Many high speed turbomachines operate above their first critical speed and thus, pass through one or more resonant frequencies during start-up. This can result in momentary vibration spikes, which is a normal condition. The magnitude of the vibration spike can depend on many factors such as the amount of rotor unbalance, how much damping is present, and other factors. The vibration level will decay as the rotor speed accelerates past the resonant frequency. The vibration levels for some stages on this machine spike momentarily during the start-up as the particular rotor accelerates through a resonant frequency. And, as noted above, the vibration on all stages spike when the compressor accelerates through the train torsional natural frequency.
In a tutorial on avoiding catastrophic failures from transients at the 2017 Turbomachinery & Pump Symposia, Patrick J. Smith of Air Products & Chemicals discussed control system settings for startup transients.
The high vibration trip set points are set at values that are lower than some of these vibration spikes because the high vibration trip set points are based on operation at the normal operating speed. To prevent the machine from tripping during start-up, the control system can be configured to automatically bypass the high vibration trips for 30 seconds. Although this is a common method to manage vibration spikes during start-up, it also eliminates the vibration protection for this period of time. To prevent unnecessary high vibration alarms and trips on start-up and still provide high vibration protection, another option is to temporarily elevate the high vibration alarm and trip set points. This method is recommended by API and is described in API-670, “Machinery Protection Systems.”
The elevated vibration alarm and trip set points need to be set higher than the rotor’s characteristic response at any resonances during the start-up, but the levels should be set as low as possible in order to maximize the machinery protection. In some machines, the alarm and trip points do not need to elevated at all. In other machines, the alarm and trip set points may need to be elevated by as much as two or even three times the normal set points. However, if the set points need to be elevated more than this, there may be a problem and this requires further review.
There are other details that need to considered as well. These include time delays, how to handle bad quality and dual instruments. In many cases vibration alarm/trip logic is configured with a time delay to prevent spurious trips. While this may be advantageous to prevent unnecessary spurious trips during steady state operation, it is less of an advantage during a start-up.
In a case study, the exponential increase in vibration on stages one and two from low vibration to the full range of the transmitters took three seconds. Tripping the machine sooner would limit the damage and possibly prevent a catastrophic failure. But, if there is a time delay on a vibration trip, the machine will run longer. That is not to say that a time delay couldn’t be configured for steady state operation. But, having no time delay on a start-up could lessen the impact of a mechanical fault.
In many cases, loss of signal or bad quality from an instrument may be configured as an alarm because it is assumed that it is an instrument issue. However, in the case of vibration, the event can happen so quickly that the vibration probe can be wiped out before a high vibration condition is detected by the control system. Configuring the control system with bad quality as a vote to trip protects against this. However, it can result in unnecessary machine trips during steady state operation due to instrument issues.
Configuring bad quality as a vote to trip only during start-up is one possible solution. Relying on a single instrument for protection also impacts the reliability. Having two instruments can improve reliability, such as the use of “x” and “y” vibration probes adjacent to each bearing rather than just a single probe. A high vibration vote to trip from both probes can prevent nuisance trips during steady state operation and while this can prevent unnecessary trips due to faulty instruments, it is not as advantageous during start-up.
In gearbox rotor bearing systems, vibration levels can be higher in one plane than in another plane. Thus, if there is a fault, the vibration in one plane can react faster in one plane verses the other. Having dual probes with nonvoting trip logic during start-up could result in tripping the machine sooner if there is a mechanical fault and again, limiting the damage. T
here are many different types of control systems and in some cases, the functionality of temporarily manipulating alarm and trip set points at start-up may be difficult to set-up. This may result in a higher cost system, more complex controls programming and may increase the potential for unnecessary trips during start-up. Having this functionality can lower the damage that is done if there is a problem during start-up. Each case needs to be evaluated on its own merits.
In the case study discussed above, several potential causes for the failure were reviewed. It was surmised that the short time the auxiliary oil pump was run prior to start-up was a major contributing factor to the failure. Compared to the previous start-up, the bearing temperature readings had not stabilized, leading to the conclusion that the rotating and stationary parts had not reached thermal equilibrium. The differential temperatures in the rotors, bearings and casing could have been significant enough to cause distortion and lead to hard rubs in the seal areas or impeller running fits which could have caused the failure. This was addressed by incorporating a mandatory longer run time for the auxiliary oil pump before starting the compressor.