Undesirable vibrations were observed in the wash oil line between two vessels during activation of the wash oil flow, most severe in the vertical section immediately downstream of the angle control valve. An earlier study had already identified slug flow as the most probable cause and proposed mitigation measures, and flow measurements at the site later confirmed that slug flow was occurring.
The probable cause was therefore established, but the urgency of the proposed restraints could not be judged, because the alternating stress amplitude actually occurring in the line, and the effect of that stress on fatigue life, had not been established. Operations also intended to raise the mass flow to around 5700 tonne/day.
Measurements and model calibration
To establish those stress levels, the line was instrumented. Accelerometers recorded displacement and frequency at six positions in the affected section, with the insulation removed, and readings were taken at two mass flows, 4800 and 4900 tonne/day. Larger differences in flow were not available at the time, and the change between the two was too small to give a clearly distinguishable vibration level. Measurements taken separately at the site over more or less the same period were kept for comparison, but only the contractor’s measurements were used as the basis for the assessment. Because the record covers a finite period, there is a small probability that higher maximum values were missed.
The measurements showed a dominant frequency of 8.4 Hz in the horizontal direction, against 4.5 Hz for the leading mode in the existing model. Two adjustments followed, based on the measured frequencies and on the support arrangement seen in the field. A guide and a rotational restraint were added near the Tee connection, where a guide support was present on the pipe but absent from the model and where piping on the dead end of the Tee had not been included. A guide function was introduced at the dummy support downstream of the angle valve to represent friction in the dynamic analysis.
The adjusted model returned modes at 6.5, 7.9 and 8.5 Hz in the direction of interest, and its displacement shape largely corresponds to what was measured. With frequencies and mode shape in line, the slug load calculated in the earlier study did not have to be changed, and the calculated displacements remained somewhat higher than the measured ones, but the calculation results are representative for the worst case.
Fatigue assessment
The assessment was made against Appendix 5 of Section VIII div 2 of the ASME B&PV code, with the highest stress amplitude arising at the branch of the Tee connection. Carried into the assessment as a stress range and assumed to act continuously at 8.4 Hz, it consumes the allowable number of cycles in 14 days. The measurement record shows the line does not behave that way. The largest displacement used in the assessment occurred once in a 518 second record, high amplitude vibration accounts for 1.5 million cycles per year rather than continuous cycling, on the assumption that the record is typical of the line’s working life, and the calculated displacement is larger than the highest measured value.
On the measured displacement, the expected fatigue life is about 120 years under the process conditions during the measurements, with an annual fatigue damage of 0.0075. The mass flow during the measurements was high in relation to normal operation. Displacements above 1 mm occur very rarely, so that figure is a reliable indication of reality, and the safety factor of 15 on cycles included in the fatigue curves has not been used in this assessment. One limitation is recorded. Because the history of the process conditions in the past is not known, the fatigue life already consumed could not be assessed.
Slug loads increase with the square of the flow increment, so the step from 4900 to 5700 tonne/day raises the calculated displacements and stresses by a factor of 1.35, and the expected fatigue life falls from about 120 years to about 6 years.
Conclusions and recommendations
The highest stress in the system under investigation occurs in the Tee connection, where the expected fatigue life is about 120 years under the process conditions during the measurements and about 6 years at 5700 tonne/day. When the mass flow is increased in the future, it is recommended to implement the measures advised in the earlier study: an X and Z restraint at node 305 and a Z restraint at node 1010, at the lower side of the dummy.
The same Tee carries an expected fatigue life of about 120 years at the flow during the measurements and about 6 years at the intended higher flow.
- Measurement set the timing of the mitigation. Calculation alone put the line in the range of days. Measured displacement and its frequency of occurrence put the expected fatigue life at about 120 years under the conditions during the measurements, and the restraints are recommended when the mass flow is increased.
- A stress result is only as good as the model behind it. The original model missed a support that was on the pipe and piping that was not included, and it was out by nearly a factor of two on frequency. The stress results were only used once the model reproduced the measured frequencies and mode shape.
- The trigger is the intended throughput increase. Because slug loads increase with the square of the flow increment, the intended increase to 5700 tonne/day carries the restraints with it, an X and Z restraint at node 305 and a Z restraint at node 1010.