Modeling and Simulation of Collision-Causing Derailment to Design the Derailment Containment Provision Using a Simpliﬁed Vehicle Model

applied sciences

Article Modeling and Simulation of Collision-Causing Derailment to Design the Derailment Containment Provision Using a Simpliﬁed Vehicle Model

In-Ho Song 1, Jun-Woo Kim 2, Jeong-Seo Koo 1,* and Nam-Hyoung Lim 3

1 Department of Railway Safety Engineering, Seoul National University of Science and Technology, Seoul 01811, Korea; [email protected] 2 Department of Rolling Stock System Engineering, Seoul National University of Science and Technology, Seoul 01811, Korea; [email protected] 3 Department of Civil Engineering, Chungnam National University, Daejeon 34134, Korea; [email protected] * Correspondence: [email protected]

Received: 5 December 2019; Accepted: 20 December 2019; Published: 22 December 2019

Abstract: As the operating speed of a train increases, there is a growing interest in reducing damage caused by derailment and collision accidents. Since a collision with the surrounding structure after a derailment accident causes a great damage, protective facilities like a barrier wall or derailment containment provision (DCP) are installed to reduce the damage due to the secondary collision accident. However, the criteria to design a protective facility such as locations and design loads are not clear because of diﬃculties in predicting post-derailment behaviors. In this paper, we derived a simpliﬁed frame model that can predict post derailment behaviors in the design phase of the protective facilities. The proposed vehicle model can simplify for various frames to reduce the computation time. Also, the actual derailment tests were conducted on a real test track to verify the reliability of the model. The simulation results of the proposed model showed reasonable agreement to the test results.

Keywords: DCP (Derailment Containment Provisions); derailment; simplified vehicle model; protective facility

1. Introduction As train speeds increase, safety becomes increasingly signiﬁcant. Railway accidents are reported around the world every day. In particular, derailment accidents cause a lot of casualties and property damage. Derailment accidents have occurred frequently in Korea. Some coaches of the Korean high-speed train were recently derailed due to an error in the operation of a turnout. The derailment accident damaged rails, PC sleepers, the turnout and coaches although it did not cause any casualties as shown in Figure1a [ 1]. In the United States, a derailment accident in a curve due to over-speed, as shown in Figure1b, caused more than 170 dead or injured and property loss of $ 9 million [2]. Secondary collision accidents following derailment result in greater damage than derailment itself due to collision with surrounding structures or bridges falling down when wheels deviate from the tracks. For these reasons, a protective facility is usually installed to minimize damage due to secondary collision or falling [3,4]. In Korea, it is a requirement to install protective facility in a sharp curve area and a bridge of over 18 (m) or other areas with high derailment risk. In the UK, the RSSB assume that, after a train is derailed, the inner wheel will be guided by the outer rail before wheels collide against the barrier walls. Based on this assumption, they suggested that the barrier wall height should be at least 350 (mm) from

Appl. Sci. 2020, 10, 118; doi:10.3390/app10010118 www.mdpi.com/journal/applsci Appl. Sci. 2020, 10, 118 2 of 18 the head of rails, and the distance of the barrier wall must be greater than 1500 (mm). In Germany, guard-rails are usually applied to passenger and freight train lines. In the case of a high-speed train line, it is assumed that derailed wheels would be guided by rails. Therefore, the barrier walls must be kept from the running wheel-sets by a distance avoiding a collision against the walls. The height of the barrierAppl. Sci. walls 2020, 9 is, xdesigned FOR PEER noREVIEW higher than the rail head [5]. 2 of 18

(a) (b)

FigureFigure 1.1. ((aa)) TheThe derailmentderailment accidentaccident ofof thethe high-speedhigh-speed traintrain inin KoreaKorea (02.2011);(02.2011); ((bb)) thethe derailmentderailment accidentaccident inin AmericaAmerica (12.(12. 2013).2013).

EvenIn Korea, though it is barrier a requirement walls are installedto install on protective bridges or facility in curves in a to sharp reduce curve damage area afterand derailmenta bridge of forover high-speed 18 (m) or other trains, areas the criteriawith high for derailment their locations risk. andIn the design UK, the loads RSSB are assume not clear, that, and after studies a train on is railwayderailed, protective the inner facilities wheel arewill insu be ffiguidedcient. by In orderthe ou toter determine rail before the wheels construction collide locations against andthe designbarrier loadswalls. of Based the protective on this assumption, facility, it is they important suggested to know that thethe derailmentbarrier wall behaviors height should of trains. be at Although least 350 it(mm) is ideal from to the check head post-derailment of rails, and the behaviors distance through of the barrier actual wall derailment must be tests, greater many than research 1500 (mm). works In generallyGermany, simulate guard-rails these are using usually multi-body applied dynamics to passenger programs and freight because train of practicallines. In ditheffi cultiescase of sucha high- as highspeed cost train and line, construction it is assumed of a testthat facility.derailed wheels would be guided by rails. Therefore, the barrier wallsJerry must Evans be kept and from Mats Bergthe running [6] discussed wheel-sets appropriate by a distance modelling avoiding choices a for collision suspension against components, the walls. wheelThe height/rail contact of the conditionsbarrier walls and is modelingdesigned no input. higher Dmitry than Pogorelovthe rail head and [5]. Viasdislav [7] considered a techniqueEven referred though to barrier as the ‘Trainwalls 3Dare method’ installed for on simulation bridges ofor trains in curves as coupled to reduce derailed damage spatial after and simplifiedderailment one-dimensional for high-speed trains, models th ofe railcriteria vehicles for their to evaluate locations safety and factorsdesign withloads dependence are not clear, on and the trainstudies operation on railway regime. protective R.Kovalev facilities and are V.N insufficient. Yazykov [ 8In] presentedorder to determine nonstiff method the construction for computing locations the nonellipticaland design loads contact of problemthe protective that canfacility apply, it is to important the wheel to rail know contact the problem.derailment Hyung-Suk behaviors Hanof trains. and Jeong-SeoAlthough Kooit is [ 9ideal] studied to check high-speed post-derailment train crashes be inhaviors three dimensions through actual using multi-bodyderailment dynamicstests, many to predictresearch the works crash generally behavior ofsimulate trains. Hyun-Woongthese using multi-body Bae [10,11] dynamics studied the programs impact forces because in the of casepractical of a collisiondifficulties with such a barrier as high wall cost using and construction a three-dimensional of a test finite facility. element model of the KTX (Korean High SpeedJerry Train). Evans Xingwen and WuMats [12 Berg,13] developed[6] discussed a half-car appropriate test specimen modelling of a vehiclechoices and for analyzed suspension its derailmentcomponents, behaviors wheel/rail through contact experiments conditions in and the modeling laboratory. input. In addition, Dmitry they Pogorelov predicted and the Viasdislav derailment [7] behaviorsconsidered of a a technique derailed high-speed referred to train as the using ‘Train dynamic 3D method’ simulations. for simulation Lirong Guo of trains [14] conducted as coupled a seriesderailed of low-speedspatial and derailmentsimplified one-dimensional test under different models test of conditions rail vehicles for to a evaluate Chinese safety train. factors This study with confirmeddependence that on gearbox the train of train operation plays an regime. important R.Kovalev role in restricting and V.N theYazykov lateral motions[8] presented of the derailednonstiff vehicle,method andfor alsocomputing the other the influence nonelliptical factors contact such as problem speed, weightthat can and apply track to were the considered.wheel rail contact Liang Lingproblem. [15] studiedHyung-Suk the effi Hancacy ofand the Jeong-Seo guard rail Koo system [9] (GRS) studied to minimize high-speed the derailmenttrain crashes of potentialin three ofdimensions trains laterally using colliding multi-body by adynamics heavy vehicle, to predict and thethe sensitivitiescrash behavior of parameters of trains. Hyun-Woong of the guard railBae system[10,11] suchstudied as thethe flangeimpact way forces width, in the and case reported of a collision the installation with a barrier height. wall Dan using Brabie a three-dimensional [16,17] studied thefinite eff ectselement on derailment model of the behaviors KTX (Korean according High to Sp theeed bogie Train). frame Xingwen and the Wu damage [12,13] degree developed of concrete a half- sleeperscar test basedspecimen on derailmentof a vehicle accidents and analyzed in Europe. its derailment Hirsch [18 ]behaviors studied the through height experiments of the barrier in wall the accordinglaboratory. to In overturning addition, they moment predicted considering the derailment the center behaviors of gravity of ofa derailed trains and high-speed impact acceleration. train using dynamicIn this simulations. study, a simplified Lirong frameGuo model[14] conducted is developed a series that can of below-speed used in simulation derailment for test design under of thedifferent protective test facility.conditions The for reliability a Chinese of the train. suggested This study model confirmed is verified that through gearbox the actualof train derailment plays an important role in restricting the lateral motions of the derailed vehicle, and also the other influence factors such as speed, weight and track were considered. Liang Ling [15] studied the efficacy of the guard rail system (GRS) to minimize the derailment of potential of trains laterally colliding by a heavy vehicle, and the sensitivities of parameters of the guard rail system such as the flange way width, and reported the installation height. Dan Brabie [16,17] studied the effects on derailment behaviors according to the bogie frame and the damage degree of concrete sleepers based on derailment accidents in Europe. Hirsch [18] studied the height of the barrier wall according to overturning moment considering the center of gravity of trains and impact acceleration. Appl. Sci. 2020, 10, 118 3 of 18 tests. In addition, the deviation prevention effects by installation of a derailment containment provision (DCP), which is one of the barrier types, are verified through the derailment tests.

