Simulation effectively sites surge- relief facilities on Saudi pipeline

J. J. Dempsey, A. H. Al-Gouhi
Saudi Arabian Oil Co.
Dhahrain

• Pump station suction and discharge header piping on the AY-1L. [17,233 bytes]
• AY-1L's high-elevation point is near Yanbu as the line begins its descent from Gunsight Pass to Yanbu. [36,145 bytes]
• Loop construction on the AY-1L progresses near Pump Station 1. [16,736 bytes]

Based on a presentation to the 16th Annual Energy-sources Technology Conference & Exhibition, Jan. 31- Feb. 3, Houston.

Pipeline hydraulic and surge analysis studies of the Saudi Aramco East-West crude-oil pipeline assisted in expanding the system's capacity by 50%.

Surge studies predicted that operational upsets, such as the trip of a pump station, cause excessive surge pressures in the pipeline system at new flow rates. Additional surge studies showed that surge-relief stations must be located downstream from each of six pump stations.

The new surge-relief stations and an increase in capacity of existing surge-relief stations protect the pipelines at the higher flow rates.

Modeling the system

The Saudi Aramco system consists of two pipelines originating in the Eastern Province and terminating at the Yanbu crude-oil terminal.

Eleven pump stations transport the crude oil to Yanbu. The pump stations are similar except for infrastructure and other support facilities.

During design of the system, computer modeling techniques and proprietary software packages determined equipment to be added to the pipelines to protect their integrity and meet capacity requirements.

The software packages TCON, developed by Scientific Software Intercomp Inc. (SSI), Houston, and PSIM, developed by Stoner Associates, Houston, predicted surge pressures in the East/West pipelines.

TCON and PSIM solve two partial differential equations describing flow in pipelines: The first (Equation 1 [205,328 bytes]; equations box) is the continuity equation; the second (Equation 2) is the force-balance equation.

The equation of state is given by Equation 3.

Bulk properties of density, viscosity, and fluid bulk modulus of the crude oil as a function of pressure and temperature were provided by laboratory analysis and covered the operating range of the pipelines.

An hydraulics analysis included temperature effects associated with pumping the crude oil, pressure reduction through control valves, and heat loss through heat exchangers. The surge analysis incorporated constant temperature profiles provided by the hydraulics study.

TCON and PSIM account for the behavior of instrumentation and equipment subjected to transient conditions. Equipment simulated included: transfer lines, header pipe, pumps, control valves, relief valves, block valves, check valves, external source-delivery, sensors, controllers, actuators, and relays.

TCON and PSIM simulate pump performance using pump affinity laws relating flow, head, and horsepower, to shaft speed.

The physical data used to model pipes include length, diameter, pipe friction factor incorporating the Colebrook correlation, pipe wall thickness, Young's modulus, and a coefficient of thermal expansion for the pipe steel.

A transient thermal simulation was considered unnecessary because the time constants associated with changes in ambient conditions were large compared to the time period simulated. Yard piping and valves were grouped into a common category because they both exhibit similar resistance to liquid flows.

Crude oil supplied to the system and removed from the system was simulated with constant pressure devices to simulate supply tanks and receiving tanks.

Instrumentation commonly used in control systems was included in the model to account for flow control, fuel control, suction-pressure control, discharge-pressure override control, suction-pressure trip logic, discharge-pressure trip logic, and back pressure control.

The surge-analysis studies incorporated a design margin between maximum allowable operating pressure (MAOP) and the peak surge pressures. This margin ensured compliance with AS-ME/ANSI B31.4, the liquid pipeline design standard, which prohibits excursions above maximum allowable surge pressure (MASP).

A 2% design margin covered the following conditions:

• Variations in local elevation from data obtained from alignment sheets
• Variations in the final location of the surge-relief stations due to construction restrictions
• Set pressure drift of field instruments as a result of changes in ambient conditions
• Error tolerances in the computer calculations
• Inaccuracies in the model due to imprecise physical data; and,
• Instrument inaccuracies in the pump station trip logic.

The design margin resulted in lower relief set points, higher predicted flow rates for the surge-relief valves, and higher predicted spill volumes.

