COMPARATIVE ANALYSIS OF DESIGN PARAMETERS FOR HIGH-SPEED RAILWAY EARTHWORKS IN DIFFERENT COUNTRIES AND A UNIFIED DEFINITION OF EMBANKMENT SUBSTRUCTURE

This paper compares design specifications and parameters for high-speed railway (HSR) earthworks in different countries (i.e., China, France, Germany, Japan, Russia, Spain and Sweden) for different track types (i.e., ballasted and ballastless), and for different design aspects (i.e., HSR


Introduction
High-speed railway (HSR) encompasses a complex reality involving many technical aspects including infrastructure, rolling stock and operations, as well as strategic and cross-sector issues including human, financial, commercial and managerial factors; it has proven to be a flexible system that can be developed under various circumstances and in different contexts and cultures (International Union of Railways [UIC], 2018a). The principal criterion in defining HSR is a commercial operating speed exceeding 250 km/h (UIC, 2016; The Council of the European Union, 1996). The implementation of an HSR project normally includes five phases; i.e., an emerging phase, feasibility phase, design phase, construction phase and operation phase. The design phase is one of the most important stages in HSR implementation. In this phase, definitions of all parameters and technical, architectural and landscaping choices necessary for the execution of works are precisely provided on the basis of applicable standards, rules and regulations, approved preliminary design documents, environmental impact assessments and reliability, availability, maintainability and safety requirements (UIC, 2018b).
Design practices of earthworks differ around the world according to specific regional/national standards. Alamaa (2016) compared regulations for the design of an HSR embankment in three European countries (i.e., France, Germany and Spain), but there has been no systematic and comprehensive comparative study on HSR earthwork design practices at the global scale. There is thus a need to enhance the safety and cost-efficiency of HSR earthworks by summarizing experiences and assimilating advanced technologies. In the present paper, earthwork specifications both in the countries where HSR technologies have been fully developed (i.e., France, Germany, Japan, and China) and in the countries where HSR technologies are developing (e.g., Turkey, Russia and Sweden) are identified, compared and analysed. The comparison and discussion focus on design parameters, material classification and other factors. The design speed, track type (ballasted or ballastless), traffic type (passenger only or a passenger-freight mix) and other factors are considered in the comparison and analysis. A unified definition of the stratified structure for the HSR embankment substructure is proposed. Since practical parameters of HSR earthworks are collected and summarized for various countries and regions and the railway substructure of an embankment is unified in the present paper, this research can serve as a reference not only for the countries lacking HSR construction experience applying HSR technologies, but also for engineers and researchers willing to develop a clearer understanding of different earthwork practices.

1.
Comparative analysis of design parameters of high-speed railway earthworks in different countries

Design service life of earthworks
The design service life is the expected service life of an earthwork element or how long the element can perform while meeting minimum functional requirements under a particular maintenance regimen. The design service life is determined by the expected working time, maintenance method, environmental conditions and life-cycle cost; it has direct implications for the operational safety and life-cycle cost (Lemer, 1996;Popović, Lazarević, Brajović, & Gladović, 2014). Some countries define design life by elements while others use the system-wide design life. Earthworks are relatively established systems that must perform consistently, as they are difficult to modify or replace. The design service lives of earthworks used in different countries are summarized in Table 1 for reference. The parameters for Spain and Sweden in the table are taken from Smekal (2012) and Alamaa (2016), respectively. Table 1 shows that the designed service life of earthworks is defined according subdivisional works in some countries, while an extensive value is defined in other countries, and the designed lifespan of main earthworks ranges from 70 to 120 years in different countries. It can be noted that the subgrade, retaining structures, and foundations are defined as the main earthworks in China, but the definition of main earthworks varies from country to country. The objectives of specific requirements towards the lifespan are therefore not exactly equivalent in different countries.