2. Wheel-Rail Contact To simulate wheel/rail contact, solid elements for wheels and shell elements for rails are used. The basic contact method of Ls-Dyna is the penalty method [19]. The concept is that when a slave node penetrates a master surface, the contact force is calculated from the amount of penetration and the contact stiﬀness. The contact force also increases in proportion to the increase of the amount of penetration. In Figure2, t represents the position vector of the slave node ns, and the master segment surface si is related to ns. If ns penetrates si, the contact point coordinates (ξc, ηc) must satisfy Equation (1).

∂r ∂r (ξc, ηc) [t r(ξc, ηc)] = 0, (ξc, ηc) [t r(ξc, ηc)] = 0 (1) ∂ξ · − ∂η · −

The initial values are estimated through a least-squares projection iteration,

ξo = 0, ηo = 0 " # " # r,ξ h in ∆ξ o r,ξ r,ξ r,η ∆η = r(ξi, ηi) t , (2) r,η r,η { − } ξi+1 = ξi + ∆ξ, ηi+1 = ηi + ∆η

From the Newton–Raphson iteration, the amount of penetration on the penetration coordinate can be calculated as (3). n o r [H] ∆ξ = ,ξ r(ξ , η ) t , ∆η r,η i i − { " − } # h i r,ξ 0 r r,ξη (3) [H] = r,ξ, r,η + · r,η r r 0 · ,ξη ξi+1 = ξi + ∆ξ, ηi+1 = ηi + ∆η When the amount of the slave node coordinate penetrates into the master surface, Equation (4) is obtained. l = η [t r(ξc, ηc)] < 0 (4) i × − At this point, the normal vector of the master surface at the contact point is:

ni = ni(ξc, ηc) (5)

If the slave node penetrates the master surface, the contact force vector becomes Equation (6).

fs = lk η i f l < 0 (6) − i i

In addition, considering the degree of freedom of ηs, the contact force is given by Equation (7).

i fm = φi(ξc, ηc) fs i f l < 0 (7)

The master segment consists of four nodes (i =1, 2, 3, 4), and the contact force ki consists of the bulk modulus Ki and the surface area Ai of the master segment. The contact force obtained in the master segment is calculated as follows:

fsiKiAi k = (8) i max(shell diaonal)

In this study, a simplified frame model is developed that can be used in simulation for design of Appl. Sci. 2020, 9, x FOR PEER REVIEW 4 of 18 the protective facility. The reliability of the suggested model is verified through the actual derailment tests.When In addition, the amount the ofdeviation the slave prevention node coordinate effects pe bynetrates installation into the of master a derailment surface, Equationcontainment (4) isprovision obtained. (DCP), which is one of the barrier types, are verified through the derailment tests.

2. Wheel-Rail Contact 𝑙=𝜂 × 𝑡−𝑟(𝜉,𝜂) < 0 (4) Appl. Sci.AtTo 2020this simulate, point,10, 118 wheel/rail the normal contact, vector ofsolid the elementsmaster surface for wheels at the and contact shell point elements is: for rails are 4used. of 18 The basic contact method of Ls-Dyna is the penalty method [19]. The concept is that when a slave 𝑛 =𝑛(𝜉,𝜂) (5) node penetrates a master surface, the contact force is calculated from the amount of penetration and slave node at each calculation cycle. The one-way contact method shows good results in the dynamic the contactIf the slave stiffness. node Thepenetrates contact the force master also surface,increase thes in contact proportion force to vector the increase becomes of Equation the amount (6). of stabilization of the vertical contact force, but it does not accurately simulate the contact between a penetration. ﬂange and a rail. Therefore, a two-way contact𝑓 =−𝑙𝑘 method𝜂 𝑖𝑓 which𝑙 <0 can simulate the contact between(6) a In Figure 2, t represents the position vector of the slave node 𝑛, and the master segment surface ﬂange and a rail was applied [20]. Depending on𝜂 the relative distance between the wheel and rail, In𝑠 addition, is related considering to 𝑛 . If 𝑛 the penetrates degree of 𝑠freedom, the contact of ,point the contact coordinates force is(𝜉 given,𝜂 ) mustby Equation satisfy Equation(7). large oscillations could occur because of the penalty method. The dynamic relaxation time was saved (1). by appropriately adjusting the distance𝑓 =𝜙 between(𝜉,𝜂 the) wheel𝑓 𝑖𝑓 and the𝑙 <0 rail before contact analysis. (7)

The master segment consists of four nodes (i =1, 2, 3, 4), and the contact force 𝑘 consists of the bulk modulus 𝐾 and the surface area 𝐴 of the master segment. The contact force obtained in the master segment is calculated as follows:

𝑓𝐾𝐴 𝑘 = (8) 𝑚𝑎𝑥 (𝑠ℎ𝑒𝑙𝑙 𝑑𝑖𝑎𝑜𝑛𝑎𝑙)

Generally, contact search methods of one-way and two-way are useful for handling contact problems. The two-way contact method calculates the contact force between a master surface and a slave node at each calculation cycle. The one-way contact method shows good results in the dynamic stabilization of the vertical contact force, but it does not accurately simulate the contact between a flange and a rail. Therefore, a two-way contact method which can simulate the contact between a flange and a rail was applied [20]. Depending on the relative distance between the wheel and rail, large oscillations could occur because of the penalty method. The dynamic relaxation time was saved Figure 2. Location of contact point when ns lies on master segment. by appropriately adjustingFigure 2. the Location distance of contact between point the when wheel 𝑛 and lies theon master rail before segment. contact analysis. Another important factor in rolling contact simulationsimulation is to reduce vibration generated by wheel rail contact between𝜕𝑟 elements. To minimize vibratio𝜕𝑟n, wheel tread and flange are finely modeled as rail contact between elements. To minimize vibration, wheel tread and flange are finely modeled as (𝜉,𝜂) ∙ 𝑡−𝑟(𝜉,𝜂) =0, 𝜉𝑐,𝜂𝑐 ∙ 𝑡−𝑟𝜉𝑐,𝜂𝑐 =0 (1) shown in Figure𝜕𝜉3 3.. ForFor simulationsimulation ofof wheelwheel railrail contact,cont𝜕𝜂act, frictionfriction coecoefficientfficient between wheel and rail is applied as 0.3 and CONTACT_AUTOMATIC_SURFACE_TO_SURFACE CONTACT_AUTOMATIC_SURFACE_TO_SURFACE keywordkeyword whichwhich is two way contactThe method initial isvalues used are to simulate estimated flangeflange through andand a railrail least-squares contact.contact. projection iteration,

𝜉 =0, 𝜂 =0

𝑟, 𝑟, 𝑟, 𝑟, = {𝑟(𝜉,𝜂) −𝑡, (2) 𝑟, 𝑟,

𝜉 =𝜉 +𝛥𝜉,𝜂 =𝜂 +𝛥𝜂 From the Newton–Raphson iteration, the amount of penetration on the penetration coordinate can be calculated as (3).

∆ , H =− {𝑟(𝜉,𝜂) −𝑡}, ,

𝑟 0𝑟∙𝑟 , , (3) H = 𝑟,,𝑟,+ Figure 3. Finite𝑟, element of wheel𝑟∙𝑟 tread, and flange.0 Figure 3. Finite element of wheel tread and flange. Figures4 and5 show the contact force and vertical displacement between wheel /rail. Oscillation 𝜉 =𝜉 +Δ𝜉,𝜂 =𝜂 +Δ𝜂 occursFigures due to 4 theand rolling 5 show contact the contact of the force finite and elements vertical and displacement it is confirmed between that in wheel/rail. a rough mesh Oscillation wheel modeloccurs largedue to vertical the rolling displacement contact of occurredthe finite elements because the and wheel it is confirmed angle and that mean in a contact rough forcemesh ofwheel fine meshmodel model large isvertical 5.4 kN displacement which is the sameoccurred as the because theoretical the wheel force. Figureangle and6 for mean a fine contact mesh model force showsof fine themesh vertical model displacement is 5.4 kN which of theis the wheel same and as the the theo displacementretical force. oscillated Figure 6 within for a fine 0.06 mesh mm. model shows the vertical displacement of the wheel and the displacement oscillated within 0.06 mm. Appl. Sci. 2020, 9, x FOR PEER REVIEW 5 of 18

Appl.Appl. Sci.Sci. 20202020,,10 9,, x 118 FOR PEER REVIEW 55 ofof 1818

Figure 4. Contact force between wheel and rail.

Figure 4. Contact force between wheel and rail. Figure 4. Contact force between wheel and rail.

