This paper deals with the evaluation of environmental impact of rail and road transport in South Korea. A framework of energy input-output analysis is employed to estimate the total energy consumption and CO2 emission in acquiring and using a life cycle of passenger and freight transport activity. The reliability of CO2 emission based on uncertainty values is assessed by means of a Monte Carlo simulation. The results show that on a passenger-kilometers basis, passenger roads have life cycle emissions about 1.5 times those of rail, while that ratio is ten times greater when the scope of evaluation regards the tailpipe. In the case of freight transport, on a million tonkilometers basis, the value for road mode is estimated to be about three times compared to those of rail mode. The results also show that the main contribution of CO2 emission for road transport is the operation stage, accounting for 70%; however, the main contribution for rail transport is the construction and supply chain stage, accounting for over 50% emission.
In order to provide a clear and complete picture of energy consumption and greenhouse gas (GHG) emission for rail and road transport, activities related to energy-GHG from the supply chains of both systems should be evaluated. Life cycle assessment (LCA) is an acceptable tool for performing this assessment. This method requires specific and detailed data on target product or service from cradle to grave. Since rail and road transport are composed largely of parts, materials and processing, it is not an easy task to gather all data. To cope with this problem, input-output analysis, which is the only method and database now available for calculating life cycle energy and GHG emission of transportation in a consistent and integrated way, is applied. Miller and Blair  stated that the fundamental information used in the input-output table deals with national economic activity representing the flow of products from each industrial sector to each sector, itself and others. The contributions of many researchers have extended the input-output framework incrementally in the direction of employing physical units particularly material and energy flows. Yoshida et al.  applied an inputoutput table for the estimated life cycle of CO2 emission intensity of a specific passenger car. Lenzen and Treloar  employs an extended input-output framework to estimate and compare an embodied energy in building material of wood and concrete. Nakamura and Kondo  evaluated the alternative waste management for the recycling of end-of-life electrical home appliances using input-output and life cycle cost analyses. Facanha and Horvath  combined the economic input-output approach and the process LCA approach to assess the environmental load of freight transport in the USA. Crawford  applied input-output analysis to quantify embodied CO2 of timber and reinforced concrete railway sleepers. Acquaye and Duffy  estimated the energy and GHG emission intensities of Irish construction and its subsectors using the input-output analysis. Currently, with regards to work, the environmental burden of the transport supply chain can be estimated by applying the merits of input-output analysis.
The objective of this paper is to investigate the net energy consumption and GHG emission of passenger and freight on rail and road transport. Input-output analysis is the approach of this work. In order to increase the reliability of the results, uncertainty analysis using a Monte Carlo simulation was performed.
The goal of this study was to evaluate the energy and GHG emission of Korea’s rail and road transportation economic activities. The scope includes energy production, supply chain activ-ity and transportation activity. End of life management was not included due to lack of sufficient technical data. The function of transport was to deliver passengers and goods from the origin to its destination. The economic activities of rail and road transportation of the year 2005, including passenger and freight traffic volume, were set as the functional unit of energy and CO2 emission calculation.
Calculations of this study were based on primary data for the year 2005 from: 1) energy balance report surveyed by the Korea Energy Economics Institute (KEEI), 2) conventional input-output table prepared by the Bank of Korea (BOK), 3) passenger and freight revenue investigated by the Korea Railroad Research Institute (KRRI), and 4) traffic volume including driving statistics for private and public transport collected by the KEEI.
