alternatives at the provincial level: e 2000 Base Scenario (2000 Base): Minimize the transportation cost, energy

consumption, or pollution emissions and give the corresponding

transportation pattern, based on the distribution of coalmines, cokemaking

plants and coke consumers in 2000 (Shanxi Statistical Yearbook, 2001).

0 2000 Transport-Minimization Scenario (2000 Transport-Min): Set plant sites to

minimize transport cost, energy consumption, or pollution emissions. First, I

set the capacity of each plant in the base scenario to 3,000,000 tonnes per

year, which is the upper limit for individual plant coke production, not the

actual production (AGS MRP Field Trip Interview, 2002). Next, I run the

GPSS and transportation tradeoff model. This calculation gives the lowest

possible transportation cost, because each plant is allowed to produce under

the maximum capacity. Then, I change the production capacity of each plant

just exceeding the production rates required by the transportation tradeoff

model. I also use the plant model to calculate the cost, energy consumption, or pollution emissions from cokemaking process and add them to the

transport part to give the total cost, energy consumption, or pollution

emissions. Therefore, this scenario gives the minimum transportation cost,

energy consumption, or pollution emissions in the supply of the total coke

demand given the locations and capacities of coalmines, the possible

locations for cokemaking plants, and the maximum cokemaking plant size (Kraines et al., 2000).

* 2000 Plant-Minimization Scenario (2000 Plant-Min): Set plant sites to

minimize plant cost, energy consumption, or pollution emissions. In this

scenario, I set the plants to maximum capacity and run the SPGPSS. I also

calculate the number of maximum capacity (3,000,000 tonne per year) plants

required to produce the total coke demand in Shanxi Province. Then, I assign

one of these maximum-sized plants to the site having the largest production

capacity calculated by the transportation-tradeoff model. I continue to assign maximum-sized plants to the sites with the largest calculated production

capacity until I have assigned the total required number of plants. I assign the

last plant a production capacity just large enough to meet the total demand. I

then run the SPGPSS again with these location conditions and calculate the

plant costs, energy consumption, and/or pollution emissions using the by the

plant process-flow model. This plant-minimization scenario is the simulation

in which small-scattered plants are closed and replaced by aggregated big-

scale plants (Kraines et al., 2000).

* Coke-Oven Technology Impact Analysis:

After the implementation of the 1997 #367 national environmental regulation,

there are two major recommended coke-oven technologies, large-machinery

coke-oven technology and nonrecovery coke-oven technology. I therefore

give three options for coke-oven technology adoption:

Option 1: All the plants use the large-machinery coke-oven technology,

Option 2: All the plants use the nonrecovery coke-oven technology, Option 3: All the plants whose capacity is more than or equal to 500,000

tonnes use large-machinery coke-oven technology; all the plants whose

capacity is less than 500,000 tonnes use nonrecovery coke-oven

technology.

* New Highway Construction and Speed Improvement Impact Analysis: the

impacts from the newly built highways and speed increase from 1990 to 2000.

* Industrial Parks Location Analysis: In future provincial planning, the Shanxi

government officials plan to develop two cokemaking industrial zones. The

tentative locations are in the Lishi-Liulin area, area, or Jiexiu area.

Using this SPGPSS, I make simulations of the establishment of these

industrial zones and compare the total cost, energy-consumption, and

pollution emissions from different choices of the industrial zones. I begin with

three location alternatives, each with two of the three industrial zones, and I

assume that each industrial zone will produce half of the current total coke

production in Shanxi:

1. Lishi-Liulin industrial zone and Linfen industrial zone.

2. Linfen Industrial zone and Jiexiu industrial zone. 3. Lishi-Liulin industrial zone and Jiexiu industrial zone. 4.3 Case Study

At the plant level, I do a case study of an individual cokemaking plant in

Shanxi Province. I focus on how a coke-plant manager can utilize the SPGPSS to conduct the valuation and comparison of the multiple plans for the plant's future operation and management, which includes selection of the location for a new plant, the potential suppliers, the transport routes and modes, and the coke- oven technologies.

For the individual plant manager, it may be inappropriate to use the

SPGPSS to do the network optimization by the transportation net-flow model, but plant managers still can utilize the analytical tools in ArcView GIS and its database linked with the professional model, such as the process-flow model, to do specific spatial analyses and multi-plan valuations. Chapter 5 IMPLEMENTATION

In the implementation of this Shanxi Province GIS -based Planning

Support System (SPGPSS), my colleagues and I collected the related data and prepared the database for the SPGPSS. I also did the work of GIS modeling, programming and processing for this system.

5.1 Data Collection and Database Creation

Our research team obtained the current GIS maps of Shanxi Province from the Australian Center of the Asian Spatial Information and Analysis Network

(ACASIAN), including the spatial data of cities and towns, roads and railways in

Shanxi Province. I thank Crissman in ACASIAN, who provided us these GIS maps. Kraines and Akatsuka first started the transportation studies in Shanxi

Province and gave me suggestions for my further work. They inserted into the

Shanxi Province GIS maps the locations of major coalmines, cokemaking plants, and coke consumers in Shanxi Province and the production, consumption, and export data as well as the capacities of major railways from the 1990 Energy

Resources Atlas of Shanxi Province (Shanxi Committee of Atlas Compilation, 1994). They also developed the formulae to calculate the transportation cost, energy consumption, and NOx emissions for each transportation link in the GIS system (Appendix A, Kraines et al., 2001). They saved those data in Microsoft

Excel spreadsheets. To create the SPGPSS and make the original work more GIS -based, I transferred the GIS data from MapInfo software to Arcview, the platform of the

SPGPSS. MapInfo and ArcView are both GIS software and widely used in the

GIS applications. I chose ArcView as the platform of the SPGPSS because

Arcview has more powerful functions and convenient tools to connect other professional models with the system itself than Mapinfo, which is an outstanding advantage for the creation of a GPSS. Arcview can read the GIS data in the format of Mapinfo and save them into the Arcview data format. In the current Shanxi GIS map, I located the 107 Town and Village

Entrepreneurships (TVEs) and 8 State-Owned Entrepreneurships (SOEs) from the Alliance for Global Sustainability (AGS) Multiregional Group (MRP) 2000 TVE

Survey and 1999 SOE Survey (AGS MRP team, 2001) using a Shanxi Province

Atlas (Shanxi Transportation Facilities Office, 2001). In ArcView, I pasted the

scanned maps from the Atlas to the background and located the selected TVEs

and SOEs by the detailed addresses provided in the surveys.

Because the surveys contain confidential information, I located the plants

by their addresses only for research and internal use. By assigning each plant a unique ID in the SPGPSS, I exclude the plant name and any other information of identification from the SPGPSS. With the unique ID for each plant, I can join the

plant's spatial data with the plant's operational and investment data from the

2000 TVE survey and 1999 SOE survey, which include the information of coke

and by-product production, suppliers, consumers, transportation, facility and

equipment, financing, employment, and pollution. To integrate the plant-specific data from the surveys into the SPGPSS, the plant managers and other users can do the different aspects of analysis of an individual plant according to their goals and needs.

