1-s2.0-s0950061814004097-main

ls

h i g h l i g h t s

prayedexperimof the innularrespon

variation of the sliding force during the cyclic loading history has been

age. The rst one consists in minimizing structural damage byadding supplemental damping devices [10,38], either viscous orhysteretic. The second strategy consists in reducing the seismicinput energy by means of seismic base isolation systems [1,17].

Dealing with the rst strategy, the development of supplemen-tal damping devices started in New Zealand about 40 years ago

idges worldwide,ns.is basedenergy. In

elements, the energy is usually dissipated by means of the slbetween two surfaces in contact, which are clamped by mthe application of hydraulic pressures, electromagnetic forin the simplest case, by means of high strength bolts. In particular,this last clamping method is, due to its simplicity, probably themost applied in civil engineering practice. In fact, by adopting highstrength bolts, it is possible to apply a constant force on one ormore surfaces in contact by simply governing the value of thetightening torque and the number and diameter of the bolts.

Corresponding author. Tel.: +39 089964342.E-mail addresses: [email protected] (M. Latour), [email protected] (V. Piluso),

[email protected] (G. Rizzano).

Construction and Building Materials 65 (2014) 159176

Contents lists available at ScienceDirect

Construction and B

evto remain safe, even though a certain amount of structural damageis accepted, in case of rare seismic events. Concerning last case, assoon as the energy balance under seismic loading conditions is con-sidered, it is clear that there are two main strategies to limit dam-

systems have been installed in buildings and brboth for seismic retrot and for new constructio

A wide category of supplemental dampersfriction for dissipating the earthquake inputhttp://dx.doi.org/10.1016/j.conbuildmat.2014.04.0920950-0618/ 2014 Elsevier Ltd. All rights reserved.on drythese

ippageeans ofces or,DampingDevicesExperimentalSlip connections

investigated by comparing also the response coming from the use of different washers: circular atwashers and cone shaped annular disc springs.The work is aimed at the investigation of friction materials to be applied within the connecting

elements of beam-to-column joints according to the double split tee conguration with friction pads. 2014 Elsevier Ltd. All rights reserved.

1. Introduction

Modern seismic resistant structures need to be designed in ordertowithstand frequent earthquakeswithout signicant damages and

[2,9,18,33]. In particular, in the past few decades, the developmentof supplemental damping systems has received a great attention ofacademics and engineers leading to the development of a numberof dissipative devices [3,6,10,19,24,33,34,37]. Many of theseKeywords:Friction

has been analyzed. The The frictional behavior of steel, brass, s The cyclic behavior of six interfaces is Static and dynamic friction coefcient The inuence of at washers or cone a An analytical model able to predict the

a r t i c l e i n f o

Article history:Received 25 November 2013Received in revised form 10 April 2014Accepted 11 April 2014Available online 20 May 2014aluminum and three types of rubber all sliding on steel is investigated.entally tested under cyclic loads.nterfaces is evaluated.springs on the magnitude of the sliding force is investigated.se of three of the investigated interfaces is proposed.

a b s t r a c t

In this paper, the friction coefcient and the cyclic response of different interfaces for friction devices areinvestigated by means of experimental tests under displacement control. In particular, six interfaces havebeen tested: steelsteel, brasssteel, sprayed aluminumsteel and three different rubber based frictionmaterials adopted, respectively, in automotive applications, electrical machines and applicationsrequiring low wearing.Static and kinetic friction coefcients have been evaluated and the inuence of the interface pressureExperimental analysis on friction materiadevices

M. Latour , V. Piluso, G. RizzanoUniversity of Salerno, Civil Engineering Department, Salerno, Italy

journal homepage: www.elsfor supplemental damping

uilding Materials

ier .com/locate /conbui ldmat

BuildFriction dampers usually fall into the category of displacementactivated dampers, because their sliding force is not dependent onthe velocity and frequency content of the excitation. The cyclicbehavior of friction dampers is usually described by means of arigid-plastic response. Therefore, the only parameter needed bythe designer is the slip force which, in turn, depends on the valueof the load normal to the surfaces in contact and on the friction coef-cient which is an intrinsic characteristic of the sliding interface. Agreat advantage of friction devices is that, they can be used to workas displacement reducers under service conditions while they candissipate the seismic input energy under severe seismic actions.

The friction coefcient depends on different phenomena, suchas adhesion, ploughing and the presence of contaminants. Themodeling of these phenomena is usually studied in tribologywhere, in order to develop theories for predicting slip forces understatic and dynamic loads, the surfaces topography, materialshardness, mechanical properties and the effects of interface layersare physically modeled. Conversely, in structural engineering, theproperties of friction materials are typically studied by followingthe experimental approach which, for seismic engineering scopesis usually retained sufcient to provide the information neededfor designing such devices.

In the technical literature several works are concerned with thecharacterization of the hysteretic behavior of sliding metallic sur-faces with different supercial treatments clamped by means ofhigh strength friction grip bolts. This case is particularly signicantfor civil engineering purposes, because the greatest part of frictiondampers developed since the 1970s to be used for dissipativebraces or links adopts this approach. One of the rst devices ofsuch type was that developed by [25] to be introduced at the inter-section of braces, which adopted asbestos brake lining padsbetween the steel sliding surfaces. One of the simplest forms offriction damper has been proposed by [36] who adopted simplebolted slotted plates located at the end of a conventional bracingmember. The brace-to-frame connection was designed to slipbefore yielding or buckling of the brace. In this device, friction isdeveloped through the sliding of steel surfaces and, in order tomaintain constant the slip load, disc spring washers were used.Another friction damper for chevron braces was proposed by [23].

An issue of paramount importance for systems using bolts aspreloading elements is the maintenance of the preloading levelduring the device lifetime. In fact, the uctuation of the bolt pre-load can lead in some cases to unstable hysteresis loops, so thatthe amount of energy dissipated can be somewhat unpredictable.In addition, under cyclic loading conditions, the wearing of the fric-tion interface can lead to the partial loss of bolt preloading and,therefore, to the degradation of the slip force.

Within this framework, in this paper, an experimental study onfriction materials to be applied to supplemental damping devices iscarried out. In particular, six different interfaces are considered:steelsteel interface, brasssteel interface, sprayed aluminumsteel interface and three interfaces adopting different types of fric-tion rubber-based materials. All specimens are clamped by meansof high strength bolts and are tested under cyclic loading condi-tions. The work is aimed at understanding the potentialities ofthe considered materials to design new friction devices to bedirectly applied within the connecting elements of dissipativebeam-to-column joints in MR-Frames [15,20,21], adopting asclamping method high strength bolts [22].

2. Friction theories

From the historical standpoint the major part of past tribology

160 M. Latour et al. / Construction andstudies have been addressed to the investigation of friction proper-ties of metals recognizing that there are two main sources of fric-tion between sliding bodies: adhesion and ploughing. The adhesioncomponent arises because when two surfaces are loaded againsteach other, asperities deform plastically leading to the formationof the so-called cold-weld junctions. Because of the intimate con-tact of these junctions, the shearing of the adhesive ties requires acertain sliding load. Regarding ploughing, it is due to the naturalsurfaces roughness, so that the relative movement between thesurfaces in contact requires that one body has to lift over the other.

