Drilling With Fluid JETS

Int J Adv Manuf TechnolDOI 10.1007/s00170-005-0398-x

ORIGINAL ARTICLE

H.-T. Liu

Hole drilling with abrasive fluidjets

Received: 23 April 2005 / Accepted: 26 September 2005 / Published online: 23 May 2006# Springer-Verlag London Limited 2006

Abstract Small-hole drilling in several materials usingabrasive fluidjets—abrasive-waterjets and abrasive cryo-genic jets with liquefied nitrogen as the working fluid—were investigated through laboratory tests, numericalsimulation, and phenomenological analysis. Drilling isaccomplished by an abrasive slurry for abrasive-waterjetsand primarily by a stream of dry abrasives for abrasivecryogenic jets as the liquefied nitrogen changes phase afterexiting the mixing tube. Scaling factors were successfullyderived through analysis of profiles digitized from holeimages to collapse the data. For abrasive-waterjets, water isnearly incompressible; a high stagnation pressure inside theblind hole was developed causing delamination inlaminates and cracking in the thermal barrier coating.The return abrasive slurry also possesses considerableresidue erosion power that could cause damage to the holeentrance due to secondary wear. Neither delamination inlaminates nor cracking in the thermal barrier coating ishowever observed when abrasive cryogenic jets were used.

Keywords Abrasive fluidjets . Erosion . Hole drilling .Composites . 3-phase flows . Abrasive cryogenic jets

1 Introduction

Drilling small diameter holes in metals, metal matrixcomposites (MMCs), ceramics, ceramic matrix composites(CMCs), composites, and layered or coated materialspresent significant challenges to existing techniques. Theuse of ceramic thermal barrier coatings (TBCs) in high-

temperature sections of jet engines is an example of anapplication where there is a need to drill a large number ofsmall-diameter and large-aspect-ratio (length-to-diameter)holes, circular or specially shaped, at shallow angles forcooling. Another example is found in drilling jet engineblades that are typically made of coated metals or exoticmaterials such as nickel aluminide (Ni/Al). Not only areadvanced materials difficult to drill, holes at shallow anglesare often required. For the examples of Ni/Al vanes andceramic-coated engine shroud parts, holes must be drilledat angles of about 25° or less for efficient fluid filmcooling. The complicated geometries of new advancedcomponents require holes to be drilled with differentshapes (convergent, divergent, and square for example), intight spaces, and at different compound angles. This posessignificant problems that often restrict design options dueto the limitations of existing drilling tools. Current methodsare limited to holes of relatively small aspect ratios,especially for bore diameters below 0.06 cm, even inhomogeneous metallic parts. Finally, there is often need fordrilling shaped holes (such as square and rhombus) onaircraft vent screens to maximize cooling efficiency.

Drilling advanced materials with “solid” drill bits isoften not possible due to material brittleness and hardness.There can also be differences in the response to the drillbit action of the drill bit action. In addition to materialproperty constraints, mechanical drills have difficulty inproducing holes that are less than 0.04 cm in diameter andat shallow angles to the surface. The use of laser drilling inmany composites, and ceramic and metal components,results in undesired surface characteristics and generallyrequires additional processing, which substantially in-creases the cost of machining. The undesirable surfacecharacteristics include: heat affected zones (HAZ), recastlayers, thermal distortion, and dross build up at the holeexit surfaces. For small bore diameters, laser drilling islimited by its beam waist to small aspect ratios. EDM(electro-discharge machining) drilling, a process forcreating precision small diameter holes with excellenttolerance and surface finish control, has several drawbacksthat include target material dependence (electrically con-

H.-T. LiuStereoVision Engineering,Bellevue, WA 98006, USA

Present address:H.-T. Liu (*)OMAX Corporation,21409 72nd Avenue South,Kent, WA 98032, USAe-mail: [email protected].: +1-253-8722300

32:(2007) 942–957

ductive), slow drilling rates, limitation on hole inclination,and others mentioned above.

Other drilling methods such as stream drilling, electronbeam drilling (EBD), and ultrasonic drilling have theadvantage that many holes can be drilled simultaneously.The stream drilling process also has an advantage thatsmall diameter holes (<0.25 mm) can be drilled withextremely tight tolerance. On the other hand, the weak-nesses to the stream drilling process are its slow drillingrate; and hazardous working fluid (acid); there arelimitations on the type of target materials, geometry ofthe workpieces, and hole inclination. With electron beamdrilling (EBD), extremely high hole production rates can beachieved. The EBD process has been shown capable ofdrilling over 1,000 holes per second, but this process alsohas limitations in terms of target thickness, workpiecegeometry (surface of revolution), and production of recastlayers. Ultrasonic drilling can produce very precisegeometries, but as the workpiece material propertiesbecomes harder, the drilling times increase exponentially.

There is considerable demand for developing an efficienttool for precision drilling complex parts such as engineshrouds and vanes, which require drilling small-diameterholes (0.06 cm or smaller) at shallow angles (25°–60°).Abrasive-fluidjets (AFJs) have emerged as promising toolsfor such applications. AFJs are generalized from abrasivewaterjets (AWJs) in that the working fluid is not justlimited to water [1]. Other fluids include, but are not limitedto, liquefied carbon dioxide and nitrogen. Depending onthe physical and chemical properties of the working fluids,these tools differ from one another in the fluid dynamicsand therefore also different in the drilling processes. Forexample, AWJs have demonstrated considerable advan-tages in drilling Ni/Al and ceramic-coated metals for jetengine applications. For drilling small-diameter and largeaspect-ratio holes, AWJs are superior to other fielded toolssuch as lasers and EDM, particularly at shallow angles.

AFJs are complex 3-phase flows consisting of high-speed air, liquid, and solid interacting closely in theirformation [2]. A phase change takes place when acryogenic fluid such as liquefied nitrogen (LN2) is usedas the working fluid. The AFJ drilling process is verycomplicated involving both liquid-solid and solid-solidinteractions with materials while removed from the work-piece by the high-speed fluid and abrasive particles. As thejet begins to pierce into the material, the return flow affectsthe incoming jet and contributes to enlarging the hole,which, in turn, affects the flow pattern and the materialremoval process. Changes in material, drilling angle, holegeometry (diameter, length, and shape), and the drillingspeed require different optimal AFJ drilling conditions.Identifying the AFJ parameters to fulfill such conditionsrequires a basic understanding of the drilling process andits mechanics. Otherwise, considerable trial and error effortwould be necessary. Therefore, understanding the fluiddynamics of the 3-phase flow inside the hole is essential forimproving the performance of AFJs for precision drilling.

This paper investigates the fluid dynamics of AFJ holedrilling via a phenomenological approach. The investiga-tion focuses on the mutual influence of the AFJ and thehole geometry on each other. Emphasis is placed ondetermining how AFJs erode the target material to form theblind hole and how the geometry of the blind hole reshapesthe flow pattern inside the blind hole. Laboratoryexperiments and numerical simulations conducted by theauthor and his colleagues on AFJ hole drilling wereexamined and further analyzed to serve as the basis fordescribing the AFJ drilling process. AFJs used in theexperiments included AWJs and abrasive cryogenic jets(ACJ) using liquefied nitrogen as the working fluid. Whenappropriate, fluidjets (FJs) such as waterjets (WJs) andcryogenic jets (CJs) in the absence of abrasives were alsoused. AFJ drilling was conducted in several materialsincluding aluminum, steel, glass, composites and lami-nates. The main goal is to understand the hole drillingprocess by identifying important mechanisms associatedwith AFJs.