Table 5.1: Coal Production, Coke Production Capacity, and Coke Consumption in Shanxi Province, 1990 and 2000 Annual increase over 1990 2000 10 years Total Coal Production for cokemaking 61,618,000 169,882,000 27.6% Total Coke Production Capacity 16,100,050 56,913,000 35.3% Total Coke Consumption 14,599,890 33,295,000 22.8% # Unit: tonnes/year Note: Total coke consumption includes consumption in Shanxi Province and exports. Source: 2001 Shanxi Statistical Yearbook, Shanxi Statistical Bureau. From 1990 to 2000, the total coal use for cokemaking in Shanxi Province increased 276% (27.6% annually), the total coke production capacity increased

353% (35.3% annually), and the total coke consumption (including the provincial consumption and exports) increased by 228% (22.8% annually). During the 10 years, the Shanxi cokemaking industry has grown rapidly. I updated the coalmine production data used in cokemaking, cokemaking plant production capacity, consumption of coke users and exports in the SPGPSS by the 2000 data from Shanxi Fifty Years 1949-1999 (China Statistics Press, 2000).

For the transportation model, I use the formulae and transportation link data for fuel efficiency, cost, energy consumption, and NOx emission from Kraines and Akatsuka (Akatsuka, 2001). The high pollution from diesel trucks, especially the particulate pollution aroused our attention during our field trips in

Shanxi Province. In the interviews with Shanxi Environmental Protection Bureau

(EPB) officials, I found that they realized the transportation pollution is another significant source of pollution, but they have not given it much attention yet because of the difficulties to make reliable measurements and conduct quantitative analyses. Because transportation pollution is a major concern from our research perspective and is also another important factor affecting the relocation and merger potential of cokemaking plants, I added the Particulate

Matter (PM) and Sulfur Dioxide(s) (SOx) emission indicators to measure the transportation emissions. I use the vehicular emission factors with respect to fuel consumption in the Philippines (Rogers et al., 1997) to estimate the PM and SOx emissions. I have not found such vehicular emission factors in China that I can use for the formulae. Because China and Philippines are both developing countries and have similar pollution-emission conditions, I use the Philippines vehicular-emissions factors in these pollution estimates. For the diesel fuel, the vehicular-emission factor of PM is 18.0 gram/liter and the vehicular emission factor of SOx is 10.8 gram/liter. I get PM and SOx emissions in units of grams for transporting one tonne of coke by each transportation link as follows

(Equation 1):

E, =EFxEff xD (1)

E,: Transportation Emission (gram/tonne) EF: Emission Factor (gram/liter) Eff: Fuel Efficiency (liter/tonne-kilometer) D: Distance (kilometer)

For the plant process-flow model, I develop the formulae to calculate the

plant cost, energy consumption, and pollution emissions. To calculate the total plant emissions, I multiply the average emissions to produce one tonne of coke by the total coke production by those plants in that year (Equation 2):

E, = AVG(E,)x TP (2)

E,: Plant Emission (kilogram/year) A VG(E,): Average Emission per tonne of coke (kilogram/tonne) TP: Total Coke Production (tonne/year)

I calculate the PM plant emission based on the average number of PM emissions permitted to be released for producing one-tonne coke in Shanxi

Province. From our interviews in the Shanxi Environmental Protection Bureau

(EPB), I learned that the average PM emissions permitted by the EPB is 1 kilogram per tonne of coke for the non-recovery coke-oven technology, and 2.5 kilogram per tonne of coke for the large-machinery coke-oven technology.

Assuming that all plants emit only the regulated maximum, I determine that for the total coke production of 33,295,000 tonnes in year 2000, the total annual PM emissions is 33,295,000 kilograms if all the plants use the non-recovery technology, or 83,237,500 kilograms if all the plants use the large-machinery technology. This, of course, is probably less than what was actually emitted, because some plants were not in compliance with the regulations in the year

2000. The average SOx emissions permitted by the EPB is 1.8 kilogram per tonne of coke produced for the non-recovery technology, and 0.4 kilogram per tonne of coke produced for the large-machinery technology (AGS MRP Field Trip

Interview, 2002). By the same method as above, I determine that the total SOx

emission is 59,931,000 kilograms if all the plants use the non-recovery

technology, or 13,318,000 kilograms if all the plants use the large-machinery technology. Because of the same total coke production, the PM and SOx emissions from cokemaking plants are the same in the three scenarios.

For the total plant cost, I sum the plant costs of all the cokemaking plants.

I calculate the three plant-cost components, operational cost, investment cost, and land cost, for each plant and then add them. Based on the economic model

(Chen, 2000) to estimate the operational cost and investment cost of different cokemaking technologies in Shanxi and the modeling of "economies of scale" of cokemaking technologies (Kraines et al., 2001), I use Equations (3) and (4) to calculate the operational and investment cost for non-recovery coke-oven technology and Equations (5) and (6)for large-machinery technology.

For non-recovery coke-oven plants:

Operational Cost = 3.0 x (0.9x PS/BPS + 0.1) + 0.16 x P (3)

Investment Cost = 1.35 x (0.9 PS/BPS + 0.1) (4)

For large-machinery coke-oven plants:

Operational cost = 19.32 x (0.9 x PS/BPS + 0.1) + 0.57x P (5)

Investment Cost = 36.81 x (0.9 PS/BPS + 0.1) (6)

Cost Unit: Million RMB per year PS: Plant Scale BPS: Base Plant Scale (300,000 tonnes per year) P: Production of Coke

In the previous work done by our team, the land cost has not been taken

into consideration. On our field trips, I noticed that non-recovery technology

consumes more land than large-machinery technology, although the operational

cost and investment cost of non-recovery coke-oven technology is much lower than that of large-machinery technology. From the interviews, I know that the average annual opportunity cost of the land for farming in Shanxi is 700 RMB

($ 84.3) per Mu. 3 For a typical non-recovery coke-oven plant of 500,000 tonnes per year capacity, the annual opportunity cost of the land of this plant used for farming is therefore 56,000 RMB ($6,747). Considering the "economies of scale" of cokemaking technologies (Kraines et al., 2001), I use Equation (7) to estimate the land cost of non-recovery coke-oven plant. For a large-machinery coke-oven plant of 1,230,000 tonnes per year capacity, the annual opportunity cost of the land of this plant used for farming is 84,000 RMB ($10,120). I use Equation (8) to estimate the land cost of large-machinery coke-oven plant.

For non-recovery coke-oven plants:

Annual land cost (RMB) = 56,000 x (0.9 x PS/500,000 + 0.1) (7)

For large-machinery coke-oven plants:

Annual land cost (RMB) = 84,000 x (0.9 x PS/1,230,000 + 0.1) (8)

PS: Plant Scale

5.2 GIS Modeling, Programming and Processing

To connect different parts of the SPGPSS, I need to consider the

database processing and model programming. To incorporate the professional

models into the SPGPSS system, I process the SPGPSS database to match the

data requirements of the professional models. The transportation NETFLOW

model needs to read six specific items of data to do the optimization

3 RMB is the Chinese currency unit. 1 RMB = 0.121 US Dollar. Mu is the Chinese area unit for land, 1 Mu = 0.165 Acre. computation, so that the SPGPSS database should have the link name, from- node, to-node, transportation cost, capacity (upper bound) and lower bound for each transportation link. In addition to those six items of data, I also put the transportation energy consumption and emission data for each transportation link into the SPGPSS database. I calculated those data from the transportation formulae presented in Section 5.1.

