Lifetime-Oriented Structural Design Concepts- P5 doc

It provides a dynamic load on the wall structure within a narrow bandwidth of frequencies determined by the train speed V.. Full scale tests performed along the high speed line Cologne-R

2.3 Transport and Mobility 77

10 2 10 3 10 4 10 5 10 6 10 7

number of vehivles and axles per year

100 200 300 400 500 600 700 800 900 1000 1100 1200 1300

Fig 2.55 Traffic records from the Netherlands recorded in 2006

periods of one or two years In this case a further increase of these transportscan be expected and it cannot be excluded that a significant percentage ofthese transports is overloaded A possible increase in the number of such vehi-cles in combination with a possible overloading has especially to be consideredfor the development of future fatigue load models

A comparable development takes place in other European countries Figure2.55 shows the vehicle weight and axle load distributions recorded in 2006 nearthe harbour of Rotterdam in the Netherlands It can be seen that the extremevalues of the gross weight and also the extreme values of the axle loads aresignificant higher than the values of the Auxerre traffic (see Figure 2.24).The shape of the distribution shows that the heavy load transports lead incomparison with the Auxerre traffic to a new shape of the distribution whichcould be taken into account by splitting the distribution into a distributionfor normal traffic and a distribution for heavy load transports

Additionally the transport industry is extremely interested in new port concepts at present In some European countries and also in some Ger-man federal states field trials take place with modular vehicle concepts, the

trans-so called Giga-Liners with gross weight up to 600 kN and a total length of25.25 m [314] Typical vehicles and the corresponding allowable axle loads areshown in Figures 2.56 and 2.57 These types of vehicles have significant highertransport capacities and can reduce the transport cost At present it cannot

be foreseen how the future traffic composition will change Some people gue that the new modular concept will reduce the total number of lorries onroads due to the higher transport capacity On the other hand it has to be

ar-Vehicles acc to the modular concept (gross weight up to 600kN)

25,25 m 16,50-18,25 m

Current trucks in Germany

Giga – Liner with gross weight of 600 kN

Giga – Liner with gross weight of 580 kN

Fig 2.57 Axle spacing and allowable axle weights of ”Giga-Liners”

considered that this new type of vehicle can not be loaded on trains, sothat it can be expected that no significant reduction of the total road trafficwill occur First investigations [201] show that especially for bridges withlonger spans the current European load model has to be modified, whenthe percentage of the new vehicles reaches 20% to 40% related to the to-tal heavy traffic Furthermore at present no information is available regardingthe driving of such vehicles in convoys, especially on routes with acclivities,and the possible overloading and wrong loading which can lead to higheraxle weights

2.3 Transport and Mobility 79

The new traffic concepts and development regarding heavy transports neednew technologies to get more detailed information about the actual trafficsituation and also a more close cooperation between the car industry and theauthorities and experts for the development of realistic traffic models TheWeight in Motion (WIM) is a technology [407, 588] for the determination ofthe weight of vehicles without requiring it to stop for weighting The systemuses automated vehicle identification to classify the type of the vehicle andmeasures the dynamic tyre force of the moving vehicle when the vehicle drivesover a sensor From the dynamic tyre load then the corresponding tyre load of

a static vehicle is estimated The most common WIM device is a piezoelectricsensor embedded in the pavement which produces a charge that is equivalent

to the deformation induced by the tyre loads on the pavements surface mally two inductive loops and two piezoelectric sensors in each monitoringlane are used

Nor-The system can be used in combination with an automatic vehicle sification system (AVC) Vehicles which do not meet the gross weight andaxle weight requirements are notified with dynamic message signs While inthe USA this systems are used in some states all over the country, in Eu-rope only in some countries these systems are used on special routes Firstfield trials with combined WIM and AVC methods take place presently inthe Netherlands The records demonstrate that besides the problem that thetotal weight of the vehicles exceed the permissible total weight there are alsocases where the permissible total weight is not exceeded, but due to wrongloading of the vehicles the weight of single axles is significantly higher thanthe permissible axle weight This can lead to excessive fatigue damage espe-cially in orthotropic decks of steel bridges and also in concrete decks Thesenew traffic records demonstrate that in the future a better cooperation be-tween bridge designers and truck producers is necessary Strategies to avoidsuch overloading of single axles could be the implementation of immobilisersystems in trucks if single axles or the total gross weight of the truck areexceeded

clas-2.3.2 Aerodynamic Loads along High-Speed Railway Lines

Authored by Hans-J¨ urgen Niemann

Shelter walls often accompany high-speed railway lines for noise tion or to provide wind shelter for the trains The walls consist of verticalcantilevered beams connected by horizontal panels The pressure pulses fromhead and tail of the train induce a pressure load on the walls, which is ingeneral smaller than the wind load However, the load is dynamic which maycause resonant amplification The load is furthermore frequent which mayrequire design for fatigue These issues are the topic of the following chapter

protec-Fig 2.58 Pressure time history at the track-side face of a 8 m high wall; at a fixed

position; V = 234.3 km/h, [573]