Embodied energy is the sum of direct and indirect energy. Direct energy is the amount of energy required to process an industrial activity, while indirect energy is defined as the amount of energy required to operate a given energy production system. To be able to estimate the embodied energy of the transport sector, including its supply chain, through the energy input-output approach, the following procedure is proposed. A basic energy input-output table, in which the transaction number between row and column identifies the energy value of output from the ith row which is used as input to the jth column, is first developed. Ton of oil equivalent (TOE) is the unit in the table. From this table, two forms of normalization energy matrices are established. G* is the energy normalization matrix of dimensions (12 × 12), in which each element expresses a proportion of the energy input per total energy output. E is a (12 × n) matrix that represents a hybrid unit for which, energy consumption of each industrial sector is normalized by its economic value. Moreover, n is the number of supply chain sectors (n = 10 for this paper). The total required economic purchase in all of the supply chain of the transport sector is determined using the Leontief formula i.e., (I-A*)-1F where I is the identity matrix, A* is the direct input coefficient matrix, and F is the vector of desired output. Therefore, the direct energy required by an industry that is processing for the
transport sector is calculated from E(I-A*)-1F. The term IG* E(I-A*)-1F represents the first level amount of energy to be processed by the energy production system. The infinite series of the energy supply chain can be acquired using the (I-G*)-1 term. So, (I-G*)-1 E(I-A*)-1F is a model for estimation of the total energy output required throughout the energy production system. Indirect energy is derived from the difference between total and direct energy.
The calculation of energy consumption through direct and indirect ways enabled us to evaluate a vector of embodied CO2. By doing so, the energy output was multiplied by a vector of CO2 equivalent per TOE unit of each fuel. The emission factors of CO2, CH4, and N2O for CO2 equivalent (eq.) for each fuel type mainly refers to the 2006 Intergovernmental Panel on Climate Change (IPCC) guidelines regarding stationary and mobility combustion. To be able to calculate the equivalent of CO2, the global warming potentials (GWP100) factors for characterizing the climate gases of CH4 and N2O were identified. IPCC , the GWP100 factors applied in this study were 21 and 310 kg CO2/kg, respectively. The factors used for greenhouse gas emission are presented in Table 1.
Furthermore, the energy conversion factors developed by KEEI [9, 10], were used to convert the mass unit of TOE to the energy unit of Joule equivalent. Then, the fuel or energy consumption in the TOE unit was easily computed to determine greenhouse emissions.
The most important implication and requirement for conducting the input-output analysis was that all minor principals of determinant (I-A*) and (I-G*) have to be positive. The (I-G*)-1 and (I-A*)-1, called the Leontief inverse matrix, used in this study have met this requirement.
The life cycle stage of the transport sector was divided into six main categories: 1) energy production, 2) raw material extrac
[Fig. 1.] Direct and indirect energy consumed by rail and road transport. a: raw material extraction, b: petrochemical product (processing base), c: petrochemical product (material base), d: non-metallic product, e: basic metal (processing base), f: basic metal (material base), g: textile, wood and pulp product, h: fabricated metal, i: manufacturing, j: construction, k: operation.
tion, 3) material preparation of petrochemical product, non-metallic metal, basic metal, textile and pulp product and fabricated metal, 4) vehicle manufacturing, 5) construction of transport facility such as bridge, and 6) operation. For this study, however, the amount of energy consumed during the energy production stage was expressed as the indirect energy for each life cycle stage. Fig. 1 identifies the quantitative relations between direct and indirect energy in line with the life cycle stage of transportation.
Indirect energy is of little significance in total energy consumption per year for the stage of petrochemical products (material base), basic metal (material base), manufacturing and construction. Conversely, a relatively large part of indirect energy is consumed by passenger rail operation, material preparation of fabricated metal, basic metal and textile products including raw material extraction. For example, indirect energy accounts for a large portion of consumption, i.e., 72%, of the fabricated metal preparation stage. In passenger rail operation, indirect energy accounted for 68% with respect to total requirements. Raw material extraction, textile and wood, including basic metals (processing base) accounted for 66.8%, 61%, and 55.5%, respectively. Even though direct energy in petrochemical product (processing base), non-metallic products, manufacturing, road passenger operation and road freight operation accounted for a great portion, indirect energy consumed during these stages provided a non-negligible total result. In summary, the results of this section point to the significant amount of indirect energy consumption versus direct energy consumption.