After processing the SPGPSS database for the specific professional models, I write the scripts (some paragraph of programming code to implement some specific functions) by ArcView's programming language, Avenue. The first script orders the SPGPSS to retrieve the needed data from the database stored in the SPGPSS, format the data according to the requirement of professional model, and write the data into the model, and execute the model. The transportation NETFLOW model generates the optimized result in a data file, which gives columns of numbers to tell which transportation links should have how many transportation flows. Although users can get the optimized results from this data file, it is very difficult for users to understand those results without showing these transportation routes and flows in GIS maps. The second script therefore reads the results back to the system and update the old database. The maps in the SPGPSS viewers automatically show the updates and changes in database. Script programming is a very important step in the whole SPGPSS creation. Only by those scripts can the SPGPSS utilize the professional models to implement some complicated tasks. With the results generated by the professional models, users can conduct further spatial analyses with the SPGPSS by using the tools embedded in the

ArcView software. By turning on or off the different themes with different spatial data, such as transportation by roads or by railways, users can clearly see the transportation flows carrying by road or by railway. With the identify tool, users can select any transportation link from the map and obtain from a table the transportation cost, energy consumption, and emissions of this link. From the legends by transportation flows, users can easily see from the map viewer how many transportation flows are carried by each transportation link. Chapter 6 APPLICATIONS

This chapter shows the results and analyses from the applications of

SPGPSS. According to the different data requirements, I designed SPGPSS specifically for two levels of research. On the first level, the provincial level, I run the SPGPSS to conduct alternative analyses and compared the results to examine the impacts from the changes in prerequisites. I focus more on the relative changes between alternatives than on the absolute results of each alternative and explore the reasons behind those changes. On the second level, the plant level, I use the detailed survey data and SPGPSS to conduct valuations and comparisons on the choices of location, transport routes and modes and coke-oven technologies for the individual cokemaking plant in Shanxi Province.

6.1 Analysis of Alternatives at the Provincial Level

Based on the year 2000 production and distribution data of coalmines, cokemaking plants, and coke consumers in Shanxi Province (China Statistics Press, 2001), I conduct analyses of alternatives to minimize:

1) Total costs from transportation and cokemaking plants;

2) Total energy consumption from transportation and cokemaking plants;

3) Total pollution emissions (PM and SOx) from transportation and cokemaking plants.

For each minimization, I run the SPGPSS to test the three scenarios described in Analysis of Alternatives of Section 4.2. They are 2000 Base Scenario, 2000 Transport-Minimization Scenario (2000 Transport-Min Scenario), and 2000 Plant-Minimization Scenario (2000 Plant-Min Scenario). The 2000

Base Scenario provides the optimized transport routes and flows based on the year 2000 production and distribution of coalmines, cokemaking plants and coke consumers. The 2000 Transport-Min Scenario provides the optimized transport routes and flows with the assumption that each plant can expand their production capacity up to 3,000,000 tonnes per year, the upper limit for individual plant coke production. This scenario can give the lowest transportation cost, energy consumption or pollution emissions. The 2000 Plant-Min Scenario provides the optimized transport routes and flows with the assumption that Shanxi Province will have several plants of 3,000,000 tonnes per year capacity to satisfy the total coke demand. This scenario can give the lowest plant cost, energy consumption or pollution emissions due to the economies of scale.

6.1.1 Minimize Total Cost

By running the SPGPSS to get the optimized transportation routes and flows for the three scenarios, I calculate the transportation cost for each scenario.

I use the plant-cost model, described in Section 5.1, to estimate the plant cost based on plant-production data in 2000. The total cost is the summation of the transportation cost and the plant cost.

(1) Total cost comparison

I assume that one of two cases of coke-oven technology is in effect: one case is that all the plants are using non-recovery coke-oven technology (Table 6.1); a second case is that all the plants are using large-machinery coke-oven

technology (Table 6.2). These two coke-oven technologies are currently

considered by the local environmental officials and plant managers to be the

most promising technologies in the cokemaking industry in Shanxi Province,

although they both have advantages and disadvantages in different aspects

(AGS MRP Field Notes, 2002). To increase production capacity and reduce

environmental pollution, plant managers in Shanxi Province are adopting the

non-recovery and large-machinery technologies as the major cokemaking

technologies.

Table 6.1: Total Cost of Non-recovery Coke-oven Technology Scenarios 2000 Base 2000 Transport-Min 2000 Plant-Min Percentage of Percentage of Percentage of Cost total cost Cost total cost Cost total cost Transportation cost 1104 15% 574 5% 1341 19% Plant cost 6456 85% 11536 95% 5768 81% Total cost 7560 100% 12110 100% 7109 100% Unit: Million Renminbi Source: Author

Figure 6.1: Total Cost of Non-recovery Coke-oven Technology

Total Cost of Non-recovery Coke-oven Technology

14000 ' 12000 10 MTransportation cost 10000 a 8000 N Plant cost

6000 D Total cost 4000 2000 0 Scenarios 2000 Base 2000 Transport-Min 2000 Plant-Min

Source: Author Table 6.2: Total Cost of Large-machinery Coke-oven Technology Scenarios 2000 Base 2000 Transport-Min 2000 Plant-Min Percentage Percentage Percentage of Cost of total cost Cost of total cost Cost total cost Transportation cost 1104 3% 574 1% 1341 5% Plant cost 33251 97% 94403 99% 26351 95% Total cost 34355 100% 94977 100% 27692 100% Unit: Millions Renminbi Source: Author

Figure 6.2: Total Cost of Large-machinery Coke-oven Technology Total Cost of Large-machinery Coke-oven Technology

100000 90000 80000 70000 Transportation C 60000 cost W500000 U Pant cscost 40000 OTotal3 cost 30000 2 20000 10000 0 Scenarios 2000 Base 2000 Transport-Min 2000 Plant-Min

Source: Author

I find that the 2000 Plant-Min Scenario has the lowest total cost in both

cases of coke-oven technology (Tables 6.1 and 6.2, Figures 6.1and 6.2). In other

words, the Shanxi Province planners could refer to these plant locations and their

transport routes and modes in the future planning to reduce the costs both from

transportation and cokemaking process. Although the transportation cost in the

Plant-Min Scenario is the highest among the three scenarios, this scenario has a

much lower plant cost than the other two scenarios. In the case of non-recovery

coke-oven technology, the plant cost inthe Plant-Min Scenario is only 50% of the plant cost in the Transport-Min Scenario, and 89% of the plant cost in the Base

Scenario.

Comparing the two coke-oven technologies, I also find that the plant cost is 85% of the total cost inthe non-recovery case, and this percentage is even higher in the large-machinery case (Tables 6.1 and 6.2). That is why the Plant-

Min Scenario has the lowest total cost due to its large savings inthe plant cost part. We can see that the plant cost is the determining factor in the total cost due to its much high percentage in the total cost. The scenario with the lowest plant cost would have the lowest total cost. Because the Plant-Min Scenario has the lowest plant cost, it finally needs the lowest total cost.

Due to the economies of scale, I find that the choice of fewer large- capacity plants in the Plant-Min Scenario (Figure B.1) has much lower plant costs than more scattered small-capacity plants in the other two scenarios. In the

Transport-Min Scenario, the plants are dispersed to be near suppliers, consumers, or major transportation routes to lower transportation cost, so that the plant cost of this scenario is the highest due to highest original investment costs of many small-scale plants in three scenarios. From the perspective of total cost minimization, the Plant-Min Scenario is the best, which actually supports the hypothesis that the large-capacity

cokemaking plants instead of the distributed small-capacity plants reduce the total cost from transportation and cokemaking process. (2) Transportation cost comparison

From the results of transportation costs, I compare the road transportation cost with the rail transportation cost. In the 2000 Base Scenario, in terms of transport flows, the road transportation accounts for 75% of the total transport flow and the rail transportation accounts for 25% (Table B.1). In terms of transport cost, the percentage of the road transport cost is 78% and the railway transport cost is 22% of the total transportation cost. So more coal and coke are transported by road than railway, due to the limited capacity of the railway transportation and the convenience and directness of the road transportation

(Akatsuka, 2001). The rail transportation comprises 25% of the total transport flows, but it only accounts for 22% of the total costs, which implies that railway transportation is, on average, slightly less expensive than road transportation.