The simplest theory to mathematically explain the origin of theadhesion component is due to [4] who state that, being adhesiondependent on the shear resistance of the cold-weld junctions, ithas to be proportional to the real contact area which, for metalswith ideal elasticplastic behavior can be assumed equal toA N=r0, where A is the real area of contact, r0 is the materialpenetration hardness and N is the load normal to the surfaces.The total friction force due to adhesion (FA) can be expressed as:

FA As Nr0 s 1

being s the force per unit of area needed to shear cold-weldjunctions.

As already stated ploughing is the friction force caused by theasperities of an hard metal penetrating in a softer metal. Accordingto Bowden and Tabor theory, this contribution is estimated asfollows:

FP nrhr0 2where n is the number of asperities, r is the half-width of the asper-ity and h is the height of the asperity. Therefore, the total slidingforce (F) due to adhesion and ploughing is given by:

F FA FP Nr0 s nrhr0 3

The ploughing component is very important during the abra-sion process but, in case of metals, it has been demonstrated thatsuch contribution is negligible compared to adhesion. Therefore,Eq. (1) explains a very important property for metals, stating thatthe ratio between the frictional force and the normal applied loadis a constant value which does not depend on the apparent area ofcontact. Practically, Bowden and Tabor theory explains two of thethree postulates of the classical theory of dry friction, stating that:

The total frictional force is independent of the apparent surfacearea of contact.

The total frictional force that can be developed is proportionalto the normal applied action.

In case of slow sliding velocities, the total frictional force isindependent on the sliding velocity.

The rst two postulates are often known as Amontons laws,after the French engineer who presented them in 1699, while thethird one, is due to Coulomb [13,26].

During slippage, the classical relationship to compute the tan-gential force acting at the sliding interface in the direction opposedto the motion is the well-known Coulomb friction equation F = lN,where F is the sliding force, N is the normal action and l is the fric-tion coefcient. The force of friction is always exerted in a directionopposed to the movement (in case of kinetic friction) or potentialmovement (in case of static friction). According to Eq. (1) the fol-lowing relationship is obtained:

l s0r0

4

where s0 is the critical shear stress of the weaker material and r0 is

ing Materials 65 (2014) 159176the hardness of the softest material. Eq. (4) provides reasonableresults for metals, but in case of rubber-based materials, such asfriction pads, the coefcient of friction should be computed

accounting for other three effects: the contact pressure (P), thesliding speed (v) and the temperature (T). Therefore, in generalthe coefcient of friction of a rubber based interface should beexpressed as l = l(P, v, T).

In case of rubber based materials, the particular structure of thematerial inuences its frictional characteristics. In fact, rubber hasa low elastic modulus and its real contact area is strongly affectedby the magnitude of the normal load, because the material adaptsto the shape of the surface asperities of the hardest material [39].The behavior of polymeric materials deviates from the classicalfriction theory. In fact, tribology of polymers is inuenced by theadhesive junctions, the shear resistance of the rubbing materialin contact and the real contact area [29]. The coefcient of frictionof polymers, depending on the considered range of applied normalload and on the type of polymer, may be represented by means ofconstant or decreasing relationships [5,28,30,32]. In particular,several mathematical relationships have been proposed to modelthe friction coefcient of steelrubber interfaces, by expressing las a function of the contact pressure (P) and of the material elasticmodulus (E). Some of them are herein reported:

whereexperi

l K 6

where

where l is the value of the friction coefcient when the pressureis inness d

dependent on the sliding velocity. This behavior is due to theviscoelastic behavior of polymers. Notwithstanding, usually formany polymeric materials, contact temperature does not varysignicantly during the loading process provided that the inuenceof velocity is negligible for a limited range (0.011 cm/s).

3. Tested friction materials

In order to investigate the frictional properties of differentinterfaces, a sub-assemblage constituted by a layer of frictionmaterial or metal placed between plates made of S275 steel [7]has been developed (Fig. 1). In order to allow the relative move-ment of the steel on the interposed material, one of the inner plateshas been realized with slotted holes.

Conversely, the other inner and the two outer plates have beenrealized with circular holes. The clamping force has been appliedby means of preloaded M20 bolts 10.9 class [8] and the holes havebeen drilled with a 21 mm drill bit. Aiming to evaluate the magni-tude of the friction coefcient and the cyclic response, severaldifferent layouts of the sub-assemblage have been considered

M. Latour et al. / Construction and Building Materials 65 (2014) 159176 161deviation from the classical friction theory has to be pointed out.In fact, in case of polymeric materials, the friction force can be1nite, a is an experimental constant and h is the Shore hard-ivided by 100. Finally, in case of rubber, another important [27]:

l l1 aPh 7E

the value of K and n have to be experimentally found; [29]:

P 1=na and b are empirical parameters to be found by means ofmental testing; [35]:

1l a b P

E

5Fig. 1. Scheme of the adopted sub-avarying three parameters: the interface, the tightening torque,the number of tightened bolts and the type of bolt washers. Inparticular, the friction properties of the following six differentinterfaces have been investigated (Fig. 2):

Steel on steel. Brass on steel. Sprayed Aluminum on steel. Friction material M0 on steel. Friction material M1 on steel. Friction material M2 on steel.

In the testing program, two different types of washers havebeen used. In the rst part of the experimental activity, circular atsteel washers have been used, while in the second part of the cam-paign a packet of steel disc springs has been interposed betweenthe bolt head and the steel plate. In addition, the experimentalanalysis has been carried out by varying the bolt tightening levelin the range 200550 N m, obtaining different values of the clamp-ing force acting on the sliding surface. The tightening torque hasbeen applied in each of the loading sequences summarized inTable 1 by means of a calibrated torque wrench. Even though thessemblage and testing machine.

ra

BuildMet

allic

Mat

eria

ls

Steel S275JR B

r M

ater

ials

162 M. Latour et al. / Construction anduse of bolt gauges allows a better calibration of the clamping force,the use of a calibrated torque wrench has been preferred, becauseit is commonly applied in the erection of bolted steel structures. Asa consequence, the obtained testing results are referred to condi-tions close to those occurring on construction site.

The main goal of the experimental program is to obtain the fric-tion coefcients of the investigated materials, both static andkinetic, for values of the normal force varying in a range leadingto sliding forces suitable for structural applications and for valuesof the velocity compatible with seismic engineering applications.In addition, the experimental analysis is also devoted to the evalu-ation of the variation of the sliding force as far as the number ofcycles of the applied loading history increases. In fact, as alreadydemonstrated by [25], the response of an interface subjected tocyclic loading conditions can substantially be of two types. The rsttype of response provides a continuously softening behavior. Inthis case, the maximum sliding load is reached during the rstcycle and in all the subsequent cycles a degrading behavior isexhibited. The second type of response is characterized by three

Rub

be

Material M0 Materi

Fig. 2. Texture of the

Table 1Summary of the tests carried out in the experimental program.

Interface Torque Number of bolts of frictio

Steelsteel 200 N m 4300 N m500 Nm

Brasssteel 200 N m 8300 N m

Sprayed aluminumsteel 200 N m 4300 N m400 Nm

M0steel 300 N m 8400 N m400 Nm500 Nm

M1steel 200 N m 8300 N m400 Nm550 Nm

M2steel 200 N m 8300 N m400 Nmss Sprayed Aluminum

ing Materials 65 (2014) 159176phases: rst an hardening response, then a steady state phaseand nally a load degradation phase.