2.3.2.1 Phenomena

As a train passes, a sudden rise and drop of the static pressure occurs tures at the trackside, such as noise barrier or wind shelter walls, in turnexperience a time variant aerodynamic load [777] It is caused by the pressuredifference over the wall sides facing the track and the rear face The load in-tensity of this aerodynamic loading is proportional to the square of the trainspeed

Struc-Figure 2.58 shows a pressure time history measured at a fixed position at

the trackside surface of a wall, 1.65 m above rail level The wall distance to the track axis is a g = 3.80 m Typically, the head pulse starts with a posi-

tive pressure which is followed by a negative pressure approximately identical

in magnitude The subsequent tail pulse is reversed and its amplitudes aresmaller unless the train is short For short vehicles, head and tail pulse maymerge and the negative pressure may dominate Additional pulses occur atinter-car gaps with amplitudes much smaller than head and tail pulses Themeasured time history clearly depends on the train speed If instead of thetime history the load pattern along the wall is considered, it becomes inde-pendent of the train speed Figure 2.59 gives an example

The pattern of the pulse sequence travels along the wall at the train speed

It provides a dynamic load on the wall structure within a narrow bandwidth

of frequencies determined by the train speed V Furthermore, a spectral composition shows that the distance Δx of the positive and negative pulses is related to the prevailing frequency Figure 2.59 gives two values of Δx mea- sured at a track distance of a g = 3.80 m at two different train speeds The

de-effect of the train speed is within the scatter of the experimental results

2.3 Transport and Mobility 81

Fig 2.59 Pressure distribution along the track-side face of a wall at two different

train speeds [573]

Fig 2.60 Full scale tests performed along the high speed line Cologne-Rhine/Main:

view of the trough; (a) measuring the train speed, (b) with measurement set-up atthe eastern wall

A spectral decomposition shows that the prevailing frequency f p is in theorder of

Depending on the natural frequencies f n of the wall or any other trackside

structure resonance may occur at a critical train speed V res ≈ 2.7Δxf n, which

in turn may cause considerable fatigue at rather few train passages The

maximal pressure amplitude measured at a train speed of 304 km/h = 84.6 m/s is ca 0.550 kN/m2 Typical wind loads are larger by a factor of 2 to

4 It has been argued that the load effect will become important only at very

high speeds beyond 300 km/h (see [617]) In fact, the aerodynamic load does

not dominate the design as long as the train speed is sufficiently below thecritical If however the critical speed is lower than the maximal track speed,resonant amplification will provide the dominant design situation

Fatigue damage occurred at protection walls along a high speed railway line

in 2003 Previous investigations e.g [36] had dealt with the static effect of thepulse and developed simplified design loads which cover the static action effect.However, they did not consider to model the loading process in view of thedynamic load effects Therefore, additional investigations became necessarywith a focus on the dynamic nature of the load One issue concerned full-scalemeasurements of the aerodynamic load patterns along the wall and over thewall height, and the relation of natural wall frequency to the critical trainspeed The following findings rely on the results of a campaign performed in

2003, see [573] The measurements were performed along a concrete wall inorder to avoid disturbances coming from the strong deformations of some ofthe walls

2.3.2.2 Dynamic Load Parameters

The streamlined shape of nose and tail, as well as the frontal area do not onlydetermine the drag of the train but also the pulse amplitudes As well, thenose length affects the distance between the pressure peaks The ERRI-report[36] identifies three typical train nose shapes and gives load reduction factors

as follows:

freight trains k1= 1, 00;

express trains with V max = 220 km/h k1= 0, 85;

high speed trains (TGV, ICE, ETR) k1= 0, 60.

