All rocks and soilsconsist of minerals thathave a distinctivescratch hardness. To

define this hardness, the Moh’shardness scale is the standardreference used. The scale isdivided into 10 increments,ranging from talc (with a hardnessof 1) as the softest, up todiamond (hardness 10) as thehardest. The scale is naturallylinear from a hardness of 1 to 9,with each mineral being able toscratch the one below it in thescale.

Among the most commonminerals, mica and calcite arevery soft (hardness 2.5 and 3,respectively), while feldspar,pyroxene and amphibole may becharacterised as medium hard(hardness 6). Quartz and garnetare very hard (hardness 7 and 7-7.5, respectively), and to a greatextent, determine the degree ofTBM cutter wear.

Cutter life can be estimated from therelative percentage of minerals of differentMoh’s hardness classes (>7, 6-7, 4-5 and<4). For coarse-grained rock and soil this ismost commonly determined bypetrographic analysis using a microscope.For fine grained rock and soil it is mostcommonly determined by X-ray diffraction(XRD), sometimes supplemented bydifferential thermal analysis (DTA). Thehigher the percentage of hard mineralsfound at the face, the more abrasive the soilor rock and the shorter the cutter life.

In addition to mineral composition, TBMperformance is also influenced by manyother textural features, such as: • grain size, shape and elongation• grain orientation, directional properties• grain interlocking• microfractures and poresThe use of Moh’s hardness therefore isrestricted mainly to preliminary estimates ofcutter wear. As far as is known, Moh’shardness has not to date been used directlyas an input in any TBM performanceprediction model.

Test methods for rockThere are several methods for estimating theabrasiveness of rocks. The most commonlyused are (Ozdemir & Nilsen, 1999 and Büchiet al. 1995):

1) The Vickers test, giving the VickersHardness Number - VHN2) The Cerchar test, giving the CercharAbrasivity Index - CAI3) The LCPC abrasimeter test, giving theLCPC abrasivity index - ABR4) The NTNU abrasion test, giving theAbrasion Value - AV/AVS

ABRASIVITY & WEAR

APRIL 2006 Tunnels & Tunnelling International 4747

-0.3 0 0.3 0.6 0.9Log10 Mohs hardness

Log 1

0 V

icke

rs m

icro

hard

ness 3

2

1

Fig 1 - Correlation betweenVickers micro-hardness (VHN)and Moh’s hardness (after Youngand Millmann, 1964).

Fig 2 - The Cerchar test

Pull to limitin 1 second

Handle

7kg (15lb) HCAD

Hard steel point Rock sample

Unitscrew

Abrasivitytesting forrock and soils

In the second of a series of three articles, B Nilsen of Norwegian Universityof Science and Technology (NTNU), F Dahl of SINTEF Rock and SoilMechanics, J Holzhäuser, of Babendererde Ingenieure, and P Raleigh, ofBabendererde Engineers, discuss existing test methods to describe theabrasiveness of rock and soils

These methods normally give a fairly reliableestimation of the abrasiveness. The greatestchallenge in most cases is to collectrepresentative samples.

Vickers hardness defines the micro-indentation hardness of a mineral, andprovides a Vickers hardness number (VHN).The hardness number is defined as the ratioof the load applied to the indentor (gram orkilogram force) divided by the contact areaof the impression (mm2). The Vickersindentor is a square based diamond pyramidwith a 130˚ included angle between oppositefaces, so that a perfect indentation is seenas a square with equal diagonals. A virtuallylinear relation has been found betweenMoh’s hardness and VHN (in log-scale), asshown in Figure 1.

As with Moh’s hardness, the use of VHN isprimarily for the purpose of preliminaryestimates of abrasivity and the expectedcutter wear.

The Cerchar test is performed byscratching a freshly broken rock surface witha sharp pin of heat-treated alloy steel (figure2). The Cerchar Abrasivity Index (CAI) is thencalculated as the average diameter of theabraded tip of the steel pin in tenths of mmafter 10mm of travel across the rock surface.The advantage of this test is that it can beperformed on irregular rock samples. TheCAI value is related directly to cutter life inthe field. CAI values vary between less than0.5 for soft rocks (such as shale andlimestone) to more than 5.0 for hard rocks(such as quartzite).

The LCPC Abrasimeter Test involves thetaking of samples of rock, soil orsynthetically created material and testingusing the 4mm-6.3mm fraction. Coarsegrained material has to be crushed andsieved and fine grained material (<4mm) is

not included in the test. An air dried sampleis placed into a cylindrical drum and arectangle steel propeller is rotated at4500rpm speed. The propeller is made ofrelatively soft steel, which can be easilyscratched with a knife. The abrasioncoefficient ABR corresponds to the weightloss of the propeller per tonne of sample.The LCPC test is mainly used for rocksamples, and a fairly good correlation existsbetween the LCPC test, Cerchar test andthe UCS of the rock tested.