2 UHP technology and previous work

In this section, a brief review of UHP technology andrelevant previous work on hole drilling is given. Relevantwork is cited or reproduced in this paper to aid in theinterpretation of the AFJ drilling processes.

2.1 Abrasive waterjets

A UHP WJ is formed by forcing pressurized tap waterthrough a small orifice via a UHP pump up to 414 MPa(60 ksi) [3]. To form an AWJ, abrasives are entrained intothe jet through a feed port just downstream of the orificewhere suction is created, i.e. the “jet pump” effect [2].There are two types of UHP pumps available commer-cially, namely, the intensifier and crankshaft pumps [4].The crankshaft pump uses a crank similar to the one in anautomobile engine; the intensifier drives the plunger with ahydraulic cylinder, usually with oil. The bulky, noisy(100 dba or higher) and inefficient (70% efficiency)intensifier pumps are now being gradually replaced byefficient (95% efficiency) and low-noise (75 to 80 dba)crankshaft pumps for most AWJ systems.

Figure 1 shows a schematic of an AWJ drilling nozzleand the primary process parameters. Water is pressurizedup to 414 MPa and expelled through a sapphire orifice toform a high-velocity, coherent waterjet. Typical jetdiameters are 0.08 to 0.76 mm, and the corresponding jetvelocities are 460 to 880 m/s. The flow of the high-velocitywaterjet into the concentrically aligned mixing tube createsa vacuum, which is used to transport abrasives from ahopper to the nozzle abrasive chamber via a suction hose.The typical abrasive material is garnet, which has flowrates of up to 1.36 kg/min. Medium and fine abrasives

943

(mesh 80 to mesh 220) are most commonly used for metal,glass, and resin composites. The abrasives are acceleratedand axially oriented (focused) in the mixing tube. Typicaltube diameters range from 0.076 to 0.230 cm with lengthsup to 15 cm.

AWJs have been used in a wide range of omnidirectionallinear cutting applications. AWJs have also been investi-gated for some machining operations, such as milling,turning, drilling, and thin wafer machining [5]. Drilling atshallow angles to produce small-diameter holes (0.05 cm indiameter) is another application that has been demonstrat-ed. The commonly used method in the industry for piercingbrittle materials such as glass or composites with an AWJ isto employ relatively low pressures (about 35 MPa) to avoidcracking or delamination. The disadvantages associatedwith low-pressure drilling include low drilling rates,limited hole depths, and a “weak” suction capability ofthe jet, which results in either inconsistent or unreliableoperation. The use of high-pressure AWJs for fast drillingof tough materials results in limited control over drillingresults and is often associated with hole widening and ablasted area at the top surface. To solve the aboveproblems, computer-controlled algorithms were developedto optimize the drilling processes for different targetmaterials. For brittle materials such as glass, the pressure ofAWJs is ramping from zero to maximum value relativelyslowly to ensure abrasives are present at high pressures. Inthe absence of abrasives at very high pressures, the glass isdestined to crack.

Hole shapes greatly depend on jet structure, targetmaterial, dwell time, and standoff distance. The sensitivityto jet structure increases as the material resistance todrilling increases. Figure 2 shows different hole geometriesthat may be obtained by drilling with AWJs. In addition,specially shaped holes rather than circular or elliptical onescan also be conveniently drilled.

Recent work on Ni/Al-vane drilling showed greatpromise and capability of the AWJ process for drillingsmall-diameter air breathing holes. As illustrated in Fig. 3,several rows of holes are drilled near the leading edges atvarious angles of entry to optimize cooling of engine vanes.The AWJ process was found to be superior to laser andelectrical discharge machining in meeting hole qualityrequirements. In addition, the same AWJ without abrasivescan be readily used as a maintenance tool to clean andunclog air breathing holes to extend the operating life of

engine vanes. Although the AWJ process produced therequired holes in several vanes for jet engine testing,several drawbacks were identified. The most important oneis the lack of the ability to quantitatively guide and predicthow drilling parameters should be changed to control thehole geometry. This was accomplished by trial and errorusing test coupons, which was acceptable for exploratorystudies but not for production runs.

2.2 Abrasive cryogenic jets

UHP cryogenic technology was initiated at the IdahoNational Environmental and Engineering Laboratories(INEEL). Working with Praxair, Inc., the author and hiscolleagues successfully developed an LN2-based cryogenicjet (CJ) and abrasive cryogenic jet (ACJ) at pressures up to240 MPa [1, 6, 7]. This was achieved by feeding LN2 froma pressurized cryogenic storage tank into an intensifierpump. A subcooler was inserted between the storage tankand the pump to ensure that no evaporation takes placebefore the CJ or ACJ exits the nozzle.

The physical and chemical properties of the CJ/ACJ andthe WJ/AWJ differ in several aspects (see Table 1). As aresult, certain aspects of the machining processes of the twotypes of jets differ considerably. Such differences can betaken advantage of for machining various materials. Forexample, Liu et al. [1] demonstrated that the non-wettingCJ/ACJ was most advantageous for machining workpiecesmade of hygroscopic materials such as plaster of Paris. Inaddition, the CJ/ACJ freezes the workpieces locally andenhances their machinability. Machining the same materialswith WJ/AWJ would fail as wetting of the materials wouldcause the newly formed edges to collapse. Another majordifference is the phase change of the LN2 into N2 as the CJ/ACJ exits the mixing tube and impinges the target surface.For hole drilling, only a very small amount of LN2 enters

PARAMETERS

WATERJET PRESSUREORIFICE DIAMETER

ABRASIVE FLOW RATEABRASIVE SIZEABRASIVE MATERIAL

MIXING TUBE LENGTHMIXING TUBE DIAMETER

STANDOFF DISTANCEDRILLING ANGLEDWELL TIME

WORKPIECE MATERIALWORKPIECE THICKNESS

COMPONENTS

HIGH PRESSURE TUBE

ABRASIVE FEED HOSE

WATERJET ORIFICE

WATERJET

MIXING TUBE

ABRASIVE LIQUID JET

DRILLING RESULTS

ENTRY HOLE DIAMETEEXIT HOLE DIAMETERDRILLING TIME

R

Fig. 1 AWJ drilling nozzle concept and parameters

Fig. 2 Hole geometries obtained with AWJ

Fig. 3 AWJ-drilled Ni/Al vane for jet engine test evaluation

944

the cavity. Because water is nearly incompressibile atultrahigh pressures, buildup of the stagnation pressureinside the blind hole is considerably stronger for the WJ/AWJ than the CJ/ACJ. In addition, the phase change ofLN2 and escape of the N2 not only reduce the return flowrate but also transform the return flow from a slurry to dryabrasives, reducing the secondary erosion on the wall of theblind holes. The effects of the above differences in fluiddynamics on hole drilling are investigated in some detail inthis paper.

2.3 Laboratory experiments

Laboratory experiments were conducted by the author andhis colleagues to investigate the AFJ drilling process [1, 8].Liu et al. [8] conducted a series of experiments to drillholes on aluminum, mild steel and glass with AWJs. Thedrilling rates of several geometric parameters such as thebore diameter, depth, and material removal were measured.In particular, the drilling rates of the hole depth, dh/dt, wereanalyzed in detail. Formulation of predictive models wasderived by properly incorporating the concept of thresholdvelocities for different material type into Bitter’s erosionmodel [9]. Relevant results are summarized and revisited inSection 3. Experimental data were further analyzed toinvestigate various aspects of AWJ drilling processes. Newresults thus obtained are presented in Section 5.