Fig. 2 presents a profile for the consumption of direct and indirect energy for rail and road modes for passenger and freight transport. These results clearly show that there was a significant amount of energy consumption from the rail transport’s supply chain and energy production activity. Energy consumption by direct or indirect activity was 0.839 and 0.974 million TOE/yr, respectively. In the case of road passengers, direct and indirect energy were 20.576 and 3.794 million TOE/yr, respectively. The direct was estimated at 0.356 million TOE/yr and indirect energy was about 0.167 million TOE/yr for rail freight transport. Road freight directly consuming energy at the point of use was 12.295 million TOE/yr and indirect energy required for delivery of goods per year was 2.915 million TOE.
In the present section, not only were the amounts of direct and indirect energy provided but also a detailed analysis regarding structural energy for both consumptions. Indeed, this was one of the advantages of the energy input-output approach developed from the current study. The type of direct energy used resulted in a significant amount and type of indirect energy at the energy production stage. For instance, as can be seen in Fig. 2, 0.11 million TOE of fuel oil and 0.21 million TOE of electricity were used to operate passenger rail transport per year. To deliver this amount of energy, approximately 0.67 million TOE was processed and converted by the energy production system. Of these, 0.24 million TOE of coal, 0.01 of crude petroleum, 0.08 of natural gas, 0.27 nuclear, 0.01 of renewable energy, 0.03 of fuel oil, and 0.02 of other energy sources are required. Likewise, freight rail transport was estimated to be using 0.17 million TOE of fuel oil and 0.01 million TOE of electricity in its operation. Calculation results showed that 0.07 million TOE was required for energy production. Passenger road transport consumed 13.291, 4.859, and 0.339 million TOE of fuel oil, other petroleum products (liquified petroleum gas, LPG) and gas, respectively, for its operation per year. According to these requirements, this model
[Fig. 2.] Life cycle energy consumption for rail and road passenger and freight transport. TOE: ton of oil equivalent. a: raw material extraction, b: petrochemical product (processing base), c: petrochemical product (material base), d: non-metallic product, e: basic metal (processing base), f: basic metal (material base), g: textile, wood and pulp product, h: fabricated metal, i: manufacturing, j: construction, k: operation.
reports that 0.172, 1.179, 0.414, 0.192, 0.317, and 0.289 million TOE of coal, crude petroleum, natural gas, nuclear, non-fuel oil and other energy sources, respectively, were consumed for operation for energy production. Considering the dimension of direct energy choice from other life cycle stages, for example, for the fabricated metals preparation stage and manufacturing stage, the requirement of electricity consumption for the fabricated metal sector give resulted in a great amount of indirect energy consumption at power generation, whereas the key direct energy consumption of the manufacturing stage were coal products, which require less indirect energy to produce. The results of this section attempted to demonstrate the influence of direct energy selection on indirect energy including total energy consumption. Moreover, the selection of energy strongly correlated with CO2 emission particularly energy based fossil fuels.
CO2 emission rate relates closely with energy consumption. To be able to address the environmental performance of rail and road, the total results for emission were divided by capita and traffic volume of passengers and freight transport.
Table 2 shows the result of CO2 calculation compared to population and traffic volume for the year 2005. According to International Transport Forum (ITF) , the South Korean population was around 48.92 million. As recorded by KEEI [9, 10] the traffic volume for the year 2005 was 45,087.2 million passengerkilometers (pkm) for rail and 560,363 pkm for road mode (estimated from value for private road transport of 477,146 pkm and for public road transport of 83,217 pkm). These numbers demonstrated the fact that passengers took a service of road transport greater than those of rail transport. In addition, these numbers implied that passenger prefer to use private transport rather than public transport. The traffic volume, which was expressed in million ton-kilometers (tkm) for road and rail mode, was 68,971 and 61,246 tkm, respectively.