The dynamic stagnation pressure of the train speed clearly governs the dynamic pressures Figure 2.61 is based on the pressures at the track-side wallsurface

aero-The diagram relates the measured pressure peaks of the head pulse, positiveand negative, to the dynamic head of the train speed:

2.3 Transport and Mobility 83

(a)

(b)

Fig 2.61 3 Effect of train speed stagnation pressure on the head pulse acting at

the track-side face of a wall; (a) positive pressure; (b) negative pressure

Figure 2.62 shows the pattern of the head pulse in terms of pressure cients The peak coefficients of±0.15 are typical for the well shaped, slender

coeffi-nose of the ICE 3 train The mean values are somewhat smaller

The detailed coefficients c p obtained for 152 train passages are:

peak pressure maximum c p = 0, 1499

mean pressure maximum c p = 0, 1380

lowest pressure maximum c p = 0, 1049

peak pressure minimum c p=−0, 1520

mean pressure minimum c p=−0, 1419

highest pressure minimum c p=−0, 1041

Fig 2.62 Pressure coefficients of the head pulse from 34 passages (at the track-side

wall face) at 1.65 m above track level

Fig 2.63 Distance between the pulse peaks and the zero crossing (ΔL1= pressure

maximum, ΔL2 = pressure minimum)

The dynamic effect is related to the distance between the pulse peaks As is

seen in Figure 2.63 a mean distance of Δx = 6.9 m is typical for the ICE 3 passing at a track distance of 3.80 m.

At a train speed of 300 km/h, the related frequency is f p = 4.5 Hz

Natu-ral frequencies of light protection walls are in the same order of magnitude.Obviously, the critical train speed may happen and its dynamic effect maybecome important

2.3 Transport and Mobility 85

Fig 2.64 Head pulse in a free flow at various distances from the track axis [98]

Fig 2.65 Head pulse in the presence of a wall

The results refer to a distance between the wall and the track axis of

a g = 3.80 m This parameter plays an important role both for the amplitude

of and the distance between peaks Figure 2.64 shows the result obtained

the-oretically regarding the pressure pulse in a free flow As the track distance a g increases, the peak amplitudes max p and min p decrease whereas the separa- tion Δx between the pulse peaks increases.

Theory predicts that in free flow without walls, the separation Δx depends linearly on the track distance a g, see e.g [98]

Experimental results can best be fitted by a slight modification:

Figure 2.65 shows the head pulse in the presence of a wall for two different

distances The measurements at a track distance of 3.80 m and 8.30 m were

performed simultaneously i.e at identical train speeds at different walls, both

8 m high The distance of the peaks at the wall decreases similar to the freeflow case However, the results indicate that the effect of the track distancebecomes non-proportional in the presence of a wall An analogous approxima-tion matches the test results

in which a g,ref = 3.8 m is used as reference.

The pressure amplitudes decrease with the inverse of the square of the trackdistance Various empirical expressions take account of this theoretical result.The following formula developed in [36] is widely accepted:

c p,max = k1

2.5 (a g + 0.25)2 + 0.025

(2.70)

Introducing the pressure at a g = 3.80 m as a reference, the peak pressure

amplitude at any distance becomes

c p,max (a g ) = c a · c p,max (3.8) =

14.1 (a g + 0.25)2+ 0.14

c p,max (3.8) (2.71)

For a g = 8.3 m, the formula gives a wall distance factor of c a = 0.333 The experimental result is in this case a decrease by a mean factor of 0.3 The

formula presented is a conservative estimate

The pressure varies over the wall height Figure 2.66 is an example of apressure pattern measured at a wall, 8 m high The pressure intensity decreases

at the upper end This end effect coincides with a shift of the pulse peaksbetween wall foot and top, meaning that they do not occur simultaneously ateach level

Figure 2.67 shows the time lag between head pulse maximum and

mini-mum as it varies over the height of a 3.5 m wall The measurements include various train speeds, the time lag has been transformed to V = 300 km/h.

The maxima occur simultaneously at each level, whereas the minimum is notsimultaneous but lags increasingly at higher levels This will in general di-minish the dynamic load effect A conservative approximation is to assumeidentical and simultaneous pulse patterns at each level Finally, the pressuremagnitudes depend on the wall height The experiments show that the pres-sures measured at low levels are higher in magnitude at high walls compared

to lower walls The pulse between the walls apparently levels out more rapidlywhen the walls are low A convenient wall height factor is:

2.3 Transport and Mobility 87

Fig 2.66 Load pattern over the height of the wall

The results refer here to H W ref = 3.50 m.

