Paper No. 2007-01-4236

Development of A Self-energizing Electro-Hydraulic Brake(SEHB)

Matthias Liermann, Christian Stammen, Hubertus MurrenhoffRWTH Aachen University, Germany

Institute for Fluid Power Drives and ControlsSteinbachstr. 53, 52074 Aachen, Germany

Copyright c©2007 SAE International

Abstract

A new hydraulic brake utilizing a self-energizing effect isdeveloped at the Institute for Fluid Power Drives and Con-trols (IFAS). In addition to a conventional hydraulic brakingactuator, it features a supporting cylinder conducting thebraking forces into the vehicle undercarriage. The brakingforce pressurizes the fluid in the supporting cylinder and isthe power source for pressure control of the actuator. Thenew brake needs no external hydraulic power supply. Theonly input is an electrical braking force reference signalfrom a superior control unit. One major advantage of theSEHB concept is the direct control of the actual brakingtorque despite friction coefficient changes.

The prototype design, presented in this paper, is done intwo phases. The first prototype is based on an automotivebrake caliper. It is set up to gain practical experienceabout the hydraulic self-energisation and to prepare thelaboratory automation environment. Active retraction isrequired for train brakes though, which cannot be donewith automotive brakes. The second prototype thereforefeatures a differential double acting braking cylinder witha pre-stressed spring for fail safe braking. A mechanicaldesign systematics is presented which helps to maprequirement specifications originating from a particularapplication to implementation in a structural design.The promising dynamic behavior of SEHB based onsimulation results is presented.

Keywords: hydraulic, self-reinforcement, self-energizing,brake, pressure control, trains, SEHB, simulation

1 Introduction

The new brake concept of Self-energising Electro-Hydraulic Brake (SEHB) combines high force-to-weight

ratio of hydraulic actuation with flexibility of digital controland efficiency by using the principle of self-energisation.It is being developed at the Institute for Fluid Power andControls (IFAS, RWTH Aachen University) within a re-search project funded by the DFG (German ResearchFoundation), see Fig. 1. The SEHB concept was firstlyintroduced in [1] and [2], with a study on the anticipatedperformance by nonlinear simulation. A first analyticalstudy on the brake dynamics on the basis of a linearisedmodel is found in [3]. SEHB offers the advantages of hy-draulic brake actuation without disadvantages of a cen-tralized hydraulic power supply. This is possible by theprinciple of self-energisation. The wheelset’s inertia mo-mentum is used by each calliper as the source of powerto supply hydraulic pressure for braking.

Figure 1: Integrated, intelligent single-wheel traction and brak-ing module (EABM), funded by German ResearchFoundation

The brake principle and its advantages over conventionalbrakes is explained in section 2

On the basis of the Performance specification given insection 3, section 4 and section 4 describe the mechani-cal and hydraulic design parameters of the first brake pro-totypes developed at IFAS. A systematics of mechanicaldesign configurations is given of which two are chosen tobe implemented as prototypes. Both prototype designs

are shortly presented in this section.

Section 6 presents simulation results using parameters ofprototype design II and discusses its dynamic brake be-havior.

2 Working principle of SEHB

The idea of SEHB is, that the pressure needed for actu-ation of a hydraulic disc brake is gained from hydraulicsupport of the friction force. Unlike conventional brakes,where the brake caliper is fixed, in the SEHB concept it ismovable tangential to the friction contact. A hydraulic sup-porting cylinder connects the calliper to the bogie struc-ture, thus fixing it between two columns of oil, see Fig. 2.

Figure 2: Principle of Self-energising Electro-Hydraulic Brake,(SEHB)

In case of braking, the friction force acts on the support-ing cylinder, causing a pressure difference. Independentof which chamber is pressurised and which is released,the configuration of four check valves assures that thelower line has the higher pressure and the upper line islow pressure. A control valve connects high- and low-pressure side with the chambers of the brake actuator.If the right flow-scheme of the control valve is active, high-pressure is applied on the piston face, while the ring sideof the brake actuator is connected to low-pressure. De-pending on the ratio between piston areas this leads to aself-energising process of brake force increase becausethe supporting pressure is always higher than the brakepressure, which would be needed to cause this support-ing pressure. This process has been studied analyticallyin [3]. If the left flow-scheme of the control valve is active,the piston face chamber is relieved to low pressure side,while the piston ring side is charged with the decreasingpressure of the supporting cylinder. The accumulator al-ways contains enough pressurised fluid from a previousbraking to enable the active lifting of the brake pads fromthe brake disk even when no pressure is left in the sup-porting cylinder.