The tests have been carried out by means of a universal testingmachine Schenck Hydropuls S56 (Fig. 1). The testing equipment isconstituted by a hydraulic piston with loading capacity equal to630 kN, maximum stroke equal to 125 mm and a self-balancedsteel frame used to counteract the axial load. In order to measurethe axial displacements the testing device is equipped with anLVDT, while the tension/compression loads are measured bymeans of a load cell. The cyclic tests have been carried out underdisplacement control for different amplitudes at a frequency equalto 0.25 Hz.

From the experimental results presented in the following sec-tions, it can be recognized that the obtained displacement historydoes not always correspond to the imposed one. The problem isdue to the ability of the testing equipment and of its electroniccontrol system in reproducing the imposed displacement historywhose accuracy reduces as far as the testing frequency increases.However, this does not constitute a problem for the interpretation

al M1 Material M2

tested materials.

n pad Number of cycles of the sequence Amplitude

10 30 mm101030 15 mm2010 15 mm101010 7.515 mm101010101010 10 mm10101010 15 mm1010

of testing results, because the actually developed displacementhistory is accurately measured. Therefore, given the testing equip-ment, the testing frequency has been selected as a compromisebetween the opportunity to accurately control the applied dis-placement history and the need to test the specimens at frequencylevels comparable to those occurring under seismic actions.

4. Experimental tests with circular at washers

The main goals of the experimental activity described in this paper are on onehand, the evaluation of the static and kinetic friction coefcients for different valuesof the normal force acting on the sliding interface and, on the other hand, theassessment of the cyclic response in order to evaluate the stability of the obtainedcycles and their energy dissipation capacity. The test results reported in the follow-ing paragraphs are represented in terms of friction coefcient, which can be easilydetermined as:

l FmnNb

8

wherem is the number of surfaces in contact, n is the number of bolts, F is the slidingforce and Nb is the bolt preloading force, dened starting from the knowledge of thetightening torque by means of the following expression:

Nb Tb0:2d 9

where Tb is the value of the tightening torque and d is the bolt nominal diameter. Inthis section the results of the experimental activity are reported pointing out themain features of all the tested frictional interfaces. For the sake of clarity, in the fol-lowing, the materials are divided into two groups: the interfaces composed by metalplates and the interfaces composed by rubber pads sliding on steel.

4.1. Metallic materials

Within this paper, three friction interfaces composed by different types of metalplates sliding on steel have been tested. In particular, the rst series of tests hasbeen carried out on a steelsteel interface made of S275JR structural steel [7] on

which three tests dened according to the loading protocol reported in Table 1 havebeen carried out. The results of the tests are represented in Fig. 3 in terms of lddiagrams, pointing out that the cyclic behavior of the steelsteel interface is quiteunstable, even though the amount of dissipated energy is signicant. In particular,in the rst test (110th cycle) the specimen exhibited a very steep initial stiffnessand after reaching the static sliding force began to slide with a lower force, pointingout that the static coefcient of friction is slightly greater than the dynamic one.After the rst loading branch, all the subsequent cycles reached values of the max-imum force greater than the initial one, showing a signicantly hardening behavior.This behavior can be due to two effects: the rst one is ploughing, the second one isthe steel strain-hardening. The former is related to the fact that, initially the twosurfaces in contact are smooth while, after the rst cycles of sliding motion, dueto the wearing of the steel in the zone under the bolt heads, the number of asper-ities increases and the surfaces become rougher. Therefore, the increase of the num-ber of asperities increases the magnitude of the ploughing component.

The second effect is related to the steel strain hardening which may affect thefriction coefcient by varying the ratio between the steel shear strength and thesteel hardness (Eq. (4)). This hardening behavior, which is quite common in caseof metallic interfaces, has been also documented in a work devoted to the studyof energy dampers [12]. In the second and third loading step, i.e. from the 11thto 30th cycle, the response of the interface is completely different. In this case, staticand dynamic sliding loads assume similar values and the maximum force experi-enced during the whole loading history is reached during the rst cycle. In fact, itis possible to observe a continuously softening behavior which can be due to thewearing of the steel and partially to the loss of the bolt preloading. The wearingof steel is due to the high stresses acting on the interface which lead to abrasionduring the motion (Fig. 4).

In addition, two experimental tests at different values of the applied preloadinghave been carried out on a steelbrass interface. Brass is an alloy of copper and zinccontaining in minor part also iron and lead. The tested material is the CuZn39Pb3.The mechanical and chemical properties of this type of brass are reported in Table 2.

The results of the two tests, reported in Fig. 5, highlight that the behavior of thesteelbrass interface is characterized by a rst sliding force quite lower comparedto the steelsteel interface but, nevertheless, as the number of cycles increasesthe sliding force also increases exhibiting an hardening behavior. This phenomenonhas already been shown in case of steelsteel interface, but in case of steelbrassinterface is less marked. This is probably due to the higher hardness of brass thatslows the abrasion process of the surfaces in contact (Fig. 4).

.5

.4

.3

.2

.1

0

.1

.2

.3

.4

.5

0

[m

olts

M. Latour et al. / Construction and Building Materials 65 (2014) 159176 163Cycles 1-10 (Ts=200Nm)

-0.6

-0.5

-0.4

-0.3

-0.2

-0.1

0

0.1

0.2

0.3

0.4

0.5

-40 -30 -20 -10 0 10 20 30 40

F/(2

nNb)

[mm]

Steel on Steel 4 Bolts Ts=200Nm f=0.25Hz

-0

-0

-0

-0

-0

0

0

0

0

0

-40 -30 -20 -10

F/(2

nNb)

Steel on Steel 4 BCycles 21-30 (Fig. 3. Friction coefcientdisplacemeCycles 11-20 (Ts=300Nm)

-0.5

-0.4

-0.3

-0.2

-0.1

0

0.1

0.2

0.3

0.4

0.5

-40 -30 -20 -10 0 10 20 30 40

F/(2

nNb)

[mm]


10 20 30 40

m]

Ts=500Nm f=0.25HzTs=500Nm) nt curves of steelsteel interface.

ras

allic

ltim

15

BuildMet

allic

Mat

eria

ls

Steel S275JR B

Fig. 4. Texture of the met

Table 2Chemical composition an mechanical properties of CuZn39Pb3 copper alloy (%).

Copper (Cu) Zinc (Zn) Iron (Fe) Lead (Pb) Yield strength (MPa) U

5560 40

Cycles 1-10 (Ts=200Nm) Cycles 11-20 (Ts=300Nm)

Cycles 21-30 (Ts=400Nm)

Fig. 6. Friction coefcientdisplacement curves for sprayed aluminumsteel interface.

0.000

0.050

0.100

0.150

0.200

0.250

0.300

0.350

0.400

0.450

1 3 5 7 9 11 13 15 17 19 21

FRIC

TIO

N C

OEF

FIC

IEN

T

SEMI-CYCLE NUMBER

Steel on Steel 4 Bolts f=0.25Hz

Ts=200 NmTs=300 NmTs=500 Nm

0

0.05

0.1

0.15

0.2

0.25

0.3

0.35

1 11 21 31 41 51 61 71

FRIC

TIO

N C

OEF

FIC

IEN

T

SEMI-CYCLE NUMBER

Brass on Steel 8 Bolts f=0.25Hz

Ts=200 NmTs=300 Nm

Fig. 7. Friction coefcient for metallic interfaces.