(a) (b)

Figure 5. ((aa)) Vertical Vertical displacement between wheel andand railrail (ﬁne(fine mesh mesh model) and and ( (bb)) Vertical Appl. Sci.displacement 2020, 9, x FOR between PEER (REVIEW a wheel) and rail (rough mesh model). (b) 6 of 18

3. SimplifiedFigure 5. Frame(a) Vertical Model displacement between wheel and rail (fine mesh model) and (b) Vertical displacement between wheel and rail (rough mesh model). For railway vehicle simulations, various modeling techniques can be used depending on their purposes3. Simplified and performances.Frame Model Simplification is not essential when simulating derailment for only one car. However, simulation for a multiple-unit train would take a tremendous computing time, For railway vehicle simulations, various modeling techniques can be used depending on their therefore simplification is necessary. purposes and performances. Simplification is not essential when simulating derailment for only one In this study, analyses were carried out using commercial finite element (FE) software, Ls-Dyna. car. However, simulation for a multiple-unit train would take a tremendous computing time, It is ideal to create the finite element model of the actual shape, but if the shape of the model is therefore simplification is necessary. changed, additional modeling work would be required, and the number of elements which is related In this study, analyses were carried out using commercial finite element (FE) software, Ls-Dyna. to a longer analysis time would increase. So in many cases of dynamic analysis a model is simplified It is ideal to create the finite element model of the actual shape, but if the shape of the model is to a level that does not significantly affect the results. changed, additional modeling work would be required, and the number of elements which is related The frame mass and momentFigure of inertia 6. Examples are assigned of simplified to the bogie. node at the center of gravity using to a longer analysis time would increase. So in many cases of dynamic analysis a model is simplified CONSTRAINED NODAL RIGIDFigure BODY 6. INERTIAExamples ofin simplified Ls-Dyna bogie.[21]. to a level that does not significantly affect the results. 3. SimplifiedSimplified Frame models Model have each node located at the center of gravity and at each suspension. The A ComparisonThe frame of Behaviorsmass and between moment th eof Real inertia Model are and assigned the Simplified to the Modelnode at the center of gravity using positionFor of railway the nodes vehicle can simulations,be changed easily. various Therefor modelinge, this techniques can be changed can be into used various depending other onmodels their CONSTRAINED NODAL RIGID BODY INERTIA in Ls-Dyna [21]. purposesby modifyingIt is necessary and performances.or adding to verify the location that Simplification the of simplified nodes. is not mo essentialdel has whenless computation simulating derailment time than forthe onlydetailed one Simplified models have each node located at the center of gravity and at each suspension. The realcar. However,model.The number Therefore, simulation of elements each for model a in multiple-unit wheelset was evaluated is trainexactly wouldun derthe thesame take same a because tremendous conditions. of using computing Tables the same 1 and time, wheelset 2 show therefore the to position of the nodes can be changed easily. Therefore, this can be changed into various other models specificationssimulatesimplification the wheel–rail and is necessary. the number interface of elinements FE analyses. [22,23]. Th Additionale number specificof elementsations of of frames bogie isare significantly referred in by modifying or adding the location of nodes. reducedsupplementaryIn this by study,using materials. the analyses simplified were model. carried As out for using railway commercial vehicles, finite two elementbogies are (FE) usually software, installed Ls-Dyna. per The number of elements in wheelset is exactly the same because of using the same wheelset to aIt vehicle. is idealSince to Ifthe createthe model number the finitedoes of not elementvehicles take modelintoincreases, account of the the actualthe effect damage shape, of the butof frame the if the frame, simpli shape rigidfication of the properties model could isbe changed,(Mat greater 20) simulate the wheel–rail interface in FE analyses. The number of elements of frames is significantly inhaveadditional numerical been modelingassigned simulations. workto the Figure would main 6 beframes. shows required, examplesThe andrealthe frame ofnumber a simplified model of elementsconsists bogie modelof which four including is main related components. tosecondary a longer reduced by using the simplified model. As for railway vehicles, two bogies are usually installed per suspensions.Bolster,analysis timeside would frame, increase. journal So in manybox, cases spri of dynamicng plate. analysis To a modelassemble is simplified these to aparts, level a vehicle. If the number of vehicles increases, the effect of the frame simplification could be greater CONSTRAINED_RIGED_BODIESthat does not significantly affect the results.and CONTACT_AUTOMATIC_ SURFACE_TO_SURFACE in numerical simulations. Figure 6 shows examples of a simplified bogie model including secondary keywords are used. suspensions.

Table 1. The specification of 3-piece bogie frame [21].

Parameters Values Gauge (mm) 1435 Wheelbase (mm) 1676 Wheel Size (mm) 860 Length (mm) 2600 Width (mm) 2350 Weight (kg) 3800 Frame Weight (kg) 1585 Frame Ixx (Kg·m2) 1420.6 Frame Iyy (Kg m2) 493.6 Frame Izz (Kg m2) 1812.2

Table 2. The comparison of elements of the 3-piece bogie.

Part Real Model Simplified Model 150,880 150,880 Wheelset (Solid element) (Solid element) 818,164 1 Bogie Frame (Solid element) (1D element) Total 969,044 150,881 Elements Primary suspension stiffness 160 kN/mm 160 kN/mm

Figure 7 shows the conditions of simulation for the derailment while driving at a speed of 13 (km/h). Appl. Sci. 2020, 10, 118 6 of 18

The frame mass and moment of inertia are assigned to the node at the center of gravity using CONSTRAINED NODAL RIGID BODY INERTIA in Ls-Dyna [21]. Simplified models have each node located at the center of gravity and at each suspension. The position of the nodes can be changed easily. Therefore, this can be changed into various other models by modifying or adding the location of nodes. The number of elements in wheelset is exactly the same because of using the same wheelset to simulate the wheel–rail interface in FE analyses. The number of elements of frames is significantly reduced by using the simplified model. As for railway vehicles, two bogies are usually installed per a vehicle. If the number of vehicles increases, the effect of the frame simplification could be greater in numerical simulations. Figure6 shows examples of a simplified bogie model including secondary suspensions.

A Comparison of Behaviors between the Real Model and the Simplified Model It is necessary to verify that the simplified model has less computation time than the detailed real model. Therefore, each model was evaluated under the same conditions. Tables1 and2 show the specifications and the number of elements [22,23]. Additional specifications of bogie are referred in supplementary materials.

Table 1. The speciﬁcation of 3-piece bogie frame [21].

Parameters Values Gauge (mm) 1435 Wheelbase (mm) 1676 Wheel Size (mm) 860 Length (mm) 2600 Width (mm) 2350 Weight (kg) 3800 Frame Weight (kg) 1585 Frame Ixx (Kg m2) 1420.6 · Frame Iyy (Kg m2) 493.6 · Frame Izz (Kg m2) 1812.2 ·

Table 2. The comparison of elements of the 3-piece bogie.

Part Real Model Simpliﬁed Model 150,880 150,880 Wheelset (Solid element) (Solid element) 818,164 1 Bogie Frame (Solid element) (1D element) Total Elements 969,044 150,881 Primary suspension stiﬀness 160 kN/mm 160 kN/mm

Since the model does not take into account the damage of the frame, rigid properties (Mat 20) have been assigned to the main frames. The real frame model consists of four main components. Bolster, side frame, journal box, spring plate. To assemble these parts, CONSTRAINED_RIGED_BODIES and CONTACT_AUTOMATIC_ SURFACE_TO_SURFACE keywords are used. Figure7 shows the conditions of simulation for the derailment while driving at a speed of 13 (km /h). Appl. Sci. 2020, 9, x FOR PEER REVIEW 7 of 18 Appl. Sci. 2020, 10, 118 7 of 18 Appl. Sci. 2020, 9, x FOR PEER REVIEW 7 of 18

Figure 7. The condition of the derailment simulation. Figure 7. The condition of the derailment simulation. Table 3 shows the simulation time for the analyses. The computing time of the real model is aboutTable 15453 3 (min),shows shows and the the simulationthe simulation simplified time time model for for the isthe analyses. abou analyst 741 Thees. (min). The computing Thecomputing simplified time time of the model of real the needs model real modelonly is about less is 1545aboutthan (min),half 1545 simulation (min), and the and simplified time the simplifiedover modelthe real model is model. about is abou741 The (min).t performance741 (min). The simplified The of simplifiedthe computer model model needs central needs only processing lessonly than less thanhalfunit simulation(CPU)half simulation is shown time intime over Table over the 4. real the model.real model. The performanceThe performance of the of computer the computer central central processing processing unit (CPU)unit (CPU) is shown is shown in Table in Table4. 4. Table 3. The comparison of simulation time. Table 3. The comparison of simulation time. Table 3. The comparison ofReal simulation Simplified time. Model Model Real ModelModelReal SimplifiedModel Simplified Model Model Computing Time(Min)Computing Time(Min) 1545Model 1545 Model 741 741 Computing Time(Min) 1545 741 Table 4.4. The specificationsspecifications of computer CPU. Table 4. The specifications of computer CPU. CompanyCompany Model Model Clock Clock Xeon(R) Company ModelXeon(R) Clock Intel Intel 3.4 GHz 3.4 GHz E5-2687WE5-2687WXeon(R) v2 v2 Intel 3.4 GHz E5-2687W v2 The simulation results for X-displacement (longitudinal) and Y-displacement (lateral) over time The simulation results for X-displacement (longitudinal) and Y-displacement (lateral) over time are shown in Figure 8. The largest difference in X-displacement is 186 (mm) at 5000 (ms) and the are shownThe simulation in Figure8 results. The largest for X-displacement di fference in X-displacement (longitudinal) and is 186 Y-displacement (mm) at 5000 (ms) (lateral) and theover largest time largest difference in Y-displacement is 103 (mm) at about 916 (ms). The differences in the simulation arediff erenceshown in in Y-displacement Figure 8. The islargest 103 (mm) difference at about in 916 X-displacement (ms). The diff erencesis 186 (mm) in the at simulation 5000 (ms) results and the are results are very slight. Consequently, it was confirmed that the simplified frame model could replace verylargest slight. difference Consequently, in Y-displacement it was confirmed is 103 (mm) thatthe at about simplified 916 (ms). frame The model differences could replace in the simulation the refined the refined real frame model. resultsreal frame are model.very slight. Consequently, it was confirmed that the simplified frame model could replace the refined real frame model.