2.3.2.3 Load Pattern for Static and Dynamic Design Calculations

The following expression summarizes the observed effects and may be applied

to static and in particular to dynamic design calculations:

c WH factor accounting for the wall height;

c p pattern of the pressure coefficient at low levels acc to Figure 2.69;

Fig 2.67 Variation of the time lag between maxima and minima of the head pulse

over the wall height transformed to V = 300 m/s

Fig 2.68 Load factor for the load distribution over the height of the wall

c z load factor accounting for the pressure variation over the wall heightacc to Figure 2.68;

c a load factor accounting for the wall distance from the track axle;

ρ mass density of air;

V train speed in m/s;

a g track axle distance;

x distance from zero-crossing of the head pulse;

z height above rail level

2.3 Transport and Mobility 89

(a)

(b)

Fig 2.69 Pattern of pressure coefficientsc pfor the ICE-3 train: (a) pressure ence between track-side and rear-side faces of the wall; (b) pressure at the track-sideface

differ-The speed of an adverse wind has to be added to the train speed where

required The load factor c z in fig 2.68 neglects the phase shift occurringtowards the top and is valid for any wall height

Figure 2.69 shows the reference load pattern The stochastic componentsuperimposed on the pressures by the boundary layer turbulence has beensmoothed out by averaging The head pulse at the track-side face (b) is sym-metric Considering the net pressure, the rear-side pressure has to be included.The measurements in ref [229] include the required data They show thatthe pressure maximum on the rear side precedes the track-side maximum.Therefore, regarding the net pressure the pulse maximum increases whereasthe minimum decreases The effect on the remaining load pattern is notnoticeable

(a) (b)

Fig 2.70 Noise protection wall (a): height 3.50 m above track level; post

dis-tance 5.00 m; lightweight panels (b) Mode shape of the 1st mode; natural frequency

f1= 4.67 Hz

The formula includes the wall distance effect on the pressure amplitude as

a constant factor It does not include the increasing distance between sure maximum and minimum In general, calculations of the dynamic loadeffect may be restricted to the head pulse It governs the dynamic amplifica-tion of the response A simple and sufficient approximation applicable to thesymmetric load pattern is

of 2.00 or 5.00 m Figure 2.70 (a) shows an example.

It is rather laborious to model the dynamic behaviour of the structure.The transient response involves large parts of the wall between recesses Theattempt was misleading to identify the dynamic response at a single pole

in a 1-D model Similarly, the natural frequencies and the relevant modeshapes cannot be identified realistically in a simplified model: as an example,the panels have to be included as 2-D plates since their torsional stiffnesscontributes considerably to the system stiffness Figure 2.70 (b) shows the 1stmode shape which is excited dominantly by the pulse load

2.3 Transport and Mobility 91

Fig 2.71 Time history of post top displacement calculated for a post in the middle

of the wall; displacement in m, positive direction outward

The natural frequencies are not well separated For the wall shown above,

the first 4 modes range from 4.67 Hz to 4.90 Hz, the 12th mode shape has a natural frequency of 6.04 Hz which is still rather close to the first one.

The post top displacement from time history calculations, s Figure 2.71indicates that the wall moves outward at the pulse maximum As it swingsback, the negative pulse amplifies the movement: the 1st inward amplitude is

ca twice the 1st outward This is a consequence of resonance

The effect of natural frequencies on the resonant amplification of the placement may be studied in a simplified manner using modal decomposition.The response time history is calculated for a static behaviour and for various

dis-natural frequencies A critical damping ratio of D = 0.05 was adopted

inde-pendent of the natural frequency The dynamic amplification of the response r

is characterized by two resonant amplification factors:

max ϕdyn=max r

rstat min ϕdyn= min r

rstat (2.75)

The Figures 2.72 and 2.73 show how the resonance factors depend on the ural frequency and the train speed, i.e the pulse time lag Both factors displayidentically that the maximal amplification is independent of the natural fre-

nat-quency with a value of max ϕdyn= 2.0 and min ϕdyn= 2.6.

The range of natural frequencies where peak resonance occurs is however

not identical in the two cases At a train speed of 300 km/h, a natural quency of 3.8 Hz provides the highest amplification of the outward displace-

fre-ment whereas the inward displacefre-ment is amplified most strongly at a natural

frequency of 4.6 Hz The wall considered suffers strong resonant vibrations.