Compaction control criteria for embankment filling
By studying the data of National Railway Administration of P.  (2008), it has been concluded that the main compaction control indices for fill materials of the subgrade include the compaction coefficient (K) or proctor density (D pr ), reaction modulus (K 30 ), second load deformation modulus (E v2 ), dynamic deformation modulus (E vd ) and first load deformation modulus (E v2 /E v1 ).
In most countries, compaction criteria for embankment fillings are determined by factors including design speed, type of track, properties of fill material, filling positions. Having collected and analysed specifications concerning compaction requirements in the abovementioned standards, these parameters and their limit values are listed in Table 2 for comparative analysis. The definitions of the structural layer will be discussed in Section 3.
It is concluded that the parameters listed in Table 2 can be classified into two categories. The compaction coefficient (K) and proctor optimal density (D pr ) are closely related to the porosity of the fill material, which represents the physical compaction state of the filling. K and D pr are measured adopting different testing methods; i.e., K is measured adopting the heavy Proctor test or surface vibration test, while D pr is measured adopting the normal Proctor test. Meanwhile, the reaction modulus (K 30 ), dynamic deformation modulus (E vd ), second  load deformation modulus (E v2 ), and first load deformation modulus (E v2 / E v1 ) are obtained by relating the applied force and corresponding deformation so as to illustrate the mechanical properties of the compacted fill material. Combined parameters chosen from the two categories that represent both the compaction state and mechanical reaction properties are rationally adopted in different countries; i.e., K, K 30 and E vd in China, K and K 30 in Japan, D pr , E v2 and E vd in Germany, and D pr , E v2 and E v2 /E v1 in France.
The physical meanings of K 30 and E v1 are similar in that they both indicate the deformation modulus derived from the first load results of the static plate load test but with different testing equipment. E vd is obtained from dynamic plate load tests while E v1 and E v2 are obtained from static plate load tests. Dynamic plate load tests are easier to conduct with smaller and lighter equipment and are thus more suited to sites where minor compaction is permissible; e.g., sidewalk zones and trenches (Mikolainis, Ustinovičius, Sližytė, & Zhilkina, 2016).

Settlement control criteria
A central problem in HSR embankment engineering is to predict settlements and their development with time (Larsson, Bengtsson, & Eriksson, 1997). Deformation of the embankment mainly comprises elastic and plastic deformations (Sayeed, 2016;Wu & Chen, 2011). Elastic deformation is a recoverable deformation generated by the train load (Trenter, 2001). When the embankment is suitably filled and the ground treatment measures are reliable, the elastic deformation is generally negligibly small. Meanwhile, plastic deformation determines the time allowed for the track laying and the post-construction settlement value of the embankment (Orr & Farrell, 2011). The deformation of earthworks can also be classified as settlement during construction and post-construction settlement by different durations. Train operation safety is closely connected to post-construction plastic deformation (i.e., settlement), so post-construction settlement is strictly controlled within a specific range to ensure riding comfort and safety in different countries (Kang, 2016). Table 3 gives the post-construction settlement limit values used in different countries.
The limit value of post-construction settlement is basically gauged with the consideration of various factors including adjustable range for track deformation, engineering and maintenance costs for settlement control. Table 3 shows that three settlement limitation methods are adopted to control earthwork deformation; i.e., rate limitation,   Post-construction settlement for transition between earthworks and abutment: ≤10-30 cm.
If ground treatment measures are adopted, post-construction settlement: ≤10 mm per 10 years.
Post-construction settlement rate: ≤3 cm per year.
Post-construction settlement for transition between earthworks and abutment: ≤5 cm.
relative limitation (deflection angle) and total limitation methods. Total settlement limitation is specified in each country to control the settlement of ballasted and ballastless tracks. Meanwhile, rate limitation is adopted for the settlement control of ballasted lines in China, Germany, France and Russia and both ballasted and ballastless lines in Japan. Relative limitation is used for the settlement control of normal sections of both ballasted and ballastless lines in Germany, ballastless lines in China, Japan and Russia, and ballasted lines in France. Settlement limitation is used for transition sections of different types of structure in China, Japan and Germany.

Country
The settlement limit values

Germany
Post-construction settlement shall not be greater than fastening adjustment range minus 5 mm.

Relative limitation
Differential post-construction settlement shall ensure track at no point exceeds 1/500 from the reference length, e.g., differential post-construction settlement ≤2 mm within 10 m long section.

Total limitation
Differential post-construction settlement limit is determined by vertical curve radius larger than 0.4v 2 (v is the design speed), i.e., the limit value is 15 mm for a HSR line with the design speed of 300 km/h.

Relative limitation
Differential post-construction settlement shall ensure track at no point exceeds 1/500 from the reference length within a certain maintenance cycle (usually 6-10 years), e.g., differential post-construction settlement ≤2 mm within 10 m long section.
Total limitations 30 mm total settlements are accepted.
Total settlements shall not exceed 3 times the value of the differential settlements after commissioning within a reference length of 40 m.