(a) Y-Displacement (Lateral) (b) X-Displacement (Longitudinal) (a) Y-Displacement (Lateral) (b) X-Displacement (Longitudinal) Figure 8. The comparison of results between the re realal and simpliﬁedsimplified models in simulation. Figure 8. The comparison of results between the real and simplified models in simulation. 4. Model Validation 4. Model Validation Appl. Sci. 2020, 10, 118 8 of 18 Appl. Sci. 2020, 9, x FOR PEER REVIEW 8 of 18 Appl. Sci. 2020, 9, x FOR PEER REVIEW 8 of 18 4.1.4. Model Derailment Validation Test Field

4.1. Derailment DerailmentDerailment Test field Field tests were carried out in a closed station area. The purpose of the field tests is to develop a DCP facility on a concrete track. Therefore, the concrete track was constructed by Derailment ﬁeldfield teststests werewere carriedcarried out out in in a a closed closed station station area. area. The The purpose purpose of of the the ﬁeld field tests tests is tois referring to the Rheda 2000 structure after removing ballasts on the track [24]. todevelop develop a DCP a DCP facility facility on a on concrete a concrete track. track. Therefore, Therefore, the concrete the concrete track was track constructed was constructed by referring by Two side barriers were constructed to block the excessive lateral movement of bogies during referringto the Rheda to the 2000 Rheda structure 2000 structure after removing after removing ballasts on ballasts the track on [the24]. track [24]. tests. The length (100 m) of concrete tracks was designed considering the maximum test speed of 60 Two sideside barriersbarriers were were constructed constructed to to block block the th excessivee excessive lateral lateral movement movement of bogies of bogies during during tests. (km/h). Figure 9 shows the view of the testing ground. Thetests. length The length (100 m) (100 of concretem) of conc tracksrete tracks was designed was designed considering consider theing maximum the maximum test speed testof speed 60 (km of /60h). (km/h).Figure9 Figureshows 9 the shows view the of theview testing of the ground. testing ground.

Figure 9. The derailment testing ground. Figure 9. The derailment testing ground. Figure 9. The derailment testing ground. 4.2.4.2. Derailment Derailment Tests Tests with with One Bogie 4.2. Derailment Tests with One Bogie DerailmentDerailment tests tests were were performed performed with with only only one one bo bogiegie at at first ﬁrst [25]. [25]. The The speed speed measured measured by by the the speedspeedDerailment sensor right tests beforebefore were derailmentderailment performed was was with 27 27 (km only(km/h)/h). one. The The bo behaviorgie behavior at first and and [25]. accelerations accelerations The speed were measured were measured measured by the by byspeeda high-speed a high-speed sensor right camera camera before and and derailment the the acceleration acceleration was 27 sensors. sensor (km/h)s. Figure. FigureThe behavior 10 10 shows shows and thethe accelerations derailmentderailment behaviors. were measured They They showedbyshowed a high-speed that that the the camerafield ﬁeld test test and and the simulations acceleration have have sensor a a similar s. Figure derailment 10 shows thebehavior. behavior. derailment After After behaviors. the the bogie bogie They was was derailed by the derailment device (approx. 420 ms), the front wheel collided with the third sleeper. showedderailed that by the the derailment field test and device simulations (approx. have 420 ms),a similar the front derailment wheel collidedbehavior. with After the the third bogie sleeper. was The rear wheel collided with the fourth sleeper at about 680 (ms) as shown in Figure 10a Thederailed rear wheelby the collidedderailment with device the fourth (approx. sleeper 420 atms), about the 680front (ms) wheel as shown collided in Figurewith the 10 a.third sleeper. The rear wheel collided with the fourth sleeper at about 680 (ms) as shown in Figure 10a

(a) (b)

FigureFigure 10. Derailment10. Derailment behaviors. (behaviors.(a) (a) Derailmenta)Derailment behavior behavior (Side (Side view); view); (b )(b Derailment) Derailment(b) behavior behavior (Front (Front view). Figure 10. Derailment behaviors.(a)Derailment view)behavior (Side view); (b) Derailment behavior (Front view) Appl. Sci. 2020, 9, x FOR PEER REVIEW 9 of 18

Figure 10b shows the behavior of the bogie after derailment and finally. Although collision between wheel and rail does not occur in the field test while the bogie was running on the sleeper, the wheel collided with the rail in the simulation. This difference in the wheel–rail impact behaviors occurred because the track components were simplified in the simulation, while real track components like tension clamps and screw spikes hindered the wheel-set from moving to the rail in the field test Appl. Sci. 2020, 910, x, 118FOR PEER REVIEW 9 of 18 Figure 11 shows the comparison of damaged sleepers between the field test and the simulation. Although there was a slight difference in damage degree, the damage of sleepers occurred at similar Figure 10b shows the behavior of the bogie after derailment and finally. Although collision locations.Figure 10b shows the behavior of the bogie after derailment and finally. Although collision between wheel and rail does not occur in the field test while the bogie was running on the sleeper, between wheel and rail does not occur in the field test while the bogie was running on the sleeper, the the wheel collided with the rail in the simulation. wheel collided with the rail in the simulation. This difference in the wheel–rail impact behaviors occurred because the track components were This difference in the wheel–rail impact behaviors occurred because the track components were simplified in the simulation, while real track components like tension clamps and screw spikes simplified in the simulation, while real track components like tension clamps and screw spikes hindered hindered the wheel-set from moving to the rail in the field test the wheel-set from moving to the rail in the field test. Figure 11 shows the comparison of damaged sleepers between the field test and the simulation. Figure 11 shows the comparison of damaged sleepers between the field test and the simulation. Although there was a slight difference in damage degree, the damage of sleepers occurred at similar Although there was a slight difference in damage degree, the damage of sleepers occurred at similar locations. locations.

Figure 11. The broken sleepers.

X-displacement and velocity (Longitudinal direction) of the bogie were obtained through the sensors attached to the bogie. Figure 12 shows similar the X-velocity and displacements between the field test and simulations. The Y-displacement (lateral direction) of the bogie was obtained using the trace of wheels. Figure 13a shows the Y-displacement results of the field test and simulations over time. The reason the Y- displacement of the simulation is largerFigure than 11. thatThe of broken the field sleepers. test is that the tension clamps and the Figure 11. The broken sleepers. screw spikes in field tests obstructed the lateral motion of the wheel as mentioned above. As shown in FigureX-displacement 13b, the distance and between velocity the (Longitudinal rail and the tension direction) clamp of theis 155 bogie (mm) were, which obtained is similar through to the the X-displacement and velocity (Longitudinal direction) of the bogie were obtained through the lateralsensors difference attached of to147 the (mm). bogie. Figure 12 shows similar the X-velocity and displacements between the sensors attached to the bogie. Figure 12 shows similar the X-velocity and displacements between the ﬁeld test and simulations. field test and simulations. The Y-displacement (lateral direction) of the bogie was obtained using the trace of wheels. Figure 13a shows the Y-displacement results of the field test and simulations over time. The reason the Y- displacement of the simulation is larger than that of the field test is that the tension clamps and the screw spikes in field tests obstructed the lateral motion of the wheel as mentioned above. As shown in Figure 13b, the distance between the rail and the tension clamp is 155 (mm), which is similar to the lateral difference of 147 (mm).

(a) (b)

FigureFigure 12. 12. TheThe comparison comparison ofof results results between between the fieldthe testfield and test the and simulation. the simulation. (a) X-Velocity (a) (Longitudinal);X-Velocity (Longitudinal);(b) X-Displacement (b) X-Displacement (Longitudinal). (Longitudinal).