In the 2000 Base Scenario (Table B.2), the coal transport flow is 20% of the total transport flows and much less than the coke transport flows, which is

80% of the total transport flows. The coal transport cost is only 27% of the total transport cost and also less than the coke transport cost, which is 73% of the total transport costs. In the optimized arrangement generated by the SPGPSS,

cokemaking plants locate nearer to the coalmines than to the coke consumers.

This pattern consequently reduces the coal transport flows and costs more than

the coke transport flows and costs. The optimized result also indicates that the

transport flows and costs can be more effectively reduced if the cokemaking

plants can be relocated nearer to the coalmines. (3) Plant cost comparison

In the case of non-recovery coke-oven technology in 2000 Base scenario

(Table 6.1), I find that the determining factor of plant cost is the operational cost, which accounts for 96% of the total plant cost. The land cost contributes a very trivial part (0.1%) to the total plant cost. As noted earlier, I use the annual opportunity cost of land used for farming to calculate the land cost. In China, all the land is owned by state. Households or enterprises only own the use right of the land. As the cokemaking plants are usually located in the suburban and rural areas, the land used by those plants was usually farming land before the peasants constructed the cokemaking plant. After the economic reform in 1978, many Town and Village Enterprises (TVEs) have emerged in the countryside.

They usually converted the land of which they own the use right from farming to the other industrial uses. It is relatively difficult to value the use right of those converted lands, because China is still in the process of land and property-right reforms, and there is no established system to value the converted land and the use rights. Because I could not obtain the data of the values of those land and use rights, I chose to use the annual opportunity cost of the land used for farming to calculate the cost of the use of those lands for coke plants. I assume that this value is far too low, but I could not determine a more appropriate measure based on the current available information.

Because of the abundant coal resources and the high quality of the coal in

Shanxi Province, the profit from cokemaking is much higher than from farming work. That is why so much land has been converted to cokemaking use and why many small cokemaking plants exist in Shanxi Province. In 1998, there were at least 1500 plants in the 5-county region in which we conducted the survey. Even today, after many plants have closed, there are at least 900 plants still in operation (AGS MRP Field Trip Interviews, 2002). Farmers have easy access to high-quality coal, which is often within 25-50 kilometers of their villages. With the implementation of national environmental regulations, many local small cokemaking plants were forced to close. These closings directly reduced the income of TVEs and local governments. In an interview during our 2002 field trip in Shanxi Province, the governor of XF County (names withheld for confidential reasons) told us that the output from the cokemaking industry was about 25% of the total output inthis county. To implement the national environmental regulations, the local officials closed about 230 low-quality modified indigenous coke plants in 2001 and about 30 high-quality modified indigenous coke plants in

2002. The total loss from closing those cokemaking plants was about 500 million

RMB (about $60 million), which greatly reduced the total output and average income of XF County. Although these forced closures can help to reduce the local environmental pollution, the hardships incurred by the local residents cannot be ignored. 6.1.2 Minimize Total Pollution Emissions and Energy Consumption

To estimate the minimization of the pollution emissions and energy consumption, I run the SPGPSS under three scenarios and calculate the transportation pollution emissions and energy consumption for each scenario. I use the Particulate Matter (PM) and SOx as the major measurements for emissions both from transportation and cokemaking plants. The PM refers to particles smaller than 10 microns in diameter, and SOx refers to S02 and other compounds in the atmosphere formed by a combination of sulfur and oxygen.

Due to the lack of the NOx data from the cokemaking process, I do not use the

NOx level as a major pollution measurement for transportation and cokemaking plants.

The PM emissions aroused the attention of national policy makers in

recent years due to their global and regional influence on radioactive forcing and

its local effects on the environment and human health. China has high PM

emissions due to the high usage rates of coal and bio-fuels (Rogers et al., 1997).

In the latest five field trips to Shanxi Province, members of our research team found that the PM emissions from heavy-diesel trucks are even greater than the

PM emissions from the cokemaking plants (except directly at the quenching car). My choice of PM emissions as the major pollution measurement is therefore

appropriate when I consider estimating the pollution both from transportation and

cokemaking plants.

I calculate the PM and SOx emissions from transportation and

cokemaking plants as described in Section 5.1. Because the PM and SOx emissions from cokemaking plants are the same in the three scenarios due to the same total coke production, the PM and SOx emissions from transportation determine the order of the total PM emissions in the three scenarios. The

Transport-Min Scenario gives the least PM and SOx emission (Tables B.3 and

B.4). Consequently, the total PM and SOx emissions in the Transport-Min scenario are the least among the three scenarios. In the comparison of energy consumption among the three scenarios, the Transport-Min Scenario also has the lowest transportation energy consumption (Table B.5).

From the emission and energy consumption perspective, I discover that the best scenario is the Transport-Min Scenario, which has the lowest total PM or

SOx emissions both from transportation and cokemaking plants. Compared with the other two scenarios, the plants in the Transport-Min Scenario are distributed to be closer to the suppliers, consumers, or major transportation routes (Figure

B.2), which effectively reduce the transportation distance, consequently, the transportation emissions and energy consumption. This pattern actually opposes my hypothesis that large-capacity cokemaking plants instead of the distributed small-capacity plants will reduce the total emissions or energy consumption from transportation and cokemaking plants. The more flexible distribution near the supplier, consumers, and major transportation routes makes small-capacity

plants release less transportation emissions and consume less energy than the

merged large-scale plants, which cannot access the resources, markets or major

transportation routes as conveniently and efficiently as those small plants. 6.1.3 Coke-Oven-Technology Impact Analysis

The type of coke-oven technology a plant manager selects has large impacts on the plant cost, energy consumption, and pollution emissions. As discussed in Section 4.2, after the implementation of the 1997 # 367 environmental regulation, there are two major recommended coke-oven technologies, large-machinery coke-oven technology and nonrecovery coke-oven technology. I suppose three options of coke-oven technology: Option 1 is the non--recovery technology, Option 2 is the large-machinery technology, and Option

3 is the mixture of non-recovery and large-machinery technologies. I assume that the cokemaking plants produce the same amount of coke annually in each of these three options.