Rate limitation:
Post-construction settlement rate within the first 25-years operation: ≤12 cm per year.

Different types of transition sections
Stiffness variations and relative displacements in a section of transition between different types of structure may lead to a localized dynamic load, greater levels of maintenance and geometric correction, increased costs of maintenance of the railway and changes to line operation (UIC, 2017). Transition sections are built for connection parts where there may be differential settlement and stiffness of the track foundation between earthworks and a bridge, tunnel or other structure beneath the railway track, between different earthwork structures and between regions subjected to different ground treatment measures (Ying & Lie, 2018). Different countries adopt different transition types and give different provisions for transition design. The design provisions of transition sections used in different countries are summarized in Table 4. Table 4 shows that the shapes of transition are trapezoids or inverted trapezoids in China and France but only trapezoids in Japan. The materials of transition sections are all made from soil (i.e., crushed stone, gravel and sand) and cement whose weight ranges are from 2% to 5%. Additionally, the required length of a transition section in China and Germany is greater than in France.

Worldwide definition of a stratified HSR embankment substructure
The components of a traditionally ballasted railway embankment can be divided into two subcategories: the superstructure and substructure (Bårström & Granbom, 2012). There are basically two categories of superstructure: ballasted and ballastless. A ballasted track uses a ballast layer to support track components and to transfer load. According to Esveld (2001), there are basically two designs principles for ballastless railway superstructures. The German slab track solution and design concept originates from experience in highway design (Esveld, 2001(Esveld, , 2003aLiu, Zhao, & Dai, 2011a). The reinforcement is placed on the neutral line of the slab and is cast onto a stiff bearing layer (Esveld, 2003b). The other method specifies the reinforcement to be placed on the top and at the bottom of the slab to increase the bending resistance (Esveld, 2001).
The substructure is the foundation of the entire railway structure (Li & Selig, 1995;Selig & Waters, 1994). In an attempt to realize the longterm stability of the structure affected by train loading, precipitation infiltration, dry-wet cycles, and freeze-thaw cycles, the substructure is designed to comply with the strength and stiffness requirements, settlement and seepage control requirements, and other special requirements; e.g., frost protection requirements in a permafrost region.

Proposal for a unified definition
A review of the literature reveals that the layers of embankment substructure are defined differently in different standards and academic articles, the same embankment substructure layer may be defined differently in different documents, while the substructure layer is defined to illustrate different layers. Substructure layers defined in one country do not match those defined in another as the whole railway structure is different. For a ballasted track railway, Giannakos (2010) defined that the blanket layer is part of the track bed layer and the track bed layer is part of the superstructure. In UIC (International Union of Railways) 719R (2008), however, a blanket layer is not regarded as part of the track bed layer, while it is regarded as part of the substructure, and the sub-ballast layer is defined as the overlap of the track bed layer and embankment substructure. Guler and Aksop (2017) consider a ballasted railway embankment substructure to be a combination of the natural soil, subgrade and sub-ballast.
By studying the documents of the National Railway Administration of P.  (2008), it has been found that even though HSR embankment substructures differ among different countries, a unified definition of the embankment substructure suitable for depicting ballasted and ballastless track railways can be obtained, as illustrated in Figure 1.
Generally speaking, the railway embankment substructure can be synoptically divided into three main parts -the prepared subgrade, upper part of the embankment and lower part of the embankmentaccording to our proposed definition. It is noted that functional layers (i.e., the blanket layer, frost protection layer and filtering layer) are included in our proposed stratified structure. These functional layers are explained in detail as follows.

Blanket layer
The blanket layer is generally laid on the top of the prepared subgrade. It provides the functions of concentrated stress reduction, frost protection, water proofing as well as seepage control (Bonnett, 2005). According to the French standards, the thickness of the blanket layer under the ballast is 20 cm or 35 cm (where the fill material is of Class S2 without geotextile). The German standards refer to a blanket layer as a protective layer whose thickness is 40 cm for the ballastless track and 70 cm for the ballasted track. According to the Japanese standards, blanket layers can be made of reinforced concrete, asphalt or graded gravel with respective thicknesses of 30 cm, 15/20 cm and 30 cm. In China, specifications (National Railway Administration of P. R. C., 2014, 2016) on general subgrade structures do not include a blanket layer; however, project tests on the laying of a blanket layer made of asphalt with a thickness of 10-30 cm have been carried out on the Harbin-Qiqihaer HSR (2014), Zhengzhou-Xuzhou HSR (2015), Beijing-Zhangjiakou HSR (2017) and Zhengzhou-Wanzhou HSR (2014).