The Y-displacement (lateral direction) of the bogie was obtained using the trace of wheels. Figure 13a shows the Y-displacement results of the ﬁeld test and simulations over time. The reason the

Y-displacement of the simulation is larger than that of the field test is that the tension clamps and the screw spikes in field tests obstructed(a) the lateral motion of the wheel as mentioned(b) above. As shown in Figure 13b, the distance between the rail and the tension clamp is 155 (mm), which is similar to the Figure 12. The comparison of results between the field test and the simulation. (a) X-Velocity lateral difference of 147 (mm). (Longitudinal); (b) X-Displacement (Longitudinal). Appl. Sci. 2020, 9, x FOR PEER REVIEW 10 of 18

Appl. Sci. 2020, 10, 118 10 of 18 Appl. Sci. 2020, 9, x FOR PEER REVIEW 10 of 18

(a) (b)

Figure 13. The comparison of results between the field test and the simulation. (a) Y-Displacement (Lateral); (b) The drawing(a) of a concrete sleeper. (b) FigureFigure 13. TheThe comparison comparison of of results results between between the the field field test test and and the the simulation. simulation. ( (aa)) Y-Displacement The(Lateral);(Lateral); impact (b) accelerations The drawing of of a concreteconcretethe bogie sleeper.sleeper. were measured by acceleration sensors installed on the bolster. Since the measured acceleration data include unnecessary frequency components such as noise,TheThe the impact data should accelerations be filtered of of the theby anbogie appropri were atemeas measured filterured to by analyze acceleration the dynamic sensors trend. installed There on on arethe variousbolster.bolster. techniques Since thethe measuredmeasured for data filtering, acceleration acceleration and data there data include areinclude acceleration-filtering unnecessary unnecessary frequency frequency criteria components forcomponents each field. such assuchIn noise,case as ofnoise,the the data vehiclethe should data collision should be filtered test,be filtered bythere an appropriateisby a antest appropri standa filterrdate tofor filter analyze handling to analyze the the dynamic vehiclthe dynamic trend.e collision There trend. data are There in various each are countryvarioustechniques andtechniques forButterworth data for filtering, data low-pass filtering, and there filtering and are there acceleration-filteringis generallyare acceleration-filtering used. There criteria is nocriteria for clear each for standard field. each Infield. caseof impact In of case the accelerationofvehicle the vehicle collision filtering collision test, for there test, derailment is there a test is standard testsa test of standa railway for handlingrd vehicles.for handling the In vehicle Europe, the collisionvehicl a 40e collisionHz data low-pass in eachdata filter countryin each is usuallycountryand Butterworth used and toButterworth evaluate low-pass high-speed low-pass filtering is filteringtrain generally body is used.generallyacceleration There used. is[26–28]. no There clear Numerical standardis no clear ofmodel standard impact validation acceleration of impact for caraccelerationfiltering bodies for was derailmentfiltering evaluated for testsderailment using ofrailway a 40 tests Hz vehicles. oflow-pass railway In filter. Europe,vehicles. But a Init 40 couldEurope, Hz low-pass lower a 40 theHz filter peakslow-pass is usually of impactfilter used is accelerationsusuallyto evaluate used high-speed into collision,evaluate train high-speedso a 180 body Hz acceleration trainlow-pass body filter a [cceleration26– was28]. used Numerical [26–28]. according Numerical model to EN15227 validation model B.2.1, forvalidation caras well. bodies for carwas bodiesFigures evaluated was 14 and usingevaluated 15 a show 40 Hzusing the low-pass comparison a 40 Hz filter. low-pass of But the it me could filter.asured lower But acceleration it the could peaks lower data of impact inthe the peaks accelerations field oftest impact with in theaccelerationscollision, simulation so a in 180results collision, Hz low-passfor soeach a 180 filteracceleration Hz was low-pass used component accordingfilter was using toused EN15227 accordingdifferent B.2.1, low-pass to asEN15227 well. filters B.2.1, (40 as Hz, well. 180 Hz). TheFiguresFigures maximum 14 and 1515acceleration showshow thethe comparison comparisonoccurred at of ofabout the the measured me420asured (ms) acceleration afteracceleration collision data data with in thein athe fieldwheelset field test test with and with the a sleeper.thesimulation simulation The results180 results Hz for filtering eachfor each acceleration is accelerationshown to component vibrate component acceleration using using diff erentdifferentas it low-passcontains low-pass high filters filters frequency (40 Hz,(40 180Hz, when Hz).180 comparedHz).The maximumThe maximumto 40 accelerationHz filtering.acceleration occurred The occurredmaximum at about at accelerationabout 420 (ms) 420 after (ms) values collision after are collision shown with a within wheelset Table a wheelset 5. and Maximum a sleeper. and a accelerationsleeper.The 180 HzThe of filtering 180 180 Hz Hz is filteringfiltering shown tois about vibrateshown twice accelerationto vibrate as high acceleration as as 40 it Hz contains filtering, as highit contains and frequency further high research when frequency compared is needed when to tocompared40 determine Hz filtering. to what 40 TheHz kind filtering. maximum of filterin Th accelerationge methodsmaximum should values acceleration be are used shown tovalues estimate in Tableare theshown5. MaximumDCP in impact Table acceleration load.5. Maximum Despite of theacceleration180 use Hz of filtering different of 180 is about Hzlow-pass filtering twice filtering, asis about high the astwice 40 tendency Hz as high filtering, ofas the40 and Hz acceleration furtherfiltering, research and is similar further is needed and research the to deviationsdetermine is needed oftowhat thedetermine experiment kind of what filtering and kind simulation methods of filterin shouldgare methods small. be usedshould to be estimate used to the estimate DCP impactthe DCP load. impact Despite load. Despite the use theofdi usefferent of different low-pass low-pass filtering, filtering, the tendency the tendency of the accelerationof the acceleration is similar is similar and the and deviations the deviations of the ofexperiment the experiment and simulation and simulation are small. are small.

(a) (b)

Figure 14. Cont. (a) (b) Appl. Sci. 2020, 9, x FOR PEER REVIEW 11 of 18 Appl. Sci. 2020, 10, 118 11 of 18 Appl. Sci. 2020, 9, x FOR PEER REVIEW 11 of 18

(c)

Figure 14. The comparison of accelerations between(c) the field test and the simulation (40 Hz low pass filter). (a) X-acceleration; (b) Y- acceleration; (c) Z- acceleration. FigureFigure 14.14. TheThe comparisoncomparison ofof accelerationsaccelerations betweenbetween thethe ﬁeldfield testtest andand thethe simulationsimulation (40(40 HzHz lowlow passpass ﬁlter).filter). ((aa)) X-acceleration;X-acceleration; ( b(b)) Y-acceleration; Y- acceleration; (c )(c Z-acceleration.) Z- acceleration.

(a) X-acceleration (b) Y- acceleration

(c) Z- acceleration FigureFigure 15. 15.The The comparisoncomparison of of acceleration acceleration( betweenc between) Z- acceleration the the ﬁeld field test test and and the the simulation simulation (180 (180 Hz Hz low-pass low-pass ﬁlter).filter). ( a(a)) X-acceleration; X-acceleration; (b ()b Y-acceleration;) Y- acceleration; (c) ( Z-acceleration.c) Z- acceleration. Figure 15. The comparison of acceleration between the field test and the simulation (180 Hz low-pass Table 5. The comparison of maximum acceleration. filter). (a) X-acceleration;Table (b) Y- 5. acceleration; The comparison (c) Z- of acceleration. maximum acceleration. Field Test Simulation Field Test Simulation Direction (40Table Hz)Field 5. The Test comparison (40Simulation Hz)of maximum acceleration.Field(180 Hz) Test Simulation(180 Hz) Direction Longitudinal (X) 4.1(40 Hz) 3.45(40 Hz) (180 11.8 Hz) (180 Hz) 10.9 Field Test Simulation Field Test Simulation LateralLongitudinalDirection (Y) (X) 4.57 4.1 4.99 3.45 11.8 10 10.9 9.5 (40 Hz) (40 Hz) (180 Hz) (180 Hz) VerticalLateral (Z) (Y) 7.14 4.57 7.59 4.99 10.7 10 9.5 11.7 LongitudinalVertical (Z) (X) 7.144.1 3.45 7.59 11.8 10.7 10.9 11.7 Lateral (Y) 4.57 4.99 10 9.5 Vertical (Z) 7.14 7.59 10.7 11.7

Appl. Sci. 2020, 10, 118 12 of 18 Appl. Sci. 2020, 9, x FOR PEER REVIEW 12 of 18

4.3.4.3. Concept of the Derailment Containment Provision (DCP)(DCP) TheThe DCPDCP isis aa facilityfacility thatthat preventsprevents aa largelarge deviationdeviation fromfrom railsrails inin orderorder toto reducereduce thethe damagedamage causedcaused byby aa secondarysecondary collisioncollision afterafter aa traintrain isis derailed.derailed. It is installed inside or outsideoutside thethe tracktrack andand guidesguides thethe wheelswheels oror axlesaxles ofof thethe traintrain toto preventprevent collisioncollision withwith surroundingsurrounding structuresstructures [[3,4].3,4]. TheThe DCPDCP is is classiﬁed classified into into three three types. types. Type Type 1 is a1 facility is a facility that guides that guides the wheels the bywheels being by installed being betweeninstalled thebetween rails. Typethe rails. 2 is aType facility 2 is that a facility guides that the wheelsguides bythe being wheels installed by being outside installed the rails. outside Type the 3 israils. a facility Type 3 that is a guidesfacility thethat axles guides from the outside axles from the railsoutside [5]. the Figure rails 16[5]. shows Figure the 16 DCPshows for the each DCP type. for Ineach this type. paper, In this the deviationpaper, the prevention deviation preven eﬀectstion of the effects DCP of Type the1 DCP were Type studied. 1 were studied.

