At a smelter, a cascade control system maintains negative pressure within the electric furnace freeboard. This system is comprised of five (5) single-input, single-output controllers. The primary controller in this system monitors the pressure of the furnace freeboard and outputs an offgas flow rate (in Nm3/h) which drives the pressure towards the setpoint defined by the operator. The output of the primary controller is cascaded to two secondary controllers that independently adjust the position of dampers to control the flow of offgas from the matte and slag ends of the furnace. A tertiary pair of controllers adjust the speed of the offgas fans to maintain the position of the dampers at a defined setpoint such that there is sufficient capability to open or close the dampers should there be a sudden increase or decrease in furnace pressure.

In mid-November, a member of the smelter’s technical staff noticed that the output of the flow controller for the slag end of the furnace was oscillating. These oscillations were particularly evident immediately following periods where the damper had been fully open. As shown in Figure 1 below, the oscillatory action of the controller resulted in significant swings in furnace pressure.

image1Figure 1. Time series plot showing the position of the offgas damper and the pressure in the furnace. 


In order to address this issue, XPS first used the PlantESP™ Loop Performance Monitoring software to determine if the oscillations were likely to be a result of a mechanical issue with the damper. Static friction (stiction) in valves / dampers can slow or inhibit proper adjustment and (as a result) can prevent the valve from accurately controlling the process, leading to oscillation. As shown in Figure 2 below, the likelihood and amount of stiction calculated during the early stages of the event were significant (54% and 10%, respectively). During this period, irregularities were observed with the process air supply in other areas of the plant and were believed to be a result of colder than normal temperatures; cold weather can increase the amount of moisture in the air supply and affect its ability to control actuators. The Process Engineer notified the instrumentation department and one of their technicians inspected the damper. There were no issues reported, though temperatures had increased by the time of the inspection.


image5Figure 2. Time series plots of the calculated stiction and oscillation during the period of poor loop performance.


By evaluating both the stiction and oscillation metrics over the period of poor control (see Figure 2), it was determined that the oscillatory action of the flow controller was also present during periods where stiction was unlikely / insignificant (likelihood and amount less than 1 %). This assessment indicated that the issue with the controller was not a direct result of a mechanical problem with the damper; rather valve stiction likely compounded with another issue (e.g. loop tuning, build-up within the offgas system, etc.) to result in the significant oscillations observed. 

Figure 3 presents time series plots of the tuning parameters recommended by PlantESP (A) as well as the calculated deviation between the recommended parameters and the existing tuning (B). The metrics handbook recommends that the tuning of a loop be investigated if the tuning deviation is greater than one (1) and PlantESP has identified more than twenty (20) model fits. During the January to November 2019 period, the calculated tuning deviation was consistently greater than two (2) and (as of November) PlantESP had identified two hundred (200) model fits. Therefore, it was hypothesised that an improvement to loop performance could be made through adjustments to the tuning parameters.



Figure 3. Time series plots of the tuning parameters recommended by PlantESP and the calculated tuning deviation over the month prior to the oscillation event.


PlantESP and LOOP-PRO™ TUNER were used in combination to identify new tuning parameters for the offgas flow controller. To begin, the tuning recommendations provided by PlantESP were compared to the existing parameters. This comparison determined that a less aggressive tuning should be implemented, a recommendation which was consistent with the advice of the Process Control Engineer. Next, the model fits were sorted based on their ‘rating’ and the most appropriate models were selected for further analysis using the LOOP-PRO TUNER software. Three tuning recommendations were developed: two moderately aggressive tunings and one moderately conservative tuning. Each tuning was implemented in the distributed control system (DCS) and monitored to determine if the oscillations were reduced / eliminated and if the pressure control system remained effective. Ultimately, it was determined that the moderately conservative tuning performed best in terms of eliminating the unwanted oscillations while still providing good overall performance. Figure 4 presents a time series plot of the offgas damper position and furnace pressure prior to the oscillation, during the period of poor control, and following the implementation of the new tuning parameters. This plot demonstrates that a significant decrease in oscillation was achieved through the implementation of the moderately conservative tuning parameters.

image8Figure 4. Time series plot of the offgas flow damper position and furnace pressure.


Figure 5 presents the output travel reported by PlantESP for the offgas flow controller before and after the new tuning parameters were implemented (November 15, 2019). This metric monitors the total movement of the output signal, indicating how much effort is required by the final control element (i.e. the damper) to maintain control. A very large output travel can cause valve damage and may be indicative of process, equipment of controller issues. In addition to improving the overall performance of the flow controller, Figure 5 demonstrates that the more conservative tuning significantly reduced output travel and (as a result) the likelihood of premature damper failure due to excessive wear and tear.

image9Figure 5. Output travel before and after loop tuning performed.


XPS is currently in the process of tuning / optimizing individual loop configurations such that PlantESP effectively directs the user to loops exhibiting poor performance. Upon completion of the investigation described above, the metric limits and weightings for the flow controller were adjusted such that future events will be detected in a timely manner, limiting the impact on the performance of the pressure control system. Based on an analysis of loop performance over a three (3) month period, the ‘Hi’ and ‘HiHi’ limits for the average absolute error metric were reduced (i.e. made more stringent) and the importance of the oscillation metric was decreased.