Without a closed loop control, with the brake valve be-ing in its right control scheme, the braking pressure wouldconstantly rise. The closed loop control acts to close thecontrol valve when the desired friction force is achieved.As a matter of principle, the SEHB needs a feedback loopto be stable. This feedback is not drawn in Fig. 2. Oneway to realize it, would be by measuring the load pressurein the supporting cylinder with pressure transducers andmagnetic actuation of the valve by a controller device. Analternative could be a hydro-mechanic feedback, wherethe pressure in the supporting cylinder is used to actuatethe spool of the control valve. The set value would be ahydraulic or mechanic actuation. In that case no electriccomponents would be needed.

While the necessity of a closed loop control might look likea drawback of SEHB at first, it is also one of its major ad-vantages: It allows the direct control of the actual brakingforce, independently of friction coefficient changes. Theload pressure in the supporting cylinder is the control vari-able of the closed loop control of SEHB. Since the loadpressure is in direct relation to the friction force, this offersthe possibility to control the actual retardation torque onthe brake disk. Conventional friction brakes only controlthe perpendicular actuator force. Since the friction co-efficient µ is influenced by parameters like speed, brakepressure and temperature, conventional brakes can onlyestimate the actual friction force Fbrake and the retarda-tion torque respectively. The retardation torque, however,is the control variable for vehicle dynamics control sys-tems like the Electronic Stability Program (ESP). This isobvious, since the vehicle dynamics is influenced by theretardation of a wheel and not by the actuator force ofthe brake calliper. Tab. 1 summarizes the advantagesand disadvantages of SEHB over a conventional hydraulicbrake.

pro conno external hydraulic in-terface

required space for movingcaliper

closed loop brake con-trol can deliver data aboutactual braking torque tosuperiour control systems(autonomous braking)constant braking torqueirrespective of friction co-efficient changes

Table 1: Pros and cons of SEHB toward conventional hydraulicbrakes

3 Performance specification ofSEHB

The performance specification for SEHB is derived fromthe target application of the brake, defined by the research

project “EABM”, funded by the German Research Foun-dation (DFG). The basic specification is based on the pre-sumption that a passengers railway car does not neces-sarily need to be heavier than a comparable road vehicle,a bus.

• Maximum speed: v0 = 120 kmh

• Maximum waggon load: m = 13.6 t

• Two pairs of individual wheels, four disc brakes

• Diameter of wheel (new / old): dwheel =920 mm/840 mm

Figure 3: Requirement specifications for design of SEHB

General performance requirements for railway brakes arenormed for specific railway types, [5]. For a maximumstopping distance of 500 m at maximum velocity and anestimated response time of 0.8 s, the brake must providea maximum deceleration of a little less than d = 1.2 m

s2 .The maximum retardation force Fd is calculated by multi-plying the mass inertia per disc brake times decelerationplus a constant force resulting from slope of s = 4 % andgravity g, Fig. 3. The rotary inertia of wheels and drives isincluded with a factor kr = 1.1 in the translatory inertia.

Fd =m

4(krd + sg) = 5822 N (1)

The maximum friction force Fbrake acting on a friction ra-dius of rf = 245 mm then yields:

Fbrake = Fddwheelnew

2 · rf= 10931N (2)

4 SEHB prototype design -hydraulic aspects

This section describes the hydraulic design parametersof the first SEHB prototype, which are a result of itera-tive steps of design and simulation analysis. Beside thecylinders, the control valve and the accumulators, alsothe springs in the brake actuator for initiation of the self-energisation and in the supporting cylinder for automaticretraction of the supporting cylinder are covered in thissection because they interact with the pressure forces andpiston friction forces

Cylinders The supporting cylinder has a piston diam-eter of d1Sup = 40 mm and a piston rod diameter ofd2Sup = 25 mm. At maximum brake force (Eq. 2), takinga transmission ratio between brake force and supportingforce of iL = 1.8 into account (Eq. 4), the supporting forceis FSup = 6072.8 N resulting in a maximum pressure ofpmax = 79.3 bar.

