PROGRESS IN FLUIDIZED BED ASSISTED ABRASIVE JET MACHINING (FB-AJM): INTERNAL POLISHING OF ALUMINIUM TUBES

Massimiliano Barletta, Stefano Guarino, Gianluca Rubino, Vincenzo Tagliaferri1 1 University of Rome “Tor Vergata”, Department of Mechanical Engineering, Via del Politecnico, 1 – 00133 Rome, Italy e-mail: barletta@mail.mec.uniroma2.it tel: +390672597168 mob: +393204394406 fax: +39062021351 ABSTRACT

This paper deals with the internal finishing of tubular components made from a high strenght aluminium alloy

(AA 6082 T6) by using a Fluidized Bed assisted Abrasive Jet Machining (FB-AJM) system.

First, the influence of abrasive jet speed, machining cycle and abrasive mesh size on surface roughness and

material removal trends was investigated performing a Taguchi’s reduced experimental plan. Second, the leading

finishing mechanisms were studied using a combined 3d profilometer-SEM analysis to monitor the evolution of the

surface morphology of machined workpieces. Finally, the circumferential uniformity and precision machining of the

inner surface of workpieces were tested by evaluating the values of the more significant roughness parameters in

different circumferential locations.

Consistent trends of surface roughness vs operational parameters were measured and significant material

removals was found to affect the workpieces during machining. Accordingly, FB-AJM was found to preferentially

machine the asperities and irregularities of the surface, hence definitely altering the overall surface morphology up

to producing more regular and smoother finishing. After all, the good circumferential uniformity and machining

accuracy guaranteed by FB-AJM even on ductile aluminium alloy workpieces ensure the applicability of such a

technology to widespread set of industrial components.

KEY WORDS Fluidized bed, Abrasive jet machining, Internal finishing, Aluminium alloys

INTRODUCTION

This paper deals with an application of a relatively novel technology, Fluidized Bed Abrasive Jet Machining

(FB-AJM), to the finishing of the inner surface of tubular workpieces made from a high yield strenght aluminium

alloy (AA 6082 T6).

Machining and internal finishing of the inner surface of tubular parts can represent a critical and time-

consuming stage of manufacturing processes every time that workpieces made from high resistance materials or

characterized by high ratios of length L to diameter D (>20) must be machined [1]. Standard machining processes

plan mostly the employment of high speed turning as first machining stage followed by grinding with appropriately

shaped tools to allow the establishment of the desired surface quality [1]. Nonetheless, when higher resistance

materials or workpieces with very long and narrow inner surfaces must be processed, several problems can affect the

conventional solutions. Severe tool vibrations and large inflexions or tool failures can compromise both machining

and finishing stage of the manufacturing process [1], hence increasing the demand for alternative advanced and non-

conventional technological solutions [2].

Several advanced and non-conventional machining solutions able to process tubular workpieces are reported in

the scientific literature, in which the mechanisms of material removal, the typologies of energy source, the cutting

tools and transfer media used in each process are detailed [3]. Accordingly, useful studies, in which a classification

of the most suitable solutions according to workpieces features is operated, can be easily found in the specialized

literature [4-6]. However, when high precision finishing has to be operated on high resistance materials or

workpieces with complex geometries, even most of the advanced polishing methods show all their limits. For

example, the techniques based on abrasive flows and jets like abrasive jet machining (AJM) [4], abrasive flow

machining (AFM) [5] and abrasive magnetic flow machining (AMFM) [6]) was found able to guarantee only

reasonable results. At the same time, among techniques based on abrasive flows and jets, AJM seems to be the most

competitive as it requires shorter start-up times as well as lower investment and running costs [1].

Several papers in which AJM is employed for finishing purposes can be found in the literature. Most part of

studies argued over the hydrodynamic characteristics of abrasive jets [7-14], hence ascertaining the influence of all

the operational variables as abrasive type, size and concentration [7-9], impact speed [10] and angle [11] on process

effectiveness. Further papers faced new problems concerning the carrier gas typologies, nozzles shape, size and

wear, jet velocity and pressure, stand-off-distance (SOD) or nozzle-tip-distance (NTD) [12-14]. This way, they

expressed the overall process performance in terms of material removal rate, geometrical tolerances and surface

finishing of workpieces as well as in terms of nozzle wear rate. Finally, an important part of papers focused their

attention on leading process mechanisms in machining of both ductile and brittle materials [15-18] and on the

development of systematic experimental-statistical approaches [19-20] and artificial neural networks [21-22] to

predict the relationship between the settings of operational variables and the machining rate and accuracy in surface

finishing.

Examining the literature studies, it seems clear that despite the lowest investment and running costs, the lack of

troubles connected to the employment of solid tools (vibrations, inflexions or failures), the low processing time and

good operational flexibility of AJM, several drawbacks due to the hydrodynamic characteristics of abrasive jet make

such process unsuitable for most of tubular shaped parts. Using AJM, precision machining can be virtually

impossible [23-24] and operative troubles like the reach of chocking regimes, significant nozzle wear, dispersed

fines in the atmosphere as well as continuous demand in abrasive replacement can arise [25-26].

