Lean Principles Don't Exclude Automation
Automation can enhance part quality, reduce physical stress faced by shop-floor workers, and improve the overall production process
By John Burg
Ellison Technologies Automation
Council Bluffs, IA
If you're not yet a cheerleader for lean principles in manufacturing and automation, perhaps you should be. There are a number of important reasons why lean makes sense (and cents), but the most important is that it might just save your job and allow your company to keep manufacturing in the US.
Some of the biggest steps in lean manufacturing come before automation, steps like developing standard work instructions, reducing waste, and involving your entire workforce in the optimization process. These measures lay the groundwork for efficiencies realized through automation; efficiencies that are not obtainable using manually loaded machine tools.
There are those who say that lean and automation do not mix. They claim that constant improvement requires flexibility in the production process, and that automation is inflexible. Their argument is strange, given the highly automated facilities that serve as models of the lean system of production. And, while I agree that constant improvement requires flexibility, I do not agree that automation is inflexible. Modern robots are highly programmable and easily redeployed within a production environment.
But there is a bigger issue here, not just about lean, but about manufacturing in general. The end result of lean, a highly consistent and controlled process, is repetitive and monotonous work. Most of us don't want to do that kind of work. If you have a very specific work flow that allows no breaks, most operators don't want that job. Repetitive motion does not give an operator the sense that he is contributing or that he matters, and these are precisely the sorts of values that lean depends upon.
And even if you're not worried about the softer side of the work environment, automation through lean might still be an economic necessity. Automation makes lean processes much more profitable and ergonomically friendly, especially when there are heavy parts, multiple interdependent steps, and takt times that are unrealistic for a human operator.
Let's look at the example of a lean manufacturing cell for pulleys. Various parts run through this system, but all are between 5 and 12" (127–305 mm) in diam with a center hole. All of the parts weigh between 30 and 50 lb (13.6–22.7 kg). The same casting is often used for multiple part numbers with machined variations. Inside and outside diameters are machined on each part, sometimes with multiple diameters. The part tolerances are ±0.0005" (0.013 mm) on the ID with concentricity of the OD to the ID of 0.001" (0.03-mm) TIR. The system takt time averages 2 min, 15 sec, but it varies by part. Parts run in lots that average 10 pieces.
You should notice, from the start, that all of these specifications would be better understood and better documented if the facility had already implemented some of the basic steps outlined above. A well-understood process is much more easily automated than a sloppy or poorly documented process. Here are the process steps:
- Inbound castings into the cell,
- Radial location for some parts using cast details,
- Machine complete OD and ID,
- Remove chips using an air blast,
- Gage critical diameters on the OD and ID on all parts,
- Spin balance,
- Mark the part with a part number and MFG. code,
- Outbound machined pulley.
A single operator would be unlikely to maintain the 2:15 takt time with these parts, at least not for long. And setups slow things down even further. Even with quick setup tactics, the time required for setup is 30–45 min. With takt times of more than two minutes and 10 pieces in a run, setup takes 50% more time than production. As a ballpark estimate, this takt time would require three or four identical manual cells to maintain the required throughput.
Though we may not worry about the human resource issues from the industrial engineering perspective, the coordination and management resources required to power these cells with people would be significant. And, with four cells and four machines per cell, the eight process steps require the operator to handle the part 18 times. If there are four cells, even the smallest parts would leave the operator to handle more than one ton (907 kg) of material per hour. Very few people will be able or willing to keep up with that pace for a whole shift and the output will be compromised on a daily basis, not to mention the ergonomic and safety concerns associated with repetitive stress.
Automation provides an alternative that can substantially increase machine utilization, decrease safety liability, and reduce the overall overhead of the production process. Many machine-tool builders offer machines that can perform multiple operations in a single holding. Lathes with dual-opposed spindles can machine both sides of the part without intervention. Multiprocess (multitasking) machines are a form of automation, and can produce complex parts with very tight tolerances, because of the ability to machine details into the part without the accumulated variation from multiple workholding steps. There are tradeoffs between multiprocess machines and simple machines. The simple machines cost less, but they often require more labor and more machines to achieve a task. We cannot always purchase the perfect machine for the job.
