Printer Friendly

Toward the untended grinding cell.

Toward the untended grinding cell

Dr Stuart Salmon, president, Advanced Manufacturing & Technology developed continuous-dress creep-feed technology while working for Rolls-Royce in England. In the early '80s, he came to the US to help us catch up to European grinding technology. We talked with him recently about the progress of grinding automation in this country, and particularly about the limits to achieving an untended grinding cell that parallels the progress we have seen in turning and machining.

Operator's diminishing role

In terms of the typical CNC grinding cell, the operator really doesn't have much of a part in the operation, Salmon feels. "He has become a watchdog. The typical CNC installation has already automated away most of his functions. In essence, the operator isn't really doing very much. Although wheel changes are often done manually, there is little incentive to automate this function, particularly in high speed applications, where making sure the wheel is running true and balanced are critical and complex functions not easily automated."

If we are reaching the limits of practical operator involvement, what's left to automate? "For those with the latest grinding installations, cut time is now minimal, so getting the part to and from the machine is the remaining area to work on to capitalize on the machine's high speed and quick grinding cycles. The challenge, therefore, is to design fixtures to hold the part and to automate the transporting of parts, machine to machine. Unless the application is particularly high volume--such as those highly automated turbine-blade root-grinding installations we've all seen or read about--there is nothing wrong with manual loading. That's still used in most cases. Most creep-feed applications today are manually loaded. It's only the turbine-blade situation where they go for the totally automated cell."

Is this turbine-blade automation really necessary? "Yes, it is easily justified because a single rotor uses something like 160 per ring, and blades must be periodically replaced; i.e., the quantities are enormous. So, the automation combines highspeed fixturing for loading/unloading with flexibility to change blade sizes and shapes without lengthy manual setup times.

"But in lesser-volume applications, automating a grinding cell would be a waste of time and money. Here, it is simply a matter of using the machine and wheels efficiently. Dressing cycles are generally well automated by now--an established cycle--or no dressing is required for the high-speed CBN electroplated wheels, the easiest to setup.

"The operator isn't really involved in monitoring the dressing operation--merely present in case something serious goes wrong: a malfunctioning CNC axis, the form on the first part is wrong, a programming error, etc. He's there to catch the catastrophic errors, not make subtle changes or adjustments. The advanced systems are self-monitoring, checking part size automatically, usually at a downstream inspection station."

Monitoring the machine

In the US, Salmon points out, this checking function is not yet on the grinder itself, something that has been offered as an option on advanced German machines for several years now and well accepted there. "The downstream automated inspection line here is usually for the aerospace people who are very nervous about exact tolerances. Although ground parts are inspected off the machine, these measurements are used to update the machine after computing trends."

Unlike turning and other machining operations here that have accepted on-machine measurement as valid feedback, the idea that you can use touch probes to measure diameters and faces on cylindrical grinders has not caught on here as yet. "That capability has been around for several years in Germany and elsewhere," Salmon admits. "People here, for some reason, seem satisfied with their success with conventional grinding methods and are not ready yet to take this step up toward adaptive control. It would require an enormous investment. The on-machine probe isn't cheap and neither is the machine it's designed for. People here are not ready to add that kind of cost to the grinding of the part."

Two different eras

Salmon sees two grinding extremes today. "Archaic shops, absolutely terrible, where they're using machines designed back when materials were easy to grind. Then, grinding was simply a matter of finishing--taking off a small amount of material over aeons of time. In contrast, today's grinder has the rigidity, stiffness, and controls to take off much more material, even with the more modern alloys.

"Those companies using these older machines are fighting two things: the newer materials and the grinding stiffness they need to remove heavy stock, and are, therefore, doubly worse off. The machine can't hack it, and they must rely heavily on the operator's skill to dress the wheel many times. As a result, they cannot expect to get anything like the productivity they could get with a more modern machine."

Adaptive controls

Adaptive controls have found some success in other metalcutting areas. What variables are being monitored in today's adaptive grinding controls? "Mostly just force measurement," he replies. "In my mind, the only successful adaptive control is one that relates force to size of workpiece; i.e., interpretive force measurement--the ability to detect when the part is going to burn or crack. But we're a long way, I think, from true adaptive-control grinding."