The size of the actuator follows from the precondition ofself-energisation, [2]:

ASup

ABA

!≤ µ · 2iL

(3)

According to Eq. 3 for a minimum friction coefficient of µ >0.14 a differential cylinder with piston diameter of d1BA =80 mm is sufficient. It has a piston rod diameter of d2BA =50 mm with 70 mm stroke.

Control valve(s) The brake valve indicated in Fig. 2 canbe realized by a

• 4/3-way control valve or a

• Combination of 2/2-way (switching) valves

A drawback of conventional 4/3-way control valves is theirleakage in the closed position. Since each side of the sup-porting cylinder brake contains just 38 ml, leakage of thecontrol valve in the closed position cannot be tolerated.Concerning the size, simulation shows, that the nominalflow should be relatively small (4 l

min @ 35 bar per land)because of the small capacities in the system. A naturalfrequency of 30 Hz is fast enough for closed loop control.The low value accentuates the fact that robust and rea-sonably priced components can be used.

Alternatively the use of four 2/2-way fast-switching valvesfor brake control has been studied and is scheduled fortrial in the prototype design. Fast switching valves as usedin anti-lock braking systems (ABS) are leakage proof inthe closed position.

Accumulators The high-pressure accumulator has astorage capacity of 11 ml, enough for retracting the ac-tuator for more than 3.5 mm. It features a pre-stressedspring that allows it to by filled starting with 15 bar. It iscompletely filled at a pressure of 21 bar. The expansiontank has a storage capacity of 141 ml. Fully charged itgenerates a system pressure of around 2 bar on the low-pressure side. The accumulator is fully charged when thebrake piston is completely retracted.

Springs Braking of SEHB results in movement of thesupporting cylinder. Therefore after every braking thesupporting cylinder should retract to the middle position.The retraction springs in the supporting cylinder are pre-stressed with 200 N to overcome friction and pull thecylinder into middle position when the switching valve,connecting the chambers of the supporting cylinder, isopened. The cumulative stiffness of the springs is 2 N

mm .

The spring in the actuator presses the brake pad on thedisk to initiate the self-energisation. Self-energisation isinitiated, when the brake force produces a certain pres-sure level in the supporting cylinder. The level is reached,when the force of the actuator caused by this pressure ex-ceeds the force caused by the initiating spring. The calcu-lation of the required spring force for successful initiationwas derived in [3]. In the present design the spring ap-plies 1920 N in middle position of the actuator and has astiffness of 16 N

mm .

5 SEHB prototype design -mechanical aspects

The previous section focused on components realizing thebrake function. This section focuses on design decisionsconcerning the mechanical implementation of these func-tions for a specific application.

For the design of the SEHB prototype the over-all conceptcan be split into subtasks, each with a range of possiblesolutions. The combination of these solutions builds thespace of possible solutions. An advantage of this theo-retic approach is, that it allows an easier documentation ofdifferent solutions through classification. Also, it system-atically reveals characteristics of different arrangements.These characteristics can be compared to design require-ments and restrictions of the target application. Followingstructural alternatives for sub-tasks of the SEHB will beconsidered in the following paragraphs.

1. Brake actuator

• Fixed caliper or pin slide caliper

• Retraction of brake pad passive or active

2. Actuator guidance

• Radial support: exact circular guidance

• Linear support: shifting friction radius

• Curve approximation / guidance mechanism

3. Alignment of supporting cylinder on vector of brakeforce

• Pivot point in or sideways of brake surface

• Axis of supporting cylinder in line with or paral-lel displaced to vector of brake force

4. Mounting orientation of supporting cylinder

• Piston rod attached to brake actuator (cylinderfixed)

• Cylinder attached to brake actuator (piston rodfixed)

5. Supporting cylinder

• Double rod cylinder

• Differential cylinder

• Arragements of 2 independent plungers

• Integrated design of 2 plungers

• Rotary actuator

5.1 Brake acutator

Two major types of brake actuators can be distinguished,the fixed and the pin-slide caliper, Fig. 4. Usually thebrake disk is axially fixed and the brake pads are mov-ably guided in the brake by a part, which supports thetangential brake force. The brake actuator generates thenormal force between brake pads and disk. To avoid lat-eral forces on the disk, both pads should apply the samenormal force. This can be achieved by two separate hy-draulic plungers with equal piston areas in a fixed caliper(left in Fig. 4), pressing from both sides. In this case thebracket guiding the brake pads and the plungers can beintegrated into one part. Another solution is a caliper con-sisting of two parts (right in Fig. 4). One part, the bracket,is connected to the bogie structure. It is responsible for lat-eral guiding the pads and conducting the brake force intothe fixed structure. The other part is guided on pin-slidesperpendicularly to the disk connected to the bracket. Itencompasses the brake disk and has a brake actuator onone side only. When the actuator is pressurized, it pushesboth pads symmetrically onto the brake disk.