I find that the plant cost in Option 1 is the lowest of the three options,

which is only about 20% percent of the plant cost inOption 2 (Table 6.3). The

major reason for this big gap is that Option 1 has much lower operational and

investment costs, although the land cost of Option 1 is higher than the land cost

with Option 2. The land cost, however, only accounts for a small part of the total

plant cost, 0.1% in the Option 1 and 0.01% in Option 2. Consequently, the plant

cost for Option 1 is still much less than that for Option 2 and Option 3. Table 6.3: Plant-Cost Comparison of Three Coke-oven Technology Options Operational Investment Land Plant Cost Cost Cost Cost Option 1 (Non-recovery technology) 6206 244 6 6456 Percentage of plant cost 96.2% 3.8% 0.1% 100% Option2 (Large-machinery technology) 24673 8573 4 3325C Percentage of plant cost 74.21% 25.79% 0.01% 100% Option 3 (Non-recovery & Large-machinery) 18512 4730 5 2324 Percentage of plant cost 79.63% 20.35% 0.02% 100% Unit: Million Renminbi per year Source: Author Figure 6.3: Plant-Cost Comparison of Three Coke-oven Technology Options Plant-Cost Comparison of Three Coke-oven technology Options

30000 U Operational Cost 25000 * Investment Cost 20000 3 Land Cost E 15000 e 10000 5000 0 Options Option 1 Option2 Option 3

Source: Author

The operational cost has the greatest impact on the plant cost (Figure 6.3).

The Option 1 has the highest percentage of operational cost in the total plant cost, because the Option 1 is more labor-intensive compared with the other two options. The investment cost is 4% of the total plant cost in Option 1. This percentage increases to 26% in Option 2. In Option 1, the non-recovery technology does not recover the by-products, and plants do not purchase and install the equipment and facilities for by-product recovery and pollution abatement. By contrast, in Option 2, the large-machinery technology requires

more initial investment inputs for the advanced machinery equipments and facilities for by-product recovery and pollution abatement. But recently, the

cokemaking industry has shown the trend to use both two technologies together.

Some non-recovery cokemaking have changed to more mechanical processes,

such as using transferring belts in coal loading and unloading. Some large-

machinery plants near Linfen are trying another simplified large-machinery technology, which does not recover by-products and does not need equipment and facilities installed for by-product recovery and pollution abatement (AGS

MRP Field Trip Notes, 2002). This trend can take advantage of the two technologies and achieve the production improvements by reducing cost, improving energy efficiency, and abating pollution.

6.1.4 New Highway Construction and Speed Improvement Impact Analysis

The highway system has developed quickly in Shanxi Province. I obtained the information presented here concerning future highway plans from the interview with one official in the Shanxi Province Transportation Planning

Department (AGS MRP Field Trip Interviews, 2002). With Taiyuan as the transportation hub, the highways in Shanxi Province form a road network linking all the counties in the province (Figure B.3). In the north-south direction, the

Datong- Highway is a major highway connecting the cities in south and north of the province. The new Taiyuan-- Highway will be built from 2005 to 2008. In the west-east direction, the Taiyuan-Jiuguan Expressway, which joins the - expressway, connects Beijing--

Tanggu expressway and Beijing- expressway, and leads to Beijing and the region of Bohai Sea rim directly. This highway will extend westwards from

Taiyuan to Lishi-Liulin area when several new highways are built in the next five years. In the Tenth Five-Year Plan, Shanxi Province will improve its highway transportation system and plans to invest at least RMB 10 billion on new highway construction in the next five years (China Tenth Five-Year Plan, 2001). During the interview, the official inthe Shanxi Transportation Planning Department told me that besides the new highway construction, another great change recently is that road conditions have been improved, especially the roads in the towns and villages. The average speed limit on the roads in towns and villages has increased from 30 kilometers per hour in 1990 to 40 kilometers per hour today.

Based on the map of Shanxi Province major highway construction in the tenth five-year plan provided by Shanxi Province Development and Planning

Committee, I added two major new highways inthe SPGPSS, Taiyuan-

Changzhi-Jincheng Highway and Taiyuan-Lishi-Liulin Highway and increased the speeds of the town-village level roads from 30 kilometers per hour to 40 kilometers per hour. This speed improvement can increase fuel efficiency from

0.056 liter/km-tonne to 0.043 liter/km-tonne for transportation on those roads

(Akatsuka, 2001). After running the SPGPSS inthe different scenarios, I obtained the results in Table 6.4.

Table 6.4: Comparisons Before and After the New Highway Construction and Road-speed Improvements

Transportation Cost (Million Renminbi/year) Scenarios Before After Decrease Percentage 2000 Base 1104 1082 2.0% 2000 Transport-Min 574 556 3.1% 2000 Plant-Min 1341 1322 1.4% Transportation PM Emissions (kilogram/year) Scenarios Before After Decrease Percentage 2000 Base 1,622,440 1,596,488 1.6% 2000 Transport-Min 802,219 788,258 1.7% 2000 Plant-Min 2,181,582 2,154,0161 1.3% Source: Author The transportation cost and pollution emissions are all decreasing after the new highway construction and road-speed improvements. In the three scenarios, the Transport-Min Scenario has the biggest decrease and the Plant-

Min Scenario has the least decrease both for transportation cost and emission pollutions, because the Transport-Min Scenario can more effectively take advantage of new highway construction and road-speed improvements to reduce the cost and emissions.

6.1.5 Industrial-Park Location-Choice Analysis

In the future provincial planning, Shanxi government officials plan to build two cokemaking industrial parks in the province to increase economies of scale and reduce environmental pollution. They have tentatively selected locations in the Lishi-Liulin, Linfen, and/or Jiexiu areas. These three places are currently the

major cokemaking industrial areas in Shanxi Province. By using the SPGPSS, I

make simulations of the several location arrangements of those industrial parks and compare the transportation cost, energy consumption, and emissions from them.

As noted in Section 4.2, I assume three combinations of cokemaking

industrial park locations: (1) Lishi-Liulin and Linfen, (2) Linfen and Jiexiu, and (3)

Lishi-Liulin and Jiexiu. In these three scenarios, I assume that each industrial

park produces half of the current total coke production of Shanxi Province. I

show the SPGPSS results in Table B.6 and maps in Figure B.4. When I compare the three combinations of cokemaking industrial parks, the Linfen-Jiexiu scenario has the lowest cost, energy consumption, PM emission, and SOx emissions from coal and coke transportation. Based on this simulation analysis by the SPGPSS, I would recommend Shanxi government choose to build two cokemaking industrial parks in the Linfen and Jiexiu areas.

From the three maps of each industrial park combination, I find that in the Linfen and Jiexiu scenario, the industrial parks locate nearer to the big coal suppliers and major highways and railways than in the other two scenarios, which consequently reduces the transportation cost, energy consumption, and emission pollution. Of course, additional factors should be taken into consideration for the industrial-park location decision, but I have used this to illustrate the type of quantitative comparisons policy makers and plant managers can conduct by

making these types of simulations with the SPGPSS.

6.2 Plant Case Studies

At the plant level, I did a typical plant case study of multi-plan valuations

and comparisons in terms of location choice, transport routes and modes and coke-oven technologies. For confidential reasons, I call this plant X cokemaking

plant.

6.2.1 Choose transport routes and modes

Taking advantage of the friendly GUI in the SPGPSS, coke managers can

choose the coal and coke transport routes for their companies. After they select the railways and roads they want to use for the company's transportation, those transport routes will be highlighted. Then, by using the statistical tool embedded

in the SPGPSS, coke managers can get the total cost, energy consumption, and emissions for transporting one tonne of coke.

I give an example of how to use the SPGPSS in the analysis of the coke transportation routes for X cokemaking plant. First, from the 2000 AGS MRP

WE Survey, I find that all the coke X plant produced in year 2000 is exported to the United of States, Germany, Japan, and South Africa. The coke transportation mode can be by truck or by railway. From the 2000 field-trip interview, I learned that X plant currently transports almost all the coke by train to

Tianjin Port, from where it is shipped overseas. The major railway they use in

Shanxi Province is the Shijiazhuang-Taijiu Line, as highlighted in Figure 6.4.