M. Latour et al. / Construction and Building Materials 65 (2014) 159176 165

As expected, there is not a signicant dependence between the two parametersin case of steel, while in case of aluminum a signicant decrease of the friction coef-cient arises for increasing values of the applied pressure. In particular, in this lastcase (Fig. 8), a regression analysis of the friction coefcient versus applied interfacepressure, based on experimental results, is proposed (R2 = 0.99).

Furthermore, in Fig. 9 the ratio between the force of i-th semicycle and that ofthe initial one is reported. The results conrm that, in case of steel, the second andthird loading sequence exhibit a rapid decrease of strength. In fact, 20% of strengthdegradation is reached after 7 and 5 cycles, respectively. Dealing with brass, theratio between the slip force of the i-th semicycle and that of the initial one showsthe hardening behavior, outlining a substantial difference between the rst andthe second loading sequence. In fact the slope of the curve corresponding to the sec-ond loading sequence is negligible revealing that, after the initial hardening phase,the friction coefcient stabilize. Concerning sprayed aluminum, it is easy to recog-nize from the results reported in Fig. 9 that, differently from the initial value of the

4.2. Rubber materials

In the following, the results of three different types of rubber materials slidingon steel are presented. The rst material described in this paper is a non-asbestosfriction material used for standard braking application, which will be called inthe following M0. The material is a blend of mineral and organic bers aggregatedby means of phenolic resin. In particular, material M0 has been tested by the pro-ducer, for mechanical engineering application, following the standard test SAE J661.In this test, which is mainly devoted to the investigation of the friction coefcient ofpads subjected to loads simulating braking conditions, the friction pad is locatedinto a machine in contact with a 11-in. drum rotating at speed of 417 rpm. Thebrake pad is clamped with a constant pressure equal to about 1 MPa. On the baseof this testing condition the typical value of the friction coefcient suggested bythe producer is 0.51.

The results of the four tests carried out according to the protocol reported in

0.00

0.10

0.20

0.30

0.40

0.50

0.60

0.0 5.0 10.0 15.0 20.0 25.0 30.0 35.0 40.0

FRIC

TIO

N C

OEF

FIC

IEN

T

Pressure [MPa]0.0 5.0 10.0 15.0 20.0 25.0 30.0 35.0 40.0

Pressure [MPa]

Steel on Steel Interface - Dynamic Friction Coefficient

= -0.0109 p + 0.6537R = 0.9989 {p [MPa]}

0.00

0.10

0.20

0.30

0.40

0.50

0.60

FRIC

TIO

N C

OEF

FIC

IEN

T

Aluminum on Steel Interface - Dynamic Friction Coefficient

Fig. 8. Dynamic friction coefcient versus interface pressure.

0.5

166 M. Latour et al. / Construction and Building Materials 65 (2014) 159176friction coefcient, there is not a signicant dependence between the force degra-dation law and the value of the pressure applied on the frictional interface.

0.5

1

1.5

2

2.5

Fi/F

1

Steel on Steel 4 Bolts f=0.25Hz

Ts=200 NmTs=300 NmTs=500 Nm0

SEMI-CYCLE NUMBER1 3 5 7 9 11 13 15 17 19 21

Fig. 9. Ratio between the force of theTable 1 are delivered in Fig. 10 showing that M0steel interface provides a verystable behavior with a signicant amount of energy dissipation and low strength

Fi/F

1

1

1.5

2

2.5

3Brass on Steel 8 Bolts f=0.25Hz

Ts=200 NmTs=300 Nm01 11 21 31 41 51 61 71

SEMI-CYCLE NUMBER

i-th semicycle and the initial one.


Cycles 41-60 (Ts=400 Nm) Cycles 61-80 (Ts=500 Nm)

-0.4

-0.3

-0.2

-0.1

0

0.1

0.2

0.3

F/(2

nNb)

[mm]

M0 on Steel 4 Bolts Ts=200Nm f=0.25Hz

-0.3

-0.2

-0.1

0

0.1

0.2

0.3

F/(2

nNb)

[mm]

[mm] [mm]


-0.3

-0.2

-0.1

0

0.1

0.2

0.3

F/(2

nNb)


-0.3

-0.2

-0.1

0

0.1

0.2

0.3

-20 -15 -10 -5 0 5 10 15 20-20 -15 -10 -5 0 5 10 15 20

-20 -15 -10 -5 0 5 10 15 20 -20 -15 -10 -5 0 5 10 15 20F/

(2nN

b)


Fig. 10. Friction coefcientdisplacement curves for M0steel interface.


Cycles 21-30 (Ts=400 Nm) Cycles 31-40 (Ts=500 Nm)

-0.25

-0.2

-0.15

-0.1

-0.05

0

0.05

0.1

0.15

0.2

0.25

-15 -10 -5 0 5 10 15 20

F/(2

nNb)

[mm]


-0.25

-0.2

-0.15

-0.1

-0.05

0

0.05

0.1

0.15

0.2

0.25

-15 -10 -5 0 5 10 15 20

F/(2

nNb)

[mm]


-0.25

-0.2

-0.15

-0.1

-0.05

0

0.05

0.1

0.15

0.2

0.25

-15 -10 -5 0 5 10 15 20

F/(2

nNb)

[mm]


-0.2

-0.15

-0.1

-0.05

0

0.05

0.1

0.15

0.2

0.25

-15 -10 -5 0 5 10 15 20F/(2

nNb)

[mm]


Fig. 11. Friction coefcientdisplacement curves for M1steel interface.


degradation. In particular, during the rst and second step of loading (140th cycle)the specimen exhibited a very regular behavior, with negligible strength degrada-tion. In addition, a negligible difference between the static and dynamic frictioncoefcient has been observed. In the third and fourth loading sequence, i.e. fromthe 41th to 60th cycle the response of the interface is characterized by strength deg-radation due to the higher initial pressure acting on the interface. In all the fourtests the behavior has been of the softening type. Therefore the force reached dur-ing the rst cycle is always the greatest of the loading history. This behavior ismainly due to the wearing of material during the sliding motion leading to the lossof bolt preloading (Fig. 10).

Afterwards, also another friction material developed by an Italian Factory forapplications to electric motors has been tested (Material M1). It is a black exiblematerial with metallic particles and it is composed by blending mineral, steel andsynthetic ber, nitrilic rubbers and phenolic resin. In this case, the average valueof the friction coefcient determined on the base of SAE J661 test proposed bythe producer is equal to 0.43. The obtained results, depicted in Fig. 11 in terms ofld curves, highlight the peculiarities of the response of material M1 which ischaracterized by a quite stable behavior for low values of the tightening torquebut, for values higher than 300 N m, it exhibits signicant stiffness and strengthdegradation. This poor performance is due to the low tensile resistance of thefriction material which, after the 4th loading sequence, collapsed with the develop-

ment of a crack at the transversal section weakened by the holes, leading to thecomplete fracture of the friction pad. Like in case of material M0, also for this rubberbased material the softening of the friction resistance occurred in all the fourloading sequences.

Therefore, also in this case there is a signicant inuence of the material wear-ing (Fig. 12) on the developed cyclic sliding force. In addition, the results obtainedfor material M1 have not pointed out a substantial difference between static anddynamic friction coefcient.