During the course of developing ACJ, laboratoryexperiments were conducted to drill access holes throughlayers of composite materials consisting of metallic skinsand relatively thick hygroscopic filler [1]. Test couponswere fabricated for the ACJ drilling experiments. Thecoupons consisted of a 9.5-mm-thick aluminum top skin, a3.2-mm-thick stainless steel second layer, a 102-mm-thickhygroscopic filler made of plaster of Paris, and a 3.2-mm-thick stainless steel bottom skin. Cutting was conducted bymounting the coupons on a turntable with the nozzlemounted on a two-axis linear traverse inside the enclosedACJ cutting and scarifying workstation [1]. To gain accessto the bottom skin, 7.6-cm-diameter holes were first cut inthe top three layers. The filler core was either pulled outmanually or scarified away by moving the nozzle towardthe rotational center. Finally, the nozzle was lowered close

to the bottom stainless steel skin to cut out a 6.4-mm-diameter hole.

The ACJ successfully cut out the access holes in the firstthree layers without disturbing the remainder of thehygroscopic filler. During cutting, the filler layer wasfrozen solid by latent heat absorption as the LN2 evaporatesinto N2, greatly enhancing its mechanical stability. As aresult, the ACJ is turned into a dry machining process asopposed to the wet AWJ counterpart. If an AWJ were used,the spent water trapped in the cavity of the coupon wouldbe absorbed by the filler material, potentially leading toconsiderable swelling of the filler material. Soaking of thefiller material by the spent water could cause the side wallto collapse or trigger a strong chemical reaction providedthe filler material is strongly acidic or alkaline. The mostdifficult step when an AWJ is used is to keep the spentwater from leaking through the bottom access hole after itis drilled through. For certain demilitarization applications,even a small amount of water leakage below the skin layerscould be disastrous. The ACJ process creates an N2-richcavity at cryogenic temperatures, greatly minimizinghazards when operating in an explosive environment.

Both the ACJ and AWJ were employed to drill holes inmetal laminates (aluminum and stainless steel) andcomposites consisting of a thermal barrier coating (TBC)on an inconel substrate.

2.4 Numerical simulation

Numerical simulation of AWJ drilling was conducted byLiu et al. [8] using a commercial code, CFD2000, byAdaptive Research (Santa Monica, USA). The code solvessteady-state coupled three-dimensional conservation ofmass, momentum (Navier-Stokes), and energy equationsusing a finite volume approach in a structured three-dimensional grid. The code utilizes the pressure implicitwith splitting of operators algorithm to resolve thepressure-velocity coupling in the Navier-Stokes equations.Computations of turbulent flow use the k-ɛ model. Thecode is capable of two-phase flow computations and hasLagrangian particle tracking with momentum coupling tothe flow-field features.

CFD2000 does not have the option of a free liquid jet inair nor the option of fluid-solid and solid-solid interaction.The problem solved was that of a liquid jet (water) withparticles injected upstream of the nozzle exit. Forcomputational efficiency, the simulation was set up as atwo-dimensional problem in cylindrical coordinates ratherthan three-dimensional, which only allows a uniformdiameter hole with a flat bottom. Liu et al. [8] conductedthree steady-state cases: 1) jet against flat surface simu-lating the initial condition at which the AWJ at operatingpressure p=345 MPa impinges the target surface withoutpenetration; 2) AWJ into a relatively shallow hole (with adepth h=1.0 mm and a diameter dn=1.02 mm); 3) AWJ intoa deeper hole (h=2.03 mm and dn=1.02 mm). For all cases,the nozzle end was a 1.0-mm-diameter tube extending1.27 mm back from the exit, and the inlet fluid velocity, νw,

Table 1 Comparison of properties of WJ/AWJ and CJ/ACJ

Properties WJ/AWJ CJ/ACJ

Working fluid Water Liquefied nitrogenTemperature Slightly above

ambientCryogenic

Phase change No LN2 evaporatesinto N2

Wetting Yes NoInertness No Chemically inertStagnation pressure High Low to very lowErosive power of spent jet Relatively high Low

945

was specified as 368 m/s at the upstream end of the 1.27-mm-long tube. The ID of the mixing tube is dm=0.51 mmand the standoff distance is SOD=0.76 mm. Particles with aspecific gravity, S=4 and uniform diameter da=70 μm wereinjected into the upstream end of the tube with an initialvelocity of 184 m/s. Lagrangian particle tracking withmomentum coupling to the flow field was not fully utilizedin this application. Instead, particles were injected into theflow field after the flow field had been integrated to steadystate, and only enough particles were injected to obtain anidea of the behavior of the particle trajectories. Thecircumferential surface extending from the OD of thenozzle tube to the target surface was specified as a zeropressure boundary.

Figure 4 illustrates gray-level-coded flow and pressurepatterns of an AWJ penetrating into a 2.03-mm-deep hole,which is one of the simulated results reproduced from [8].The results serve as the basis for correlating the fluiddynamics with the drilling process inside the blind hole.The images in the figure show only the right half of theAWJ, the mixing tube and the hole. The left edge of eachimage corresponds to the centerline of the AWJ, the mixingtube (top), and the hole (bottom). The areas in gray to theright of the jet represent the walls of the mixing tube (top)and the hole (bottom), respectively. The standoff distanceof 0.76 mm is shown as the gap between the end of themixing tube and the target surface.

Figure 4a presents the patterns of the streamline (solid)and the abrasive trajectory (dashed). The two coincide witheach other inside the mixing tube. The high-speed waterflows in the hole along the core of the hole, reverses, turnshorizontally in the gap, and then exits radially. After theflow turns horizontally, the streamlines collapse and theflow is confined into a thin layer along the end of the mixing

tube. The maximum downward (positive u) and upward(negative u) flows occur immediately downstream of theend of the mixing tube and along the hole perimeter next tothe hole entrance, respectively, as illustrated in Fig. 4b. Theparticle trajectories show essentially the same trend of thestreamlines except that there is overshooting at locationswhere the streamlines change directions. As a result,particles are forced toward the hole bottom beforereversing, toward the wall before leaving the hole, andtoward the bottom of the mixing tube before turninghorizontally. At these locations, secondary erosion or wearis anticipated. Only the erosion at the hole bottom isdesirable, whereas the erosion at the wall near the holeentrance and the wear at the bottom of the nozzle are not.Figure 4c shows the gray-level pressure field. Maximumpressure is at the stagnation points on the bottom of thehole, as expected. Similar flow and pressure patterns wereobtained for the shallow hole (h=1.02 mm) except that forthe deep hole considerably large portion of the hole bottomis exposed to the stagnation pressure.

The CFD simulation was conducted specifically forAWJ. When the simulation was carried out, the CFD2000code did not have the capability for modeling 3-phase flow,particularly the phase change of the working fluid. It isanticipated that the results would be very much differentfrom the above for the ACJ. As most LN2 evaporated intoN2 before entering the blind hole, the stagnant pressureinside the blind hole is considerably reduced. The returnflow will mainly consist of abrasives without the organizedflow pattern as that of the return slurry flow, as shown inFig. 4a,b. The differences in the fluid dynamics of the twoworking fluids are expected to have considerable effects onthe erosion process applied to hole drilling.