As alternative to AJM, a relatively novel hybrid technology, namely FB-AJM, was proposed and tested for the

first time, by Barletta et al. on tubular workpieces (L/D≤20) made from high strength stainless steel [27]. In this

technology, the peculiar fluidized bed hydrodynamic is used to improve the abrasive feeding system, the uniformity

of abrasive distribution through the workpiece and the recovery and self-regeneration of abrasive during machining.

This way, it was stated that precision and uniform machining as well as accurate surface finishing could be achieved

even on inner surfaces of tubular parts [27]. Nevertheless, several concerns regarding the possibility to extend such

technology to different and widespread metal alloys as well as to machine very long and narrow tubular parts

(L/D>20) were left still unresolved.

In this context, this paper tries to demonstrate the applicability of FB-AJM to internal finishing of high strenght

aluminium alloy (AA 6082 T6) using workpieces with ratio L/D higher than 20. The development of a ‘built ad hoc’

experimental set-up, the execution of tests on various workpieces, the interpretation of the evolution of material

removal and surface roughness parameters over process variables and the establishment of reliable correlations

between process settings and machining effectiveness are also detailed. For such purpose, a Taguchi’s reduced

experimental plan was executed to evaluate the influence of the main operational variables, i.e. abrasive mesh size,

machining cycle and jet speed, on mass removal and surface roughness. As a result of the preliminary experimental

plan, first consistent trends of material removal and average roughness according to operational variables were

found. A standard test condition was also identified and, subsequently, applied to study the evolution of roughness

parameters of machined surfaces according to mesh size of abrasive media, aimed at achieving best finishing

operative conditions. Moreover, the leading process mechanisms, which determine the material removal and,

concurrently, the achievement of a better surface finishing were interpreted in the light of the evolution of surface

morphology studied by using a combined SEM – 3D profilometer analysis. Finally, the circumferential uniformity

and precision of FB-AJM of the workpiece inner surfaces were tested in different locations, thereby definitely

assessing the reliability and reproducibility of such technology.

METHODS

Fluidized beds assisted abrasive jet machining (FB-AJM) system.

A sketch of the experimental apparatus employed to perform FB-AJM on the aluminium alloy tubes is shown in

Figure 1, with its main characteristics specified in Table 1. A full depiction of the experimental apparatus is reported

in a previous related paper [27]. Here, it is worth remarking that the apparatus consists of a compressor, two

fluidized beds and related systems of three way valves, nozzle and Venturi’s pipes, a couple of abrasive tanks and

more valves, tubes and measuring and controlling instruments. Besides, the employment of a specific control system

allows to carry out all the operative settings and regulate air and abrasive fluxes.

Supplying compressed air at moderate pressure (<1 bar) as working fluid to one of the two fluidized beds, the

abrasive, preventively fed to the fluidized bed from the abrasive tank, is suspended (i.e. taken in a fluid-like state)

under the action exerted by the aerodynamic push of air. Simultaneously supplying further compressed air at high

pressure (up to 16 bar) to the three-way valve located ahead of Venturi’s pipe related to the chosen fluidized bed, the

suspended abrasive is quickly sucked up the fluidized bed and jetted to the workpiece through the designed ad hoc

nozzle. Afterwards, the abrasive, once got through the workpiece, is restored into the second fluidized bed through a

dust collector system (cyclone) located on top of it. The collected abrasive is then fluidized by an additional

supplying of working fluid at moderate pressure into the second fluidized bed. Flowing back of abrasives and

stronger disturbances of the abrasive jet to the hydrodynamic of the second fluidized bed are averted by the

employment of star valves at the basis of the cyclone. The continuity of the operation is guaranteed by the

connection between the two fluidized beds, with the level of abrasives inside them kept constant according to the

principle of communicating vessels. Finally, it stands to reason that the roles of the two fluidized beds and annexes

can be automatically reversed as shown in Figure 2. Accordingly, the abrasive jet would departed oppositely, being

supplied into the workpiece from the opposite side and with the outgoing abrasive collected by the first fluidized

bed.

Two air filters on top of the fluidized beds prevent the system from massive elutriations of abrasive, although

they allow the evacuation of the dispersed fines (produced by the machining) and fluidization air in excess. The flux

of air and fines are then treated by using ‘absolute’ air filters. Besides, to ensure a constant amount and quality of

abrasives during machining, fresh abrasives are continuously loaded from the hoppers to the fluidized beds. Finally,

the draining of dispersed fines and the reloading of fresh abrasives guarantee a sort of self-regeneration to FB-AJM

machining capabilities.

Basic machining principle.

When the abrasive is jetted into the workpieces, it machines the metal by impinging it at high speed and impact

angle. In particular, the abrasive jet preferentially acts on the peaks and asperities sticking out from metal surface by

ploughing or, even, cutting off them. As a consequence, a progressively smoothing of workpiece surface associated

with remarkable material removal can be expected.