Hydraulics analysis

Fig. 1 [51,559 bytes] shows the hydraulic gradelines for the AY-LL (Abqaiq-Yanbu) pipeline. The vertical axis represents elevation in feet; the horizontal axis, distance in miles between pump station No. 1 and Yanbu crude-oil terminal. The pipeline elevation profile and a profile of the MAOP are superimposed on Fig. 1.

The hydrostatic gain through each of the 11 pump stations and the hydrostatic loss through the pressure-reducing stations are clearly shown.

The hydraulics study revealed that the AY-1 and AY-IL pipelines can transport the design flow rates between Pump Station 1 and Yanbu crude-oil terminal.

At Pump Station 6, a portion of the crude oil is diverted to air coolers under summer conditions to limit crude-oil temperature.

The back pressure controls at pressure-reducing stations maintain a positive pressure at the low-pressure point. The pressure set points depend on flow which is measured at Pump Station 11 and transmitted to the pressure-reducing stations (PRSs).

An override circuit at PRS-1/PRS-IA stations compares the low-pressure set point with a pressure signal fed forward from a pressure sensor located at the low-pressure point.

If pressure, monitored at the low-pressure point, falls below a minimum value, the control circuit causes the pressure-control valves to close to increase the back pressure at the pressure-reducing stations and raise pressure at the low-pressure point.

A pipeline test conducted on the AY-IL pipeline at a flow rate of 72% of design identified differences between steady-state pressures predicted by TCON and pressures obtained from field measurement.

Overall, there was good agreement between TCON's predicted pressures and the field measurements. The maximum difference between predicted pump discharge pressure for the 11 pump stations and measured pump discharge pressure was approximately 2%.

Surge analysis

Operational upsets studied during surge analysis included trips of pump units, trips of pump stations, and closure of pressure-control valves at the pressure-reducing stations.

To guarantee that the pipeline system cannot be completely shut-in, a design condition establishes that motor-operated isolation valves at the battery limit of each pump station and on the bypass line around each pump station be locked open and that the power be disconnected.

Fig. 2 [51,851 bytes] shows the stabilized hydraulic profile for the AY-1L pipeline following a trip at Pump Station 11, Pump Stations 1 through 10 are operating at or close to their high discharge-pressure limits.

The flow rate in the pipeline reduces from 3 million b/d to 56% of design flow rate.

Fig. 3 [49,720 bytes] shows the stabilized hydraulic profile for the AY-1L pipeline following a trip at Pump Station 5. Pump Stations 1 through 4 are operating close to their high discharge-pressure limits.

Pump Stations 6 through 11 are operating at or close to their low suction-pressure limits. In general, following a trip at a pump station, upstream pump stations operate on high discharge-pressure override and downstream pump stations operate on low suction-pressure override.

The flow rate in the pipeline reduces to 54% of design flow rate.

Acoustic transit times

An upset in a pipeline propagates from the source of the upset at the acoustic wave speed. This speed depends on three properties: mass density, the properties of the pipe, and the method of anchoring the pipe.

The relationship for a pipe which is anchored throughout is defined in Equation 4.

An analysis using Equation 4 reveals that the acoustic transit time between pump stations is short. It will be difficult for the pump station operators at adjacent pump stations to take corrective action after the pressure surge has arrived at the pump station.

They must react quickly to sudden changes in suction and discharge pressure. If they are unsuccessful, upstream pump stations will trip because of high discharge pressure, and down-trip due to low suction pressure.

Relief-valve simulation

TCON and PSIM contain a standard mathematical model of a Grove Flexflo relief valve.

The Grove relief-valve coefficients were adjusted during the study to provide the Cv flow coefficient characteristic, opening speed, and opening characteristic of a Danflo relief valve.

The Danflo is designed to modulate while tracking the incoming surge pressure. The valve opens approximately linearly with rising pressure. Anticavitation protection was not required on the surge-relief valves because the period of actual operation is infrequent and short.

Using cavitation trim was avoided to reduce the possibility of blockage in the trim.

TCON and PSIM use Equations 5 and 6 to model the Grove Flexflo Model G887 relief valve.

The Danflo Bulletin 400-40 uses Equation 7 to represent the Danflo relief valve. Substituting the Danflo flow factor (F) in Equation 7 into Equation 5 yields Equation 8.

Because the Danflo valve is linear up to the fully open position, the coefficient b = 1. Typically, the relief valve is fully open at MAOP.