Fig. 4 Results of CFDsimulation of AWJ drilling(p=345 MPa, dm=0.51 mm,dn=1.02 mm, h=2.03 mm,SOD=0.76 mm, S=4,da=0.07 mm) [8]. a Streamlineand trajectory. b +/−ucomponent. c Pressure

946

3 Relevant formulation and simple model

For the AWJ drilling, the waterjet flow rate, the abrasivemass flow rate or concentration, and the abrasive velocityare among the key governing parameters. Based on thepressure and velocity relationship, a Bernoulli velocity νBis defined as

νB ¼ffiffiffiffiffi2p

ρ

s(1)

where p and ρ are the jet pressure and water density,respectively. The waterjet flow rate, Q, is given by

Q ¼ π4Cdd

2o

ffiffiffiffiffi2p

ρ

s¼ π

4Cdd

2oνB (2)

where Cd and do are the overall discharge coefficient andthe orifice diameter, respectively.

Recent measurements of the velocities of water dropletsand abrasives in an AWJ, using a dual-disc apparatus, haveshown that the velocities of the water droplets in a WJwithout abrasives agree well with the Bernoulli velocity[10] when the abrasive feed port is closed. The velocities ofthe water droplets reduce to about 93% of the Bernoullivelocity when the feed port is open. In an AWJ, the averageabrasive velocity depends on the abrasive mass flow rate orconcentrationCa, defined as (lb/min)/gpm or 0.12 (kg/min)/(l/min). The best-fit average velocity of the abrasivesdecreases hyperbolically with Ca, or

νa ¼ νo þ ab

bþ Ca(3)

where a, b, and νo are fitted coefficients. Figure 5 illustratesthe measured average abrasive velocities for severalabrasive mass concentrations. Note that νw corresponds tothe water-droplet velocity as Ca vanishes.

From the results of cutting tests, threshold velocities ofthe water droplets and abrasives were measured for severalmaterials [10]. It was demonstrated that material removalwould not occur until the threshold velocities of the waterdroplets or the abrasives are exceeded. For example, thethreshold velocities of theWJ for Lexan and aluminumweremeasured to be about 600 m/s and greater than 800 m/s,respectively. For Lexan, the threshold velocity of abrasivesin an AWJ using Barton 220-mesh garnet is 190 m/s, whichis about one third of that of the WJ. It is surprising that thethreshold velocities of the abrasives change only margin-ally for materials from Lexan, and aluminum (≈210 m/s), tostainless steel (≈230 m/s). This is one of the reasons that theAWJ is more powerful and efficient than the WJ foreroding away materials.

A simple model describing the AWJ process, as shown inFig. 6, was developed based on the work of Bitter [9]. The

model relates the volume removal rate, V�to the change of

the product of the abrasive mass flow rate,m�and the kinetic

energy per unit mass of the abrasive, ν2a�2;with a correction

for the material strength, σf, or

V�¼ m

�ν2a

2σf: (4)

By assuming that the diameter of the hole is proportionalto the diameter of the mixing tube dm of the AWJ, Eq. 4 canbe rewritten in the following form:

h�¼ 2m

�v2a

πd2mσf; (5)

where h�

is the drilling (depth) rate. Recent experimentaland CFD results have shown that Eq. 5 is inadequate formodeling the AWJ drilling process [8]. A revision to Eq. 5was recommended by incorporating the effects of threshold

Fig. 5 Abrasive velocities as a function of abrasive mass concen-tration where p=345 MPa [10]

dm

h

hdA

v

vo

Fig. 6 Schematic used to describe the drilling process

947

velocity νt and material type (ductile or brittle) into Bitter’serosion model [10]:

h�¼

2m�ν2a

πd2mσ0f; for νa � νt

0; for νa < νt

;

8><>: (6)

where σ0f ¼ σf for ductile materials and σ0

f ¼ νaσf forbrittle materials.

In most AWJ drilling processes, the velocity of theabrasive particles determine the material removal rates andnot the velocity of the water carrying the abrasive particles.To simplify the analysis in determining the abrasiveparticle velocity, it is assumed that the water within thehole was stagnant, thus the abrasive particles aredecelerating due to viscous drag. Using Newton’s methodthe particle velocity can, therefore, be determined by

madνadt

¼ ma@νa@t

þ νa@νa@x

� �¼ �CdρAa

v2a2; (7)

where ma is the mass of the abrasive particle, Cd is thedrag coefficient, ρ is the fluid density, and Aa is the cross-sectional area of the particle. Consider a steady AWJ,@νa@t ¼ 0 and Eq. 7 becomes

dνaνa

¼ �CdρAa

2madx: (8)

Integrating Eq. 8 across the depth of the hole, we have

νa ¼ νa0e�K2h; (9)

where K2 ¼ CdρAa

�2mað Þ: Substituting Eq. 9 into Eq. 6,

then

h�¼ K1νa02e�2K2h; for νa � νt

0; for νa < νt;

�(10)

where K1 ¼ 2m�

d2mσ0f. It is interesting to point out that Eq. 10

has the same form as the best-fit curves of the drilling testresults reported in [8], or

po�p

� �nh�¼ k1e�k2h; for νa � νt

0; for νa < νt

�(11)

where k1 and k2 are two parameters in which the pressure,water, abrasive flow rates, abrasive properties, and othersetup configurations are lumped together, and n depends onthe material type. From the regression results, k1=1.574,0.606, and 2.023 and k2=0.0440, 0.0460, and 0.0366 foraluminum, steel and glass, respectively. Comparing Eqs. 10and 11, we have k1 ¼ K1ν2a0 po

�p

� �nand k2=2K2. These

equations have established the trend of exponentialdecrease in the drilling rate with increasing depth.

4 Laboratory facilities

4.1 AWJ facility

AWJ drilling experiments were conducted in a laboratoryfacility equipped with a computer-controlled AWJ system.The experimental setup used in the majority of thisprogram is shown schematically in Fig. 7. This systemconsisted of using a 3-axis Adept I manipulator which wasmodified for AWJ machining applications. This systemwas used for most of the parametric drilling tests conductedin this program. The nozzle used during this program was aWaterjet Technology intelligent nozzle system, whichallows for quick nozzle cartridge changes and processmonitoring of the critical AWJ process parameters. A 50 hpultrahigh-pressure (UHP) pump was used to obtain theUHP water for the drilling operations. A PC-based processcontrol system was used for all pressure ramping tests. APC-based data acquisition system was used for storing allof the pressure and reaction forces. A break-throughdetector was used to detect the point at which the AWJ firstpierces through the workpiece. The piercing time wasrecorded by the manipulator’s control system, and thendownloaded to the process PC for analysis.

4.2 ACJ facility

The cryogenic facility used for hole drilling included anLN2 supply and delivery subsystem and a machiningstation. Figure 8a shows a photo of an 11,360-liter LN2

storage tank, a cryogenic pump, subcoolers, and an in-linecooler. The ACJ machining station, as illustrated in Fig. 8b,consists of a linear traverse, a nozzle assembly, an abrasivehopper, and a personal computer that control the movementof the traverse. The system was developed for applying CJsand ACJs for various machining and surface preparationapplications [1, 7]. An enclosed ACJ cutting and scarifyingworkstation equipped with a 2-axis linear traverse, a

Fig. 7 AWJ drilling set-up

948

rotating table, and a waste collection port was constructedto assure safe operations [1].

5 Results

Previous and new results of laboratory experiments andCFD simulation are examined to investigate the fluiddynamics of the AFJ for hole drilling in several materials.Previous results including those of AFJ drilling in metals,glass, and composites were presented to show the macroaspects of the drilling process, namely, the drilling rates inthe hole depth and in the average diameter [1, 9]. Some ofthe relevant results are reproduced in previous sections.Further analysis of previous work was subsequently

conducted with the goal of examining the local erosionpattern as the AFJs enter and continue penetrating into theblind hole. In addition, new results of AFJ drilling on metallaminates and composites are also added to demonstrate theeffects of properties of the working fluid on the multi-phaseflow pattern inside the blind hole, which in turn affect thedrilling process. The results of CFD simulation arespecifically used to correlate the multi-phase flow patternspredicted by the CFD code and the drilling mechanisms.