In FB-AJM, the effectiveness of operation must be ascribed to the peculiar characteristics of the abrasive jet. In

specific, the assistance of the fluidized beds, differently from what happens in standard AJM, allows the

establishment of a sort of fast transport regime inside the workpiece during machine [27]. In this way, the radial

distribution of the abrasive jet across the section of the workpiece is guaranteed to be the best and all flux deviations

and anomalies as well as peculiar abrasive patterns inside the workpiece are minimized. Beyond that, the uniformity,

precision and accuracy of FB-AJM can also be ascribed to the above mentioned reversibility of abrasive direction,

which allows to homogeneously distribute the machining force of abrasive jet all over the metal surface without

interrupting finishing operations.

Material and experimental procedure.

From an aluminium alloy (AA 6082 T6) rod six meter long and 20 mm in diameter, a set of workpieces 350 mm

long was cut. Afterwards, each workpiece was machined by internal longitudinal high speed turning, with inner

diameter being opened at 10 mm.

Angular red-brown alumina (Al2O3) mesh size in the range of 24 to 220, supplied by Smyris Abrasivi Srl as

finely divided powders, was employed as abrasive. Hardness of 2300 HV and factor shape around 0.67 were

declared by abrasive manufacturers. Abrasive jet velocities in the range of 5 to 50 m s-1 and machining cycles in the

range of 1 to 8 were investigated. Each machining cycle consisted of the transit of 5 kg of abrasives into each side of

the workpiece (namely side ‘d’ and ‘r’, Figure 3), respectively. According to the jet velocity and abrasive size, each

machining cycle could last from 3 to 10 min. The amount of abrasives passing through the workpieces was checked

by measuring the rotation speed of the star valve at the bottom of each cyclone.

Table 2 displays the reduced Taguchi’s design, which was developed to examine the influence of operational

parameters (i.e. machining speed, abrasive mesh size and machining cycle) on the experimental response ‘surface

roughness’ and ‘material removal’. A L18 mixed 3-6 level design with the experimental factors, abrasive mesh size,

abrasive jet speed and number of machining cycle, was scheduled. Each test was replicated four times to ensure the

reliability and reproduceability of the experimental results. Once defined the best settings of operational parameters

(four machining cycles and 10 m s-1 as jet velocity), they were employed in the second set of experiments, where

FB-AJM was reiterated on workpieces using, step by step, smaller abrasive size as reported in Table 3.

A strict experimental protocol was followed in order to ensure the reliability of the material removal and surface

roughness measurements. Material removal was estimated using a Sartorious Model BP 211-D as weighing

instrument with a resolution of 0.01 mg. To minimize any disturbances during tests, workpieces were enclosed in a

box to prevent air currents. Two special stainless steel slabs were kept separately and used as standard weights to

calibrate the weighing instrument each time it was used. These special efforts allowed the scales to perform the

measurements within the required accuracy and with good reproduceability.

The weighing procedure was as follows. The workpiece initial weight was taken with particular caution as to

surface polishing. Then, in order to obtain a comparison between the workpiece weight before and after the

machining, very fine tissue (Kimwipes) was used to clean the surface after the test was performed. After that, the

workpieces were put into a moderate temperature (60 °C) dryer. Following this, a further passage with the tissue was

performed and weighing was carried out as soon as possible thereafter to avoid the influence of dusts or oxidations.

The measurements were designed to last long enough to produce the stabilization of the results shown on the display

of the electronic scales. Each workpiece was measured several times and, if the difference between the two

following measurements failed to agree within 0.5 mg, the measurements were repeated until agreement within this

range was attained in successive determinations.

To measure the evolution of surface morphology during machining, surface roughness measurements were

performed by using a standard profilometer (Taylor Hobson model CLI 2000) based on an inductive probe.

TalyMap software release 3.1 was used to process the experimental data and extrapolate roughness parameters.

During roughness measurements, four areas of 7x2 mm2 were scanned along the four directions displayed in Figure

3. A total amount of 200 profiles, each one with a resolution of 1000 points per mm, was acquired to depict the

surface morphology at various step of machining.

Finally, a field emission scanning electron microscope (Leo model Supra 35) was used to take picture and

detect the evolution of surface conditions during machining at various levels of magnification.

RESULTS AND DISCUSSION

The analysis of process parameters: a statistical approach

Figure 4 reports the main effect plots (MEPs) of material removal and average roughness according to all the

operational variables, in which the raw response data (i.e. the means of the response variable for each level of a

factor) are reported. As can be seen, the main effect on the mean responses (i.e. the difference between the means of

the response variable between two different levels of a factor, namely, δ-value), material removal and average

roughness, is quite larger for the operational variable ‘mesh size’. In particular, the greater the increase in mean

diameter of abrasives distribution, the greater the increase in material removal and the worse the expected values of

average roughness. Accordingly, remarkable main effects on response ‘material removal’ are reported for both

‘machining cycle’ and ‘jet speed’. Instead, much smaller main effects on response ‘average roughness’ are detected

for both ‘machining cycle’ and ‘jet speed’. In particular, an increase in machining cycle and in jet speed produces

faster wear of the machined substrates, with the expected values of average roughness keeping mostly unchanged.

The analysis of variance (ANOVA) confirms the results of the analysis of mean (ANOM) reported in Figure 4.