At full open, Q₁ = Q₂ and Equations 5 and 6 are expressed as Equation 9 which yields the adjusted valve coefficient (C).

An adjustment (Equation 10) is required to account for the different values of the Grove Flexflo Model G887 valve-recovery coefficient (C_VG) and the Danflo recovery coefficient (C_VD).

The accuracy of the Danflo valve simulation was verified by comparing the flow rate and pressures computed by TCON with the theoretical Danflo flow rates and pressure calculated from Equation 7. The results, as Fig. 4 [40,341 bytes] shows, closely agree.

Surge-relief protection

A surge analysis revealed that a trip of a pump station or closure of control valves at the pressure-reducing stations greater than 70% of design flow rates produces excessive transient pressures in the pipelines.

Fig. 5 [47,739 bytes] shows the peak surge pressures predicted for each pipeline section of the AY-1 L pipeline following a trip at Pump Stations 1 and 11. Surge pressures would exceed both MAOP and MASP if surge-relief systems did not operate to protect the pipelines following a trip at many of the pump stations.

Fig. 5 reveals that surge pressures in each section of the pipeline are less than MAOP following a trip at Pump Station 1. The other 10 pump stations will trip on low suction pressure because of a rarefaction wave which propagates to each downstream pump station.

Fig. 5 also shows that a trip at Pump Station 11 is highly disruptive to pipeline operations. Peak surge pressures are greatest, and MAOP is exceeded in all sections of the pipeline.

Pump Station 10's discharge pressure increases suddenly as the surge pressure is propagated upstream from Pump Station 11 to Pump Station 10.

Rapid corrective action is required to reduce pump station discharge pressure to prevent a cascading series of events from causing excessive pipeline pressure in each pipeline section and high discharge-pressure trips at the upstream pump stations.

Surge-relief stations

From the initial surge studies that have been described, designers implemented a method to achieve the following:

• Reduce the number of surge-relief stations.
• Determine their optimum locations to protect the integrity of the pipeline.
• Reduce long-term maintenance requirements resulting from installation of the surge-relief stations.

When this phase of the study began, the number of relief stations and their locations and capacities were unknown.

Operational upsets were simulated with surge-relief facilities deactivated and activated. The first cases analyzed surge pressures resulting from a trip of Pump Station 11, near the western end of the pipeline system. Highest peak transient pressures occurred after a trip of Pump Station 11.

Transient pressures were within MAOP between Pump Stations 10 and 11.

Similarly, a trip at Pump Station 10 does not result in excessive pressures between Pump Stations 9 and 10. Surge-relief facilities are not required in these sections of the pipeline.

Further analysis for the remaining pump stations revealed that trips at each pump station would result in upstream pressures exceeding MAOP and MASP. The initial surge analysis identified seven pipeline sections which required surge protection.

Station locations

Consideration of two options determined the location of each surge-relief station:

• The first locates the surge-relief stations at the suction side of pump stations.

Intuitively, this is the preferred location because the relief station is located at the source of the pressure surge. The relief valves can react immediately to divert flow and reduce the magnitude of the surge pressures.

&3149; The second option positions the surge-relief stations downstream of selected pump stations at the locations where the highest transient pressures are predicted to occur.

The accompanying box shows the advantages and disadvantages of both options.

The design team agreed that remote location of the surge-relief stations was safer and easier for long-term operation of the pipelines. The team incorporated a design margin between the local steady-state pressure at the relief station and the normal discharge pressure of each upstream pump station.

The surge-relief station does not infringe on the normal operating window of the pipeline system. Operational changes at design conditions will not cause the surge-relief valves to operate.

Only severe changes at or near design conditions, such as a trip of a pump unit or pump station will cause the surge-relief valves to operate to protect the pipelines.

A control strategy is used when pump stations are bypassed to reduce discharge pressure override set points and prevent local pressures at the relief stations from exceeding the relief-pressure set points.

This strategy prevents inadvertent operation of the relief valves at reduced pipeline flow rates due to lower friction losses between the upstream pump station and the remote relief station.

At the completion of this phase of the study, it was concluded that six surge-relief stations are required to protect the pipelines between pump stations.

Simulation results

Fig. 6 [190,601 bytes] displays typical pressure and flow-time responses at two surge-relief stations. The stations are activated following a trip at Pump Station 11 on the AY-1L pipeline system. The pipeline is operating at design flow rates.