One advantage of a fixed caliper is, that it has no movingparts except the pistons. The hydraulic connection canbe stiff piping instead of flexible hose. With the conceptof SEHB, during braking there is always a movement be-tween brake actuator and pressure source, though. Ad-vantages of a pin-slide caliper over a fixed caliper are:

Figure 4: Fixed and pin-slide caliper brake

• Reduced installation space

• Reduced weight

• Less external sealings

• Only one contact surface between piston and brakepad reduces heat transfer into fluid

The retraction of the brake piston is necessary to lift thebrake pads off the disk. In many brake designs the re-traction is realized passively. The slight unevenness ofthe disk pushes back the brake piston while at the sametime it is pulled by the elasticity of a deformed sealing or aspring. The gap of typically 40− 60 µm between pads anddisk of conventional automobile brakes is produced by thesealing elasticity. For trains it is common practice to havea gap of 2 − 3 mm, which is guaranteed by pre-stressedsprings. The reason for the huge gap of train brakes is thatthere is no stiff connection between caliper and wheelset.The brake pads are hanging on links mounted to the bogiewhich can move vertically or tilt in relation to the wheelsetdue to the spring-damper system connecting them. Toprevent frequent or permanent touch between brake padsand disk, which can lead to glazing of the brake lining, thegap is set to a higher value. For safety reasons, as fail-safe state a complete deflated system must be assumed.Therefore an active retraction is needed, which uses thepressurized fluid in the high pressure accumulator fromthe previous braking for active retraction. Fig. 5 showsthe difference between passive and active retraction of thebrake piston.

5.2 Actuator guidance

A unique characteristic of SEHB is the brake pad move-ment in direction of the brake force, producing the neces-sary hydraulic power in the supporting cylinder, which in

Figure 5: Passive and active retraction of single and double act-ing cylinder

turn is used for brake actuation. This movement must beguided in some way to avoid an overlap of the brake padsbeyond the disk. While it may be considerable in some ap-plications to allow small overlap, an exact or approximatedcircular guidance is certainly required in most cases. Twooptions to realize this guidance are depicted in Fig. 6

Figure 6: Circular guidance through radial bearing or slider

The most obvious way to realize circular guidance is toconnect the caliper with a radial bearing to the brake shaft,as shown left in Fig. 6. The radial bearing appeals to besimple at first but it has some significant disadvantages.Firstly it produces loss in the drive train. Moreover, it mustbe designed separable, in applications where it cannot bemounted on one of the axletree’s ends. The shaft radius ofa trains wheel-set is typically around 180 mm, while for thepurpose of the bearing 40 mm would be sufficient. The at-tachment of the caliper on the wheel-set is contrary to to-day’s service procedures. Before changing the wheel-setthe brake would have to be disconnected from the shaft.Therefore alternative solutions for circular guidance arenecessary. Right side in Fig. 6 depicts a solution using aslider guided in a circular groove. The suitability of this so-

lution in the rough environmental conditions that a bogieis exposed to is questionable. A third solution using onlysimple joints is the application of a Watt-I 6 link mecha-nism. The Watt-I linkage, where the first beam (0;6) isfixed, is shown in Fig. 7.

Figure 7: Watt-I linkage

With part (1) and (3) being straight beams crossing eachother in the middle and (2), (4) and (5) each half as longas (1) and (3), beam (5) is guided as if rotating aroundpoint (P), which is the intersection of (0) and (5). Thismechanism is applied on SEHB in Fig. 8. The advantageof this 6 link mechanism is, that no machine parts have tobe near (P). The brake can be completely mounted to thebogie.

