NAMRC 35 Demonstrates Manufacturing's Vitality
The intellectual vigor of manufacturing is illustrated by research presented at the annual conference of SME's North American Manufacturing Research Institution
By Brian J. Hogan
The annual meeting of the North American Manufacturing Research Institution of the Society of Manufacturing Engineers, NAMRC 35, was held in Ann Arbor, MI, May 22–25 at the University of Michigan. An international forum on manufacturing research, NAMRC is described by SME as an annual conference of the international community of researchers contributing to the furthering of manufacturing technology.
At the NAMRC welcoming ceremony, keynote speaker David Taub, executive director, GM Research & Development, General Motors Corp. (Detroit), gave an address entitled "Fostering Innovation in American Industry," covering a range of manufacturing topics, including analysis of competitive pressures rising from globalization of the automotive industry.
"The automotive industry is global. In fact, just two years ago, GM created more vehicles outside the US than inside," notes Taub, who prior to joining GM in 2001 worked eight years at Ford Motor Co. (Dearborn, MI) after spending 15 years in R&D at General Electric Co. (Fairfield, CT).
What manufacturers do means more to economic development that most industries, he adds, noting that vehicle manufacturing accounts for 4% of the US GDP and represents 11% of shipped goods. "Simply put, manufacturing matters! We've never counted it, but Tom Stevens, the head of our Powertrain organization, says General Motors drills more holes, and grinds more centimeters than any other company in the world, and I suspect that that's a true statement," says Taub.
Automotive OEMs employ some 500,000 highly paid workers in the US, which creates an interesting dynamic, because the auto industry currently has about 20% over-capacity. "It affects profitability," Taub says. "In that environment, the most innovative companies are going to win."
Automotive manufacturers need to compress current product-development design cycles. "What's the next play? The next play is speed," Taub says. "In the hypercompetitive market we find ourselves in, a product is considered stale in 18 months. If you go back just 15-20 years ago, the product-development cycle was six years. We're now in a marketplace where you've got to be much faster. It's not just designing the product with these tools, we have to get to the point where we're designing our manufacturing facilities fully. It's not just art-to-part, but it's art-to-plant. Today's tools simply do not allow that.
"There are some fundamental innovations and research required for us to move to this vision. We're looking to be able to virtually launch factories. We are getting there with product; we're lagging there in plant. We also want to manage our plant through simulation and real-time simulation, marrying those technologies."
Seventy papers on a tremendous range of subjects were presented at NAMRC 35. The following excerpts illustrate the work discussed during the conference.
Researchers Elijah Kannatey-Asibu Jr. and Bhumika Lathia of the Department of Mechanical Engineering at the University of Michigan (Ann Arbor) discussed Laser Forming by Shock Peening. Speaking in general terms, laser forming eliminates the dies associated with traditional metal forming and thus significant forming costs. It's flexible and independent of tool inaccuracies that result from wear and deflection. Laser forming can provide precise deformation in confined or inaccessible locations by remote application, and there's a minimal heat-affected zone or material degradation, when compared to flame bending.
Traditional laser forming involves three principal mechanisms: temperature gradient, upsetting, and buckling. The temperature gradient mechanism evolves from processing conditions that generate very high temperature gradients in the material. Thermal expansion results in initial bending of the sheet away from the heat source. But the constraint of the cooler surrounding material restricts free expansion of the heated layer, resulting in compressive thermal stresses that induce plastic compressive strain. On cooling, the heated surface layer contracts more than the lower layers, and the sheet bends toward the heat source.
Buckling arises in relatively thin sheets, where the ratio of the diam of the heated area to the sheet thickness is of the order of 10. The temperature gradient in the thickness direction is relatively small, and the heated region tends to expand. This expansion is hindered by the surrounding material, and thermal compressive stresses develop in the sheet, which may lead to buckling when a critical stress value is reached. Upsetting occurs when uniform heating of a localized zone is achieved through the thickness of the sheet. Process parameters may be similar to those of buckling, except for the diameter of the heat source area, which is relatively small. A common feature of all three mechanisms is the generation of strains by thermal stresses.
Laser shock peening imparts compressive stresses on the surface of a material. It produces compressive residual stresses that extend deeper into the material than with traditional shot peening, and the stresses are produced with less cold work. It's normally done using a thermoprotective coating or absorbing layer of black paint or tape applied to the part's surface. A layer of dielectric that's transparent to the laser beam, usually water or glass, is placed atop the absorbing layer.
The laser beam instantly vaporizes the absorbing layer, producing plasma that expands rapidly and creates very high pressures on the workpiece, as a result of the recoil momentum of the ablated material. The pressure causes compressive stresses in the material. Pressures on the order of GPa are achieved.
