Resolution of Step Change in Vibration Levels of a Steam Turbine

Abstract This is a Steam turbine unit of three casings (HP/IP, LP1 and LP2) connected to a Generator through a rigid coupling. The rated power of the Generator is 660 MW. The unit was commissioned in 2014 and undergone a major outage in February 2018. During the outage, the generator stator windings were replaced and coupling bolts between all rotors were modified and replaced with new design (hydraulically fitted coupling bolts) to ensure smooth assembly/ disassembly procedure. After the maintenance, the unit was started up in March 2018 and it was running at baseload with normal vibration levels below alarm set points. It was reported that sudden increase of vibration at LP2 turbine was noticed on 16th July 2018 followed by another step change on 26th July 2018. Maximum recorded vibration levels reached 130 μm PP (Alarm 120 μm PP) at LP2 BRG (towards LP1 side)). While the machine was still under service and during the period of January to April 2019, several vibration step-changes events occurred with the highest vibration levels at LP2 Bearings (both ends). All these events were recorded only on DCS trends whereas there was no online diagnostics system to capture the dynamic behavior of the machine. A check list of maintenance activities was carried out by the customer during the planned outage to ensure the machine integrity and rule out some suspects (coupling and hold down bolts tightening, alignment check between LP2 & Generator, bearings clearances check etc.). A multichannel analyzer was hooked to collect the data after the outage. The machine was started up with the same vibration levels!! The review of data during the transient mode (Startup) and at steady state indicated a clear symptom of unbalance. Also, the historical review of DCS data for various step changes showed that the vibration changes were localized to LP2 turbine mainly. This case study is focused on how the unbalance was detected using the transient as well as steady state vibration plots which was matching with the inspection findings of broken LP-2 rotor blades of last two stages. Also, this case study is focusing on how the process data from DCS data helped in identifying the root cause for the broken LP rotor blades (abnormal extraction temperature, abnormal change of LP-2 differential expansion etc.) which helped in fixing the issue permanently.