February 2019

Maintenance and Reliability

Resolving vibration problems in a crude booster pump

The pumps referred to in this article are 2,600-HP vertical pumps utilized as booster pumps for the shipment of crude oil.

The pumps referred to in this article are 2,600-HP vertical pumps utilized as booster pumps for the shipment of crude oil. Each suction tank has two pumps (pump A and pump B) acting in parallel and receiving flow from a 60-in. pipe header (FIG. 1).

FIG. 1. Layout of the crude booster pumps.
FIG. 1. Layout of the crude booster pumps.

 

The pump suction and discharge pipes are 42 in. The pumps are rated at 45,000 gallons per minute (gpm), with a minimum flow requirement of 12,000 gpm. Each pump has a separate 16-in. recycle line that returns to the 42-in. suction line. The pumps operate at 890 revolutions per minute (rpm) and are driven by electric motors.

High vibrations caused frequent damage to the concrete pipe supports, as shown in FIG. 2. They also led to frequent pipe cracks at instrument line connections.

FIG. 2. Damaged pipe supports in the pump recycle line.
FIG. 2. Damaged pipe supports in the pump recycle line.

 

Vibration survey

Vibration data was measured using a vibration analyzera and recorderb (TABLE 1).

 

Vibration levels were highest at location AO (orifice flange) and AV (suction vent), with levels close to 1 in./sec RMS. Vibration near PZV at a frequency of 16.4 Hz was 0.47 in./sec, which is in the correction zone (FIG. 3). This PZV had the most recent pipe crack. Vibration levels with a closed motor-operated valve (MOV) (pump A) were much lower.

FIG. 3. Piping vibration severity chart (EDI report 85-305).
FIG. 3. Piping vibration severity chart (EDI report 85-305).

 

The vibration recorder was used to record vibration before and after opening the recycle line’s MOV valve. Vibration levels changed from a 1-g peak to a 35-g peak when opening the MOV for pump A (FIG. 4).

FIG. 4. Recorder acceleration data near the orifice of pump A.
FIG. 4. Recorder acceleration data near the orifice of pump A.

 

For pump B, the vibration near the orifice plate reached 17 g, even with the MOV closed (FIG. 5). This indicated that liquid was possibly flowing through the MOV.

FIG. 5. Recorder acceleration data near the pump B orifice.
FIG. 5. Recorder acceleration data near the pump B orifice.

 

Flow measurement

Since MOV passing was suspected, an ultrasonic flow measurement was conducted on the pump B recycle line. The flow measurement showed that 9,450 gpm was passing through the closed-recycle MOV during the operation (TABLE 2). This measurement was 27% of the pump flow, indicating a significant energy loss due to the passing recycle MOV valve, causing both energy loss and high piping vibration due to cavitation.

 

Energy losses

The loss in horsepower is even more severe since efficiency varies with flowrate. The recycle valve passing has caused both pumps to run when one would be enough to deliver the required flow of 74 bph (15% above rated flow) and at higher efficiency. The energy loss when using two pumps would be 55% (2,412 + 2,422 – 2,288)/(2,412 + 2,424), leading to an additional cost of $775,000/yr for one set of two pumps (TABLE 3).

Due to the high passing rate, cavitation is expected at the recycle MOV, which acts as an orifice when the MOV is closed. Since this flow is always present, high vibration is expected and was confirmed by vibration readings in the earlier study. This high vibration will cause a pulsating flow, which will excite the piping acoustic and mechanical resonances. Pulsation was evident when observing the pressure gauge needle oscillations of over 5 psi at suction and discharge pipes. The associated vibration is the leading cause for pipe support damage and small branch connection failures.

Due to the significant passing in the recycle MOV, the following was recommended:

  1. Perform a bypass MOV passing survey for all 10 crude booster pumps
  2. Inspect all MOVs showing excessive passing rates. 

The benefits of the two recommendations were as follows:

  • Identify passing rates on all MOVs to prioritize which ones should be inspected first, based on severity
  • Optimize pump system efficiency by identifying which pumps had zero or minimal passing rates to determine whether pump A or pump B would provide better system flow, based on the pump with less passing rate across the bypass MOV
  • Inspect the MOV to determine the failure mode and mechanism resulting in excessive passing, as identifying the failure mechanism would possibly lead to a design or material change to improve the life of the valve shut-off integrity.
FIG. 6. Hydraulic simulation using modeling software.<sup>c</sup>
FIG. 6. Hydraulic simulation using modeling software.c

 

Hydraulic simulation

The hydraulic system was modeled using software.c Several scenarios were performed to study the system under various flow conditions (FIG. 6). Orifice k-factors were calculated based on literature.2 In the software used, a quadratic function was fitted to the pump curve in the expected flow region. A key parameter in evaluating cavitation was the cavitation index, as defined by Eq. 1.

             (1)

where:

Pu = Upstream pressure at the orifice

Pd = Downstream pressure at the orifice

Pv = Crude oil vapor pressure assumed at 4.6 psia or –10.1 psig.