Figure 8: Circular guidance of the caliper by a 6 link mechanism

5.3 Alignment of supporting cylinder onvector of brake force

A support of the brake force by the supporting cylinderwith a vertical or horizontal displacement leads to stress

in the parts between brake bracket, which holds the brakepads, and its links to the wheelset and bogie. This factis illustrated in Fig. 9. The brake force vector lies in thecenter of the brake disk tangential on the friction radius.By principle, the supporting cylinder carries only an ax-ial load. Therefore, a vertical displacement between thesupporting cylinder and the brake force (see left side inFig. 9) leads to a bending momentum in the bracket, whichis compensated by additional bearing forces in the jointsconnecting the bracket to the wheelset and the supportingcylinder. Also flexural stress is caused in the bracket. Bothnecessitates stronger links and components. A horizontaldisplacement (right side in Fig. 9) adds another momen-tum in vertical direction.

Figure 9: Point of application of supporting and brake force

The attachment of the supporting cylinder(s) directly ontothe guiding bracket of the brake pads would be ideal tominimize flexural stress in the brake.

5.4 Mounting orientation of supportingcylinder

The supporting cylinder should carry only axial forcessince lateral forces increase the friction force of the cylin-der and cause wear of the guide sleeves. The easiest wayto eliminate transverse forces is the use of spherical joints,which can be mounted on the piston rod ends but not onthe cylinder. Only a differential cylinder offers the possi-bility to be fixed by two spherical joints, one mounted tothe piston rod and the other to the bottom end of the cylin-der. The double rod cylinder is normally mounted over twolinkage stubs attached to both sides of the cylinder. Theyallow a one-dimensional tilting only and have to be alignedprecisely to prevent transverse forces. For the generationof the supporting pressure, it does not mind whether thepiston rod is attached to the caliper and the cylinder to thebogie or the opposite way. For the mechanical design itmakes a difference because the piston rod end and thesperical joint are much smaller than the cylinder. Advan-tages resulting from the attachment of the cylinder to thecaliper are:

• Integrated design possible of supporting cylinderand caliper

• Single serviceable unit

• Simple mechanical connection to bogie

• No hydraulic lines between bogie and caliper

Advantages resulting from the attachment of the pistonrod to the caliper are (see upper right of Fig. 10):

• Compact caliper

• Hydraulic valves mounted on cylinder are shock pro-tected

• Reduced unsprung mass

• Connection of supporting cylinder to bracket withinbrake radius less complex

5.5 Supporting cylinder design

The type of design of the supporting cylinder has not yetbeen discussed. Fig. 10 is an overview over possible con-figurations for different supporting cylinder types.

Figure 10: Configurations using single or double rod cylinders

To facilitate bi-directional braking, both, a double rod or adifferential cylinder, come into question, see solution (1)to (4) in Fig. 10. However, one requirement of a brake fortrains is that the braking effect is independent of the vehi-cles direction of motion. Therefore, the supporting cylin-der must have equal piston areas.

The double acting cylinder can be divided into twoplungers as shown in solution (5) to (10). Using plungersoffers greater design flexibility, which can be useful whenaiming on large scale integration. Plungers are also cost-efficient for mass production. Another advantage is that

the diameter of a plunger compared to a double rod cylin-der with the same piston area is much smaller.

A very interesting solution is the integration of twoplungers into one actuator as shown in (11). It looks likea differential cylinder but it features two equal piston ar-eas. Pressurizing chamber A leads to a pushing force.Pressurizing the ring area in chamber B leads to a pullingforce. This design has significant advantages. It is verycompact in length and not much wider than the double rodcylinder. Spherical joints can be mounted on both ends.A drawback is from a production viewpoint that it requiressmall tolerances. The cylinders have to fit very well intoeach other.

To complete this systematics, rotary actuators (12) and(13) can be used to support the brake force. At the firstglance they may facilitate very compact solutions, allowinglarge scale integration. Though, a difficulty is the ratherhuge leakage compared to a cylinder. It could lead to anearly exhaustion of the supporting cylinder during a brak-ing.

5.6 Prototype I on basis of automobilebrake calliper

By mapping the requirement specifications with charac-teristics of the design options presented in this section, itis possible to choose a set of configurations coming intoquestion for the first prototype design. Of course bound-ary conditions such as project time line and already ex-istent parts also have influence on the final configuration.Therefore the prototype design of SEHB is done in twophases. The purpose of prototype design I is the demon-stration of the self-energizing effect and its successfulclosed loop control. The verification of the simulationmodel by measurements serves as a basis for further de-tailed theoretical studies. It will presumably trigger studyof new aspects which were not observed by simulationso far. In this way measurements, validating the SEHBconcept are obtained at an early project phase. The firstprototype is designed on the basis of a conventional auto-mobile brake, as seen in Fig. 11.