As can be seen in Table 2, the first two rows compared the contribution of CO2 from tailpipe and from life cycle. Note that tailpipe was a consideration of fuel consumption and its contribution on CO2 emission at the operation stage. The difference between tailpipe and life cycle for rail transport was from 80% to 450% and from 24% to 32% for road transport. Although the majority contribution from both modes was from tailpipe, this study demonstrated the importance of CO2 in all life cycle stages. Passenger movement from source to destination by road mode emitted a total amount of CO2 seventeen times that of the rail mode. Goods delivered between sources by road mode caused an emission of CO2 33 times greater than that of rail mode. In regards to traffic volume, total rail and road passenger emission was 67.24 and 95.685 g CO2-eq./pkm, respectively. In other words, total road passenger emission was approximately 1.5 times larger than rail emissions on the pkm basis. Chester and Horvath  showed that emission was between 50-410 g CO2/ pkm for road and between 90-140 g CO2/pkm for rail transport. Considering the freight traffic volume, road mode produced emissions 2.1 times larger than rail mode due to road mode’s emitting on a 607.07 g CO2-eq./tkm basis, while the value for rail mode was 284.67 g CO2-eq./tkm. On a capita basis, rail mode generated 62.71 kg CO2-eq. while road mode emitted 1,110.78 kg CO2-eq. Prior research reported that road and rail modes emitted 78,330 and 820 million kg CO2-eq. for the year 2005, while results for normalization per capita were 1,622.07 and 16.98 kg CO2-eq./ yr for road and rail modes .
A crucial problem of energy input-output contributed to the existence of uncertainties in the utilized data. In other words,
uncertainty of data may cause a certain degree of error for energy and CO2 evaluation. To increase the reliability of the data used, a Monte Carlo simulation procedure, performed with software called Cristal ball (Oracle, Redwood Shores, CA, USA), was used to carry out this work. The uncertainty value, described by the standard deviation divided by the mean, was a term for the communication reliability of the result. To perform the Monte Carlo simulation, input parameters had to be specified for the uncertainty distribution. The Batch Fit tool, contained in the crystal ball, was employed for best fit probability distribution regarding each piece of technical data . Lognormal was the most appropriate type of distribution for the current study. The total number of trial runs performed using the Cristal Ball software was 10,000 times. Table 3 shows the distribution type, including the coefficient of variation or the uncertainty value acquired from simulation. The coefficient of variation (CV) was given by the ratio of the standard deviation to the mean IPCC . The small CV indicated the less variable and more stable.
Table 3 presents the CV of CO2 emission from each stage of the life cycle. In the case of passenger transport, the CV value of the total process by rail mode, estimated at 0.0553, was considerably small compared to the value for road mode, which was estimated at 0.0754. For freight transport, the CV value of rail and road modes yielded values of 0.0548 and 0.0801, respectively. Consideration of the uncertainty of most of the life cycle stages
gave small values from 0.0515 to 0.0903. The large uncertainty happened at the stage of raw material extraction, and at the preparation of the petrochemical product (material base) and the non-metallic metal stage. The uncertainty values for these stages were from 0.1886 to 0.1971 for the preparation of the nonmetallic metal stage, from 0.1033 to 0.1648 for the preparation of the petrochemical product (material base) and 0.1553 to 0.1877 for the raw material extraction stage. In order to reduce the uncertainty level, the number of industrial sectors aggregated in those life cycle stages should be lessened.
Fig. 3 depicts the distribution of CO2 emission induced by the passenger and freight transport for rail and road modes estimated from the Monte Carlo simulation. The X-dimension provides the CO2 emission per basis. The Y-dimension shows the probability of each value of the total CO2 emission. From Fig. 3(a-d), the certainty value ranges for a 95% certainty interval of passenger rail, passenger road, freight rail, and freight road for the year 2005 were 58.62-77.45 kg CO2-eq./capita, 945.01-1,289.98 g CO2- eq./capita, 55.91-69.71 kg CO2-eq./capita, and 646.89-906.26 g CO2-eq./pkm, respectively. Fig. 3(e-h) presents the certainty value range for a 95% certainty level of freight rail and road transport on a traffic volume basis. The CO2 emission was 62.86-83.11 g CO2/pkm for rail mode and 81.32-111.01 g CO2-eq./pkm for road mode. On the basis of tkm, emission for rail and road mode were 254.06-315.29 and 509.50-716.10 g CO2-eq.