In their experiments, the University of Michigan researchers used an outer Type I PVC frame for holding water. It measured 229 x 203 x 10 mm. The frame's base is covered with a 3.175-mm-thick 6061 aluminum plate to protect the PVC from the laser. An aluminum block with clamps is mounted on the base to hold the low-carbon-steel workpiece, which measures 76 x 76 x 1.6 mm. After the workpiece is spray-coated with black paint and dried, the frame is filled until the workpiece is covered by a 2-mm-deep layer of water.
A PRC Model 300/2 fast axial flow CO2 laser used in the experiments normally generates two continuous wave output beams in D-mode [combined TEM00 (60) and TEM01 (40) modes] with maximum power output of 1.2 kW for each beam. For the experiment, the laser was set in the pulsed mode, and generated pulses of up to 5 kHz frequency, with a 50% duty cycle. For each set of conditions, 10 or 20 scans were made over the same path before the bend angle was measured. Experiments compared traditional laser forming and laser peening; peening was also done with only a protective laser and no water, and scans offset 0.5 mm from each other.
In general, the bend angle is found to increase with incident power. With higher ablation pressures, residual stresses induced by the resulting shock waves will increase in magnitude. Experiments demonstrate that the bend angle obtained with laser shock peening was 5.95°, while that for traditional forming was 2.45°. In other words, forming by shock peening can increase the resulting bend angle by more than 100%. Increasing workpiece traverse velocity increases the angle, and increasing average power of the laser beam pulses increases the bend angle in straight-line sheetmetal bending.
A presentation by Mohammad Malekian and Simon S. Park of the Department of Mechanical and Manufacturing, Schulich School of Engineering, University of Calgary, Calgary, Alberta, was entitled Investigation of Micro Milling Forces for Aluminum. They point out that micro end milling offers a high rate of material removal while being able to create the same complex geometries achieved by macro end milling. Their paper examines a method of predicting the micromechanical milling forces for aluminum workpieces.
Micro milling can be translated into two cutting regimes: shearing-dominant cutting and plowing dominant. Various experiments were performed to verify micro milling forces for aluminum 7075. There are several critical issues associated with micro machining; they result mainly from the miniaturization of the components, tools, and processes. One challenging problem in micro machining is the measurement and prediction of cutting forces. Accurate prediction and measurement of cutting force indicate the state of a machining operation, and allow process engineers to select the optimal process parameters to achieve productivity goals and maintain accuracy.
The experimental micro CNC machining center platform was based on conventional column-and-knee-type machines. A 300-W electric motor spindle capable of speeds to 80,000 rpm was used. Three linear tables, equipped with cross-roller linear bearings and stepper motors, were used to actuate the X, Y and Z axes, where the position accuracy of the X, Y stage is ±1.3 µm. Control was achieved via an open-architecture control system. Micro tools used were tungsten carbide micro end mills with micrograins (i.e. 600 nm). Nominal diameter of the carbide end mill was 500 µm, but actual diam was approximately 400 µm.
Experimental results reveal that simple scaling of the macro cutting model cannot be used for micro machining. The shearing-dominant cutting region, where the chip loads were greater than the critical chip thickness, showed a linear relationship. The sharp-edge cutting force model was used to predict forces by identifying the cutting constants. Instantaneous peak forces for half-immersion down-milling tests were used to derive explicitly the equations for the constants.
It was observed that the edge coefficients for micro tools are much smaller than those of macro tools. These coefficients can be used to predict the micro milling forces at high feed rates. When cutting is done at low feed rates, namely below the critical chip thickness, cutting forces become nonlinear, and the conventional theory of cutting forces can no longer be used. An empirical powering force model is proposed by the researchers that considers material characteristics, edge radius, and critical chip thickness. Tests were performed to verify the proposed model.
The micro world is of increasing importance to manufacturing engineers. One technique for producing micro features and products is ultrasonic machining. In a paper entitled Experimental Study of Tool Wear in Micro Ultrasonic Machining, X. Hu of Sauer-Danfoss (US) Co. (Plymouth, MN), Z. Yu, School of Mechanical Engineering, Dalian University of Technology (Dalian, Liaoning, China), and K. P. Rajurkar, Center for Nontraditional Manufacturing Research, University of Nebraska-Lincoln, look at wear characteristics of a rotational cylindrical tool.
Micro ultrasonic machining (USM) is scaled down from conventional USM by reducing tool dimensions, abrasive particle size, and vibration amplitude. Either the tool or the workpiece is vibrated at an ultrasonic frequency (usually 20–40 kHz) and an amplitude of several microns. With the application of force between the tool and workpiece, material is hammered and removed from the workpiece by abrasive particles in the slurry. The hammering of abrasive particles on the tool also causes tool material loss, resulting in tool wear.