(a) (b) (c)

FigureFigure 16.16. TypesTypes ofof derailment derailment containment containment provision provision (DCP). (DCP). (a) ( DCPa) DCP Type Type 1, (b 1,) DCP(b) DCP Type Type 2, (c) 2, DCP (c) TypeDCP 3.Type 3.

4.4.4.4. Derailment Test withwith aa WagonWagon (DCP(DCP IsIs Installed)Installed) TheThe nextnext derailment derailment experiment experiment was was conducted conducte withd with one wagon.one wagon. The DCP The of DCP Type of 1 wasType installed 1 was andinstalled analyzed and analyzed to evaluate to evaluate the derailment the derailment prevention prevention effects and effects vehicle and behaviorvehicle behavior when the when DCP the is installed.DCP is installed. The specifications The specifications of the wagon of arethe shownwagon Tables are 6shown–8[ 23 ,29Tables]. Additional 6–8 [23,29]. specifications Additional of bogiespecifications are referred of bogie in supplementary are referred in materials. supplementary materials.

Table 6. The speciﬁcation of welded bogie frame. Table 6. The specification of welded bogie frame.

ParametersParameters Values Values GaugeGauge (mm)(mm) 1435 1435 WheelbaseWheelbase (mm)(mm) 1800 1800 Weight (kg) 4500 LengthWeight (mm)(kg) 4500 3183 WidthLength (mm)(mm) 3183 2256 Frame WeightWidth (kg)(mm) 2256 1961 Frame Ixx (Kg m2) 1563 Frame Weight · (kg) 1961 Frame Iyy (Kg m2) 1114 Frame Ixx ·(Kg·m2) 1563 Frame Izz (Kg m2) 2574 · Frame Iyy (Kg·m2) 1114 Table 7.FrameThe speciﬁcation Izz (Kg of· wagonm2) body2574 frame.

Parameters Values Table 7. The specification of wagon body frame. Weight (ton) 5.0 Ixx Parameters(Kg m 2) Values 4166 · Iyy (Kg m2) 58,835 Weight (ton)· 5.0 Izz (Kg m2) 62,610 Ixx (Kg··m2) 4166

2 Table 8. TheIyy speciﬁcation (Kg·m of) bogie58,835 coil spring. Izz (Kg·m2) 62,610 Outer Inner Spring constant 42.35 kg/mm 76.36 kg/mm Table 8. The specification of bogie coil spring. Solid height 155 mm 138 mm Free length 270Outer mm Inner 228 mm Spring constant 42.35 kg/mm 76.36 kg/mm Solid height 155 mm 138 mm Appl. Sci. 2020, 9, x FOR PEER REVIEW 13 of 18

Free length 270 mm 228 mm

The primary suspension spring of the welded bogie consists of an inner coil spring and an outer Appl. Sci. 2020, 10, 118 13 of 18 coil spring. The outer spring operates at low loads and the inner spring operates when the load is increased and compressed to a certain displacement. Figure 17 shows the wagon frame. The left is a The primary suspensionrealspring shape frame of the model welded composed bogie consists of finite of elements an inner and coil the spring right andis a simplified an outer frame model. The coil spring. The outer springwagon operates used in at derailment low loads tests and is theshown inner Figure spring 18. operates when the load is Appl. Sci. 2020, 9, x FOR PEER REVIEW The wagon model consists of two main13 of parts, 18 bogie and body frame. To connect the bogie and increased and compressed to a certain displacement. Figure 17 shows the wagon frame. The left is a real frame, three beam elements were used. One beam element was for the center pivot and two elements shapeFree frame length model composed 270 mmfor side of finite bearer. 228 elements mm and the right is a simplified frame model. The wagon usedAppl. in derailment Sci. 2020, 9, x FOR tests PEER is shownREVIEW Figure 18. 13 of 18 The primary suspension spring of the welded bogie consists of an inner coil spring and an outer coil spring. The outer spring operates at low loads and theFree inner length spring operates 270 mm when the 228 load mm is increased and compressed to a certain displacement. Figure 17 shows the wagon frame. The left is a real shape frame model composedThe of primary finite elements suspension and springthe right of theis a welded simplified bogie frame consists model. of an The inner coil spring and an outer wagon used in derailment testscoil isspring. shown The Figure outer 18. spring operates at low loads and the inner spring operates when the load is The wagon model consistsincreased of two and main compressed parts, bogie to anda certain body displacement. frame. To connect Figure the 17 bogie shows and the wagon frame. The left is a frame, three beam elements realwere shape used. frame One beam model element composed was forof finitethe center elements pivot and and the two right elements is a simplified frame model. The for side bearer. wagon used in derailment tests is shown Figure 18. The wagon model consistsFigureFigure of 17. two 17.Di Different mainﬀerent parts, models models bo gie of the and wagon wagon body frame. frame. To connect Figure the 18.bogie View and of the test wagon. frame, three beam elements were used. One beam element was for the center pivot and two elements for side bearer. To verify that the simplified frame model and the real frame model have same behaviors, two models were evaluated under the same testing conditions in simulation.

Figure 17. Different models of the wagon frame. FigureFigure 18. 18.ViewView of the of test the wagon. test wagon.

To verify that the simplifiedThe wagonframe model model and consists the real of frame two mainmodel parts, have same bogie behaviors, and body two frame. To connect the bogie and models were evaluatedframe, under the threeFigure same beam 17. testing Different elements conditions models were ofin used.the simulation. wagon One frame. beam element F wasigure for 18. theView center of the test pivot wagon. and two elements for side bearer. To verifyTo verify that that the the simpliﬁed simplified frame frame model model andand the real real frame frame model model have have same same behaviors, behaviors, two two . modelsmodels were were evaluated evaluated under under the the same same testing testing conditionsconditions in in simulation. simulation. (a) (b)

. (a) (b)

(a) (b()c )

Figure 19. Comparison between the real frame model and the simplified frame model.

(a) Longitudinal displacement (b) Lateral displacement (c) Vertical displacement

(c)

Figure 19. Comparison between the real frame model and the simplified frame model.

(c) (a) Longitudinal displacement (b) Lateral displacement (c) Vertical displacement Figure 19. ComparisonFigure 19. Comparison between thebetween real frame the real model frame and model the and simplified the simplified frame model.frame model. (a) Longitudinal displacement (b) Lateral displacement (c) Vertical displacement. (a) Longitudinal displacement (b) Lateral displacement (c) Vertical displacement Appl. Sci. 2020, 9, x FOR PEER REVIEW 14 of 18 Appl. Sci. 2020, 9, x FOR PEER REVIEW 14 of 18 Appl. Sci. 2020, 10, 118 14 of 18 Figure 19 shows a comparison of displacement between the real frame model and the simplified Figure 19 shows a comparison of displacement between the real frame model and the simplified frameFigure model. 19 Displacement shows a comparison was measured of displacement at the cent betweener of the the frame, real frameand X, model Y, and and Z displacements the simplified showframe similar model. curves. Displacement Since comparison was measured of the at two the modelscenter of showed the frame, the similarand X, behavior,Y, and Z displacements the reliability frameshow similar model. curves. Displacement Since comparison was measured of the at two the centermodels of showed the frame, the similar and X, Y,behavior, and Z displacements the reliability ofshow the similarvehicle curves.models Sincewas verified comparison by comparing of the two with models the showed field test the and similar the simplified behavior, themodel reliability given inof thisthe study.vehicle models was verified by comparing with the field test and the simplified model given ofin thethis vehicle study. models was verified by comparing with the field test and the simplified model given in this study.The speed measured by the speed sensor right before the derailment was 52 (km/h). The first collisionThe after speed derailment measured occurred by the speed at about sensor 350 ms. right The before front thewheel derailment collided withwas the52 (km/h).seventh The sleeper. first collisionThe after speed derailment measured occurred by the speed at about sensor 350 ms. right The before front the wheel derailment collided waswith 52the (km seventh/h). The sleeper. first Figurecollision 20 after shows derailment that the collision occurred occurred at about at 350 the ms. seventh The front sleeper. wheel After collided the first with collision, the seventh the vehicle sleeper. proceedsFigure 20 and shows collides that the with collision the installed occurred DCP. at theThe seventh lateral displacement sleeper. After is the limited first collision, by the DCP the sovehicle that Figureproceeds 20 showsand collides that the with collision the installed occurred DCP. at theThe seventh lateral sleeper.displacement After is the limited first collision, by the DCP the vehicleso that theproceeds wheel anddoes collides not depart with from the installedthe rail within DCP. Thea certain lateral distance. displacement The configuration is limited by at the the DCP moment so that of collisionthe wheel with does the not DCP depart is shown from the in Figurerail within 21, and a ce itrtain is checked distance. that The collisions configuration with theat the DCP moment occur atof thecollision wheel with does the not DCP depart is shown from the in railFigure within 21, and a certain it is checked distance. that The collisions configuration with atthe the DCP moment occur ofat thecollision similar with location. the DCP is shown in Figure 21, and it is checked that collisions with the DCP occur at the the similar location. similar location.