Figure 6.4: Coke Transportation Choices of X Cokemaking Company By Railway By Highway

Source: Author Due to the limited railway capacity, the continuing production growth of X plant, and the rapid development of the highway system, I consider the alternative of transporting X plant's coke by truck on the highways. If all the coke of X plant is transported by the roads (highways are highlighted in Figure 6.4), the corresponding cost, energy consumption, and emissions for transporting one tonne of coke by railway or by truck are shown in Table B.7. For the coke transportation of X cokemaking plant, railway is a much better choice than truck, especially from the aspects of energy consumption and pollution emissions (Table B.7). Currently, the cost of railway transportation is less than road transportation, but the increasing price of railway transportation and the limited capacity could restrict the use of railway transportation to some extent. Because of using heavy diesel trucks for road transportation, the energy consumption in the all-by-truck case is about 10 times that in the all-by-train case, the PM emissions is about 6 times, and the SOx emissions is about twice. By contrast, the coke export by railway can effectively reduce the pollution emissions from transportation.

6.2.2 Selection of a new location In the location-choice analysis, the manager of the X cokemaking plant plans to move the plant to a new location or expand the plant capacities in place nearer to suppliers. I assume the plant would not change its suppliers

(coalmines) and coke consumers. From the 2000 AGS MRP TVE survey, I find that the X cokemaking plant has four major coal suppliers, two of them are located 30-40 kilometers away from the current plant, and another two are located 80-100km away from the current plant. The plant transports all of their coal from the suppliers by truck. From the comparison of the old location scenario (Table B.8) and new location scenario (Table B.9), I determine that by

relocating the plant to the place of one supplier, which stays nearest to the center

of the road system (Figure 6.5), the cost, energy consumption, and pollution

emissions of the X plant's coal transportation all decrease. The SPGPSS system

efficiently gives the evaluations of transportation cost and other measures according to the locational choices of the users. To compare those results, users can decide which is the best locational choice from the transportation

perspective.

Figure 6.5: Locational Choice of X Cokemaking Plant

Source: Author 6.2.3 Choose coke-oven technology

I also can estimate the different plant cost with the different adoption of coke-oven technologies (Table B.10). Currently, X cokemaking plant is using non-recovery technology. In 2000, the plant scale was 600,000 tonnes. The plant production was 500,000 tonnes (AGS MRP TVE 2000 Survey).

By using the plant-cost formulae, I estimate the operational cost, investment cost, and land cost with the adoption of large-machinery coke-oven technology. I find that the adoption of the large-machinery technology is much more expensive than the non-recovery technology, although non-recovery technology consumes more land and has a higher land cost than the large- machinery technology.

Using the SPGPSS, I conduct two levels of research with the different data requirements. Based on the results of alternative analyses on the provincial level, I optimized the plant locations and transport flows in terms of the total cost, energy consumption and pollution emissions, and valued the impacts from coke-oven technologies, new highway construction and speed

improvements, and industrial park establishments. The SPGPSS also can help coke managers on the choices of individual plant location and transportation

issues. Chapter 7 CONCLUSION

I created a Shanxi Province GIS-based Planning Support System

(SPGPSS) for the transportation and industrial plant location studies of the cokemaking sector in Shanxi Province. By integrating database, map viewer, scripts, and professional models in the GIS environment, the SPGPSS is able to optimize plant locations, transport routes and modes under the different scenarios on the provincial level, and also compute the corresponding cost, energy consumption, and pollution emissions in the transportation process.

Policy makers and industrial organizations can utilize the SPGPSS to value the economic and environmental impacts from different policy possibilities and assist their planning decisions on location rearrangements and structural changes. On the plant level, I used the GIS functions and tools to conduct spatial analyses and

evaluations for an individual plant in the planning of transport routes and new

location. The coke managers can compare different transportation and location

plans for an individual plant and choose the best plan according to their

requirements. By simulating alternatives and further comparative analyses in the

GIS environment, the SPGPSS is capable of assisting and supporting real-time decision-making in the planning process.

By the applications of SPGPSS, I tested my hypothesis that combining

plants into several large-capacity plants or industrial parks is preferable to having them distributed throughout the region. From the perspective of total cost

minimization, the merged large-capacity cokemaking plants and industrial parks instead of the distributed small-capacity plants would reduce the total cost from the transportation and cokemaking process. From the perspective of total energy consumption and pollution emission minimization, however, the merger of dispersed small-capacity plants to large-capacity plants and industrial parks would increase the total energy consumption and pollution emissions. These conclusions are based on the assumptions and model optimizations I used in this study.

I also found that the type of coke-oven technology used has a great impact on the plant cost, energy consumption, and pollution emissions, which, consequently, can affect the plant location and transportation choices a plant manager makes. The transportation cost and pollution emissions both decreased after new highway construction and road-speed improvements were completed, especially in the transport-minimization scenario. In the industrial- park location analysis, the simulation results indicate that the choice of the Linfen and Jiexiu cokemaking parks would reduce transportation cost, energy consumption and pollutions more than the other two alternatives due to their closer access to big coal suppliers and major highways and railways. I also used the SPGPSS to assist managers at individual plants make decisions on their choices of location, transport, and coke-oven technology based on the detailed plant information from the surveys and interviews. APPENDIX A: Transportation Cost, Energy Consumption, and NOx Emissions for Each Transportation Link in the GIS system Kraines and Akatsuka in Tokyo University first started the transportation studies in Shanxi Province. They inserted into the GIS maps the locations of major coalmines, cokemaking plants, and coke consumers in Shanxi Province and the production, consumption, and export data as well as the capacities of major railways from the 1990 Energy Resources Atlas of Shanxi Province (Shanxi Committee of Atlas Compilation, 1994). They also developed the formulae to calculate the transportation cost, energy consumption, and NOx emissions for each transportation link in the GIS system (Kraines et al., 2001).

Table A.1: Transportation-cost Coefficients for Diesel Trucks Item Diesel Truck Units Reference unit capacity 10 tonnes (Survey Team 1999) weight factor () 0.5 - (Akatsuka 2001) operator wages (Cp) 0.75 RMB/tonne-hr (Kraines and Akatsuka 1999) fuel efficiency (Eff) 0.025 - 0.055 liter/tonne-km (Kraines and Akatsuka 1999) fuel cost (Cf) 2.7 RMB/liter (Kraines and Akatsuka 1999) vehicle cost 200,000 RMB (Kraines and Akatsuka 1999) vehicle lifetime 10 year (Kraines and Akatsuka 1999) investment cost (in) 0.05 RMB/tonne-km (Shanxi Province Statistics Office 1999b) loading cost 0.5 RMB/tonne (Akatsuka 2001)

Table A.2: Transportation Cost Coefficients for Diesel and Electric Trains Item Diesel Electric Units Reference unit capacity 2300 3000 tonnes (Shanxi Province Statistics Office 1999a) fuel efficiency (Eff) 0.005 0.0113 liter/tonne-km (Shanxi Province Statistics Office 1999a) contracted cost (Ct) 0.11 0.11 RMB/tonne-km (Kraines and Akatsuka 1999) Table A.3: Transportation Energy Consumption Coefficients for Diesel Trucks and Trains Item Diesel Truck Diesel Train Units Reference weight factor (j 0.5 (Akatsuka 2001) fuel efficiency (Eff) 0.025 - 0.055 0.005 liter/tonne-km (Kraines and Akatsuka 1999) diesel heat value(Hd) 9.2 9.2 Mcal/liter (Kraines and Akatsuka 1999)

