Online transient analysis optimises asset utilisation - a case study |
Written by AIA |
Traditional condition monitoring systems are designed for “steady state” operation of the turbine. But a turbine is rarely operating under these conditions. Loads, pressures, temperatures and vibration are changing rapidly, especially during startups, shutdowns and “bumps” in the night.
Extracting dataSince the turbine’s operating parameters are constantly changing, it is important to monitor overall machine health and not just changes in vibration. In transient analysis, turbine engineers and operators have easy access to continuous, real-time vibration information, allowing them to closely monitor the condition of turbines during critical startup and shutdown periods.
Through the “live turbine dashboard” in the control room, users have dual monitors to view the development of seven different plot types: Orbit, Shaft Centerline Bode, Nyquist, waveform, spectrum, and cascade — on up to eight bearings. The data are updated up to five times per second.
FIG 1 THE TURBINE GOES FROM HEAT SOAK TO RUNNING SPEED IN THE TOP PLOT. IN THE SECOND PLOT LEFT THE USER HAS MOVED THE “REGION OF INTEREST “RECTANGLE TO A SLIGHT “BUMP” HE WANTS TO INVESTIGATE, AND CLICKED EXTRACT. FIG 2 THE ANALYST NOW HAS DETAILED INFORMATION CALLED FOR IN FIG 3 STORED DATA THAT WERE EXTRACTED SHOW THAT VIBRATION AMPLITUDE DID NOT REDUCE AFTER THE FIRST CRITICAL SPEED. FIG 4 ORBITS, SHAFT CENTRELINE AND SPECTRUM PLOTS SHOW THAT SHAFT MISALIGNMENT MAY HAVE CAUSED HIGH VIBRATION. In the past, it was necessary to have a vibration collection strategy for monitoring turbines because of the large volume of data generated. But those collection strategies sometimes missed critical data. Today, setting up data-capture simply requires checking a box on the “configuration” screen. If that box is checked, transient data are collected. A “region of interest” selector enables turbine engineers and operators to pick the window where they want to extract detailed information (Figures 1, 2). No time is wasted searching through megabytes of data. Aiding turbine analystsTransient data help analysts to zoom in on any anomaly in turbine operation in the last 60 hours. The following case study highlights how this detailed data can benefit turbine analysts. When starting up, a turbine is attempting to pass through the critical resonance. Vibration should have begun to decrease at that point. Instead, it took an abrupt turn and vibration began increasing greatly with speed (Figure 3). The Bode plot (upper right) shows that the critical resonance speed of the machine was changing — it was shifting to the right. For the critical to shift, damping, stiffness or mass must be changing . It is reasonable to assume that mass did not change, but that damping or stiffness of the system has changed . The operator noticed that the turbine did not pass through the critical speed. (In some cases the anomaly is not too great and the operator could continue to increase speed, passing through the critical speed). In this case, the operator was not getting past the critical resonance. Vibration continued to increase, so a decision was made to shutdown the turbine. Simply re-starting this turbine, with no remedial action, will cause the same result. Other plots will reveal what action must be taken for a successful startup. The orbit shows that the normal dynamic shaft motion has abruptly detoured (Figures 3, 4). A normal orbit would follow the green path. The shaft was not moving freely in an elliptical path as we would normally expect. The shaft centerline plot in the lower left should not be one dot, as the turbine came up to speed (Figure 4). Ideally the shaft centerline position within the bearing should move up and to the right. However, if the shaft is misaligned or improperly loaded, it would be forced and held in an abnormal position. The spectrum plot shows a high 2x turning speed peak — another indication of misalignment or improper loading. All the above show that the turbine needs to be re-aligned and checked for binding and improper loading. The above technology provides the user a powerful tool for predicting the future health of machines and developing an optimization strategy to reduce maintenance, thus enhancing their plant's return on assets (ROA). -- Shailesh Naik |