To facilitate capturing the images of the drilled hole, themetal workpieces were sectioned along the maximumdiameter of the bores, whereas the transparent/translucentglass workpieces require no additional preparation. Imagesof AWJ-drilled holes in aluminum, mild steel, and glasswere captured with a digital camera. The profiles of holeswere measured from the captured images for furtheranalysis. Sample images of drilled holes and the measuredprofiles were illustrated in [8]. The drilling rates in the holedepth for these materials were derived from the measuredprofiles. Nonlinear regression of the drilling rate wasconducted based on the improved erosion model of Bitter[8, 9]. The consistency of the model and the regressionresults are examined in Section 4. These results haveestablished the trend of exponential decrease in the drillingrate with increasing depth.

The laboratory data were revisited and further analyzedto interpret the erosive mechanisms from the fluid dynamicpoint of view. The physical and chemical properties of theAFJ and the results of the CFD simulation are used to aid inthe interpretation. New results on AFJ drilling in metallaminates and composites are presented to provide insightto the drilling performance resulting from differences in thefluid dynamics of AFJs. Additional evidence is providedby examining the different performances of WJ and CJ forcoating removal.

Fig. 8 Abrasive cryogenic jet system for machining and surface preparation. a LN2 delivery subsystem. b Sketch of UHP ACJ machiningworkstation

Fig. 9 Images of hole drilled in aluminum at p=138 and 241 MPa.a t =20, 35, 50, 70, 90, 110, 130, and 150 s. b t=50, 60, 70, 80, 90and 100 s

949

5.1 Drilling in metals and glass

Experimental results reported by Liu et al. [8] were furtherprocessed and analyzed to understand the AWJ holedrilling process. In particular, holes drilled with AWJs onaluminum (type 6061) and float glass workpieces areexamined qualitatively and quantitatively. The AWJ wasconfigured to have a 0.18 mm/0.51 mm (sapphire diameter/mixing-tube diameter) nozzle combination. The mixingtube had a length of 5.08 cm. Barton garnet abrasives witha size distribution of 120-mesh were used. Typical abrasivemass flow rate ranged from 0.03 to 0.1 kg/min. Thestandoff distance was set at 0.64 mm. It should be pointedout that pressure ramping from zero to the operatingpressure was required for glass drilling. Note that a UHPAWJ void of abrasives due to clogging often induces

cracking in glass. The pressure ramping is to ensure thatabrasives begin to flow into the nozzle after the AWJ hasreached a fraction of the operating pressure. Such ananomaly is attributed to the large stagnation pressure thatexceeds the tensile strength of glass (see discussion in thenext section).

Aluminum workpieces were cut in half along the axes ofthe hole array that was otherwise inaccessible photograph-ically. Images of the halved holes were captured using adigital camera (e.g., Canon Power Shot G5 with 5 megapixels). To enhance the contrast between the cavities andthe background, the hole cavities were carefully filled withplaster of Paris. The profiles of the holes were thenmeasured digitally to quantify the results. It should bepointed out that axes of individual holes are not expected tofall onto the cutout planes of the aluminum workpieces,resulting in low biases of the diameters measured from theimage. Cutting the aluminum workpieces in half couldslightly chip the hole edges; therefore subsequent polishingof the finish surface was used to reduce such an effect. Forthe glass workpieces, images of the blind holes werecaptured using back illumination.

Figures 9 and 10 illustrate, respectively, several imagesand profiles of AWJ-drilled holes in aluminum for p=138and 241 MPa and drilling durations, t, up to 150 seconds.The abscissa and ordinates in Fig. 10 are, respectively, thebore diameter d and axial distance from the hole entrance lwith the horizontal scale distorted to exaggerate the changein the diameter with the axial distance from the holeentrance. In general, there is a consistent trend that the borediameter increases with the depth, reaches a maximum, andthen decreases toward the bottom of the blind hole. Thebottom of the blind hole is pointed in shape. At a givenaxial distance, the bore diameter increases with the drillingduration, as expected. A similar trend is observed for holesdrilled in other ductile materials such as mild steel [8].

Figures 11 and 12 illustrate, respectively, hole imagesand profiles in float glass for p=138, 241, and 345 MPa anddurations up to 150 s. It is evident from Figs. 9, 10, 11, 12that the image and edge qualities are visually andquantitatively better for holes in glass than for those inaluminum. Cutting the aluminum workpieces in halfdegraded the image and edge qualities.

Fig. 10 Diameter profiles of aluminum holes measured fromimages shown in Fig. 9. a p=138 MPa. b p=241 MPa

Fig. 11 Images of holes drilledin float glass at p=138, 241, and345 MPa. a t=30, 45, 60, 75, 90,and 105 s. b t=40, 50, 80, 90,100, 120, 150 s. c t=50, 60, 70,80, and 100 s

950

There are noticeable differences between the geometryof the holes drilled in the two materials because thematerial properties and wear resistance differ. For example,holes in glass (but not in aluminum) experience considerablewear in the vicinity of the hole entrance. As a result, theglass hole entrance is rounded into a funnel shape. Such awear pattern is induced by the return slurry flow shown in

Fig. 4a,b, the results of the CFD simulation. Apparently,the difference in the wear resistance to the return slurryflow between the two materials is responsible for thedifferent degrees of wear observed in Figs. 9 and 11. Basedon the findings that the drilling rate decreases exponentiallywith depth and that the entrance wear increases withpressure, the optimal strategy for glass and materials withlow erosion resistance would be to use lower pressures toimprove the quality of hole drilling.

Attempts were made to collapse the data to remove thedependency on the drilling duration. Further analysis of thedata has shown that the bore diameter correlates well withln (t) for a given AWJ pressure. Figures 13 and 14 replotFigs. 10 and 12 by replacing l with l/lmax in the ordinatesand d with d/ln(t) in the abscissa. All of the profilescollapse reasonably well using the scaling factors lmax andln(t) or

d lð Þ lnðtÞ� ¼ f l lmax= Þ;�(12)

where l is the axial distance from the hole entrance, lmax isthe hole depth, and t is the piercing time in seconds. Theprofile of the scaled diameter, d(l)/ln(t), is a function of theAWJ pressure and piercing parameters.

The measured hole profiles shown in Figs. 13 and 14have demonstrated that they are not strictly parabolic inshape as illustrated in Fig. 6. In particular, the actual profiledoes not have the maximum diameter at the hole entrance.Furthermore, secondary erosion by the return slurry flowshould be considered especially for materials with lowthreshold velocities. Refinements will be needed to takethese considerations into account in future modelingefforts.

Fig. 12 Diameters of glassholes at three pressures forseveral drilling durations.a p=138 MPa. b p=241 MPa.c p=345 MPa

Fig. 13 Scaled diameter profiles of aluminum holes. a p=138 MPa.b p=241 MPa

951

5.2 Drilling in laminates

Based on the results of the CFD simulation, the axialvelocity component u decelerates to zero and reverses itsdirection near the bottom of the blind hole (Fig. 4b). Alarge stagnation pressure is therefore developed inside theblind hole (Fig. 4c). In metals, the stagnation pressurewould have no effect on the drilling process. In glass,however, the tensile strength is relatively low. Largestagnation pressure could induce cracking under certaincircumstances. In fact, drilling holes in glass with UHPAWJs void of abrasives, even momentarily, often results incracking of the glass workpiece. Although the exact causeof such a problem is not yet understood, large stagnationpressure developed inside the blind hole is primarilyresponsible for such an anomaly. Pressure ramping is aworkable remedy to assure that abrasives begin to flow intoAWJs at start up.