The test rig features a brake disk driven by a hydraulicvariable displacement motor in secondary control modein combination with a flywheel. The brake shaft ends withthe brake disk. As a caliper a pin-slide caliper with passiveretraction of the brake actuator by elasticity of a squareseal is used. A circular guidance is realized by polymerplastics sleeve bearings. The point of brake force appli-cation into the supporting cylinder is on the opposite sideof the caliper initially on the brake radius. For testing pur-poses it can be adjusted vertically with a slider mecha-nism ±10 mm. The bending momentum in the part con-necting the caliper to the pivot and the supporting cylinderis relatively high. However, only a small flexural stressis caused because of the small gap between supporting

Figure 11: Prototype I on the basis of a conventional automobilecaliper

piston and disk. The supporting cylinder is a double rodcylinder. The arrangement (2) of the cylinder between testbed and caliper shown in Fig. 10 was used because an at-tachment of the cylinder onto the caliper would have beencomplicated.

5.7 Prototype II featuring double-acting dif-ferential cylinder

Since the research project, within which the brake is de-veloped, aims at an integrated module consisting of driveand brake, the prototype design should not only provethe SEHB concept, but also hold out the prospect of suc-cessfully implementation in a train. This regards safety(fail-safe), installation space, brake functions (wear adjust-ment, gap between pads and disk, service brake, parkingbrake, etc.) and comfort (performance of closed loop con-trol control, noise). Therefore prototype II is designed witha double acting brake actuator for active retraction and afail safe braking concept realized by a prestressed springas shown in Fig. 12.

The point of brake force application into the supportingcylinder is in-plain above the brake disk. The attachmentof the supporting cylinder above the caliper leads to areduced supporting force compared to the braking force.The gear transmission ratio is

iL =Fbrake

FSup=

rsup

rfric= 1.8 (4)

The bending momentum in the part connecting the caliperto the pivot and the bows connecting it to the support-ing cylinder is lower than in prototype I. Flexural stressis minimized because the supporting force is in-plain withthe disk. The supporting cylinder is a double rod cylinder.

Figure 12: Prototype II featuring a double acting brake actuator

The arrangement (1) of the cylinder between test bed andcaliper shown in Fig. 10 was used.

6 Simulation of SEHB Prototype II

A DSHplus simulation model of the presented prototypedesign is used to predict the brake’s dynamic behaviorand facilitates the study of influencing parameters suchas piston friction, pressure dependent bulk modulus andcontained air in the fluid.[4]. Fig. 13 illustrates the layoutof the model. It is comprised of a hydraulic section withcylinders, accumulators and valves, and a signal sectionincluding the state dependent friction coefficient and theclosed loop control.

controlunit

friction coefficient � �P|�P|

area- & levertransmission

factor

�F

Fref

expansiontank

high-pressureaccumulator

Fbrake

brake returnchamber

brakechamber

brake valve

PT2

brake actuator

supportingcylinder

controlunit

Factual

pHP

pLP

controller

FSup

joint levergear transmission

ratio iL

Figure 13: Layout of the brake simulation model

Simulation parameters The parameters of the cylin-ders, valves, accumulators and springs were derived inSec. 4 from the requirement specifications. In additionto the already discussed design parameters, the follow-

ing parameters were used for the presented simulationresults:

The friction of the cylinders is modeled on the basis ofa Stribeck Curve, featuring a brakeaway force at the be-ginning of a motion, a region of mixed friction, where thefriction force decreases with rising speed, and a region ofviscous friction starting at higher velocities. The support-ing cylinder is parametrised with 50 N breakaway force, amixed friction of 30 N at 0.1 m

s and a viscous friction of10 Ns

mm . The breakaway force of the brake actuator is es-timated to be higher (200 N) mainly because of the largerdiameter. The prediction of the friction forces is very un-certain because of its pressure dependency which is notincluded in the model.