Table 4 presents data detailing how each life cycle stage of rail and road contributed to the embodied CO2 emission for a 95% confidence interval on capita, pkm, and tkm base. Operation stage for passenger and freight on rail and road mode accounted for a large portion of the life cycle of CO2 emission, 47.28-48.73% for passenger rail, 88.54-88.92% of passenger road, 47.37-48.75% of freight rail and 78.88-80.14% of freight road. The construction stage accounted for 17.82-18.79% of passenger rail, 3.66-3.80% of passenger road, 17.85-18.81% of freight rail, and 6.81-6.85% of freight road. Material preparation of textile and wood products accounted for 8.80-9.37% of passenger rail, 2.24-2.46% of passenger road, 8.77-9.36% of freight rail, and 4.05-4.56% of freight road mode. Fabricated metal preparation accounted for 6.93-7.15% of passenger rail, 1.43-1.53% of passenger road, 6.93-7.14% of freight rail, and 2.56-2.80% of freight road mode. The construction phase of rail mode had large CO2 emissions compared to the other phases of the life cycle. It should also be emphasized that the impact from construction, including other passenger rail supply chain activity, accounted for a great portion of emission, 51.27-52.72% with respect to the life cycle. The International Union of Railways (UIC)  reported CO2 emissions shared by traffic infrastructure in different case studies of rail transport were 94.5% for the Switzerland-study and 65-72% for the US-study. Within this analysis, the main CO2 emission for road was operation; however, for rail transport, the main emis-
sion was the construction and supply chain. This result gave an idea related to previous research by UIC  and Hendrickson  which stated that although the rail mode tends to have a lower environmental impact compared to that of road mode, the environmental effects of the infrastructure and supply chain to accommodate such a shift are substantial if capacity expansion is required.
This work provided a more complete picture of energy and GHG emission for rail and road transport for the year 2005 which should enable practitioner to compare the results when the improvement regarding energy and emission performance is made. In this study, the input-output analysis and approach is used. Given the modeling approach used for the analysis, with regards to the first result, not only the amounts of direct and indirect energy were provided but a detailed analysis on structural energy for both consumptions was also given. Second, the results show that there were significant differences between tailpipe and life cycle in terms of emission analysis. In line with the comparison of total emission, rail mode generated 130.88 kg CO2-eq. while road mode emitted 1,879.86 kg CO2-eq. on per capita basis. In addition, on a traffic volume basis, the results showed that on a pkm basis, passenger road emitted life cycle emissions about 1.5 times greater than that of rail, while that value was ten times greater when the scope of evaluation was tailpipe. In the case of freight transport, on a tkm basis, it was estimated that road value was about three times that of rail. Moreover, the results also demonstrate that the main contribution of CO2 emission from road transport was in the operation stage, which accounted for 70%; however, it was the construction and supply chain stage which accounted for over 50% of the emissions concerning rail transport. This indicates that attention should also be focused on those supply chain activities having a potentially sizeable impact on CO2 emission, particularly passenger and freight rail transport. To quantify the reliability of the CO2 emission computation, the uncertainties of the CO2 emission of each life cycle stage for passenger and freight transport were analyzed by means of the Monte Carlo simulation. The uncertainty values for passenger and freight on rail transport were considerably small, being 0.0553 (5.53%) and 0.0548 (5.48%), respectively. In the meantime, the uncertainty values arisen from passengers and freight road transports were 0.0754 and 0.0801.