Laminates could be an ideal material to investigate theeffects of stagnation pressure on hole drilling as the bondingstrength of the adhesive is considerably weaker than thetensile strength of the sheets that form the laminates. AWJdrilling tests were conducted using aluminum and stainlesssteel shim stocks consisting of multiple 0.076-mm sheets.Tests were conducted using 2.54 cm by 2.54 cmworkpieces

for three pressures, p=34.5, 103.4, and 206.8 MPa, and fourratios of jewel diameter to mixing-tube diameter, dn (mm)/dm (mm) = 0.127/0.508, 0.178/0.508, 0.178/0.762, and0.229/0.762. 220-mesh Barton garnet, with a mass flow rateof 0.09 kg/min, was used as the abrasive. Test results showthat delamination between sheets begins at pressures of103.4 MPa and higher, indicating that bonding strength ofthe adhesive is overcome by stagnation pressures developedunder these conditions.

Many of the delaminated workpieces were separated intotwo halves. Let us examine the images of a typical AWJ-drilled hole in an aluminum workpiece with a nominalthickness of 2.4 mm. The AWJ was operated at 207 MPaand the nozzle combination dn/dm was 0.178/0.762.Figure 15a–c show the top, rear, and side views of thetop half of the workpiece, respectively. Delamination ismanifested in the form of a large popped-up bubble manytimes larger than the size of the hole. The bubble is ellip-tical in shape (Fig. 15a,b). Once the bubble is initiated, itcontinues to grow in size until delamination reaches one ofthe edges of the workpiece, as demonstrated in Fig. 15a,cwhere the bubble reaches the top edge of the workpiece. Atand beyond that point, the stagnation pressure built up inbetween sheets is vented. Under the same experimentalcondition, similar results were obtained for a 2-mm-thick

Fig. 14 Scaled diameterprofiles of glass holes.a p=138 MPa. b p=241 MPa.c p=345 MPa

Fig. 15 AWJ-drilled hole inaluminum laminate (scale inmm). a Top view. b Rear view.c Side view

952

stainless steel workpiece, as illustrated in Fig. 16a–c. Thedelaminated stainless steel workpiece was also separatedinto two halves. The above similarity is somewhat expectedas the adhesives used for the two laminates have about thesame bonding strength although the tensile strengths ofaluminum and stainless steel sheets differ.

As discussed in Section 2.2, considerable reduction inthe stagnation pressure can be achieved by using an ACJ asthe working fluid LN2 changes phase after exiting themixing tube. The ACJ would provide an ideal opportunityto investigate the effect of stagnation pressure on holedrilling in laminates. Subsequent tests were conducted bydrilling holes in aluminum and stainless steel laminatesusing the ACJ described in Section 4.2. Except for thedifference in the working fluid, other experimentalparameters were kept the same as those used in AWJdrilling tests. This would ensure a meaningful comparisonof the test results for assessing the effects of the fluidproperty on the hole drilling process. Figures 17 and 18illustrate the images of the ACJ-drilled holes in thealuminum and stainless steel workpieces. They correspondto the AWJ-drilled holes shown in Figs. 15 and 16. It isevident from Figs. 17 and 18 that the ACJ does not causedelamination and therefore no bubble is developed in thelaminates.

The above results have demonstrated that the phasechange of the working fluid, LN2, in the ACJ hasconsiderably modified the behavior of the ACJ, ascompared to that of the AWJ. Such a modification affectsthe erosion process of the ACJ for hole drilling. Most of theN2 evaporates from LN2 as the jet exits the mixing tubeand, upon impinging onto the target surface, escapes theblind hole. Only a small amount of LN2 that continues toevaporate enters the blind hole. The stagnation pressureinside the blind hole therefore reduces proportionally.When the stagnation pressure has reduced below the

bonding strength of the adhesive, no delamination isexpected to take place. Evidently, this occurred during theACJ drilling tests although the stagnation pressure insidethe blind hole was not measured. There are means ofcontrolling the rate of phase change of ACJs such asheating the workpiece to adjust the stagnation pressureinside the blind hole. However, phase change must not takeplace inside the mixing tube in order to maximize theacceleration of abrasives and therefore the erosive power ofACJs.

The return flows inside the blind hole also differ betweenthe ACJ and AWJ. Because of the reduction of LN2

entering the blind hole, the flow patterns inside the blindhole deviate from those shown in Fig. 4. As LN2 enteringthe blind hole continues to evaporate, the return flowmainly consists of dry abrasives and N2. Material erosion islargely performed by dry abrasives rather than by theabrasive slurry of AWJs. In the absence of a slurry flow, thereturn flow of N2 is incapable of trapping and channelingthe abrasives because of the vast density differencebetween the two media. As a result, the trajectories of theabrasives inside the blind hole are much less organized forACJs than for AWJs. The secondary erosion by the exitingabrasives of ACJs is more or less a random process. Duringdrilling tests, it was discovered that spent abrasives in ACJsare indeed dry and their exit angles are very large. On theother hand, spent abrasives of AWJs are confined in aslurry jet with relatively small exit angles.

In Figs. 15 through 18, the diameters of entrance holesare consistently larger than those of the exit holes. This is aresult of the relatively short dwell time after the laminateswere broken through and is consistent with the holeprofiles in metals and glass that the bottom of the blind holeis pointed (Figs. 9 and 11). In other words, the diameter ofthe exit hole can be readily enlarged to match that of theentrance hole by increasing the dwell time.

Fig. 16 AWJ-drilled hole instainless steel laminate (scale inmm). a Top view. b Rear view.c Side view

Fig. 17 ACJ-drilled hole inaluminum laminate (scale inmm). a Top view. b Rear view.c Side view

953

5.3 Drilling in composites

A composite material consisting of a ceramic-basedthermal barrier coating or TBC (0.8 mm thick) on aninconel substrate (1.6 mm thick) was selected for AFJdrilling to further investigate the fluid dynamic effects onhole drilling. The porous and granular TBC has relativelylow tensile strength as compared to that of inconel.Figure 19 illustrates several images of AWJ-drilled holesin the TBC-inconel plate. The granular structure and sizedistribution of the TBC can be readily visualized from theimages. Figure 19a shows four holes drilled in the platewith a diameter of about 0.6 mm and a spacing of 3.2 mm.The dark-color area located in the top, right and bottomedges of the image corresponds to the original surface ofthe TBC prior to drilling. Severe delamination induced bythe AWJ has led to formation of cracks around individualholes (Fig. 19b). Relatively large pieces of TBC eventuallybreak away from the surface as cracks developed inneighboring holes connect to one another (Fig. 19a - light

color area). In addition, a considerable amount of TBC isremoved from the immediate vicinity of the hole entranceresulting from secondary wear induced by the return slurry,exposing a part of the inconel substrate (see particularly thethird hole from the left). Figure 19c represents themagnified image of the 1-mm-diameter hole shown inFig. 19b. Exposition of the inconel substrate resulting fromthe secondary wear at the hole entrance by the return slurryis evident. Apparently, high toughness of the inconelsubstrate tends to magnify the effects of the stagnationpressure and return slurry as they have more time todamage the porous and granular TBC before breakthroughof the blind hole takes place.