The parameters of the fluid simulate the behavior ofHLP 46 hydraulic fluid. Pressure dependency of the bulkmodulus and the influence of contained air is accountedfor. The bulk modulus has a significant influence on theperformance in the phase of initiating the self-energizingeffect of the brake, as proved by simulation. The simula-tion results shown below were yielded for an undissolvedair content of 0.01%.

The mechanical stiffness of the brake caliper, brake lin-ings and brake disc is estimated to be 75 kN

mm . Atthe beginning of a simulation a clearance of 3 mm isparametrised between brake pads and disc.

The frictional force between brake pads and disk is cal-culated using a characteristic diagram. It was derived inthe context of this research project from test data suppliedby a manufacturer of brake linings, [3]. Fig. 14 shows thefriction coefficient in relation to velocity and pressure usedfor the simulation.

Figure 14: Characteristic diagram of friction coefficient over ve-locity and pressure

The friction model facilitates a more realistic simulation.The friction coefficient rises while the vehicle is deceler-ating. Therefore the brake controller will act to minimizethe resulting brake force deviation. For very low braking

pressures, as they occur in the initiation phase, the frictioncoefficient and the self-energisation respectively is lowerthan for higher pressure. This leads to more realistic eval-uation of the rise time of the brake.

Simulation results The simulation results provide evi-dence of the high dynamics of the brake. Special focus isgiven on the system’s dead time td, the time constant Tc ofthe closed loop control dynamics and times tretract for re-tracting of the brake actuator and tregen for regeneration ofthe supporting cylinder. The dead time td can be definedas the period between brake demand and 90 % achieve-ment of the set value. The time constant Tc is definedsimilar to the time constant of a first-order delay system.It is the time between the intersections of a tangent onthe steepest part of the control variable trajectory with thestarting and the set value.

The response of the brake can be demonstrated partic-ularly well with sudden changes in the reference inputvalue. This is not intended to be the simulation of a typicalrail vehicle braking operation, which, of course, is not sud-den for reasons related to passenger comfort and safety.The achievable brake dynamics plays an important role forwheel slide protection performance and constitutes one ofthe main advantages of hydraulic systems over pneumaticbrakes.

The simulation is divided into three phases, see Fig. 15:

1. Initiating braking with maximum braking forceFbrake = 10931 N, t = 1s − 2.6s

2. Quick changes of reference value, Fbrake = 5000 N−10931 N, t = 2.6 s − 3 s

3. Venting the brake and setting the air clearance,Fbrake = 0 N, t = 3 s − 4 s

At t = 1 s, the control valve opens fully due to the highcontrol deviation. Because of the spring in the actua-tor the clearance of 1.5 mm on each side of the disk isoverridden in only 0.890 s. From that time on the self-energising effect is initiated starting with a rise of pressurein the supporting cylinder, which is consequently moving.After a total of td = 1.27 s 10 % of the reference value isachieved. The process of self-energisation speeds up dueto the higher bulk modulus of the oil at higher pressures.In this phase contained air is most troublesome becauseit leads to longer dead times. The set value is achievedafter t = 1.42 s, which is in the range of todays pneumaticsystems.

At t = 2.6 s two step changes in the reference valuedemonstrate the high closed loop brake dynamics ofSEHB. Fig. 16 is the enlargement of the upper chart inFig. 15 showing the step changes in the set value and thecontrolled brake force in response.

Figure 15: Simulation of braking with SEHB prototype II

Figure 16: Simulation of braking with SEHB prototype II

The time constant Tc of the closed loop control is verysmall compared to pneumatic systems. It is slightly dif-ferent for falling (31 ms) and rising (42 ms) set value. Thetime until 10% of the reference step is reached is between66 and 87 ms. For a comparison: In this period the brakedisk has not finished one revolution at full vehicle speedof 33 m

s . It is interesting to notice that this change in brakeforce does not significantly exhaust the supporting cylin-der, as can be seen in Fig. 15. This is important for slid-ing protection algorythms, where the brake force set valuechanges quickly for a period of time.

At the moment of setting the brake force set value to 0 kNthe servo-valve opens in negative direction and relaxesthe compressed fluid of the brake piston to the expansiontank. The brake actuator releases, as shown in Fig. 15.During venting the brake the high-pressure line is con-

nected to the surface of the brake actuator piston ring,increasing the relaxing effect. The surface of the brakeactuator piston ring is smaller than the surface of the pis-ton face, which means that less volume flow is requiredfor the return stroke. The brake control opens the servo-valve negatively as long as needed for lifting the actuator3 mm in total, which is the predetermined air clearance of1.5 mm on each side. The time for complete retraction istretract = 0.41 s.