High tool wear rate is a major drawback of this process. Also, the machined feature's dimensional tolerance cannot be guaranteed when the reduction of tool length, or the alteration of its shape, is not known.
The researchers looked at a micro tool with a simple cylindrical shape, and used stainless steel, tungsten, and cemented carbide tools. In their experimental setup, a workpiece is attached to a vibrator driven by an ultrasonic generator. A micro tool, fixed to a mandrel rested on a V-shaped block, is rotated by a DC motor with adjustable rotational speed. Contact force between the tool and workpiece (via abrasive particles) is regulated by implementing a closedloop force-control strategy with force-signal feedback from an electronic balance. Variation of the contact force is determined within 10% by the force control. Polycrystalline diamond micro powder is used as the abrasive material, and vibration amplitude is adjusted by varying the input voltage.
During the experiments, tool flanging (extended material built around the edge of a tool bottom) and rounding were repeatedly observed. Flanging was only seen for AISI 316L and tungsten tools. Data indicate that abrasive slurry plays a crucial role in flanging for these materials. Flanging is strongly related to the speed of tool engagement. When low machining speed (0.35 µm/sec) is used, the tips of tools made from 316L and tungsten are easily altered into a flanged shape. No flanging was observed for cemented carbide tools, because the cemented carbide strongly resists deformation or displacement of material.
After a tool tip goes fully into the workpiece, a rounded tool bottom is the most commonly observed tool shape. The tool is first chamfered, then rounded, and the tool tip becomes sharpened over time. Rounding is inevitable, but the speed of rounding is affected by the wear resistance of tool material (when process parameters are constant). Cemented carbide resists rounding best, 316L is least resistant.
Larger static load, vibration amplitude, and particle size result in higher longitudinal tool wear rate. Overall, when compared with the three preceding parameters, the influence of tool rotational speed and workpiece material on tool wear rate is insignificant.
Cutting tool wear implies a gradual loss of tool material at workpiece/tool-contact zones, and impacts the dimensions, surface finish, and surface integrity of the workpiece. Researchers Antonio J. Vallejo, Department of Engineering, Ruben Morales-Menendez and Luis E. Garza-Castanon, Center for Innovation in Design and Technology, Technologico de Monterrey Campus Monterrey (Mexico), and J. R. Alique, Computer Science Department, Instituto de Automatica Industrial (Madrid, Spain), describe an indirect means of monitoring tool condition in their paper: Pattern Recognition Approaches for Diagnosis of Cutting Tool Wear Condition. The paper proposes use of new approaches with some intelligent features—pattern recognition, learning, knowledge acquisition, and inference from incomplete information.
Their experiments were performed on a three-axis machining center; the cutting tool studied was a 63-mmdiam Komet face mill F511 with five TiCN/TiN-PVDcoated grade inserts (designated BK-84). Flank wear was selected as the criterion to evaluate the tool's life, and measured according to ISO 868802. Cutting tool flank wear (VB) was defined as new (0 less than or equal to Vbmax 75 µm), half-new (75 µm less than or equal to Vbmax 150 µm), half-worn (150 µm less than or equal to Vbmax 250 µm), and worn (250 µm less than or equal to Vbmax).
Feature vectors that represent the cutting tool condition were computed from vibration signals using Mel Frequency Cepstrum Coefficients (MFCC). Three pattern- recognition approaches were used: Hidden Markov Models (HMM), Artificial Neural Networks (ANN), and Learning Vector Quantization (LVQ). A database was built with 110 experiments; 23 used a new cutting tool, 35 a half-new tool, 29 a half-worn, and 23 a worn cutting tool. A MonteCarlo simulation for the system training/testing steps was implemented. Results for each approach were generated.
A classic test in a diagnosis system is to identify two alarms due to a bad classification of the phenomena being examined. Typical alarms are: False Alarm Rate (FAR) and False Fault Rate (FFR). The FAR condition falsely identifies a tool as damaged, while FFR describes a damaged tool as good.
New ideas based on MFCC were used to characterize the vibration signals between the cutting tool and workpiece. Of the three classical algorithms for pattern recognition: HMM, ANN, and LVQ, the best result was obtained in the HMM-based algorithm. It recognized the cutting tool wear condition with 84.24% accuracy, and generated only an 8.48% FAR. The researchers believe they have proven that it's possible to exploit MFCC representation for vibration signals, and an HMM-based approach for a diagnostic system, to monitor tool condition. They recommend that an HMM-based algorithm, and a second diagnosis via the LVQ-based approach, might yield improved results. They are doing new experiments to apply this methodology in a cutting-tool wear-condition monitoring system with different tools and materials.
This article was first published in the July 2007 edition of Manufacturing Engineering magazine.