Figure 20a shows images of a portion of two rows ofholes drilled with the ACJ in the same TBC-inconel plate.The bore diameter is about 1 mm. The two rows of holesare staggered to facilitate three different spacings betweenholes: 3.4, 5.1 and 5.5 mm. Careful examination of theworkpiece shows no evidence of delamination andcracking of the TBC around the holes. Secondary wear

Fig. 18 ACJ-drilled hole instainless steel laminate (scale inmm). a Top view. b Rear view.c Side view

Fig. 19 AWJ-drilled holes oncomposite, thermal barrier coat-ing (TBC) on inconel substrate.a TBC flaked off from delami-nation (see b). b Crack on TBCresulting from delamination.c Irregular wear in vicinity ofhole entrance

954

by return dry abrasives of the ACJ rather than by theabrasive slurry of the AWJ also reduces considerably fromcomparison of the images shown in Figs. 19b and 20b. Theabove results further demonstrate the advantages ofreducing the stagnation pressure through phase change ofthe ACJ in alleviating delamination of porous coatingssuch as the TBC.

For delicate applications such as aircraft depainting, it ispreferable not to use abrasives so that only the old corrodedpaint is removed leaving the underlying primer untouched.The use of abrasives also tends to damage the thinaluminum/composite substrate. Efforts have been madewith some success to apply both WJs and CJs for aircraftdepainting [7, 11]. Of particular interest is the effect ofphase change of CJs on the performance of paint removalas compared to that of WJs. Although the information isnot directly related to AWJ hole drilling, it helps furtherunderstand the roles of phase change of the working fluidin affecting the fluid dynamics and therefore materialremoval without the complication of the presence ofabrasives.

A series of tests was conducted by Liu et al. [11] to useWJs and CJs to remove aircraft paint samples furnished byBoeing Commercial Company. The paint was directlyapplied to a thin aluminum substrate without a primer. Bothjets used a Flow International (WA, USA) fanjet nozzleoperating at 138 MPa. The nozzle standoff distance was1.3 cm and the jets traversed at 6.4 cm/s from left to right.Figure 21a,b illustrate images of two sets of strips after asingle passage of the two jets. There are striking differencesbetween the visual appearances of the traces. In essence,the strips for the WJ are wider with more rugged edges thanthose for the CJ. Careful examination of Fig. 21a showsthat pieces of paint flakes that appear in light color arelifted but loosely attached to the edges of the strip.

Based on the properties of the two jets, an attempt toexplain the observed depainting differences is in order. Asthe primary WJ approaches the point of impingement, theaxial velocity decelerates to zero. The kinetic energy of thejet converts into potential energy by means of developing astagnation pressure. After impingement, the WJ turns 90degrees (i.e., the secondary WJ consisting of the spentwater), the potential energy then converts back to kineticenergy by accelerating the horizontal velocity of a plane jetmoving radially outward. As the secondary WJ spreadsoutward, the velocity decreases with the radial distance inorder to satisfy the continuity condition and to work againstthe drag force. The paint is removed from the substrate

provided the velocity of the primary WJ exceeds thethreshold velocity of the paint. After turning 90 degrees,the secondary WJ, consisting of the spent water, can stillachieve very high velocity and possess considerableresidue erosion power. However, the impingement gen-erates strong turbulence and the velocity of the jetspreading radially outward is highly fluctuating and rapidlydecelerating with the radial distance. As a result, paintremoval only takes place locally where the jet velocityexceeds the threshold velocity of the paint, leaving behindstrips with rugged edges. Laboratory and numerical studieshave demonstrated that the impinging jet as it turns 90degrees generates a strong (primary) vortex inside theviscous boundary layer [12, 13]. An adverse pressuregradient is formed slightly ahead of the primary vortexcore. The presence of the adverse pressure gradient causesflow separation in the viscous boundary layer, which inturn generates a counter rotating secondary vortex beneaththe primary vortex. The interaction between the primaryand secondary vortices induces an upward and outwardmotion that helps lift the delaminated paint chips from thesurface [14].

For the CJ, the LN2 begins to evaporate as soon as itexits the mixing tube. There is essentially no secondary CJ(in the liquid state) formed after impingement as most ofthe N2 escapes the target surface. In other words, the paintis removed mainly by the primary CJ. In the absence ofpaint removal by the secondary CJ, traces are relativelynarrow with smooth edges, as illustrated in Fig. 21b.

6 Summary and discussion

The processes of hole drilling in several materials with twoAFJs, AWJs and ACJs were investigated to understandrelevant erosion mechanisms. Emphasis was made tounderstand the effects of fluid dynamics inside blind holeson the drilling processes. The investigation was carried outby revisiting and further analyzing several series of AFJ-drilling tests in aluminum, stainless steel, glass, aluminumand stainless steel laminates, and a ceramic-based thermalbarrier coating. Results of numerical simulation of AWJdrilling using a computational fluid dynamic code,CFD2000, were also used to aid in interpretation of therole of fluid dynamics in hole drilling processes.

Fig. 20 ACJ-drilled holes oncomposite, thermal barrier coat-ing (TBC) on inconel substrate.a Absence of crack and TBCflaking next to holes. b Reducedentrance wear

955

6.1 AFJ flow pattern

The AWJ flow pattern inside the blind hole, as derivedfrom the CFD model, is reproduced in Fig. 4 [8]. Theabrasive slurry, the primary source of erosion, enters intothe blind hole, decelerates toward the bottom, and reversesits direction to form a return flow (Fig. 4a). The maximumaxial velocity of the return water takes place near the sidewall toward the hole entrance (Fig. 4b). The abrasivetrajectories lag the streamline forcing abrasives in thereturn slurry toward the wall of the blind hole and the endtip of the mixing tube. The above flow pattern indicates thatsecondary erosion induced by the return slurry couldenlarge the bore diameter in the vicinity of the holeentrance. The tip of the mixing tube is also experiencingconstant bombardment of the spent slurry. Inside the blindhole, a large stagnation pressure is created by the velocityreversal of water near the bottom of the blind hole.

The corresponding flow pattern of ACJs inside the blindhole is expected to differ from that shown in Fig. 4. The LN2

in ACJs begins to change phase after exiting the mixingtube. Most of the LN2 has evaporated into N2 beforereaching the bottom of the blind hole. In other words, theprimary mode of erosion is achieved by a high-speed streamof dry abrasives. In the absence of LN2 inside the blind hole,the return flow, consisting of spent abrasives bouncing fromthe interior surface of the blind hole and colliding with oneanother, is not as coherent as the return abrasive slurry ofAWJs. In other words, the secondary erosion is more or lessrandomly distributed along the interior surface rather thanconcentrated in the vicinity of the hole entrance.

6.2 Hole profiles in metals and glass

Laboratory results of AWJ-drilled holes in aluminum,stainless steel, and glass have established an exponentialdecrease in the drilling rate with the depth [8]. Furtheranalysis of the data has established a simple relationship,expressed in Eq. 12, to collapse the hole profiles, indicatingthat AWJ drilling processes are repeatable. Such a findingis of fundamental importance because repeatability is thekey to improving process precision.

6.3 Performance of AWJs and ACJs

The effects of the fluid dynamics on hole drilling processesare further demonstrated by test results using AWJs and

ACJs. The use of aluminum and stainless steel laminates, aceramic-based TBC on an inconel substrate, and aircraftpaint on an aluminum substrate rather than metals and glasshas magnified the effects that are otherwise difficult toobserve.