In the example described above, there are processes that cannot be accomplished in a single machine, or not with the same quality. Part marking is generally performed as an auxiliary process by a specialized device. Multitasking machines can mark parts, but not with the speed and print quality of a laser marker or pin-stamping machine. Balancing is another good example; the spin-balancing process requires a specific machine designed for that task and no other. Gaging is often best done outside the machining process. Most machine tools have probe-based internal gaging systems, but the positioning error of the machine tool will also affect the gage. An error in the machine-tool positioning while machining may be reproduced when gaging the part, leaving the error undetected. External gaging systems are not influenced by the machine's positioning error, and also may be easier to calibrate than the probe-based system.
By far the most time-consuming operations encountered when making the part are the machining-process steps: turning, facing, and boring. Spin balancing, gaging, and laser marking are each twice as fast as the machining process. These process speeds will determine the balance of machines required per cell.
The illustration on page 63 shows a CAD-generated layout for the automated lean cell developed to manufacture the family of cast-iron pulleys described above. The cell includes the following equipment:
- Two Mori Seiki ZT2500Y multifunctional twin-spindle machines,
- One Fanuc Robotics M710i50T Toploader overheadmounted six axis robot with IR integrated vision system,
- One Etamic gage station with data recording and automated offset adjustment for the Mori Seiki machines,
- One Spin balancing machine, and
- One Trumpf laser marker.
The cell also includes multiple inbound and outbound part conveyors, and a PC-based cell control with user HMI for cell communication with the operator.
Normal operation in this cell requires only one operator. His or her job is to monitor the cell, assure quality, set up the CNC machine, load raw castings on inbound conveyers, and unload outbound parts. The cell can handle one-piece lots, and the two machining centers allow the operator to safely set up one machine while the robot and the other machine continue production.
On the inbound side, the operator loads the parts without concern for radial orientation. Next, the operator selects the part number at the HMI for input to the cell controller. The robot and the cell controller subsequently control the recipe for the part and the process steps. Obviously, the robot cannot miss steps or perform them in the wrong order. The robot is equipped with a Fanuc Robotics 2-D vision system that locates the radial orientation of parts requiring orientation, and adjusts the robot's position at the load point of the first operation, so that correct orientation is maintained throughout the entire machining process.
The Mori Seiki machine has two spindles. When the robot is called, it carries a raw casting ready for processing to the machine. When the doors on the machine open, the robot loads the raw casting into the first-operation spindle/workholding, and immediately unloads a finished part from the second-operation spindle. As soon as the robot has cleared the machine, the door closes and machining begins. The robot moves to the spin-balancing process, unloads this machine, and deposits the next part to be balanced.
Once it has been balanced, the part is cleaned, gaged, laser-marked, and conveyed out of the cell. Each of these process steps uses a different machine. The gage monitors trends in machined dimensions, and uploads offsets to the Mori Seiki ZT2500Y to maintain dimensional integrity even as cutting tools wear. Next, the finished part is placed on the outbound conveyor or on an inspection slide for verification. The two inbound and two outbound conveyors can queue ten or more parts each, allowing a minimum of 40 min of untended operation.
An automated system like this can be justified in terms of cost, quality, safety, throughput, and machine utilization, and all of these should be considered when calculating the ROI necessary to justify a system. But it's also useful to think at a more basic level about the kind of manufacturing that we aspire to do. It is not possible to manufacture without people. People design our products. People decide how best to make our products. And people value our products enough to purchase them. In any foreseeable future, people will continue to make and service the machines used in manufacturing. And, when it makes sense, people do the manufacturing themselves. But in the end, there are better and more meaningful things for a person to be doing than carrying pieces of metal from one process to the next. If such activity is the highest form of employment that we hope to provide, it is unlikely that we will provide it in the US.
If you haven't applied lean principles to your machining processes, get to work. You will be surprised at how quickly your mental and social investment will be paid back by process improvement. But don't stop after you've attained some improvement through the application of simple technologies. Constant improvement means that you must keep looking for opportunities to cut and slice your costs and improve your throughput. Many times you won't need automation. Lean thinking and process improvement will be sufficient to achieve the takt time you require.
There will be cases like the pulley production operation described above, however, when a very lean process needs automation to go the distance. Automation can provide an excellent return on investment.
This article was first published in the July 2009 edition of Manufacturing Engineering magazine.