Figures 5 and 6 display contribution percentages, degrees of freedom and Fisher’s values for each operational

variable investigated for responses ‘material removal’ and ‘average roughness’, respectively. In the former case

(Figure 5), all the experimental factors were found to be significant, with their Fisher’s values quite larger than their

correspondent Fisher’s values tabulated. Furthermore, percentage of contribution for each experimental factor was

found to be rather high, with values ranging from 20 to 40 %. Instead, the percentage of contribution of factor ‘error’

is quite low (close to 10%), hence stating the good reliability of the experimental procedure. In the latter case

(Figure 6), all the Fisher’s values were found to be significantly higher than correspondent Fisher’s values tabulated.

Nevertheless, examining the percentage of contribution, only the experimental factor ‘mesh size’ was found to

significantly influence the response ‘average roughness’ (more than 90 % of contribution) in according with

indications provided by ANOM. Instead, the other two experimental factors, jet speed and machining cycle, even if

p-values close to 0 were calculated, can be definitely considered as factors with a minor influence on response

‘average roughness’, as they have low contribution percentages. Finally, the percentage of contribution of factor

‘error’ is very low, hence confirming, once more, the reliability of the experimental procedure.

MEPs and ANOVA provide only partial suggestions in setting the operational parameters in FB-AJM. From the

above analyses, a decrease in mesh size of abrasives is required when more accurate surface finishing are desired,

with concurrent lower wear phenomena. Nevertheless, statistical instruments are not enough accurate in suggesting

the best choice for machining cycle and jet speed. In both cases, trends reported in MEPs state that larger values of

machining cycle and jet speed produce faster wear phenomena, without sensibly influencing the accuracy of surface

finishing. However, slight improvements in average roughness can be detected from Figure 4 for progressively

higher values of machining cycle and, above all, for lower values of jet speed.

As support to the indications provided by MEPs and ANOVA, the surface morphology of a sample treated in

three different zones with the same abrasive, but with a different number of machining cycles and with different

abrasive jet speeds was studied and the measured morphologies reported in Figure 7. The raw 3D maps highlight the

better overall behaviour of the surface treated with enough large number of machining cycle (4) and using low jet

speed (10 m/s).

A useful interpretation of the 3d maps can be provided. By comparing panel a, b, c and d in Figure 7, it can be

noticed that surface finishing definitely improves until two machining cycles elapse (panel b), whereupon it remains

almost constant. Furthermore, the improvement in surface finishing goes with a remarkable change in overall

morphological aspect of the machined surface. If two more machining cycles are performed (panel c), the overall

surface finishing does not improve significantly, but the degree of morphological uniformity still increases. If all the

eight machining cycles are completed (panel d), neither significant variation in surface finishing and in uniformity of

morphological aspect is achieved. These considerations explain that finishing mechanism varies from initial

selective removal of surface asperities (peaks) to a subsequent more uniform removal of material all over the

machined surface with further slight decrease of asperities and, after that, to a sort of stabilization of surface

roughness and morphology. Accordingly, material removal can be expected very fast during the first machining

cycles, hence becoming gradually slower, when the surface roughness is reduced, and approaching to a steady value,

when the stabilization of machined surface is reached.

These results are in good agreement with the data available in the literature. In FB-AJM of stainless steel tubes,

a fast improvement of average roughness of the machined surface with a concurrent increase in material removal

was attained in short order [27]. Afterwards, a stabilization of finishing process was observed, with progressively

slower improvements in surface finishing and the establishment of a standard amount of material removal per unit of

time and surface roughness [27]. In both magnetic abrasive jet machining [24] and standard abrasive jet machining

[19], surface roughness approached quickly to asymptotic values, strongly related to the settings of operational

variables. Therefore, in the first part of machining the material removal is very quick, with this value sensibly

diminishing, when asymptotic conditions were approached [19, 24]. In both abrasive [28] and magnetic [29] flow

machining, average roughness also approached to asymptotic values. Nonetheless, in flow machining processes

material removal tended to keep constant during all the finishing operations [28-29].

Such difference in material removal trends between flow and jet machining processes can be probably ascribed

to the dissimilar finishing mechanisms involved. In particular, in FB-AJM, material removal mostly comes from the

surface asperities and irregularities (panel a), hence deeply changing the morphology of machined surface (panel b).

Therefore, when the stabilization of surface morphology is achieved (panel c), the mass loss from the surface is

strongly slowed down, hence approaching to a steady value (panel d). Consequently, FB-AJM can be mentioned as

no pressure copying process. On the other hand, in flow machining, material removal occurs from both the peaks

and valleys of machined surface from the beginning, hence leaving substantially unchanged the overall aspect of

morphology of machined surface and guaranteeing uniform mass loss during all the process [28-29]. In fact,

differently from FB-AJM, flow machining processes is mentioned as pressure-copying process [27-29].