While the brake pad is lifted from the disk, the supportingcylinder regenerates through the integrated pre-stressedsprings. With the given spring parameters, the regenera-tion takes place within tregen = 1 s. For cases in which thistime for complete regeneration is not allowable, the strokeof the supporting cylinder must be long enough to providebraking power for several brakings.

7 Conclusion

This article focuses on the major hydraulic and mechan-ical design aspects of a self-energizing hydraulic brake.For each mechanical aspect, such as kind of brake actu-ator, actuator guidance, geometric configuration and kindof supporting cylinder design, different solutions are dis-cussed indicating their specific advantages and disadvan-tages. The matrix of possible solutions in combinationwith practical boundary conditions leads to a prototype de-sign in two steps. Goal of the first step is the verification ofthe SEHB principle using parts from an automotive brake.With a second prototype design, the specific demands ofa brake for railways are met. The simulation of the brakeis used iteratively in the design process and helps for theunderstanding of the interdependency of design parame-ters. Simulation results showing the brake dynamics arepresented on the basis of the second prototype designand fulfill the basic requirements. Future work will focuson implementation of the prototype and verification of itsmodel with measured data.

8 Acknowledgments

The authors thank the German Research Foundation(DFG) for funding this project.

References

[1] Matthias Liermann, Christian Stammen Self-Energising Hydraulik Brake for Rail Vehicles[Selbstverstarkende hydraulische Bremse furSchienenfahrzeuge - Intelligentes, Integriertes

Einzelrad-Antriebs-Brems-Modul], Olhydraulik undPneumatik O+P 10/2006, 500-507

[2] Matthias Liermann, Christian Stammen, Huber-tus Murrenhoff Development of A Self-energisingElectro-Hydraulic Brake (SEHB) for Rail Vehicles,The Tenth Scandinavian International Conference onFluid Power, SICFP’07, May 21-23, 2007, Tampere,Finland

[3] Matthias Liermann, Christian Stammen, HubertusMurrenhoff Pressure tracking control for a self-energizing hydraulic brake, Symposium on PowerTransmission & Motion Control, PTMC’07, Sept 12-14, 2007, Bath, England

[4] DSHplus Fluid Power Simulation Software by FluidonGmbH, Aachen, www.fluidon.de

[5] DIN EN 13452-1 Railway applications – Braking –Mass transit brake systems - Part 1: Performancerequirements, 2005

9 Contact

Dipl.-Ing. Matthias Liermann, born 1977 in Essen,Germany, studied Mechanical Engineering at the Techni-cal University Aachen.

Since 04/2004 he participates as scientific staff in re-search projects at the Institute for Fluid Power and Con-trols (IFAS), Technical University Aachen. His special in-terests are modeling and control of hydraulic systems.His main project is the the development of the Self-energizing Electro-Hydraulic Brake (SEHB) together withDr.-Ing. Christian Stammen.

Email: matthias.liermann@ifas.rwth-aachen.de

Dr.-Ing. Christian Stammen, born 1976 in Neuss,Germany, studied Mechanical Engineering at theTechnical University Aachen.

Since 04/2001 he participates as scientific staff inresearch projects at the Institute for Fluid Power andControls (IFAS), Technical University Aachen. Hefinished his Ph.D thesis 01/2005 about methods forcondition monitoring of hydraulic systems which wasrewarded with the Gertraude-Holste-Reward. He initiatedthe development of SEHB and was rewarded for his workon SEHB with the “Wissenschaftspreis NRW” in 05/2007.

Email: christian.stammen@ifas.rwth-aachen.de

Univ.-Prof. Dr.-Ing. Hubertus Murrenhoff is ExecutiveDirector of the Institute of Fluid Power Drives andControls (IFAS) in conjunction with the correspondingchair at the University of Technology (RWTH) Aachen,Germany. The renamed Institute IFAS was the formerIHP headed by Professor Backe until 1994.

Email: hubertus.murrenhoff@ifas.rwth-aachen.de

Web:www.ifas.rwth-aachen.de