The last rubber material tested is a friction material developed by an Italian Fac-tory for applications where low wearing is necessary (Material M2). In fact, thematerial is characterized by a very high ShoreD hardness and, therefore by a goodresistance to abrasion. Similarly to material M1, it is a black hard and stiff materialcomposed by a blending mineral and synthetic ber, nitrilic rubbers and phenolicresin.

The mean value of the friction coefcient declared by the producer for contactpressure equal to 1 MPa is equal to 0.36. The results delivered in Fig. 13 point outthat M2steel interface provides a response characterized by three main phases.In the rst loading sequence, i.e. cycles from 1th to 10th, during the rst cycle afterthe linear branch a peak corresponding to the static friction is reached and then theinterface slides showing a slight hardening behavior. In the subsequent cycles,increasing values of the sliding force are exhibited. In the second test, namely cycles

Rub

ber

Mat

eria

ls

Material M0 Material M1 Material M2

Fig. 12. Texture of the materials after the test.

-0.25

-0.2

-0.15

-0.1

-0.05

0

0.05

0.1

0.15

0.2

0.25

-20 -15 -10 -5 0 5 10 15 20

F/(2

nNb)


-0.25

-0.2

-0.15

-0.1

-0.05

0

0.05

0.1

0.15

0.2

0.25

-20 -15 -10 -5 0 5 10 15 20 25

F/(2

nNb)


5

.2

5

.1

5

0

5

.1

5

.2

5

0

[m

ts T

168 M. Latour et al. / Construction and Building Materials 65 (2014) 159176Cycles 1-10 (Ts=200Nm) [mm]

-0.2

-0

-0.1

-0

-0.0

0.0

0

0.1

0

0.2

-25 -20 -15 -10 -5

F/(2

nNb)

M2 on Steel 8 BolCycles 21-30 (Fig. 13. Friction coefcientdisplacemCycles 11-20 (Ts=300Nm) [mm]

5 10 15 20 25

m]

s=400Nm f=0.25HzTs=400 Nm) ent curves for M2steel interface.

00.05

0.1

0.15

0.2

0.25

0.3

FRIC

TIO

N C

OEF

FIC

IEN

T

SEMI-CYCLE NUMBER

M0 on Steel 4 Bolts f=0.25Hz

Ts=200 NmTs=300 NmTs=400 NmTs=500 Nm

0

0.05

0.1

0.15

0.2

0.25

FRIC

TIO

N C

OEF

FIC

IEN

T

SEMI-CYCLE NUMBER



0

0.05

0.1

0.15

0.2

0.25

1 6 11 16 21 26 31 36 41 46 1 6 11 16 21

1 6 11 16 21

FRIC

TIO

N C

OEF

FIC

IEN

T

SEMI-CYCLE NUMBER


Ts=200 NmTs=300 NmTs=400 Nm

Fig. 14. Friction coefcient for rubber interfaces.

0.00

0.10

0.20

0.30

0.40

0.50

0.60

Pressure [MPa]

M0 on Steel Interface -Ratner & Sokolsakya

0.00

0.10

0.20

0.30

0.40

0.50

0.60

Pressure [MPa]


0.00

0.10

0.20

0.30

0.40

0.50

0.60

0.0 5.0 10.0 15.0 20.0 25.0 30.0 35.0 40.0 0 5 10 15 20 25 30 35 40

0 5 10 15 20 25 30 35 40

Pressure [MPa]


Fig. 15. Dynamic friction coefcient versus interface pressure.


from 11th to 20th, the behavior is very stable and the same value of the sliding forceis reached in all the cycles. Only in the third loading sequence a softening behavioris exhibited where the slip resistance progressively reduces as far as the number ofcycles increases.

Similarly to the case of metallic materials, also for rubber interfaces the valuesof the friction coefcients have been evaluated by means of Eqs. (8) and (9) (Fig. 14).This gure points out the different behaviors exhibited by the three materials. Incase of material M0, in all the loading sequences the friction coefcient starts froma value of about 0.25 and decreases as far as the number of cycle increases. It is use-ful to observe that, in the considered tests, a constant frequency has been chosenand, therefore, the change of amplitude at each loading step results in a changeof the sliding velocity from 7.5 mm/s to 15 mm/s. Consequently, it can be alsoobserved that the friction coefcient of material M0 is not signicantly affectedby the sliding velocity.

The behavior of material M1 is similar to that of material M0 in terms ofdegradation. In fact, the friction coefcient starts from a value of about 0.2 and suc-cessively quickly decreases. In the rst loading sequence, corresponding to the min-imum tightening torque, the lowest friction coefcient has been obtained. Probablythis anomalous result is due to an imperfect application of the tightening torque.Conversely, in case of material M2, a different behavior has been observed. In fact,due to the initially hardening behavior, the experimental values of the static anddynamic friction coefcients are not coincident, being the static friction coefcientequal to 0.158 while the dynamic friction coefcient has an average value equal to0.178. In addition, in contrast with the outcomes of tests on M0 and M1 frictionmaterials, during the rst loading sequence, starting from a static friction coef-cient equal to 0.158, increasing values have been observed. In the second loadingsequence an almost constant friction resistance has been obtained and, nally, inthe third loading cycle sequence a softening behavior has been pointed out.

In order to provide a tool for estimating the kinematic friction coefcient of thetested rubber materials, as for metallic interfaces, the dependence between the fric-tion coefcient and the interface pressure has been examined. It is worth to notethat the average value of the friction coefcients obtained in the sliding tests onthe three materials by varying the interface pressure in the range 1333 MPa is con-siderably lower than the values given by the producer, equal to 0.510.430.36 formaterials M0M1M2, respectively, which are dened in correspondence of aninterface pressure equal to 1 MPa. This behavior can be easily explained consideringthe friction laws usually adopted in technical literature to represent the depen-dence between the friction coefcient and the normal pressure for rubber based

materials. In fact, Eqs. (5)(7) show that in case of rubbers, usually the coefcientof friction decreases as far as the interface pressure increases and tends to becomeconstant at very high pressures. The range of interface pressure values 1333 MPainvestigated in this work are aimed at structural engineering applications, whileSAE J661 standard test is, in truth, conceived for automotive applications, wherethe interface pressure is signicantly lower. Therefore, in Fig. 15, the law proposedby [27] is used to carry out a regression analysis by considering both the point givenby the producer and the points found in this testing program. In any case, it is worthto say that for structural applications a constant value of the friction coefcient canbe assumed with sufcient approximation.

Moreover, in Fig. 16 the ratio between the slip forces of the i-th semicycle andthat of the initial one are reported. For material M0 the results highlight that forbolt torques greater than 300 N m the force decreases much more quickly.

Fi/F

1

0

0.2

0.4

0.6

0.8

1

1.2

1 6 11 16 21

SEMI-CYCLE NUMBER



1

el 8 Bolts f=0.25Hz

Fig. 17. Possible combination of disc springs [31].

170 M. Latour et al. / Construction and Building Materials 65 (2014) 1591760

0.2

0.4

0.6

0.8

1

1.2

1 6 11 16 21 26 31 36 41 46

Fi/F

1

Fi/F

1

SEMI-CYCLE NUMBER


0

0.2

0.4

0.6

0.8

1

1.2

1.4

1 6 1

M2 on SteSEMI-CYC

Fig. 16. Ratio between the force of the16 21

Ts=200 NmTs=300 NmTs=400 NmLE NUMBER

i-th semicycle and the initial one.