Watching panel a, c and e in Figure 7, further considerations upon the influence of jet velocity on surface

finishing can be done. More specifically, by comparing the evolution of surface morphology varying jet speed, while

keeping constant both abrasive mesh size and number of machining cycles, remarkable differences in achievable

surface finishing and morphology can be noticed. If low jet speed values are employed (close to 10 m/s), the best

finishing, with minor morphological irregularities affecting the machined surface, can be reached. In fact, in such

operative conditions, fast transport regime of abrasive through the workpiece being finished went on, determining

the activation of material removal mostly from asperities and irregularities of machined surface and, concurrently,

the progressive establishment of a smoother surface (panel c). To the contrary, if higher jet velocities were set, the

positive influence of fluidized beds and of nozzle design on the hydrodynamic of the fluid-abrasive mix went off,

with abrasive likely exhibiting irregular jumbling into the workpieces. With jet velocity set very close to the highest

value (close to 40 m s-1), the degree of irregular jumbling was expected to increase. As a result, the accuracy of

finishing operations decreased because of material over-removal caused by the aggressive strikes of the jumbling

abrasives, which worsened the machined surface, reducing at a large extent the morphological uniformity (panel e).

Even these results are in good agreement with the data available in the literature [24, 28-31]. An optimized

value of 13 m s-1 for jet velocity was analytically found by Kim et al. [24] in the attempt of modelling a magnetic

abrasive jet machining system and a similar value was found to be the best in FB-AJM of stainless steel tubular parts

by Barletta et al. [27]. Moreover, a self-limiting condition imposed by abrasive jet speed on accuracy of theoretical

finishing achievable was already stressed by one of authors in a previous study related to fluidized bed machining of

complex shaped aluminum components [32] as well as in further studies related to abrasive jet machining processes

[23, 25-26].

In the light of results coming from statistical approach and from previous morphological analysis, a standard

machining condition can be deducted for FB-AJM of AA 6082 T6 tubular parts. Four machining cycles and jet

velocity set at 10 m s-1 should guarantee the best performance and accuracy of finishing process, minimizing, at the

same time, the connected troubles as well as the operative costs.

The analysis of surface morphology.

Figure 8 displays the trend of roughness parameters vs. mesh size by using standard machining conditions, that

is, 4 machining cycles and 10 m/s as abrasive jet speed. Panels a and b in Figure 8 report the trends of amplitude

parameters and Figure 9 reports the evolution of roughness profile according to abrasive size. As can be seen, using

abrasives with smaller mesh size, a smoother and uniform surface finishing was progressively achieved. In specific,

the first machining step performed with abrasive mesh size 24 reset the starting topography, with the initial cutting

marks produced by high speed turning quickly disappearing. The next steps with abrasives of smaller size

progressively reduced the surface asperities and irregularities of machined surface up to obtain an almost flat

finishing. As a result, average roughness Ra stepped from 3-5 μm to 0.6-0.7 μm and peak to valley distance Rz

stepped from 20-40 μm to 5-6 μm, with very small deviations around the mean values. Consequently, ‘improvement

ratio’ (i.e. ratio between starting and final roughness) in the range of 4 to 7 could be claimed.

These results are in good agreement with data available in the scientific literature and, particularly, with

roughness trends achieved by Barletta et al. [27] using FB-AJM on tubular workpieces made from hardened

stainless steel and by Kim et al. [24] using MAJM on SUS 304 circular tubes, where progressive improvements of

surface finishing lowering abrasive mesh size were observed. Nevertheless, in those cases, better average roughness

values (as low as 0.015 in FB-AJM and 0.2 in MAJM ) could be achieved, being stainless steel substrates much

more suitable to abrasive jet finishing than the softer aluminium substrates. On the other hand, if a comparison is

performed with results achieved by Jain et al. [29] on aluminium substrates using AFM, the better overall

performance of FB-AJM can be highlighted. In fact, AFM allowed achieving average roughness very close to 1 μm,

starting from an average roughness around 2 μm with an ‘improvement ratio’ of just 2.

Panel c in Figure 8 displays the trend of spacing Rsm vs abrasive mesh size. A continuous decreasing in spacing

was observed going toward the employment of abrasive with smaller mesh size. In fact, the relatively longer

wavelength components of the roughness profile induced onto the surface by high speed turning (Figure 9, untreated

roughness profile) shortened during FB-AJM. The transformation of surface morphology can be ascribed to the

features of FB-AJM process. In fact, as mentioned in previous section and as known from the literature [23-27], in

jet machining the abrasives are lined along the lines of the jet fluid, making sort of abrasive chains. The abrasive

chains machine the surface, removing the material and performing the finishing operation. As the abrasive chain

moves through the workpiece at a very high velocity, the chain has no time to make its shape suitable to the shape of

surface being machined. Accordingly, the chain of abrasives behaves like a solid tool, removing and displacing

material mostly from the peaks of the uneven surface, flattening it and generating roughness profile with shorter

wavelength components (Figure 9, from 24 mesh size to 220 mesh size roughness profiles).