A surge pressure wave propagates upstream of Pump Station 10 following the trip at Pump Station 11.

The discharge pressure at Pump Station 10 increases until pipeline pressure exceeds the discharge-pressure trip set point causing Pump Station 10 to trip.

A cascading series of similar events causes all other pump stations to trip. During these events, the surge-relief stations operate to protect the pipelines from exceeding MASP.

Figs. 6a-c display a time response of upstream pressure, flow, and spill volume for the surge-relief station located downstream of Pump Station 7.

Fig. 6a shows that pipeline pressure at the surge-relief station suddenly increases following the trip of Pump Station 11. Pipeline pressure exceeds the pressure-relief set point, and the surge-relief station operates to limit pipeline pressure at the relief station.

Fig. 6b shows that the surge-relief valves open to relieve pipeline pressure and divert crude to the relief-station tank. Fig. 6b also shows that the relief valves open a second time after they reseat because of a secondary pressure-surge wave.

Fig. 6c shows the resulting spill volume as it accumulates in the surge-relief tank.

Figs. 6d-f display a time response of upstream pressure, flow, and spill volume for the surge-relief station located downstream of Pump Station 1.

Fig. 6d shows pipeline pressure at the surge-relief station increases following the trip of Pump Station 11. Pipeline pressure exceeds the pressure-relief set point, and the surge-relief station operates to limit pipeline pressure at the relief station.

The surge-relief valves operate to relieve pipeline pressure and divert crude to the relief-station tank.

Fig. 6f shows the total crude diverted to the surge-relief tank.

Tank sizing; valve testing

The surge-analysis studies called for six new remote surge-relief stations installed downstream from Pump Stations 1, 2, 3, 5, 6, and 7 to protect the East/West pipelines.

The tank-sizing criteria to account for oil-spill volumes derived from the larger value of the following methods:

• Two surge-relief events per pipeline plus 20%; or,
• One spill from each pipeline, plus one 6-in. surge-relief valve stuck open, plus 20% of this total.

The total time to close the stuck open surge-relief valve derived from the time to react to the alarm at the control room of the pump station, the travel time to reach the remote surge-relief station, and the time to locate and isolate the stuck-open valve.

Discussions with the relief-valve vendor indicated that this is a conservative design basis because the most likely cause of a relief valve to fail open is debris stuck in the seat of the valve. Experience indicates, according to the vendor, that such a failure may result in 1% leakage.

A Danflo surge-relief valve performance test was held at Saudi Aramco facilities to verify the vendor's published data under high pressure-drop conditions.

Cavitation and valve-recovery coefficients were calculated for the axial flow valves with and without anticavitation trim.

The first test used a prototype 12-in. Danflo valve with two-stage, anticavitation trim. The second test used a 6-in. Danflo valve with standard trim. The prototype valve was tested at Daniel's flow test facilities under ANSI/ISA S75.02 flowing with water at low pressure drop.

For the purposes of the test, the Danflo valves replaced an existing surge-relief valve located at PRS-2A pressure-reducing station.

The test yielded an average flow coefficient which was 85% of the rated flow coefficient. Calculations show that the reduced capacity resulted from the valve's being operated at greater than choked flow conditions.

During the second test, the 12-in. prototype valve was replaced with a 6-in. valve with standard trim. The results of this test were inconclusive because of the valve's higher capacity.

At valve openings greater than 50%, severe vibration occurred on the valve and the pressure-relief station's skid. Precise positioning of the valve plug was impossible.

The valve was tested at approximately 50% open, however, and the measured flow rate exceeded design requirements by 72%.

Another study

An additional study, not described here, identified downgrade strategies implemented by supervisory control and data acquisition (scada) and DCS computer systems monitoring and controlling the pipeline system.

The downgrade strategies maximize system throughput by changing pump discharge and suction pressures and fuel set points at all pump stations. The new set points are issued whenever a trip occurs at a pump station and flow rates exceed 70% of design throughput.

The downgrade strategies minimize disruption to flow caused by operational upsets and maximize pipeline throughput.

Acknowledgments

The authors thank Saudi Aramco for permission to publish this article and engineers at Brown & Root Braun and Fluor Daniel who participated in phases of this study.

THE AUTHORS