Conversely, in case of material M1, due to the low resistance to abrasion, the resultsshow that in all the considered cases a very quick strength degradation is obtained.Therefore, material M1 cannot be considered adequate for structural applicationswith high values of the interface pressure. Dealing with material M2, the resultsshow, as expected, a high resistance to abrasion leading to a low strength degrada-tion which, in the considered cases, is always less than the 20%.

5. Experimental tests with disc springs

The application of the clamping force in a friction device bymeans of preloaded high strength bolts requires two design issuesof paramount importance to be considered: the keeping of the pre-loading force during the lifetime of the friction damper and the

reduction of the loss of the preloading level during the slidingmotion. In fact, during the motion, the sliding force can decreasedue to the wearing of the friction material. In order to keep thepreloading force during lifetime and to reduce drawbacks comingfrom material consumption, the use of disc spring washers canbe suggested.

Such a type of washer is an high resistance cone shaped annularsteel disc spring which attens when compressed and returns to itsoriginal shape if compression loading is released. In this way, if thebolt had to lose the preloading level due to the wearing of the fric-tion material, the disc spring would restore the force by keepingthe bolt shaft in tension. The particular shape of cone washersallows single springs to be combined in different ways dependingon the desired loaddisplacement curve. In particular, they can bestacked (Fig. 17):

In the same direction (parallel stacking): in this way a stifferspring is obtained.

In alternating direction (series stacking): in this way a moreexible spring is obtained.

Spring stack of parallel sets in series: in this way a proper com-bination of washers located in parallel and in series allow toobtain a predetermined value of the stiffness.

In order to evaluate the effect of disc springs on the variation ofthe sliding force under cyclic loading conditions, further tests havebeen carried out on material M0 and material M2 by applying acouple of disc springs stacked in series under the bolt nut. In par-ticular, cone washers with internal diameter, external diameter,height and thickness equal to 21, 45, 6.4 and 5 mm, respectively,

-0.3

-0.2

-0.1

0

0.1

0.2

0.3

0 5-20 -15 -10 -5 10 15 20

F/(2

nNb)

[mm]

M0 on Steel 4 Bolts Ts=200Nm f=0.25Hz (Disc Springs)

Fig. 18. M0 on steel interface with disc springs forcedisplacement curve.

Friction coefficient versus cycles

0

0.05

0.1

0.15

0.2

0.25

0.3

1

FRIC

TIO

N C

OEF

FIC

IEN

T

SEMI-CYCLE NUMBER

M0 on Steel 4 Bolts Ts=200Nm Circular Washers vs Disc Springs

Disc SpringsCircular Washers

M0 on Steel 4 Bolts Ts=200Nm Circular Washers vs Disc Springs

1.0

6 11 16 21 26 31 36 41

YC

M0 on Steel 4 Bolts Ts=200Nm

M. Latour et al. / Construction and Building Materials 65 (2014) 159176 1710.0

0.5

1 6 11 16

SEMI-C1.5

2.0

2.5

Ei/E

1Dissipated energy: ratio b

Fig. 19. M0 on steel interface circuRatio between i-th sliding force and initial one

0

0.2

0.4

0.6

0.8

1

1.2

1

F i/F

1[k

Nm

]

SEMI-CYCLE NUMBER


6 11 16 21 26 31 36 41

21 26 31 36 41

LE NUMBER

Circular Washers vs Disc Springs

Disc SpringsCircular Washersetween i-th cycle first cycle lar washers versus disc springs.

have been used. Such a kind of washer completely attens under aload equal to 130 kN. In order to compare the cyclic behavior thesame loading protocol has been adopted. Conversely, materialM1 has not been tested again because of its poor behavior with cir-cular at washers. In particular the following additional tests havebeen carried out:

M0steel interface: Bolts torque = 200 N m, 10 cycles withamplitude of 7.5 mm and 10 cycles with amplitude of 15 mm.

M2steel interface: Bolts torque = 200 N m, 10 cycles withamplitude equal to 15 mm.

Regarding friction material M0, the new specimen exhibited abehavior similar to that obtained in the test with circular atwashers, but lower degradation was obtained (Fig. 18).

The comparison between the two tests in terms of frictioncoefcient shows that in the test with disc springs a lower valueof the initial friction coefcient has been obtained, because thefriction pad has not been substituted after the test with at wash-ers. Conversely a more stable behavior in terms of strength degra-dation and energy dissipation capacity has been exhibited (Fig. 19).

Regarding friction material M2 (Fig. 20), starting from a force ofabout 120 kN, after the rst sliding, the new specimen exhibited aslightly hardening behavior.

The comparison with the test with at washers points out thatthe friction coefcient is almost unaffected, it starts from a staticvalue of about 0.15 and then increases to about 0.19. Similarly, inboth tests, there is an increase of strength and energy dissipationcapacity, with respect to the rst cycle, of about the 20% (see Fig. 21).

Therefore, on the base of the obtained results, it seems that theapplication of disc springs provides only a slight improvement ofthe response under cyclic loads. Nevertheless, as aforementioned,the application of disc springs in practice can be useful for keepingthe preloading level during the lifetime of the dissipative device. Infact, as already demonstrated by [14], tightened bolts are usuallyaffected by a short term relaxation, which occurs within the rst12 h after joint assembly, and a long term relaxation, which usuallycontinue asymptotically during the time. Furthermore, whengroups of bolts are tightened, there is also a loss of pretensiondue to a mutual effect, i.e. after any bolt tightening the boltspreviously tightened experience a loss of pretension. All theseeffects can be reduced by means of annular cone disc springs.

-0.25

-0.2

-0.15

-0.1

-0.05

0

0.05

0.1

0.15

0.2

0.25

-20 -15 -10 -5 0 5 10 15 20

F/(2

nNb)

[mm]

M2 on Steel 8 Bolts Ts=200Nm f=0.25Hz (Disc Springs)

Fig. 20. M2 on steel interface with disc springs forcedisplacement curve.

Friction coefficient versus cycles

0

0.05

0.1

0.15

0.2

0.25

FRIC

TIO

N C

OEF

FIC

IEN

T

SEMI-CYCLE NUMBER

M2 on Steel 8 Bolts Ts=200Nm Disc Springs vs Circular Washers


1 3 5 7 9 11 13 15 17 19 21

M2 on Steel 8 Bolts Ts=200Nm Disc Springs vs Circular Washers

1.4

12-CY

M2 on Steel 8 Bolts Ts=200Nm

172 M. Latour et al. / Construction and Building Materials 65 (2014) 1591760.0

0.2

0.4

0.6

0.8

1.0

1.2

4 6 8 10

E i/E

1[k

Nm

]

SEMI

Dissipated energy: ratio

Fig. 21. M2 on steel interface circuRatio between i-th sliding force and initial one

0

0.2

0.4

0.6

0.8

1

1.2

1.4

1 3 5 7 9 11 13 15 17 19 21

F i/F

1

SEMI-CYCLE NUMBER


14 16 18 20 22CLE NUMBER

Disc Springs vs Circular Washers

Disc SpringsCircular Washers between i-th cycle first cyclelar washers versus disc springs.