Table A.4: Transportation Energy Consumption Coefficients for Electric Trains Item Electric Train Units Reference power efficiency of electric trains (Eei) 0.0113 kWh/tonne-km (Shanxi Province Statistics Officel 999a) power efficiency of power plants (Pf) 0.30 - (Sadakata 2000)

Table A.5: Transportation NOx Emission Coefficients for Diesel trucks, )iesel Trains, and Electric Trains Item Value Units Reference NOx emission per unit engine power (Noe) 8.0 g/kWh (Faiz et al. 1996) engine power per heat value of diesel (Ef,) 0.15 - (Global Network 2001) heat value of diesel fuel (Hd) 9.2 Mcal/liter (Transport ation Ministry of Japan 1999) NOx emission of power plants (Nop) 0.976 g/Mcal (Bernstein et al. 1999) power efficiency of electric trains (Eei) 0.0113 kWh/tonne-km (Shanxi Province Statistics Office 1999a) power efficiency of power plants (Pf) 0.30 (Sadakata 2000)

Formulae to calculate the transportation cost, energy consumption, and NOx emissions

Truck transport cost [RMB / tonne] = Cf * Eff* di (1+ _) + C, * di / v + In* di

Train transport cost [RMB / tonne] = Ct* di

Diesel Truck and Train energy consumption [Mcal/ton] = Eff(v) VHd V d V (1+)

Electric Train energy consumption [Mcal/ton] = Efe Vdi Pf

Diesel train and truck transport NOx [g NOx / tonne] = Noe * Efc * Ef* Hd * d,

Electric train transport NOx [g NOx / tonne] = No, * Efe * di I Pf APPENDIX B: Application Results

Table B.1: Road Transportation vs. Rail Transportation (2000 Base Scenario) Transport Flow Transport Cost Percentage Percentage Road Transport 75% 78% Road Transport - Coal 15% 22% Road Transport - Coke 60% 56% Rail Transport 25% 22% Rail Transport - Coal 6% 5% Rail Transport - Coke 19% 17% Source: Author

Table B.2: Coal Transportation vs. Coke Transportation (2000 Base Scenario) Transport Flow Transport Cost Percentage Percentage Coal Transport 20% 27% Coal Transport-Road 15% 22% Coal Transport -Rail 5% 5% Coke Transport 80% 83% Coke Transport-Road 61% 56% Coke Transport - Rail 19% 17% Source: Author

Table B.3: PM Emissions from Transportation and Cokemaking Plants, 2000 Scenarios 2000Base 2000Transport-Min 2000Plant-Min Large-machinery technology - From Transportation 1,622,440 802,219 2,181,582 From Plant 83,237,500 83,237,500 83,237,500 Total PM 84,859,940 84,039,719 85,419,082 Non-recovery technology From Transportation 1,622,440 802,219 2,181,582 From Plant 33,295,000 33,295,000 33,295,000 Total PM 34,917,440 34,097,219 35,476,582 Unit: kilogram/year PM = particulate matter Source: Author Table B.4: SOx Emissions from Transportation and Cokemakin Plants, 2000 Scenarios 2000Base 2000Transport-Min 2000Plant-Min Large-machinery technology - From Transportation 1,198,883 597,763 1,514,819 From Plant 13,318,000 13,318,000 13,318,000 Total SOx 14,516,883 13,915,763 14,832,819 Non-recovery technology _ _ From Transportation 1,198,883 597,763 1,514,819 From Plant 59,931,000 59,931,000 59,931,000 Total SOx 61,129,883 60,528,763 61,445,819 Unit: kilogram/year Source: Author

Table B.5: Transportation Energy Consumption, 2000

Scenarios 2000 Base 2000 Transport-Min 2000Plant-Min Transportation Energy Consumption 792,347,689 370,842,199 1,058,481,736 Unit: 1000 kcal/year Source: Author

Table B.6: Comparison of Three Cokemaking Industrial Park Scenarios Scenarios Lishi-Liulin & Linfen Lishi-Liulin & Jiexiu Linfen & Jiexiu Cost (RMB/year) 1,885,228,826 2,160,621,076 1,883,268,865 Energy (1000kcal/year) 1,921,549,427 2,079,814,965 1,773,941,756 PM (kg/year) 3,492,874 3,995,512 3,372,174 Sox (kg/year) 2,252,170 2,552,780 2,208,906 Source: Author

Table B.7: Coke Transportation of X Cokemaking Plant Cost Energy PM SOx (RMB/1000-tonnes) (kcal/tonne) (g/tonne) (g/tonne) All by train 26387 9384 33 50 All by truck 72973 97544 191 112 Source: Author Table B.8: Locational Choice of X Plant: Old-location Scenario Cost Energy PM SOx (RMB/1000-tonnes) (kcal/tonne) (g/tonne) (g/tonne) Supplier 1 10311 11339 23 15 Supplier 2 9569 11606 23 15 Supplier 3 18630 22720 45 31 Supplier 4 14715 20096 38 25 Total 53225 65761 129 86 Source: Author

Table B.9: Locational Choice of X Plant: New-location Scenario Cost Energy PM SOx (RMB/1000-tonnes) (kcal/tonne) (g/tonne) (g/tonne) Supplier 1 0 0 0 0 Supplier 2 6069 7360 14 4 Supplier 3 15130 18474 36 25 Supplier 4 21733 29149 58 35 Total 42932 54983 108 64 Source: Author

Table B.10: Plant-cost Comparison of Different Coke-oven Technologies Non-recovery coke-oven Large-machinery coke-oven technology technology Operational cost 85,700,000 332,292,000 Investment cost 2,565,000 90,105,000 Land cost 66,000 45,300 Plant cost 88,331,000 422,442,300 Unit: Renminbi/year Source: Author Figure B.1: 2000 Plant-Min Scenario for Total Cost Minimization

Legends (1000 tonnes/year)

Cokemaking Plant 03000 Coal Mines A0 - 385 A 386-1358 g 359 -3000 A 3001 - 6058 Coke Users * 0-252 E253 -979 9980-2798 279-116895 Coal Road Transport Flow -500 500 - 1000 N 000 - 2000 000 - 3000 000O- 15590 Coal Rail Transport Flow A1 -500 00-1000 000-2000 000-3000 00-18857 Coke Road Transport Flow \/1-500 00-1000 000 - 2000 000-3000 000 - 15590 Coke Rail Transport Flow i/- 500 N 500 -1000 000 - 2000 000 -3000 W.-I...;iE000- 18857 =Towns S " Shanxi-border

Source: Author -vmnviwp-w

Figure B.2: Particulate Emissions in the 2000 Transport-Min Scenario

Legends (1000 tonnes/year)

Cokemaking Plant 03000 Coal Mines -A 0-385 A 386 -135 8 A 1359 -30 00 A 3001 - 6058 Coke Users S0 -252 S253 -979 S380 - 2798 2789 - 11695 Coal Road Transport Flow /-500400 -1000 "/000- 2000 S000 - 3000 000 - 15590 Coal Rail Transport Flow /\/1-500 /500 - 1000 "t000 - 2000 000-3000 000 - 18857 Coke Road Transport Flow b\/1-500 500 - 1000 "000 - 2000 000 - 3000 %W000-15590 Coke Rail Transport Flow i - 500 NV500 - 1000 "/000 -2000 ANW000 -3000 W ~ E 000O-16857 Towns N Shanxi border