Frost protection layer
The main role of the frost protection layer is to dewater the subgrade, thus protecting frost-sensitive geomaterials and preventing frost damage (Lichtberger, 2005). In a the seasonally frozen region, the frost protection layer is generally laid with a thickness determined by the frost depth. Materials of the frost protection layer commonly are characterized by high permeability (DB Netz AG, 1999;German Institute for Standardization, 1988;Japanese Standard, 1999;Liu, Niu, Niu, Lin, & Lu, 2011b;National Railway Administration of P. R. C., 2014. In UIC 719R (2008), the frost protection layer is defined as a specific structural layer above the prepared subgrade. From a boarder perspective, however, all rail substructure layers below the frost penetration depth must be designed so as to be protected against frost during the entire design life; e.g., fill materials are frost-resistance materials. With the reference to other standards and considering functionality, we recommend the frost protection layer be a combination of layers -a blanket layer, prepared layer (or part of it), and upper part of the embankment (or part of it)below the frost penetration depth, as shown in Figure 1.

Filtering layer
The most vital role of the filtering layer is to prevent intrusion of a filling layer with small grains into an adjacent filling layer with large grains. When grain sizes of two adjacent filling layers are unable to meet the requirement that D 15 < 4d 85 , a filtering layer is generally employed between them. Geosynthetics are commonly used as the materials for the filtering layers.

Layer thicknesses of the embankment substructure
The thicknesses of layers of the embankment substructure are determined to meet the deformation and strength requirements considering the defined quality of the fill material and compaction criteria. Owing to the diversity of embankment substructures, fill material compaction criteria and calculation approaches, the thickness of subgrade layers varies from country to country. The fill material and thickness of the embankment substructure in different countries are compared in Table 5. Table 5 demonstrates that on the whole, the total thickness of the prepared subgrade and upper part of the embankment is more than 2.5 m, while the thickness of the prepared subgrade is more than 0.4 m. A blanket layer is placed at the top of the prepared subgrade in Japan (with a thickness of 0.15-0.3 m), Germany (with a thickness of 0.4 m) and France (with a thickness of 0.14 m), while it is not required in China or Russia.

Design width of the embankment substructure surface
In most countries, the width of the embankment substructure surface is determined by the design speed, number of lines, type of track, distance between tracks, widening of a curved track, width of the substructure shoulder, requirements for maintenance, type of cable trough change to "  and the foundation of the poles of the overhead contact system and other factors (Cui, 2018). The width of the embankment substructure surface excluding the widening of a curved track in different countries is illustrated in Table 6. Table 6 shows that the width of the embankment substructure surface mostly ranges from 6.9 m to 9.0 m for a single-track railway and from 11.6 m to 14.5 m for a double-track railway. It has also been found that the width of the embankment substructure surface is usually larger for a ballasted line than for a ballastless line in China but that the widths are identical in Germany; this is because accumulated ballast is designed on the surface, which increases the width of the substructure surface. On the whole, French HSR lines have the widest surface, while Japanese HSR lines have the narrowest. One reason for the narrow surface in Japan is that a trench is usually dug between two lines not only to drain water away but also to obstruct ballast.

Concluding remarks
Important design parameters for HSR earthwork design have been summarized and a comparative analysis has been conducted on their limit values and related specifications in certain countries. The HSR embankment substructure compositions in different countries were sorted for comparison, and a unified definition of the embankment substructure structure that is applicable to different countries was proposed. It is noted that the design parameters of some countries used for comparison in this paper were acquired from national standards while some were obtained from technical specifications of one or several HSR lines. At the same time, we mostly collected and analysed railway construction standards written in different languages. Even though most of the standards have been translated into English, some information may have been lost in translation. Another complication might be that railway organizations refer to the same thing but have different definitions. There is a need for further comparison of the technical and economic rationality of earthwork techniques in different countries and numerical analysis on how each variable affects strain and shear strengths specifically.