6. Mathematical modeling of friction materials

In order to provide a practical tool to be employed in the designand modeling of the response of friction pads to be applied tosupplemental damping devices, a mathematical model able to pre-dict the cyclic behavior of the friction materials investigated in thispaper has been carried out. Aiming to the use of such materials infriction dampers for seismic design applications, the requirementsto be satised regard the coefcient of friction which, under cyclicloading conditions, should be as high as possible in order to reducethe size of the damping devices and, in addition, the stability of thecyclic behavior. Therefore, on the basis of the experimental resultspreviously presented, in the following, only three interfaces havebeen selected: steelsteel, sprayed aluminumsteel and materialM0steel, i.e. those characterized by the higher values of the fric-tion coefcient and by hardening or slightly softening response.

Dealing with steelsteel interface, as aforesaid, its behavior ischaracterized by two different phases. In a rst phase, due to theprogressive ploughing of the surfaces in contact, the friction coef-cient quickly increases reaching a constant value, which is main-tained before the achievement of a softening phase due to thematerial wearing and the partial loss of the bolt preloading force.In this phase, the friction coefcient begins to decay with a lawwhich, as pointed out by testing results, is not dependent on theforce applied to the interface.

Therefore, the modeling of the hardening/softening rules of thesteelsteel interface can be carried out starting from the testresults, by reporting the values of the friction coefcient versus aparameter able to represent the progressive wearing and degrada-tion of the material. In this work, the cumulated value of theenergy dissipated is chosen as reference parameter. The result ofthe analysis is reported in Fig. 22 where the values of the kinematic

Fig. 22. Regression analysis of experimental data for steelsteel interface.

M. Latour et al. / Construction and Building Materials 65 (2014) 159176 173Fig. 23. Regression analysis of experimental data for sprayed aluminumsteel interface.

friction coefcient observed at the i-th cycle divided by the initialvalue of the friction coefcient are reported versus the cumulatedenergy both for the hardening and for the softening phase.

Consistently with the experimental behavior, the modeldescribing the coefcient of friction under cyclic loading condi-tions is dened by means of the three branches reported inFig. 22. The rst one modeling the hardening phase, characterizedby a linear approximation, the second one modeling the steadystate phase, assuming a constant value, and the third one describ-ing the softening phase, characterized by an exponential law:

lilk0:0076Ediss;tot0:335 0< Ediss;tot687kNmHardening phase

lilk1 87< Ediss;tot6123kNmSteady state phase

lilk e0:0025Ediss;tot Ediss;tot >123kNmSoftening phase 13

Therefore, starting from the knowledge of the initial value of thedynamic friction coefcient, the cyclic forcedisplacementresponse of the steelsteel interface can be predicted by meansof Eq. (13).

Concerning the modeling of the behavior of the metallic inter-face made of sprayed aluminum sliding on steel, it can be easilycharacterized. In fact, as demonstrated by the sliding tests the ini-tial value of the friction coefcient of this material stronglydepends on the interface pressure, but the degradation law doesnot signicantly depend on the pressure applied to the interface.Therefore, in this case, a regression analysis of the data of the threeexperimental tests has been carried out following the same proce-

lilk 1 0:000328Ediss;tot with Ediss;tot kN m 14

A model able to describe the degradation law of material M0has also been dened by exploiting the experimental data comingfrom the sliding tests. In case of material M0, as aforesaid, theinitial value of the friction coefcient always decay providingincreasing values of the degradation rate as far as the contactpressures increase. In fact, due to the low value of the ShoreD hard-ness, material M0 is subjected, under cyclic loadings, to signicantwearing which lead to a progressive deterioration of the frictioncoefcient.

Therefore, starting from the four tests previously reported, amathematical model able to predict the value of the friction coef-cient under cyclic loading histories has been developed. To thisscope, a multiple linear regression analysis of the experimentaldata has been carried by considering two independent variables:the cumulated value of the energy dissipated and the contact pres-sure at the interface calculated by means of Eq. (12). The followingresult has been obtained (R2 = 0.916):

lilk10:00216Ediss;tot0:000284p with Ediss;totkNm and pMPa

15

The results of the regression analysis are represented in Fig. 24where the agreement between experimental and predicted valuesis shown.

In order to show the ability of the proposed models to

174 M. Latour et al. / Construction and Building Materials 65 (2014) 159176dure specied for the steelsteel interface. The obtained resultsshow that the degradation law of the sprayed aluminumsteelinterface is described by the following equation which provides asmooth decay law of the friction coefcient (see Fig. 23):

Fig. 24. Regression analysis of experim

reproduce the response of the different frictional interfaces, somesimulations have been carried out by exploiting the mathematicallaws provided by Eqs. (13)(15) and the values of the friction coef-cient dened by equations reported in Figs. 8 and 15. In Fig. 25ental data for M0steel interface.

-0.4

-0.3

-0.2

-0.1

0

0.1

0.2

0.3

0.4

[m

olts

hem

Build-0.6

-0.4

-0.2

0

0.2

0.4

0.6

F/(2

nNb)


-40 -30 -20 -10 0 10 20 30 40

-20 -15 -10 -5

F/(2

nNb)

M0 on Steel 4 B

[mm]

Fig. 25. Application of the proposed mat

M. Latour et al. / Construction andthe good agreement between experimental and predicted cyclicresponse is pointed out.

7. Conclusions

In this work, an experimental program devoted to the investiga-tion of the frictional properties of six different materials to be usedin supplemental damping devices adopting dry friction as source ofenergy dissipation has been carried out. To this scope a sub-assem-blage has been tested by varying the friction interface, the numberof preloaded bolts, the tightening torque and the type of washers.In particular, three metallic friction interfaces and three rubberfriction pads on steel have been tested.

Regarding metallic friction interfaces, the main results of theexperimental program can be summarized as follows:

Steel on steel interface exhibited high coefcient of friction, buta quite unstable behavior which is initially characterized by asignicantly hardening behavior and successively by a rapidlysoftening behavior.

Brass on steel interface exhibited a quite low value of thefriction coefcient and a stable behavior characterized by asignicant hardening.

Thermally sprayed aluminum on steel interface exhibited anhigh value of the friction coefcient and a cyclic response withsmall degradation.

Concerning rubber friction pads, the following main observa-tions can be made:

Material M0, which is a rubber based material developed forautomotive applications exhibited a very stable behavior andhigh energy dissipation capacity also under high values of thepreloading level.0 5 10 15 20

m]

Ts=500Nm f=0.25Hz

[mm]

atical model to some experimental tests.

ing Materials 65 (2014) 159176 175 Material M1, which is a rubber based material developed forelectrical machines, exhibited pinching behavior, low frictioncoefcient and a rapidly degrading behavior.

Material M2, which is a hard rubber based material developedfor applications where low wearing is necessary, developed aquite low value of the friction coefcient, but a very stablebehavior and high energy dissipation capacity.

The work herein presented is part of a research program aimingto the use of such materials in friction dampers to be adoptedwithin the structural detail of innovative dissipative beam-to-col-umn connections, based on bolted double split tee conguration,for seismic design applications [22]. To this scope, an high valueof the coefcient of friction is desired in order to reduce the sizeof the friction damper. In addition, a stable cyclic response is alsorequired. Therefore, on the basis of the experimental results previ-ously presented, three interfaces have been selected seem moreappropriate: steelsteel, sprayed aluminumsteel and materialM0steel, i.e. those characterized by the higher values of the fric-tion coefcient and by hardening or slightly softening response.For such interfaces, mathematical models to predict their cyclicresponse has been developed and compared with experimentaltest results showing their accuracy. These models are based onan initial value of the friction coefcient and on a law describingits variation as a function of the energy dissipated during the load-ing history. Depending of the friction interface, the initial value ofthe friction coefcient can be either dependent on or independentof the applied pressure due to bolt preloading.