The NTNU abrasion test (AV/AVS)A methodology for estimating the drillabilityof rocks by percussive drilling wasdeveloped at the Engineering GeologyLaboratory of the Norwegian Institute ofTechnology (NTH) in the early1960’s (Lien, 1961). Abrasiontesting of crushed rockparticles <1mm, as illustrated inFigure 3, was then introducedtogether with the Brittlenesstest and the Sievers-J miniaturedrill test for estimating thedrillability parameters DRI(Drilling rate index) and BWI (BitWear Index).

Since the early 1980’s, thetests have been used mainly forpredicting hard rock TBM wearperformance according to themethod developed by theNTH/NTNU Department ofBuilding and ConstructionEngineering (in 1996, as resultof a merger, NTH changedname to NTNU - the NorwegianUniversity of Science andTechnology - and theNorwegian method now is

referred to as the NTNU method), Bruland,Dahlø & Nilsen (1995). For TBM cutter wearprediction, a test piece of cutter steel is usedinstead of the tungsten carbide test pieceused for percussive drilling estimation, andthe parameter CLI (cutter life index) iscalculated instead of BWI. The NTNUprognosis model has been continuouslyrevised and improved as new tunnelling datahas become available, and is now based ondata from about 250km of bored tunnels inNorway and many other countries aroundthe world (NTNU-Anleggsdrift, 1998).

Since the introduction of the NTNUdrillability/boreability testing, acomprehensive database representingabout 3,000 different rock samples has beenestablished. Today, SINTEF Rock and SoilMechanics is operating the drillabilitylaboratory in co-operation with NTNU.

The Abrasion Values AV/AVS representtime dependent abrasion of tungstencarbide/cutter steel caused by crushed rockpowder. The same test equipment as for theAV is used to measure the AVS, but insteadof the tungsten carbide test pieces used forAV, the AVS test uses test pieces of steeltaken from a cutter ring.

The two tests are defined as follows:AV: The Abrasion Value is the mean value ofthe measured weight loss in milligrams of 2 -4 tungsten carbide test bits after 5 minutes,i.e. 100 revolutions of testing, by using anabrasion apparatus and crushed rockpowder.AVS: As described for AV, but with 1 minute,i.e. 20 revolutions of testing.

For the AVS-test, the standardNTNU/SINTEF test procedure (shown infigure 3), is as follows:• A representative rock sample consisting

ABRASIVITY & WEAR

48 Tunnels & Tunnelling International APRIL 2006

Fig 3 - Principle sketch of the NTNU abrasion test

30mm

10mm

rr=15mm

AV: Tungsten carbideAVS: Cutter ring steelSAT: Cutter ring steel

AV: 100rev./5 minAVS: 20 rev./1 minSAT: 20rev./rev min

Vibratingfeeder

Suctionassembly

Flow rate~80g/min

Rotationsteel disc

Rock (AV/AVS)or soil (SAT)

<1mm

Weight 10kg

S

F

Fig 4 - Miller test to determine the SlurryAbrasivity (Miller Number) and the Slurry AbrasionResponse (SAR Number)

S

FNormal force:F=22.24N

Reciprocatingmotion: 200mmback and forth

Test slurry:for example: bentonite slurry + sand

Standardisedsteel block

of approx. 2kg is used for preparation ofabrasion powder.

• Crushing is done gently in several crushersteps to avoid excessive production offines. The initial crushing is performed in ajaw crusher with the outlet openingadjusted to 10mm. Further crushing isperformed using a smaller laboratorycrusher in minimum 2 steps. The outletopening is adjusted to approx 3mm-4mmprior to the first crusher step.

• The crushed material is sieved on a 1mmquadratic mesh. The fraction < 1mm istransferred to a suitable pan and thefraction > 1mm is crushed again afteradjustment of the outlet opening toapprox. 1mm. This process is repeateduntil the grain size distribution is 99%<1mm and 70 ± 5 % < 0.5mm.

• The crushed powder is mixed thoroughlybefore pouring it into the funnel on thevibrating feeder connected to theabrasion apparatus. The test apparatus isset-up by starting the rotation of the steeldisc together with the suction assemblyand gradually adjusting the vibratingfeeder until a thin and uniform layer ofabrasion powder covers the track.

• 2 – 4 cutter steel test pieces are preparedby grinding them to the specifieddimensions. The grinding of the testsurface is a critical step and extra care isneeded to avoid overheating. The edgesof the test surfaces are ground, honedand visually examined to make sure thatthey are smooth and straight. The testbits must also be absolutely clean and drybefore weighing to the nearest 0.001g.

• One of the controlled test pieces isclamped under the load and placedgently on the steel disc. The test surfaceshould be horizontally aligned with thesteel disc, as it should otherwise beadjusted by the clamping of the test pieceand the suspension of the load.

• Testing time is 1 minute, i.e. 20revolutions. The amount of abrasionpowder fed onto the steel disc should besufficient, but not excessive. It is thereforeimportant to adjust the vibrating feederduring the test in order to avoid steelagainst steel abrasion or a pile of powderin front of the test piece. The operatorshould also make sure that the test pieceruns in the middle of the track and that asingle point of it does not bear directlyagainst the steel disc.