The cavitation level was acceptable for a cavitation index larger than 1, and values larger than 1 were often needed. The cavitation index depended highly on pump suction pressure, which determines the pressure downstream of the orifice. A low value of –2 psig was assumed, which was used to design the orifice. During full recycle, the flow from the tank was zero; hence, there were no piping losses from the tank. The piping losses from the tank to the pump suction were estimated at 4 psi–5 psi. As per the terminal instruction manual, the pump would not start if the suction pressure was less than 5 psig. For this reason, a 5-psig suction condition was used for the full recycle operation.

It was observed from the system simulation table that the cavitation index, when using a single-stage orifice, varied between 0.2 for a full recycle and 0.13 for a low pump flow of 7,000 gpm (TABLE 4). The 7,000 gpm produced the worst cavitation and was therefore used for designing the multistage orifice. The passing MOV also caused a low cavitation index, close to full recycle cavitation.

 

The multistage orifice will cause less pressure recovery to avoid cavitation. The pressure is reduced in stages to avoid going below the liquid vapor pressure, which causes vapor to form. This vapor will collapse as the pressure is recovered a short distance downstream, and this collapse will cause high noise and vibration. FIG. 7 shows the effect of the multistage orifice system. The staged pressure drop will make it less likely that the discharge pressure falls below the liquid vapor pressure. The same concept is used in anti-cavitation valves.

FIG. 7. Effects of the multistage orifice system.
FIG. 7. Effects of the multistage orifice system.

 

A multistage orifice solution was simulated using US Nuclear Regulatory Commission (NRC) NUREG 6031. The system required at least four orifice plates, each with many slotted holes. The flow scenario with the most critical cavitation was used to design the orifice. Other scenarios were checked and found to be acceptable, as well. Once installed, this multistage pressure-reducing system decreased noise levels significantly below 70 dB.

When designing multistage orifices, the thickness is important to minimize fatigue-induced failures (FIG. 8). The distance between stages is needed to prevent vapor collapse on the downstream plate and to allow for pressure recovery. A distance above 1 diameter is needed to prevent damage on a downstream orifice, and 2 or more diameters are needed for full pressure recovery. The number of holes should also be evaluated based on the gap between the holes. A very small gap may lead to early fatigue damage. For this reason, the last orifice plate stage was selected as a 7-hole stage, compared to 19 holes for the other stages.

FIG. 8. Sketch of the proposed multistage orifice.
FIG. 8. Sketch of the proposed multistage orifice.

 

Takeaways

The system was found to have serious cavitation issues at the orifice plate, as well as at the passing MOV valves. The passing MOV causes fluid cavitation, resulting in severe piping accelerations, which could damage the piping supports and excite small branch piping attachments, causing fatigue failures. The cavitation will excite piping mechanical and acoustic frequencies, causing vibrations away from the pump, as well. Passing MOV and using recycle during flows between 20 Mbph and 40 Mbph can lead to serious energy loss.

Recommendations

Summarized are recommendations for eliminating vibration and ensuring safe operation:

  • Perform ultrasonic flow measurements on all pump recycle lines.
  • Inspect and correct passing recycle MOV valves.
  • The current bypass line should be checked for erosion, especially at the bend and pipe after the orifice at the nine o’clock position.
  • Install permanent ultrasonic flowmeters on recycle lines to monitor valve passing. This approach is preferred to the ultrasonic leak detector, which will not be able to quantify the passing rates.
  • Modify the operating procedure so that the recycle MOV is closed at a lower flowrate to minimize recycling, while keeping pump vibration in the acceptable range. This point was identified as a potential serious energy loss.
  • Replace the single orifice with a multistage orifice, as outlined under the hydraulic study.
  • Check fluid temperatures during operation to ensure that the recycle operation will not adversely increase the temperature above the 82°C (180°F) limit.
  • Install special clamps with damper material on the bypass line to reduce vibration further.
  • After installing the multistage orifice plate, conduct vibration surveys on the pump, piping and small branch connections to ensure that vibrations are within limits. HP

ACKNOWLEDGMENTS

A special thanks to Saudi Aramco maintenance engineer Muhammad Qureshi, process engineers Syed Nisar and Ghulam Farooq, valve engineer Omar Amri, piping engineer Victor Giron Valencia, and structural engineering specialist Emad Abu Aisheh for their support. Valuable discussions with senior consultant Nabeel Al-Odan and pump engineer Amer Dhafiri were very useful.

LITERATURE CITED

  1. Wachel, J. C., et al., “Vibrations in recprocating machinery and piping systems,” EDI Report 85-305, Engineering Dynamics Inc., San Antonio, Texas, 1985.
  2. Idelchik, E., Handbook of Hydraulic Resistance, Begell House, Inc., New York, New York, 2008.

NOTES

a Refers to Emerson’s CSI 2140 vibration analyzer

b Refers to Mide’s Slam Stick X vibration recorder

c Refers to Bentley’s PlantFLOW software

The Authors

Related Articles

From the Archive

Comments