Source: Author Figure B.3: Highway System in Shanxi Province

N

W+ E

S

Source: Author Figure BA: Different Industrial-Park Locations in the PM Emission Minimization Linfen and Jiexiu Scenario Legends (1000 tonnes/year) 1 Cokemaking Industrial Parks Coal Mines A 0 -385 A 386 - 135 8 A 135 - 3000 A 3001 - 6058 Fe' Coke Users * 0-252 M253-979 980-2798 2798- 11685 Coal Road Transport Flow 1-500

000-2000 000 - 3000 M - 15580 Jiexiu Coal Rail Transport Flow /V 00 --100000

"000 - 3000 000 -16857 Coke Road Transport Flow - 500 -1000 00-3000 - 15590 Coke Rai Transport Flow / i- 500 000 - 000 -3000 Linfen -is165 PETowns //Shnxiborder

Lishi-Liulin and Linfen Scenario Lishi-Liulin and Jiexiu Scenario

-Lishi-Liulin

Linfen "NJiexiu BIBLIOGRAPHY

Alliance Global Sustainability (AGS) Multiregional Planning (MRP) Team. 1999. 1998 AGS MRP Town ship and Village Enterprises Survey.

AGS MRP Team. 2000. 1999 AGS MRP State-owned Enterprises Survey.

AGS MRP Team. 2001. 2000 AGS MRP Township and Village Enterprises Survey.

AGS MRP Team. 2001. 2001 AGS MRP Field Trip Notes.

AGS MRP Team. 2001. 2001 AGS MRP Field Trip Interviews

AGS MRP Team. 2002. 2001 AGS MRP State-Owned Enterprises Survey.

AGS MRP Team. 2002. 2002 AGS MRP Field Trip Notes.

AGS MRP Team. 2002. 2002 Field Trip Interviews.

Akatsuka, Takeyoshi. 2001. Modeling and Evaluation of the Transportation Sector in the Coke-Making Industry of Shanxi Province, China (in Japanese). Master's Thesis. University of Tokyo.

Center of Human Ecology and Environmental Sciences (CUEH), University of Geneva. 1998. AIDAIR-GENEVA Project. The Internet. (http://ecolu- info.uniqe.ch/recherche/EUREKA)

Chen, Hao. 2000. Technological Evaluation and Policy Analysis for Cokemaking: A Case Study of Cokemaking Plants in Shanxi Province, China. Master's Thesis. Technology and Policy Program, Massachusetts Institute of Technology, Cambridge, MA.

China State Development and Planning Committee. 2001. China Tenth Five- Year Plan. China Peoples Press.

Department of Industry and Transportation Statistics, National Bureau of Statistics, P.R.China. 2001. China Energy Statistical Yearbook (1997-1999). China Statistics Press.

DiPasquale, Denise, and William C. Wheaton. 1996. Urban Economics and Real Estate Markets. Englewood Cliffs, NJ: Prentice Hall, pp. 91-123

Editorial Committee of Shanxi Fifty Years. 1999. Shanxi Fifty Years (1949- 1999). China Statistics Press. ESRI Inc. 1996. Using ArcView GIS. New York: John Wiley & Sons.

ESRI Inc. 1996. Avenue: Customization and Application Development for ArcView GIS. New York: John Wiley & Sons.

ESRI Inc. 1997. Understanding GIS, The Arc/Info Method. New York: John Wiley & Sons. ESRI Inc. 2000. Using Arc/MS. New York: John Wiley & Sons.

Fang, Jinghua. 2002. Email notes from Professor Fang Jinghua of Taiyuan University of Technology in December 2, 2002.

Gesellschaft f1 r Angewandte Fernerkundung (GAF) mbH Company. 1998. Bolivia - GIS-Based Decision Support System for Transportation Planning and Infrastructure Management. The Internet. (http://www.qgaf.de/bolivia-gis/)

Gittings, B. M., T. M. Sloan, R.G. Healey, S. Dowers, and T. C. Waugh. 1993. Meeting Expectations: A View of GIS Performance Issues, Geographical Information Handling. London: Wiley. pp. 33-45.

Healey, R. G. 1991. Database Management Systems, Geographical Information System: Principles and Applications. London: Longman. pp. 251-267.

Hua, Zugui. 2001. Presentation by Chinese National Coal Industry Import and Export Corp. Coke Summit 2001 (Intertech's 5* Annual International Coke Forum). Pittsburgh, PA.

Li, Shaojin, and Weiqing Shen. 1995. The Production and Pollution Protection of Modified Coke Ovens (in Chinese). Shanxi: Science and Technology Press.

Kennington, Jeff L., and Richard V. Helgason. 1980. Algorithms for Network Programming. New York: John Wiley.

Kraines, S. B., T. Akatsuka, L. Crissman, K. R. Polenske, and H. Komiyama. 2001. "Pollution and Cost in the Coke-making Supply Chain in Shanxi Province, China - Applying an Integrated System Model to Siting and Transportation Tradeoffs." Journal of Industrial Ecology. In Press.

Leedy, Paul D. 1989. Practical Research Planning and Design. New York: Macmillan Publishing Company.

MCNC. 2000. The Environmental Decision Support System. The Internet. (http://www.iceis.mcnc.orq/EDSS/EDSSPage.htm) Manheim, Marvin L. 1979. Fundamentals of Transportation Systems Analysis. Cambridge, Mass.: MIT Press.

Mejia-Navarro, M., and A. L. Garcia. 1995. Integrated Planning Decision Support System (IPDDS). The Internet. (http://www.lance.colostate.edu/-mario/ipds.html)

Onions, C. T. 1955. The Oxford University Dictionary on Historical Principles. Oxford: The Clarendon Press.

Polenske, Karen R., and Francis C. McMichael. 2002. A Chinese Cokemaking Process-flow Model for Energy and Environmental Analyses. Energy Policy 30 (2002). pp. 865-883.

Rees, John, and Howard A. Stafford. 1986. "Theories of Regional Growth and Industrial Location: Their Relevance for Understanding High-Technology Complexes." In Technology, Regions, and Policy, edited by John Rees. Totowa, NJ: Rowman & Littlefield. pp. 23-50.

Rogers, Peter P., Kazi F. Jalal, Bindu N. Lohani, Gene M. Owens, Chang-Ching Yu, Christian M. Dufournaud, Jun Bi. 1997. Measuring Environmental Quality in Asia. The Division of Engineering and Applied Science, Harvard University and the Asian Development Bank.

Shanxi Committee of Atlas Compilation. 1994. Energy Resources Atlas of Shanxi Province (in Chinese). Shanxi: Science Press.

Shanxi Statistical Bureau. 2001. 2001 Shanxi Statistical Yearbook. China Statistical Press.

Shanxi Transportation Facilities Office. 2001. Shanxi Province Atlas (in Chinese). Shanxi: Science and Technology Press.

Tolley, Rodney. 1995. Transport Systems, Policy and Planning: A Geographical Approach. New York: Wiley.

U.S. Department of Transportation (DOT). 1998. Portland Metro, Oregon - GIS Database for Urban Transportation Planning. The Internet. (http://tmip.tamu.edu/clearinghouse/docs/gis/metro/)

War Gaming and Simulation Center. 1998. The Planning Decision Support System (PDSS). The Internet. (http://www.ndu.edu/wgsc/USERGui.html)