6.3.1 Holes drilled in laminates

Holes drilled in laminates consisting of multiple 0.076-mmaluminum and stainless steel sheets were made to investigatethe effects of fluid dynamics on hole drilling processes.AWJs operating at pressures higher than 103.4MPa inducedsevere delamination inside the laminates, creating pop-upbubbles many times larger than the size of holes. Delam-ination is caused by the build up of stagnation pressure thathas overcome the bonding strength of the adhesive betweensheets of aluminum and stainless steel. The elliptical bubblescontinue to grow in size once initiated until delaminationreaches one of the edges of the workpiece and the stagnationpressure is vented. On the other hand, holes drilled withACJs show no evidence of delamination and bubbleformation in the laminates. It is evident that phase changeof the LN2 in the ACJs has reduced the stagnation pressurebelow the bonding strength of the adhesive.

6.3.2 Holes drilled in TBC on inconel substrate

AWJs and ACJs were also applied to drilling holes in TBCon an inconel substrate. The porous and granular TBC haveconsiderably lower tensile strength than that of the inconelsubstrate. Test results show that severe delamination in theTBC has led to development of cracks around the AWJ-drilled holes. Large pieces of TBC break away from thesurface as cracks which have developed in neighboringholes connect to one another. Also, a considerable amountof TBC is removed from the immediate vicinity of the holeentrance resulting from secondary wear induced by thereturn slurry, exposing a part of the inconel substrate. Hightoughness of the inconel substrate worsens the damage ofthe stagnation pressure and return slurry to the porous andgranular TBC. On the other hand, the ACJ does not causedelamination and cracking in the TBC as the stagnationpressure reduces due to phase change of the LN2. Inaddition, the transformations of the return flow from acoherent abrasive slurry to a stream of incoherent dryabrasives in the ACJ has significantly reduced the damageto the TBC in the vicinity of the hole entrance.

Fig. 21 Stripping of aircraftpaint with WJ and CJ using afanjet nozzle [1]. a WJ. b CJ

956

6.3.3 Aircraft depainting

The results of removing aircraft paint on a thin cladaluminum substrate using WJs and CJs equipped withfanjet nozzles have further helped understand the erosionpower and pattern of the two jets in the absence ofabrasives. Both jets have been found to be capable ofremoving the paint without damaging the aluminumsubstrate. Paint strips removed by WJs are wider withmore rugged edges than those removed by CJs. For the twojets, paint is removed primarily by the impinging action aslong as the velocities of the WJs and CJs exceed thethreshold velocity of the paint. The spent WJs, after turning90 degrees, however, still possess considerable residueerosion power to lift and remove paint that survives theimpinging action of the primary WJs. On the other hand,the LN2 in CJs changes phase after exiting the mixing tube.After turning 90 degrees, the spent CJs that consist ofmostly N2 have very little residue erosion power and areincapable of removing additional paint left on the work-piece; although, the higher paint removal rate of WJs is anadvantage over CJs. The rugged edges left behind by WJsare undesirable because coating removal cannot beprecisely controlled. And additional clean-up steps arerequired to remove partially detached paint chips beforerepainting. Therefore, CJs with low secondary erosionpower are advantageous for certain applications whereprecision, rather than the rate of coating removal, is animportant consideration.

6.4 Future work

The current findings that the AFJ hole drilling processesare repeatable and that the bore diameter increases as ln(t)indicate that a more accurate analytical model of AFJmodel may be developed. The incorporation of thresholdvelocities of target materials and material types in themodel described in Section 3 has resulted in limitedimprovement. Additional improvements must be made totake into account several other factors including thenonparabolic hole profiles, 3-phase flow state, secondarywear by the return flow, and other fluid-solid and solid-solid interactions. Of particular interest is an accuratemodel describing the erosive wear long after breakthroughof the blind hole has taken place. Such a model would havea direct application to maximizing the life of mixing tubesas a means to improve the performance of AWJ precisionmachining. It would be useful to determine whether theincrease of the bore diameter of the mixing tube alsofollows a simple rule similar to that expressed in Eq. 12.

Although CJs/ACJs have shown advantages over WJs/AWJs to minimize damage to various materials during holedrilling, CJs/ACJs are considerably bulkier and moreexpensive, and more hazardous to operate than WJs/AWJs.Research and development of a novel WJ/AWJ system to

emulate the phase changing characteristics of the CJs/ACJsis currently underway. Such a system could be switchedback and forth to work as conventional WJs/AWJs or asemulated CJs/ACJs, greatly enhancing the versatility of theultrahigh-pressure technology for precision machining,surface preparation, and other industrial and militaryapplications.

Acknowledgements This work was sponsored partly by ArmySBIR Contract No. DAAJ02-97-C-0025 and an IRD funding fromStereoVision Engineering. The laboratory experiments were con-ducted at Waterjet Technology, Inc. (WTI) where the author served asa senior scientist. Some of the experiments reported in this paperwere performed jointly by the author and his colleagues Pete Milesand Nick Cooksey. The author would like to thank Praxair Inc. forinstalling the cryogenic facility at WTI as an in-kind contribution toseveral SBIR projects.

References

1. Liu H-T, Fang S, Hibbard C (1999) Enhancement of ultrahigh-pressure technology with LN2 cryogenic jets. Proceedings ofthe 10th American waterjet conference, Houston, Texas,August, pp 14–17

2. Hashish M (1995) Advances in fluidjet beam processing.Proceedings of the 8th American waterjet conference, Houston,Texas, 26–29 August, pp 487–503

3. Summers D (1995) Waterjet technologies, 1st edn. Taylor &Francis, London, pp 882

4. OMAX Corporation (2005) A comprehensive overview ofabrasivejet technology. http://www.omax.com/about_abrasive.php. Cited 10 March 2006

5. Momber AW, Kovacevic R (1998) Principles of abrasivewaterjet machining. Springer, Berlin Heidelberg New York,pp 398

6. Dunsky CM, Hashish M (1994) Feasibility study of machiningwith high-pressure liquefied CO2 jets. ASME proceedings ofsymposium on nontraditional manufacturing processes in the1990s, Chicago

7. Liu H-T, Bulter T (1998) A vanishing abrasive cryogenic jet forairframe depainting. Proceedings of the 14th internationalconference on jetting technology, Brugge, Belgium, 21–23September, pp 519–533

8. Liu H-T, Miles P, Veenhuizen SD (1998) CFD and physicalmodeling of UHP AWJ drilling. Proceedings of the 14thinternational conference on jetting technology, Brugge, Belgium,21–23 September, pp 15–24

9. Bitter JGA (1963) A study of erosion phenomena - part I. Wear6:5–21

10. Liu H-T, Miles P, Hibbard C, Cooksey N (1999) Measurementsof water-droplets and abrasive speeds in waterjets and abrasivewaterjets. Proceedings of the 10thAmericanwaterjet conference,Houston, Texas, 14–17 August

11. Liu H-T, Fang S, Hibbard C (1999) Enhancement of ultrahigh-pressure technology with LN2 cryogenic jets. Proceedings ofthe 10th American waterjet conference, Houston, Texas, 14–17August

12. Didden N, Ho CM (1985) Unsteady separation on a boundarylayer produced by an impinging jet. J Fluid Mech 160:235–256

13. Bilanin AJ, Teske ME, Hirsh JE (1978) Neutral atmosphericeffects on the dissipation of aircraft vortex wakes. AIAA J16:956–961

14. Liu H-T, Hwang PA, Srnsky RA (1992) Physical modeling ofground effects on vortexwakes. AIAAJAircraft 29(6):1027–1034

957