From the evolution of surface roughness profiles in Figure 9, it can be seen that the cutting edges of the abrasive

tend to machine the peaks of the surface and to reduce all the other irregularities on machined surface. Moreover, the

lack of built up of further micro-scratches on the slopes of the starting profile witnesses that the material is removed

mostly from the peaks, leaving untouched the valleys. Consequently, the final surface consists of an accumulation of

shorter wavelength micro-scratches superimposed on a surface made definitely flat by FB-AJM, as 3D maps

reported in Figure 10 display and trend in slope Rλq in panel d in Figure 8 confirms. Smaller the abrasive size, which

was employed to finish the surface of the workpiece, smaller the dimension of the micro-scratches occurring (Figure

10 from panel b to panel g), thereby determining an improvement of overall quality of surface finishing. This result

is confirmed also by the trend of skewness Rsk and kurtosis Rku, which assumed progressively more uniform values

when smaller abrasive sizes were employed (Figure 8 panel e and f), hence witnessing the approach to a symmetric

and normal profile.

In accordance on what has just been mentioned and with experimental observations reported in the previous

section (Figure 7), FB-AJM cannot be classified as pressure-copying processes. Pressure-copying processes embrace

operations, which machine under constant pressure the small irregularities from longer wavelength components of

machined surface, while not significantly altering the longer wavelength features. In FB-AJM, the longer

wavelength features of surface morphology are the first to be machined. Therefore, FB-AJM belongs to process able

to selectively cut or plough the sharper asperities of the machined surface, hence leaving substantially unaffected the

valleys.

Figure 11 shows the evolution of surface morphology according to abrasive sizes by using SEM images at 500X

as magnification. The abrasive cutting marks produced by FB-AJM change the starting surface texture produced by

high speed turning (panel a). Accordingly, the dominant finishing direction of abrasive jet can be easily seen on the

surface, with its main morphological features being strictly linked to the abrasive size (from panel b to panel g). In

fact, employing abrasive media with smaller size, the surface is finished with higher quality. Concurrently, the

peculiar machining marks of FB-AJM become smaller but more uniformly distributed all over the machined surface.

As mentioned, the machining marks are a result of the abrasive displacement in the form of chains along the

workpiece surface due to the machining fluid. After the employment of progressively smaller abrasive media, the

larger irregularities of the surface go almost flat and the surface texture made by high-speed turning is definitely

replaced by the peculiar micro-scratched texture of FB-AJM. This result is in good agreement with experimental

findings in other abrasive jet and flow machining systems, where a peculiar surface texture is always produced after

finishing or machining operations [23-24, 29].

Figure 12 shows the SEM photographs of machined surface (standard condition, 120 as mesh size) at

progressively higher magnification levels. As previously seen, before FB-AJM, the surface presented uneven cutting

marks caused by the aggressive material removal due to the previous high-speed turning process. After finishing, the

surface resulted to be definitely smoothed, with the effect of abrasive displacement due to abrasive jet being visible

in the cutting marks parallel to the machining direction (panel a and panel b in Figure 12).

The effect of material displacement and microcutting action of abrasive edges is clearly visible in panel c and d

in Figure 12, where it is even possible to note the small grooves created by the abrasive impact and the consequent

built up of adjacent crests (panel c). At the same time, the microcutting action with the residual microchips standing

still on the surface can be also observed (panel d).

The embedding of abrasive splinters can be easily seen in panel e and f in Figure 12, where larger splinters of

abrasive remained trapped and partially stuck out from the metal surface are noticeable. However, embedding

phenomena of alumina splinters all over the machined surface can observed from the first moment of FB-AJM.

During the machining, a rapid saturation of surface with abrasive splinters occurred. Nevertheless, employing

progressively smaller abrasive size, the average size of splinters embedded in the surface decreased and,

consequently, a better aesthetic result could be attained. Similar results were found in previous studies about

fluidized bed machining, in which massive embedding phenomena of harder abrasive splinters into the softer

substrates (alumina splinters in aluminium and brass substrates [32, 33], copper splinters in polyamide 6 and 66

substrates [34]) are recurrent phenomena.

Figure 13 shows the uniformity of the roughness profile all over the machined surface. By observing the trends

of average roughness, it can be seen that the surface is uniformly smoothed all around the internal circumferences,

with no significant differences in average roughness rising among all the four investigated locations (Figure 3). This

is in a deep disagreement with experimental findings reported in the literature for the other abrasive jet machining

systems [23-26], where slight deviations from ideal running of the systems determined relevant unevenness of the

surface being machined. The difference between FB-AJM and the other systems can be probably attributed to the

claimed uniform distribution of abrasives across the section of the workpiece in FB-AJM which, as said before,

avoiding any flux disturbance, any peculiar pattern of abrasives into the workpiece and any anomaly of jet fluid,

guarantees the finishing uniformity all around the internal circumference of the machined tubes.

CONCLUSION

Experiments were carried out on long and narrow circular tubes made from AA 6082 T6 as workpieces, and some

results were obtained, as follows.

From the preliminary Taguchi’s experimental plan, it can be deducted: (i) the machining capability of the

developed FB-AJM system is verified by the consistent material removal and average roughness trends (MEPs)

according to leading operational variables; (ii) a standard machining condition (jet velocity of 10 m s-1 and 4

machining cycles) for an effective machining of inner surface of workpieces can be deducted by combined

experimental (evolution of raw 3d morphology) and statistical considerations (ANOM and ANOVA).

