Nearly a decade ago, a collaborative project with Duke Energy, Clemson University and Dranetz-BMI tried to change the focus from the quality of power to the quality of the process.
Given that different processes have different susceptibilities to the characteristics of the supply voltage, it seemed like a good idea to look at what the effects were on the product being produced. The process could be the flow of information in a data center, such as financial information from a cash register seeking a credit card approval request from the central banking computer or the weld and painting quality in an automotive plant, which can be compromised by the quality of the supply.
The project focused on the operation of a textile mill, which, like most continuous stream processes, uses electricity to power the thousands of sensors. The sensors provide feedback to dozens of programmable-logic controllers (PLCs), which control hundreds of adjustable-speed drives; this process maintains the proper parameters within the system to generate a quality product. In a textile fiber mill, plastic pellets are melted under controlled temperature and pressure conditions, then extruded through a spinnet as a thin fiber. The fiber is wound through five to seven drums to, typically, stretch the fiber until it is five times thinner. Speed sensors (digital tachometers, resolvers, encoders) and torque sensors (load cells) are at each drive location to provide feedback to the control system to compare against the “recipe.” It is then spun into yarn and wound onto the take-up reel, which contains hundreds of yards of product before being hot-swapped to another reel. This stretching process is done by precisely controlling the speed and torque of each drum in the stage using PLC-controlled, ASD-fed motors, which are each turning at slightly faster rates to stretch the fiber.
Before multiple fibers are spun into yarn, the fiber does not have adequate tensile strength to withstand variations in the torque that may occur if one drum suddenly turns at a slower rate, producing slack, and then abruptly returns back to normal speed. The fiber may either be stretched too thin and degrade the product quality or snap completely, which results in a lengthy clean up and restart effort that can take several hours. Even though drives are often vulnerable to trip out due to overcurrent indications during sags, the sags may affect the sensors themselves. If the sensor produces the wrong output signal compared to the parameter being monitored, the control system will adjust unnecessarily and possibly detrimentally to the process quality.
Voltage sags that were occurring once a month at one particular South Carolina textile mill were causing more than $100k in lost production. A textile processor simulator was developed, and this “Webstand” was subjected by a voltage-sag generator to sags of varying depths, duration, phase angle of initiation and the number of voltage phases involved. When the system was subjected to sags, there were changes in the sensors’ output due the effect on the power supply to the sensor’s circuitry. If the sensors’ output misrepresents the actual parameter values during the sag, the control system will try to respond to it and make those unnecessary corrections, which may, in turn, affect the quality of the product or cause product breakage. This is shown in the figures above, with Figure 1 showing the voltage sag and effect on the current. Figure 2 shows the effect on the sensors.
The output of one of the load tension sensors doubles in value during the first cycle of the sag, then abruptly reduces to less than its nominal value. This very high sensor rate was interpreted by the control signal as a binding condition that resulted in a system shutdown, shown by the output decay of the speed sensors.
The global economy has put pressure on the production manager to improve productivity and product quality. He or she should not have to rely on limited or no information about why the process tripped offline or why the product quality has been degraded. Regardless of having actual power quality knowledge, the effect of the power quality on the quality of the process is what really matters.
BINGHAM, a contributing editor for power quality, can be reached at 732.287.3680.