Focusing the attention on the three interfaces providing thebest results. Even though the initial friction coefcient of thermallysprayed aluminum on steel reduces as far as the applied pressureincreases, it can be recognized that it gives rise to the highest valueof the initial friction coefcient. In addition, such interface is ableto provide quite stable hysteresis loops with only small

degradation. Therefore, the authors believe that the use of frictionpads made of steel plates with thermally sprayed aluminum canrepresent a good solution to improve the bending performanceand the cyclic response of beam-to-column joints equipped withfriction dampers. This issue will constitute the forthcomingdevelopment of the planned research activity.

References

[1] Aiken I, Clark P, Kelly J. Design and ultimate-level earthquake tests of a 1/2.5scale base-isolated reinforced-concrete building. In: Proceedings of ATC-17-1seminar on seismic isolation, passive energy dissipation and active control. SanFrancisco, California; 1993.

[2] Aiken I, Nims D, Whittaker A, Kelly J. Testing of passive energy dissipationsystems. Earthquake Spectra 1993;9(3).

[3] Alonso J. Mechanical characteristics of X-plate energydissipators. Berkeley: University of California; 1989.

[4] Bowden F, Tabor D. The friction and lubrication of solids: Part I. Oxford: OxfordUniversity Press; 1950.

[5] Bowers R, Clinton W, Zisman W. Frictional behavior of polyethilene,polytetrauorethylene and halogenated derivatives. Lubr Eng 1953;9:2049.

[6] Cahis X, Bozzo L, Foti D, Torres L. An energy dissipating device for seismicprotection of masonry walls. Taormina, Italia: Lingegneria Sismica in Italia;1997.

[7] CEN. Eurocode 3: Design of steel structures Part 11: general rules and rulesfor buildings; 2005.

[8] CEN. Eurocode 3: Design of steel structures Part 18: design of joints; 2005.[9] Christopoulos C, Filiatrault A. Principles of passive supplemental damping and

seismic isolation. Pavia: IUSS PRESS; 2006.[10] Constantinou M, Soong T, Dargush G. In: Research MC, editor. Passive energy

[18] Kelly J, Skinner R, Heine A. Mechanisms of energy absorption in special devicesfor use in earthquake resistant structures. Bull NZ Soc Earthquake Eng1972;5(3):6388.

[19] Kobori T, Miura Y, Fukuzawa E, Yamada T, Arita T, Miygawa N, et al.Development of hysteresis steel dampers. In: Earthquake engineering tenthworld conference; 1992. p. 23416.

[20] Latour M, Rizzano G. Experimental behavior and mechanical modeling ofdissipative T-stub connections. J Struct Eng 2012;138(2):17082.

[21] Latour M, Piluso V, Rizzano G. Cyclic modeling of bolted beam-to-columnconnections: component approach. J Earthquake Eng 2011;15(4):53763.

[22] Latour M, Piluso V, Rizzano G. Experimental analysis of innovative dissipativebolted double split tee beam-to-column connections. Steel Constr2011;4(2):5364.

[23] Mualla I, Belev B. Seismic response of steel frames equipped with a newfriction damper device under earthquake excitation. Eng Struct2002;24(3):36571.

[24] Nakashima M. Strain-hardening behavior of shear panels made of low-yieldsteel: test. J Struct Eng ASCE 1995;121(12):17429.

[25] Pall A, Marsh C. Response of friction damped braced frames. J Struct Div1981;108(6):131323.

[26] Persson B. Sliding friction. Berlin: Springer; 2000.[27] Ratner S, Sokolskaya V. The inuence of the hardness of rubber on its

coefcient of static friction without lubrication. Rubber Chem Technol1956;29:82933.

[28] Rees B. Static friction of bulk polymers over a temperature range. Research1957;10:3318.

[29] Schallamach A. Friction and abrasion of rubber. Wear 1958;1:384417.[30] Schallamach A. The load dependence of rubber friction. Phys Soc 1952:65861.[31] Schnorr. In: Kleiner EF, editor. Handbook for disc springs. Heilbronn: Adolf

Schnorr GmbH; 2003.[32] Shooter K, Thomas R. Frictional properties of some plastics. Research

1952;2:5339.[33] Skinner R, Kelly J, Heine A. Hysteresis dampers for earthquake resistant

176 M. Latour et al. / Construction and Building Materials 65 (2014) 159176dissipation systems for structural design and retrot. State of NewYork: University at Buffalo; 1998.

[11] Faella C, Piluso V, Rizzano G. Experimental analysis of bolted connections:snug versus preloaded bolts. J Struct Eng 1998;124(7):76574.

[12] Grigorian C, Yang T, Popov E. Slotted bolted connection energy dissipators.earthquake engineering research center. Berkeley: University of California;1992.

[13] Halling J. Principles of tribology. London: Macmilln Education LTD.; 1978.[14] Heistermann C. Behaviour of pretensioned bolts in friction

connections. Lulea: Lulea University of Technology; 2011.[15] Iannone F, Latour M, Piluso V, Rizzano G. Experimental analysis of bolted steel

beam-to-column connections: component identication. J Earthquake Eng2011;15(2):21444.

[16] Kelly F, Sha W. A comparison of the mechanical properties of re-resistant andS275 structural steels. J Constr Steel Res 1999;50:22333.

[17] Kelly J. Aseismic base isolation: a review. In: Proceedings of 2nd US nationalconference on earthquake engineering. Stanford, CA; 1979. p. 82337.structures. Earthquake Eng Struct Dynam 1975;3:28796.[34] Soong T, Spencer Jr B. Supplemental energy dissipation: state-of-the-art and

state-of-the-practice. Eng Struct 2002;24:24359.[35] Thirion P. Les coefcients dadherence du caoutchouc. Rubber Chem Technol

1948;21:50515.[36] Tremblay R, Stiemer S. Energy dissipation through friction bolted connections

in concentrically braced steel frames. ATC 17-1 Semin Seism Isolation1993:2;55768.

[37] Whittaker A, Bertero V, Alonso J, Thompson C. UCB/EERC-89/02 earthquakesimulator testing of steel plate added damping and stiffnesselements. University of California, Berkeley: College of Engineering; 1989.

[38] Whittaker A, Uang C, Bertero V. UCB/EERC-88/14 an experimental study of thebehavior of dual steel systems. University of California, Berkeley: College ofEngineering; 1990.

[39] Zhang S. State-of-the art of polymer tribology. Tribol Int 1998;31:4960.

Experimental analysis on friction materials for supplemental damping devices1 Introduction2 Friction theories3 Tested friction materials4 Experimental tests with circular flat washers4.1 Metallic materials4.2 Rubber materials

5 Experimental tests with disc springs6 Mathematical modeling of friction materials7 ConclusionsReferences

1-s2.0-s0950061814004097-main

Documents

seismic retrot

seismic loading conditions

seismic base isolation

case of rare seismic

i g h t sprayedexperimof

cyclic loading history

simplest case

cyclic behavior