Application SummaryAluminum extrusion machines often spend too much time loading a fresh billet and performing some "housekeeping" tasks. Although this dead time is necessary, it limits the number of billets that can be processed. Machine owners strive to minimize the dead time by fine-tuning the control system. Specialists who fine-tune the machines are expert in balancing a number of critical parameters to ensure smooth operation of the hydraulic ram, minimize system pressure spikes, and guarantee a safe machine.
Harold Baker of Harold Baker Consulting, Bartlette, Ill., has a lot of experience fine-tuning the best machines used by the Aluminum Company of America, Morris, Ill., as well as other aluminum-extrusion manufacturers. Baker monitors system hydraulic pressure in several locations while adjusting time delays, ram position sensors, pressure sensors, pump commands, and flow control valves. The data acquisition system he uses must be able to accommodate up to 15 inputs, record hours of data without interruption, and work reliably on the factory floor. In addition, the data must be easily recalled in a graphic format to be included with customer reports.
Possible Solution
Baker had been using a well-known brand of data acquisition system since the middle 1970's. He upgraded to their newer models in the early 1990's, but had to work around its limit of four input channels for pump signals and pressure sensors. "I always had to compromise my choice of measurements," says Baker. "I was really frustrated that I could not monitor every pressure and pump signal that I thought was important — simultaneously. I had to run several tests repeatedly with different combinations of sensors to establish the time and pressure profile I needed." Another drawback was the weight of the instrument, about 76 pounds. This made it extremely difficult to transport between various sites.
IOtech's Solution
The equipment's strict limitations prompted Baker to investigate another data acquisition system that provided more measurement channels, was more portable, and had the robustness to operate without failure in the rather harsh factory environment. He evaluated the IOtech and was pleasantly surprised to find more than he had expected. Says Baker, "The nice thing about the DaqBook is that I can put it in my roll-around computer case and carry it on the plane with me, so I don't have to check it in with baggage. By comparison, sometimes the old system didn't survive the plane trip, in spite of it being packaged in a foam-padded travel case. Too many times I arrived at a site with a dead instrument that had to be returned for repair."
Besides portability, Baker likes the ability to play back the data he needs for including in reports for his customers. Says Baker, "It's easy to log onto any portion of the recorded data stream and select the chart of interest with a simple "copy screen" command and insert it directly in my Microsoft® Word document." This makes a very professional-looking report. The old system used a 3.5-inch disk and could store only a few seconds of data. The DaqBook stores hours of data.
The larger number of input channels on the also makes Baker's job easier. He can monitor 8 transducer inputs and 6 pump command inputs and watch the system's response while adjusting the variable volume pumps and fine-tuning the valves. Moreover, during the cycle, he can record the pressure at the end of an extrusion while the system decompresses, the ram retracts, the container opens, and the butt shears. The next part of the cycle is just as critical to record, that is, when the container closes, the loader moves up, the ram travels forward and crushes the billet, and then extrudes. "The dead time is when nothing is being extruded," says Baker. "So I examine the dead cycle to see where I can find delays. Timers and pump commands can be adjusted to speed up the dead cycle but not shock the system."
When running 30 billets per hour, and fine-tuning takes from one to five seconds per charge off the press in dead-cycle time, at the end of the day, week, or month, it amounts to a huge amount of time saved. Several more billets could be processed, but the press is effectively just sitting there. "A prime example is a 7300-ton UBE press that Alcoa owns," says Baker. "It processes 18-inch diameter by 54-inch long billets weighing more than 1,000 pounds. So at the end of the month, you can often push through more than one additional billet by reducing the dead cycle." It's quite an art, claims Baker, and for the longest time, people did not worry about dead cycle. "But when you start looking at a second here and another second there, it adds up quickly," continues Baker. "I removed multiple seconds from numerous machines, caused by such things as air in the main ram. When loading a new billet, the ram crushes it against the die and builds pressure on the side cylinders. This commands the system to close the main ram prefill valve, which in turn prevents the ram from running at extrusion speed. When the full pump commands are present at the same time, air in the main ram is under compression, and you can see it on the recording. The just puts everything in front of you. You don't have to guess what the problem is."
Dead-cycle time depends on the size of the machine and where the dead cycle is measured. This has always been debatable, but for Baker, a cycle is pressure off, pressure on. The machine should have a short dead cycle, but operate without components banging and slamming. Says Baker, "I fixed an extrusion machine that had such a huge spike that the pressure was knocking the pipes off the butt shear. Every clamp on the butt shear piping was broken." He connected the and found that the original program contained a full-pump command, but as the shear passed a slow-down proximity switch, the pump was supposed to destroke. The proximity switch logic was not sealed for slow down, so it just ignored the switch and returned to commanding full stroke. The shear was slamming at full pump volume generating a huge spike in the hydraulic system. But by spreading out the command with the , Baker could see that the slow-down command was present for only a few milliseconds, and then returned to full command. So he changed the program to respond to the slow-down switch, and sealed the command until it reached the top of the shear. That took out the spike that shot clear off the scale, past the 5000-psi pressure transducer limit.