From the subsequent experimental tests performed employing progressively lower abrasive mesh size, it can be

deducted: (i) a relevant improvement in surface roughness can be achieved and a fast approach to a sort of

asymptotic condition for all the roughness parameters can be claimed; (ii) an improvement of average roughness

from 3-5 mm to values as small as 0.6-0.7 mm can be quickly attained; (iii) the establishment of a progressively

more regular surface morphology is stated by trends of hybrid roughness parameters, with their values approaching

to more favourable and asymptotic conditions.

Such results state that fluidized bed hydrodynamic improves the characteristics of the jet fluid and,

consequently, of the abrasive configuration and distribution through the workpiece, leading to better overall

finishing results.

An interpretation of material removal mechanism is gained by examining the changes in surface texture at a

microscopic level using a combined analysis of 3d surface profilometry and scanning electron microscope. In

particular, it can be deducted: (i) the action of abrasive cutting edges against the surface is observed to both cut

asperities into the surface and to displace material; (ii) a remarkable embedding phenomena of the harder abrasive

splinters into the softer aluminium matrix can be underlined; (iii) in agreement with the trend of spacing Rsm, a

smoothing of the longer wavelength components of profiles can be claimed when smaller and smaller abrasive size

are used, with the resulting surface being an accumulation of abrasive cutting marks (smaller wavelength

components) superimposed on the flattened morphology.

Accordingly, in FB-AJM, the abrasive jet, being not so flexible to adapt its hydrodynamic shape to follow the

irregularities of surface being machined, removes material essentially from the sharper peaks and asperities, hence

changing the morphology and stating, once more, that FB-AJM does not belong to the category of pressure-copying

finishing processes.

Finally, because of the consistent distribution of abrasive across the section of the workpiece during FB-AJM,

good uniformity all around the circular shape of the workpiece can be attained.

REFERENCES

[1] M.C. Shaw, Principles of Abrasive Processing, Oxford Series on Advanced Manufacturing, Oxford University Press, 1996

[2] J.R. Davis, ASM Metal Handbook, Machining, American Society for Metals, ASM International, 9th edition, 1989

[3] N. K. Jain, V. K. Jain, Modelling of material removal in mechanical type advanced machining processes: a state-of-art review, Int. J. Mach. Tools Manufacture 41, 2001, 1573–1635

[4] P.M. Khodke, D.J. Tidke, Abrasive jet machining: a state-of-art review, Journal of Institution of Engineer 77, 1996, 1–8.

[5] R.K. Jain, V.K. Jain, P.M. Dixit, Modelling of material removal and surface roughness in abrasive flow machining process, Int. J. Mach. Tools Manufacture 39, 1999, 1903–1923.

[6] G.Z. Kremen, E.A. Elsayed, V.I. Rafalovich, Mechanism of material removal in the magnetic abrasive process

and the accuracy of machining, Int. J. Prod. Res. 34, 1996, 2629–2638. [7] B.R. Lawn, A model for brittle solids under fixed abrasive conditions, Wear 33, 1975, 369–372. [8] M. Hutching, A model for the erosion of metals by solid particles at normal incidence, Wear 70, 1981, 269–

281. [9] D.B. Marshall, A.G. Evans, M.E. Gulden, J.L. Roubort, R.O. Scatergood, Particle size distribution effects on

the solid particle erosion of brittle materials, Wear 71, 1981, 363–373. [10] J. Wolak, Parameters affecting the velocity of particle in an abrasive jet, Trans. ASME, J. Eng. Mater.

Tech. 99, 1977, 147 [11] G. Sundararajan, An empirical relation for the volume of crater formed during high velocity oblique impact

tests, Wear, 97, 1984, 9–16. [12] P.K. Sarkar, P.C. Pandey, Some investigations on abrasive jet machining, J. Inst. Eng. (India) 56, 1976,

284–287. [13] M. Hashish, M.P. DuPlessis, Theoretical and experimental investigation of continuous jet penetration of

solids, Trans. ASME, J. Engng. Ind. 100, 1978, 88–94. [14] M. Hashish, M.P. Du Plessis, Prediction equations relating high velocity jet cutting performance to stand-

off distance and multipasses, Trans. ASME, J. Engng. Ind. 101, 1979, 311–318. [15] I. Finnie, Erosion of surfaces by solid particles, Wear 3, 1960 87–103. [16] J.G.A. Bitter, A study of erosion phenomena: I and II, Wear 6, 1963, 5–21. [17] G.P. Tilly, Two stage erosion mechanism of ductile erosion, Wear 23, 1973, 87. [18] B.R. Lawn, A model for brittle solids under fixed abrasive conditions, Wear 33, 1975, 369–372. [19] V.C. Venkatesh, T.N. Goh, K.H. Wong, M.J. Lim, An empirical study of parameters in abrasive jet

machining, Int. J. Mach. Tools Manuf. 29, 1989, 471. [20] A.P. Verma, G.K. Lal, A theoretical study of erosion phenomenon in abrasive jet machining, Trans.

