OR WAIT null SECS
As the oil and gas reservoirs are ageing, the trend in the North Sea shows a gradual decline of reservoir pressure. This has initiated numerous upgrade, revamp and retrofit projects, which involve rotating equipment such as motors, gears, and centrifugal compressors.
In these projects, it may not be possible to perform complete string tests at OEM facilities; hence prediction of rotordynamic behaviour becomes vital in ensuring smooth commissioning and operation.
As the demand for higher pressure and power is increased, revamps often include compressor modifications on the existing skid, baseplate, or piping, thus placing constraints on the machine layout. This may lead to sub-optimal and non-standard design choices, which in turn increase the importance of accurate prediction of the rotordynamic properties (Figure 1).
Most critical turbomachinery is purchased according to the American Petroleum Institute (API) standards which contain general guidelines for buyers, including datasheets, terminology, definitions, guidelines on rotordynamic analysis to be performed by the vendor, Undamped Critical Speed Map (UCSM), synchronous unbalance response analysis, stability analysis and shop test vibration limits. Some manufacturers make exceptions to the API and some buyers specify additional requirements (such as the Shell Design and Engineering Practice — DEP).
However, there are certain limitations to the API standards. Since they are general standards, the guidelines are quite conservative to cover a large span of machinery. This can have implications on rotordynamic analyses of newly designed units. In particular, the following design features of machinery should be treated with precaution: advanced machinery elements, frequency range, manufacturing process, modelling assumptions and large machines with stator interaction.
As pressure ratios have increased, the hole pattern seal has gained popularity due to its sealing properties and positive rotordynamic effects for high-pressure centrifugal compressors. API 617 states that a rotordynamic stability analysis should “. . . include the dynamic effects from all sources that contribute to the overall stability of the rotating assembly as appropriate . . .” , but information on how advanced machinery elements should be considered in the analysis is limited. It is therefore required that the contract be well defined with clear guidelines on how these extended analyses should be addressed.
For shop and string testing purposes, the frequency range defined in API should also be scrutinized. The definition in API regarding the non-synchronous (outside operating range) frequency range states that “. . . data shall cover 0.25 – 0.80 times the maximum continuous speed . . .” In this range, any discrete peak in the vibration spectra shall not exceed 20% of the acceptance level.
However, research and practical experience have established that many rotordynamic phenomena occur at frequencies lower than 0.25X, such as diffuser stall or bearing flutter (Figure 2). It is therefore recommended to increase the range to 0.05 – 0.8 to ensure that the low frequency content is captured. If the root cause of the source is known, it could also be possible to accept levels above 20%.
Due to the increased demand for natural gas, many North Sea projects rebuild the offshore process to allow higher gas flow at lower pressures. An example of this is a recent project where two compressor trains were revamped to increase speed and power within an existing casing.
This was realized by manufacturing new gear rotor sets with higher gear ratio, i.e., by modifying the number of gear teeth. During commissioning offshore, both the main and spare gear were damaged, which led to delays and immense costs for the operator.
The Root Cause Analysis (RCA) that followed concluded that the damage was due to surface contact fatigue caused by poor manufacturing of the gear. The final grinding of the gear was not done properly, which led to residual tensile stress and reduced hardness of the gear flanks. The marginal design of the rotor could not tolerate such deficits; hence high loads rapidly wore out the teeth. This example stresses the need for an increased focus on the manufacturing process — even during procurement (Figure 3).
When performing rotordynamic predictions, API standards often require the analysis to include several scenarios to take into account all possible excitation sources. However in some cases, the modelling assumptions are not sufficient to get the full picture.
Referring for example to torsional rotordynamic analysis of VFD (Variable Frequency Drive) driven rotor trains, API requires at least 10% separation margin between the torsional natural frequencies and any excitation sources. These include running speed(s), one and two times electrical line frequency, harmonic and inter-harmonic torsional excitation from the electric motor drives.
While it is well known that VFDs can be a source of inter-harmonic torsional excitation within the driven equipment, the DC link should provide decoupling of frequency components carried through the current between the motor side and the line side.
The DC link “decoupling power” is in reality finite, thus frequency components can travel through. This effectively links the mechanical system’s torsional dynamic to the grid which the train is connected to. In a typical offshore installation, the grid itself is small and considered to be “weak.” As a result, it effectively constitutes a bridge where dynamic excitation can easily travel and where subharmonic frequencies can easily couple.
During commissioning of new VFD motors, numerous instances have been observed where turbogenerators have tripped due to high vibration — without apparent reason. In these cases, troubleshooting is difficult because the real nature of the problem is often misunderstood or overlooked. However the problem could easily be avoided at the design stage by performing dedicated simulations to study the effect of introducing a new motor drive on the existing rig setup.
Another trend in the North Sea is replacing original electric motors with new and more energy-efficient models. When performing these rebuilds, it is imperative to perform structural and rotordynamic calculations to assess the effects the new motor will have on the system. Presumably, the new motor has a different mass both on rotor and stator, which may adversely affect the dynamic behaviour. At worst, structural natural frequencies can produce excessive vibration levels and thus generate subsequent operational difficulties. For the operator, these problems can be minimized through such actions as:
• Measurements before the redesign begins to map the dynamics of the existing skid and driver
• Finite Element (FE) modelling of the existing skid which is tuned to the above measurements
• Rotordynamic calculations on new motor to locate critical speeds of the rotor and bearing system
• FE calculations of the new motor on the old skid to assess possible skid natural frequencies
• Extraction of impedance terms, i.e., stiffness and damping of the new bearing pedestals on the old skid
• Rotordynamic calculations with the above impedance terms used together with the rotor and bearing system to determine how it affects the motor
• Measurements of the new motor on the old skid during commissioning activities to verify calculations and to rectify premature issues before the unit is handed over to operations.
To the operator, the above actions are always a good investment and generally constitute a comparatively small cost in a rebuild budget. Further, if performed by third parties, it will yield unbiased results which can help both operators and vendors reach successful handover to operations.
Staffan Lundholm is a Senior Consultant for Energy, Machinery Dynamics at Lloyd’s Register Consulting, a company which provides risk management and engineering dynamics services.
Stefano Morosi is a Consultant for Energy, Machinery Dynamics at Lloyd’s Register Consulting. For more information, visit www.lr.org/consulting or contact firstname.lastname@example.org.