• Test pieces from 2 – 4 parallel tests arerinsed and dried thoroughly beforeweighing. The weight loss is calculated,and the results should normally notdeviate by more than 5 units (mg).

Test methods for soilFor soils the situation is quite different.

There are very few test methods to describethe abrasive characteristic of soils. Typically tests are limited to describe thehardness of minerals such as the VickersHardness Number (VHN), Mohs hardness,quartz content and abrasive mineral content(AMC), but grain size of the soil is not takeninto account.

Additionally there exist some abrasivitymodel tests for soils, such as the LosAngeles Abrasion Test, the Nordic Ball MillTest (NBMT) and Dorry’s Abrasion Test,which were developed to study the abrasionof aggregates to be used in road pavementworks (Gudbjartsson and Iversen, 2003). • The Los Angeles Abrasion Test rig

consists of a rotating circular drum (0.7mdiameter ; L = 0.52m) which is filled with48mm diameter cast iron spherical ballsalong with the aggregates (5-10kg). Thecylinder is rotated at a speed of 30 to33rpm for 500 to 100 revolutions. Thenthe material is sieved through 1.7mmsieve and the passed fraction isexpressed as a percentage of the totalweight of the sample. This value is called“Los Angeles abrasion value”.

• The Nordic Ball Mill Test, which iscommon in Scandinavia and Iceland, issimilar to the L.A. abrasion test.

• Dorry’s abrasion test uses the resistanceof aggregates to surface wear byabrasion induced by a rotating steel plateand is determined by measuring thevolume loss of the aggregate specimen.

The former three tests are suitable tomeasure the abrasion of soil grains due toabrasion induced by steel or by contact toother soil grains but they are not valid todetermine the abrasion of steel induced bysoil, which is the case in TBM tunneling.

On a Slurry-TBM, abrasion can have anadverse impact on the slurry dischargecomponents, pipes and pumps. Particularlyon long tunnel drives severe abrasion canoccur due to the long period of exposure ofthe discharge components to flowing slurrymixed with excavated soil.

In the USA there is a standardized test,called the Miller test (ASTM G75-01), whichwas originally developed in the oil industryfor deep vertical borings, but deals with asimilar abrasion problem as on Slurry-TBMdrives (figure 4). This test can be used tocollect data from which the relativeabrasivity of a slurry related to astandardized steel surface can be known,additionally the response of differentmaterials to an abrasive slurry can beinvestigated.

The test consists of a tray covered with alayer of Neoprene on the bottom and filledwith the test slurry (e.g. bentonite slurry +soil). A standardised steel block is dippedinto the test slurry and is loaded with a fixed

weight (22,24N is applied as a normal force).The steel block is driven in a reciprocatingmotion through the test slurry for 6 hours.The mass loss of the steel block ismeasured and gives the Miller Numberwhich is an index of the relative abrasivity ofslurries in terms of wear of a standardreference material. The wear damage on thestandard wear block is worse as the MillerNumber gets higher.

If materials other than the standard steelblock are used the measured mass lossindicates the Slurry Abrasion Response(SAR Number).

As described above, only few abrasiontest methods are available for soils. Theyprovide information on the abrasioncharacteristic of minerals within the soil andof slurry-soil mixtures, which is importantinformation, but limited to specific aspectsof the abrasion problem.

As will be described in Part 3 next month,a new attempt has been made for anabrasion test for soils, the NTNU SoilAbrasion Test (SAT), which describes theabrasiveness of soils in a more objectiveway. The initial testing has given quitepromising results, and the test is believed tohave a great potential for soft ground. T&T

ABRASIVITY & WEAR

49APRIL 2006 Tunnels & Tunnelling International

1. L Ozdemir & B Nilsen, 1999.“Recommended laboratory rock testingfor TBM projects”. AUA News 14:2, 21-35.2. E Büchi, JF Mathier & Ch Wyss, 1995.“Rock abrasivity – a significant cost factorfor mechanical tunnelling in loose andhard rock”. Tunnel 5/95, 38-44.3. R Lien, 1961. "An indirect test methodfor estimating the drillability of rocks". Dr.thesis, NTH Dept. of Geology, 90p. 4. A Bruland, TS Dahlø & B Nilsen, 1995."Tunnelling performance EstimationBased on Drillability Testing". ISRMCongress, Tokyo, 1995, 123-126.5. JT Gudbjartsson & K Iversen, 2003.“High-Quality wear-resistant pavingblocks in Iceland“. InternationalConference in Concrete Block Paving,Sun City, Oct 2003.6. ASTM G75-01 Standard Test Methodfor determination of slurry abrasivity (Millernumber) and slurry abrasion response ofmaterials (SAR number). Sep 2001.7. NTNU-Anleggsdrift, 1998. "Hard RockTunnel Boring". Norwegian University ofScience and Technology, Dept. of Buildingand Construction Engineering, Report 1B-98, 164p.8. BB Young & AP Millmann, 1964.“Microhardness and deformationcharacteristics of ore minerals”. Trans.Inst. Min. Metal, 73:437-466.

REFERENCES