ASME, J. Manuf. Sci. Eng. 118, 1996, 564. [21] S.S. Rangwala, D.A. Dornfeld, Learning and optimization of machining operations using computing

abilities of neural networks, IEEE transactions on Systems, Man and Cybernetics 2, 1989, 299-314. [22] G. Chryssolouris, M. Guillot, Modelling of machining process using neural network, Trans. ASME, J. Eng.

Ind. 112, 1990, 122-131. [23] N. Ramachandran, N. Ramakrishnan, Abrasive jet machining upcoming technology in metal processing: a

review, J. Mater. Process. Technol. 39, 1993, 21–30. [24] J-D. Kim, Y-H Kang, Y-H Bae, S-W Lee, Development of a magnetic abrasive jet machining system for

precision internal polishing of circular tubes, Journal of Materials Processing Technology 71, 1997, 384-393 [25] R. Balasubramaniam, J. Krishnan, N. Ramakrishnan, Investigation of AJM for deburring, Journal of

Materials Processing Technology 79, 1998, 52-58 [26] R. Balasubramaniam, J. Krishnan, N. Ramakrishnan, A study on the shape of the surface generated by

abrasive jet machining, Journal of Materials Processing Technology 121, 2002, 102-106 [27] M. Barletta, V. Tagliaferri, Development of an abrasive jet machining system assisted by two fluidized

beds for internal polishing of circular tubes, Int. J. Mach. Tools Manufacture, approved May 2005, available on line on www.scienceserver.com

[28] H. Yamaguchi, T. Shinmura, Study of the surface modification resulting from an internal magnetic abrasive finishing process, Wear 225-229, 1999, 246-255.

[29] V.K. Jain, S.G. Adsul, Experimental investigation into abrasive flow machining, International Journal of Machine Tools & Manufacture 40, 2000, 1003-1021.

[30] Zhu J., Grace J.R. and Lim C.J., Tube wear in gas fluidized bed – I. Experimental findings, Chemical Engineering Science, 45, 1990, 1003-1015.

[31] Zhu J., Grace J.R. and Lim C.J., Tube wear in gas fluidized bed – II. Low velocity impact erosion and semi-empirical model for bubbling and slugging fluidized beds, Chemical Engineering Science 46, 1991, 1151-1156.

[32] M. Barletta, A new technology in surface finishing: fluidized bed machining (FBM) of aluminium alloys, Journal of Materials Processing Technology, approved September 2005, available on line on

www.scienceserver.com. [33] M. Barletta, G. Costanza, R. Polini, Al2O3 thin coating of AA 6082 T6 components using a fast regime

fluidized bed, Thin Solid Films, approved November 2005, available on line on www.scienceserver.com. [34] M. Barletta, A. Gisario, V. Tagliaferri, Electrostatic spray deposition (ESD) of polymeric powders on

thermoplastic (PA66) substrate, Surface and Coatings Technology, approved November 2005, available on line on www.scienceserver.com.

List of Figures

Figure 1. Fluidized bed assisted abrasive jet machining (FB-AJM) system

Figure 2. The description of the reversible machining system. Panel (a): Direct machining; Panel (b): Reverse

machining.

Figure 3. A sketch of the workpiece: the two abrasive jet directions, namely, ‘d’ and ‘r’, and the different

circumferential locations (1, 2, 3 and 4) along which roughness measurements were performed

Figure 4. ANOM on L18 mixed level Taguchi’s experimental design.

Figure 5 Results of ANOVA on response ‘Material Removal’.

Figure 6 Results of ANOVA on response ‘Average Roughness’.

Figure 7 The influence of machining speed on surface flatness of machined substrates

Figure 8: Roughness parameters vs abrasive mesh size in standard machining condition

Figure 9: Roughness profile vs abrasive mesh size: analysis of 2d morphology

Figure 10: 3d maps vs abrasive mesh size: analysis of 3d morphology

Figure 11: SEM images of surface machined with different abrasive mesh sizes

Figure 12: SEM images at various magnification level: machining in standard condition with 120 as mesh size

Figure 13: Average roughness uniformity vs abrasive mesh size: analysis of four different zones

List of Tables Table 1: Specification of FB-AJM system

Nozzle F 10 mm Abrasive feed hose F 10 mm Discharge pipe F 10 mm Working fluid (Fluidization air + Jet Fluid) Air, max 8-10 atm, 130 m3/h Pressure gauge 0-16 atm Fluidized bed Diameter F 140 mm, height 1200 mm Packed bed height 180 mm (5 kg) Workpiece size Length 350 mm, Inner diameter 10 mm Workpiece material AA 6082 T6

Table 2: Taguchi’s reduced experimental plan

Experimental Factors Experimental Levels Abrasive Size (MS) Jet Velocity (m s-1) Machining Cycle

I 24 10 2 II 40 20 4 III 60 50 8 IV 80 - - V 120 - - VI 220 - -

Table 3: Second set of experiments

Experimental Factors Experimental Levels Abrasive Size (MS) Jet Velocity (m s-1) Machining Cycle

I 24 10 4 II 40 - - III 60 - - IV 80 - - V 120 - - VI 220 - -