Just by measuring the force on the wheel? "Transducers also measure the reactive forces in the workpiece, either buried in the table or wheelhead, strain-gage sensors in wheelmotor bearings, or pressure-pocket sensors in hydrostatic machines. This is still in its infancy in the US, with the most work being done in Japan, plus pockets of activity in Germany.

What will come after force-based adaptive control? "Measuring force tells us a couple of things. Obviously, a force on the machine structure causes deflection because we don't have infinitely stiff machines. Right now, the majority of adaptive-type control systems are predicting size error, based on knowing the machine's stiffness and a given grinding force. Next, you must relate that force to the specific grinding situation and determine if that tells you anything about how well you are grinding.

"This is where we need some R&D. For a given grinding-force increase, how should you effect change? There are number of ways for decreasing the force: speed up the wheel, dress the wheel, decrease infeed rate, etc. But, which is correct? We need research to tell us what our response should be for a given situation, and that decision must be made in a very short period of time.

"So, for the moment, when the reactive forces exceed a predetermined limit, we merely sound an alarm and ask the operator to react, and we use his judgment. The closest we come to adaptive control is simply grinding the part, measuring the forces, and continuing until something is basically wrong with the part. This establishes a force level not to exceed the next time. The adaptive control merely sets a trigger level, and you must decide what to do before this level is exceeded. You establish a buffer.

"Thus, a next step would be incorporating into the control a more appropriate reaction in real time, based on grinding research. Although many companies claim to have adaptive control, none are at this level yet. In my mind, their trigger-type control is actually detrimental because you're not running to the optimum line that lies somewhere within that buffer region. This means incorporating a certain amount of error at all times, instead of operating much closer to optimum grinding conditions with true adaptive control."

Closing the creep-feed gap

A large part of Salmon's work has been providing companies with training sessions in creep-feed methods and grinding automation. Recently, he has noticed an improvement in operator education as more people learn how to push grinding beyond part finishing to more of a part-producing process. "People are starting to become much better educated in grinding basics--grit size, wheel speed, etc--and understanding how these factors affect the process."

Creep-feed grinding is picking up noticeably, Salmon observes, based in part on the response to his grinding demonstrations at recent major tool shows (IMTS-88 and WESTEC '89). "It is going beyond aerospace and high tech to smaller job shops and the more mundane manufacturing industries. The productivity incentive is getting to them, the savings in deburring costs, the cutter-sharpening cost improvement over milling or broaching, and the fact that a CNC-grade grinder can also move into superabrasives and take on contouring tasks."

Lights-out grinding?

Is untended, completely lights-out grinding attainable, and does it really make sense? "Yes, it can be done, but it's not for the everyday workshop. To my knowledge, only the highly automated turbine blade installations (Rolls Royce, GE, Pratt & Whitney, etc), and the Briggs & Stratton cylindrical grinding of crankarms are examples of this level of automation. Because of the type of material used in turbine blades, no other process can yield both the quantities and accuracies required. You have to grind them, and you need the automation for uniformity and productivity."

But who else needs an untended grinding cell? "Well, this level of automation is not needed for a great majority of industry. Job shops and companies with a fairly expensive CNC grinder will more likely be doing small batch quantities, and the added cost of automation would not be economically viable."

Thus, it seems that unlike the case being made for untended turning, grinding technology is more complex and still requires some operator involvement or supervision in the metal-removal portion of the operation. So, why even try to eliminate the operator? Why not keep him involved, and let him also be responsible for loading parts, except for those high-volume applications that can clearly justify automation?

So what should we be working on, if not to get that operator into early retirement? "Fixture design can assist him in loading/unloading better than he can with older loading methods such as magnetic tables and collet-chucking arrangements. These required the operator to do a lot of playing around to get the part into the right position; i.e., lengthy setup times for each part. With fixtures designed to speed up the positioning of the part, loading time can be minimized and cutting time maximized. To accomplish this, the people in tool design must be much more aware of what's involved in CNC grinding to achieve productivity."

Can cantilevers cut it?

Salmon sees one basic flaw in all of grinding-machine design today. "To my mind, it's evolving in the wrong direction. The original machines were used strictly for finish grinding--easy-to-grind materials and not much material to be removed. They worked quite well, but had a rather weak structure. "With the advent of superabrasives and creep-feed grinding, these machines needed to be beefed up significantly.