Figure 20. FirstFirst collision collision with with concrete concrete sleeper. sleeper. Figure 20. First collision with concrete sleeper.

FigureFigure 21. CollisionCollision with with the the DCP. DCP. Figure 21. Collision with the DCP. Comparing damagedamage ofof the the track, track, Figure Figure 22 shows22 shows that that there there is a di isﬀ erencea difference in the degreein the ofdegree damage of damagebetweenComparing experimentbetween damageexperiment and simulation, of the and track, simulation, but Figure the collision 22 but shows the locations collisionthat ofthere the locations experimentis a difference of the and experimentin simulation the degree wereand of simulationreproduceddamage between were well reproduced because experiment the well failure and because occurredsimulation, the atfailure similarbut theoccurred locations. collision at similar locations locations. of the experiment and simulation were reproduced well because the failure occurred at similar locations. Appl. Sci. 2020, 9, x FOR PEER REVIEW 15 of 18 Appl. Sci. 2020 10 Appl. Sci. 2020, 9, x, 118FOR PEER REVIEW 15 of 18

Figure 22. Comparison of location of broken DCP. Figure 22. Comparison of location of broken DCP. Figure 22. Comparison of location of broken DCP. In the field test, only the seventh and eighth sleepers were damaged. In the simulation, damage In the ﬁeld test, only the seventh and eighth sleepers were damaged. In the simulation, damage occurredIn the at field the test,edge only of thethe ninthseventh and and tenth eighth sleeper sleeperss as werewell damaged.as the seventh In the and simulation, eighth sleepers.damage occurredIn the at field the edge test, of only the ninththe seventh and tenth and sleepers eighth sl aseepers well as were the seventhdamaged. and In eighth the simulation, sleepers. However, damage occurredHowever, at the the initial edge collision of the ninthoccurred and at tenth the same sleeper location.s as well Figure as the 23 showsseventh broken and eighthsleepers. sleepers. occurredthe initial at collision the edge occurred of the atninth the sameand location.tenth sleeper Figures as 23 well shows as brokenthe seventh sleepers. and eighth sleepers. However, the initial collision occurred at the same location. Figure 23 shows broken sleepers.

Figure 23. Comparison of location of broken sleepers. Figure 23. Comparison of location of broken sleepers. Figure 23. Comparison of location of broken sleepers. TheThe laterallateral displacementsdisplacements ofof thethe simulationsimulation andand fieldfield teststests areare shownshown inin FigureFigure 2424.. The simulation andand derailmentThederailment lateral experimentsdisplacements experiments showed showed of the asimulation maximuma maximum and diff fidifferenceerenceeld tests of 7.97%are of shown7.97% and 8.07%inand Figure 8.07% at the 24. at front The the simulation andfront center and points,andcenter derailment points, respectively, respectively, experiments and showed and showed showed the same a maximum the trend. same Figure trend.difference 24 Figure also of indicates 247.97% also andindicates that 8.07% if the that DCPat the if isthe front installed, DCP and is centeritinstalled, guides points, theit guides wheelrespectively, the of thewheel derailedand of showedthe bogiederailed the in samebogi the distancee trend.in the Figuredistance shown 24 Figureshown also indicates 25 Fi.gure From 25. that the From if result the the DCP ofresult the is installed,simulationof the simulation it andguides experiment, and the wheelexperiment, itof was the it confirmedderailed was confirmed bogi thate in if thatthe the distance DCPif the isDCP installed,shown is installed, Figure it could 25. it couldFrom prevent theprevent a result large a ofdeviationlarge the deviation simulation of the of derailed andthe derailedexperiment, vehicle. vehicle. it was confirmed that if the DCP is installed, it could prevent a large deviation of the derailed vehicle.

Figure 24.24. The comparison of Y displacement (lateral) between the ﬁeldfield testtest andand thethe simulation.simulation. Figure 24. The comparison of Y displacement (lateral) between the field test and the simulation. Appl. Sci. 2020, 9, x FOR PEER REVIEW 16 of 18 Appl. Sci. 2020, 9, x FOR PEER REVIEW 16 of 18 Appl. Sci. 2020, 10, 118 16 of 18

Figure 25. Available displacement between the DCP and the wheel. Figure 25. Available displacement between the DCP and the wheel. Figure 25. Available displacement between the DCP and the wheel. TheThe laterallateral accelerationacceleration for for the the initial initial 1 1 s wassecond compared was compared for a period for a of period colliding of colliding with the sleeperswith the andsleepers DCP,The and becauselateral DCP, acceleration the because DCP isthe designedfor DCP the isinitial designed based 1 onsecond based lateral was on motion. lateralcompared As motion. a for result a As period of a theresult comparison,of ofcolliding the comparison, with the two the accelerationthesleepers two acceleration and curves DCP, because were curves similar, the were DCP as similar, shownis designed as Figure shown based 26 Figure. on lateral 26. motion. As a result of the comparison, the two acceleration curves were similar, as shown Figure 26.

Figure 26. The comparison of Y-acceleration (lateral) between the field test and the simulation. Figure 26. The comparison of Y-acceleration (lateral) between the field test and the simulation. AccelerationFigure 26. The of thecomparison first collision of Y-acceleration was measured (lateral) as between 0.79 g whenthe field the test front and the wheel simulation. collided with Acceleration of the first collision was measured as 0.79 g when the front wheel collided with the the DCP at 0.5 s and acceleration of the simulation was 0.85 g. Acceleration of second collision was DCP at 0.5 s and acceleration of the simulation was 0.85 g. Acceleration of second collision was measuredAcceleration as 1.15 g of and the acceleration first collision of was the simulationmeasured as was 0.79 1.42 g when g. the front wheel collided with the measured as 1.15 g and acceleration of the simulation was 1.42 g. DCPThe at 0.5 overall s and trend acceleration of collision of acceleration the simulation was foundwas 0.85 to be g. similar. Acceleration This shows of second that the collision experiment was The overall trend of collision acceleration was found to be similar. This shows that the andmeasured the simulation as 1.15 g haveand acceleration similar behaviors of the derailingsimulation and was colliding 1.42 g. at the same time. experimentThe overall and the trend simulation of collision have similaracceleration behaviors was derailing found to and be colliding similar. atThis the sameshows time. that the 5.experiment Conclusions and the simulation have similar behaviors derailing and colliding at the same time. 5. Conclusions Protective facilities such as barrier walls are installed in a dangerous zone in order to reduce 5. Conclusions damageProtective after the facilities derailment such of as a train,barrier but walls research are installed on protective in a facilitiesdangerous with zone derailment in order behaviorsto reduce ofdamage railwayProtective after vehicles the facilities derailment is insu ffisuchcient. of as a Dynamictrain,barrier but walls simulationresearch are installed on is protective the most in a e dangerousfacilitiesfficient way with zone to derailment check in order the derailment behaviorsto reduce behaviorofdamage railway after in vehicles order the derailment to designis insufficient. aof protective a train, Dynamic but facility. research simulation However, on protective is if the we facilities most use a efficient model with derailment including way to acheckbehaviors car andthe bogiederailmentof railway frame behaviorvehicles consisting inis order ofinsufficient. finite to design elements, Dynamic a protective computing simulation facility. time is However, of the simulation most if efficientwe could use a way takemodel considerableto including check the a time.carderailment and Therefore, bogie behavior frame a simplified in consistingorder frame-modelingto design of finite a protective elements, technique facility. computing which However, can time be if used weof usesimulation in thea model design includingcould phase take of a theconsiderablecar derailmentand bogie time. protectiveframe Therefore, consisting facility a simplified was of proposed.finite frame-mo elements, Afterdeling thecomputing technique modeling time which of a simplifiedof can simulation be used frame in could the model, design take a full-scalephaseconsiderable of derailmentthe derailmenttime. Therefore, test wasprotective conducteda simplified facility to frame-mo verifywas pr theoposed.deling model. techniqueAfter the modelingwhich can of be a used simplified in the designframe model,phaseFrom of a full-scalethe the derailment comparison derailment protective results test was of facility simulation conducted was pr and tooposed. verify derailment Afterthe model. the tests, modeling we can deriveof a simplified the following frame conclusions:model,From a full-scale the comparison derailment results test wasof simulation conducted and to verify derailment the model. tests, we can derive the following conclusions:From the comparison results of simulation and derailment tests, we can derive the following (1) Since the analysis of post-derailment behaviors of trains takes an excessive period of time, the conclusions: simplified frame model using NODEL RIGID BODY INERTIA in Ls-Dyna was proposed. Appl. Sci. 2020, 10, 118 17 of 18

(2) In order to verify the reliability of the simplified model, actual derailment tests were conducted. The post-derailment behaviors were captured with a high-speed camera and lateral acceleration was measured. As a result, the simplified frame model reproduced the derailment behaviors well. (3) The impact accelerations were measured by acceleration sensors. When the data of the field test and the simulation results were compared during every stage of derailment, the acceleration curves and the maximum impact accelerations were similar. (4) The deviation prevention effects of DCP after derailment were verified through an experiment and simulation. DCP prevents large lateral deviation of wheels and a simplified frame model reproduced the derailment behaviors well when the DCP is installed

In this study, a freight wagon was used to validate the simpliﬁed frame model but the protective facility will be constructed in a high-speed train line. Therefore, a further study is necessary to evaluate design loads and locations of the facility using a high-speed train model with the same modeling technique.