"Yet, all this time, we have retained a cantilevered spindle arrangement. The latest machines have a rotary dressing device and a rotary grinding wheel, all moving in a manner that can never be perfectly balanced. Therefore, there is always some vibration when we try to move that large mass of grinding head to within tenths of thousandths of an inch of a relatively small workpiece supported by a giant cast-iron or granitan base.

"It would be far better to hold the wheel stationary and bring the workpiece to it! But, that goes against the grain of all grinding machines being designed today."

Steadying part deflection

Speaking of the effect of wheel on part, we've seen a variety of steady-rest devices on European cylindrical grinders that either measure or support the part in an attempt to compensate for this effect. Will this work? "Part of their approach, I believe, is to allow the part to deflect and measure that deflection. In cylindrical grinding, we normally want to creep up on final size without deflecting the part. So we incorporate rough-grind, medium-grind, and fine-grind cycles to keep the part running concentrically.

"What we could be doing instead is plunging the grinding wheel at much higher force into the workpiece, allowing it to deflect (even to the point of permitting some wheel breakdown) to get tremendous stock-removal rates. In fact, wheel infeed could go beyond final size as the part deflects, but when the wheel retracts, the part would also retract to leave the few thousandths you need for finish grinding to size after pausing to redness the wheel. With an on-machine measuring system (or a good experential data base), you could achieve this.

"A number of steady systems are available that compensate for wheel forces, including manually adjusted and CNC controlled to keep the steady always in the right position so that you don't have to stop the process and move the steady the amount that has just been ground off. But these are relatively advanced technologies compared to anything being done in this country at the moment."

Best is yet to come?

In your view, what US industries are doing the best grinding today?

"The one's with money!" he jokes. "I'd love to say automotive, but I can't. A few companies (one in hand tools that I can think of) are turning to abrasive-machining processes over traditional chipmaking processes. Right now, the best example of leading-edge grinding and automation is at Briggs & Stratton, grinding crankshafts for small-horsepower engines." (See box, Untended CBN grinding, the best of both worlds.)

Salmon feels the grinding-automation examples he's seen in the automotive industry merely automate grinding machines and technology that is decades old. "I don't think the automotive people in this country realize how this part of their manufacturing technology affects the performance of their engines. Both Schaudt and Landis have similar new cam grinders that manufacture a cam profile much more accurately with respect to the bearing diameter than ever before. They are based on holding the cam by the bearing diameter, not the centers; i.e., instead of moving the cam in and out of the wheel, the wheel moves in and out with the cam shape. The result is a much more accurate cam profile with respect to the bearing diameter upon which that cam will actually run."

Thus, the automakers focused on productivity improvement and not on the extra warranty miles more accurately ground profiles could add to their engines? "Apparently. It's clear from the aerospace examples we've seen that once your competitor does something like this, you really cannot afford not to follow. The US jet-engine makers did, and now can boast of engines that produce better mileage, go longer between servicing, and run quieter. Achieving re-entrant type cam forms in the auto industry would be similarly beneficial. It would enable them to keep up with imported engines. But, this is something that is extremely difficult to justify financially--to jot down on a spreadsheet and show exactly how and when it would pay off. It involves such emotional things as the Japanese lead in engines that doesn't translate well into financial terms."

PHOTO : Two examples of machine-mounted touch probes on cylindrical grinders to monitor part size either during or immediately following grinding.

PHOTO : Advanced controls on this Schaudt (Stuttgart, West Germany) cylindrical grinder can manipulate up to nine axes, including automatic loaders. Shown here are movements of the Z axis, workpiece transverse; X axis, wheel in-feed; V axis, dressing-tool in-feed; U axis, readjustment of steadyrest for wheel-size change; and Q axis, readjustment of distance between centers.
COPYRIGHT 1989 Nelson Publishing
No portion of this article can be reproduced without the express written permission from the copyright holder.
Copyright 1989 Gale, Cengage Learning. All rights reserved.

Article Details
Printer friendly Cite/link Email Feedback
Author:Sprow, Eugene E.
Publication:Tooling & Production
Date:Jun 1, 1989
Previous Article:The case for optical CMMs.
Next Article:Automated grinding boosts crankshaft quality.

Related Articles
EDM today and tomorrow.
Untended machining - a boost to productivity and profits.
Software automates punch manufacture.
Innovation for global competitiveness.
On the move with the EDM.
Multi-tasking grinding.

Terms of use | Copyright © 2017 Farlex, Inc. | Feedback | For webmasters