Supplementary Materials: The data of the bogie used to support the findings of this study have been deposited in the National Digital Science Library in Republic of Korean repository. http://www.ndsl.kr/ndsl/ search/detail/report/reportSearchResultDetail.do?cn=TRKO201000018779, http://www.ndsl.kr/ndsl/search/detail/ report/reportSearchResultDetail.do?cn=TRKO201300032730, and http://www.ndsl.kr/ndsl/search/detail/report/ reportSearchResultDetail.do?cn=TRKO200300002651. Author Contributions: Conceptualization, I.-H.S., J.-W.K., and J.-S.K.; methodology I.-H.S., J.-W.K. and J.-S.K.; software I.-H.S., J.-W.K.; validation I.-H.S., J.-W.K. and N.-H.L.; writing—review and editing, I.-H.S. and J.-S.K.; supervision, J.-S.K., N.-H.L.; project administration, J.-S.K.; funding acquisition, N.-H.L. All authors have read and agreed to the published version of the manuscript. Funding: This research was supported by a grant (19RTRP-B122273-04) from Railway Technology Research Program funded by Ministry of Land, Infrastructure and Transport of the Korean government. Conflicts of Interest: The authors declare no conflicts of interest.

References

1. Aviation and Railway Accident Investigation Board. Investigation Report of KTX Train Derailment at Gyeongbu Expressway Gwangmyeong Station; Transport and Maritime Affairs; ARAIB/R 11-2; Ministry of Land: Sejong-si, Korea, 2011. 2. Planning Report on Development of the Concrete Track Technology with the Deviance Containment Provision; Transport and Maritime Affairs; Infrastructure R&D Report; Ministry of Land: Sejong-si, Korea, 2016. 3. Booz Allen Hamilton. Report on the Findings of: “Current Practice and Effectiveness of Derailment Containment Provisions on High Speed Lines”; Issue 1, Ref: R00673; HSL-Zuid Organisation: Zoetermeer, The Netherlands, 2004. 4. Rolling Stock and Vehicle Intrusion Protection for High-Speed Rail and Adjacent Transportation Systems TM 2.1.7. Available online: http://www.tillier.net/stuff/hsr/TM%202.1.7%20Intrusion%20Protection%20081025% 20TM%20Excerpt%20A.pdf (accessed on 22 December 2019). 5. Bae, H.-U.; Park, B.-J. Consideration on the Concept of DCP by Foreign Case Study. In Proceedings of the Autumn Conference & Annual Meeting of Korea Society for Railway, KSR2014A382, Jeju, Korea, 30 October–1 November 2014; pp. 1155–1159. 6. Evans, J.; Berg, M. Challenges in simulation of rail vehicle dynamics. Int. J. Veh. Mech. Mobil. Veh. Syst. Dyn. 2009, 47, 1023–1048. [CrossRef] 7. Pogorelov, D.; Yazykov, V.; Lysikov, N.; Oztemel, E.; Arar, O.F.; Rende, F.S. Train 3D: The technique for inclusion of three-dimensional models in longitudinal train dynamics and its application in derailment studies and train simulators. Int. J. Veh. Mech. Mobil. Veh. Syst. Dyn. 2017, 55, 583–600. [CrossRef] 8. Kovalev, R.; Yazykov, V.N.; Mikhalchenko, G.S.; Pogorelov, D.Y. Some Aspects of Wheel-Rail Contact Modeling and Optimization of Running Gears. Mech. Based Des. Struct. Mach. Int. J. 2003, 31, 315–334. [CrossRef] 9. Han, H.-S.; Koo, J.-S. Simulation of Train Crashes in Three Dimensions. Int. J. Veh. Mech. Mobil. Veh. Syst. Dyn. 2003, 40, 435–450. [CrossRef] 10. Bae, H.U.; Yun, K.M.; Moon, J.; Lim, N.H. Impact Force Evaluation of the Derailment Containment Wall for High-Speed Train through a Collision Simulation. Adv. Civ. Eng. 2018, 2018, 2626905. [CrossRef] Appl. Sci. 2020, 10, 118 18 of 18

11. Bae, H.U.; Yun, K.M.; Lim, N.H. Containment capacity and crashworthiness estimation of derailment containment wall on high-speed train. Proc. Inst. Mech. Eng. Part F J. Rail Rapid Transit 2018, 232, 680–696. [CrossRef] 12. Wu, X.; Chi, M.; Gao, H. Post-derailment dynamic behavior of a high-speed train under earthquake excitations. Eng. Fail. Anal. 2016, 64, 97–110. [CrossRef] 13. Wu, X.; Chi, M.; Gao, H. The Study of post-derailment dynamic behavior of railway vehicle based on running tests. Eng. Fail. Anal. 2014, 44, 382–399. [CrossRef] 14. Guo, L.; Wang, K.; Lin, J.; Zhang, B.; Chen, Z.; Song, X.; Du, G. Study of the post-derailment safety measures on low-speed derailment tests. Veh. Syst. Dyn. 2016, 54, 943–962. [CrossRef] 15. Ling, L.; Dhanasekar, M.; Thambiratnam, D.P.; Sun, Y.Q. Minimising lateral impact derailment potential at level crossings through guard rails. Int. J. Mech. Sci. 2016, 113, 49–60. [CrossRef] 16. Brabie, D. On the Influence of Rail Vehicle Parameters on the Derailment Process and its Consequences. Ph.D.Thesis, KTH, School of Engineering Sciences, Stockholm, Sweden, 2005. 17. Brabie, D.; Andersson, E. Post-derailment dynamic simulation of rail vehicles-methodology and applications. Veh. Syst. Dyn. 2008, 46, 289–300. [CrossRef] 18. Hirsch, T.J.; Harris, W.J.; James, R.W.; Lamkin, J.; Zhang, H. Analysis and Design of Metrorail-Railroad Barrier System; Research Report 3780-2 on Research Project TTI-3780; Texas Transportation Institute: College Station, TX, USA, 1989. 19. LS-DYNA Theory Manual, version 971; LSTC: California, CA, USA, 2007. 20. Lee, J.-H. Study on Prediction of Collision-induced Derailment of Virtual Testing Model Using Rolling Contact. Master’s Thesis, Seoul National University, Seoul, Korea, 2012. 21. Livermore Software Technology Corporation (LSTC). LS-DYNA Keyword User’s Manual; Livermore Software Technology Corporation: Livermore, CA, USA, 2007; Volume 1. 22. Ge, X.; Wang, K.; Guo, L.; Yang, M.; Lv, K.; Zhai, W. Investigation on Derailment of Empty Wagons of Long Freight Train during Dynamic Braking. Shock Vib. 2018, 2018, 2862143. [CrossRef] 23. Na, H. A Study on Integration Method for Effective Connection of Tile Two Korea’s Railway Systems (4th Year); Research report of Korea Institute of Construction & Transportation Technology Evaluation and Planning; National Digital Science Library: Daejeon, Korea, 2004; Research No.1500000416. 24. Bae, H.-U.; Min, K.H.; Kang, Y.S.; Lim, N.H. Test-field and Test-infra Construction for Derailment/Collision of Railway Train. Korean Soc. Civ. Eng. 2017, 65, 20–23. 25. Kim, J.H.; Bae, H.U.; Kim, J.U.; Song, I.H.; Lee, C.O.; Lim, N.H. Post-derailment Behavior of Casting Bogie by Full Scale Test. J. Korea Soc. Railway. 2018, 21, 815–829. [CrossRef] 26. Severson, K.J.; Tyrell, D.; Perlman, A.B. Rail Passenger Equipment Collision Tests: Analysis of Structural Measurements; ASME RTD; U.S Department of Transportation: Washington, DC, USA, 2000; Volume 19. 27. Jacobsen, K.; Tyrell, D. Rail Car Impact Test with Steel Coil: Collision Dynamics. In Proceedings of the 2003 IEEE/ASME Joint Railroad Conference, Chicago, IL, USA, 24 April 2003; pp. 73–82. 28. Jang, H.-J.; Shin, K.-B.; Han, S.-H. A Study on Crashworthiness Assessment and Improvement of Tilting Train made of Sandwich Composites. World Acad. Sci. Eng. Technol. Int. J. Mech. Mechatron. Eng. 2012, 6. [CrossRef] 29. You, W.H. Design Optimization Study on Bogie Mechanism; Research report of Korea Railroad Research Institute; KRRI-Research-01-51; National Digital Science Library: Daejeon, Korea, 2001.

© 2019 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).