Case Study II: Surface line and gas-oil separator optimization for a Venezuelan heavy oil field (Hamaca)
Engagement Background: The Venezuelan Hamaca oil field was one of four major projects underway in the Orinoco belt which involved the extraction of extra-heavy crude and bitumen deposits. The 9° API extra-heavy Hamaca crude oil was diluted with naphtha to 16° API in order to lower its viscosity and enable the fluid to flow through the producing system.
Starting from an initial design supplied by the Hamaca partners, PDVSA, Chevron, Texaco and Conoco Phillips, EnSys Yocum was engaged to rework the original design plan and investigate operating feasibility, safety and stability based on a 36 inch diameter well-gathering trunkline flowing to a manifold which divided the production between seven low pressure first stage separators, followed by large 2nd-stage surge tanks operating at atmospheric pressure. EnSys Yocum simulated the system at the initial year one production target of 90,000 bpd and at a year three forecast production target of 190,000 bpd.
Use of EnSys Yocum Upstream Simulators: Production Facilities Simulation (PRODSIM) was employed for the Hamaca trunkline system to identify flowline issues and assess how the system could be optimally expanded to handle a number of wells flowing into a 17 mile long 36 inch diameter trunkline. PRODSIM was employed to calculate the system pressure and temperature profiles along the 36 inch line, originating at multiphase pumps located at the well site, flowing through the trunkline, to a riser and into two 24 inch diameter looped lines, which acted to divide flow and reduce pipeline velocity before entering the manifold. The manifold further divided the multiphase flow between seven first stage separators operating at pressure of 60 psig followed by second stage surge tanks at atmospheric pressure. To assess the effect of potential GOR increases with time, EnSys Yocum evaluated several scenarios by case study analysis, initially 90 scf/barrel GOR, increasing to 250 scf/barrel GOR, and eventually reaching 475 scf/barrel GOR, reflecting more wells drilled (ultimately 120 in total) and declining reservoir pressures.
An important factor in re-working the initial design was to assess the viability of increasing production by reducing wellhead pressure. A key to achieving this was to reduce the well head multiphase pump discharge pressure and corresponding pump suction pressure. Although increasing the back pressure on the 36 inch line could be considered counter intuitive, the calculated net effect was to reduce the trunkline pressure drop by moving the flow regime out of slug flow. The effects of doing this were greatest for the low GOR case. Nevertheless, the recommended EnSys Yocum re-work plan was balanced to consider the initial production conditions in conjunction with the long term conditions as the Hamaca field was further developed.
EnSys Yocum Gas-oil Separator Simulation (GOSPSIM) was employed for the gas-oil separator calculations for the high viscosity Hamaca crude oil based on estimates of the gas bubble size distribution and liquid residence time. GOSPSIM calculated the cutoff gas bubble diameter based on the liquid residence time and terminal velocity of rising gas bubbles based on Stoke’s Law. The corresponding gas-carry-under is calculated based on the volume of gas bubbles which remain entrained in the liquid, in turn based on GOSPSIM’s calculation of the cutoff diameter— the diameter below which gas bubbles will carry under with the separator liquid bottoms.
A wide array of separator internals options are available in GOSPSIM that influence separator performance. These include an inlet diverter impingement plate of variable efficiency, downflow baffles, inlet cyclonic vortex tubes, gas outlet mesh pads, liquid zone perforated plates, wave plates and structured packing, as well as provision for an electrostatic coalescer and internal and external separator heaters. These equipment devices are simulated by GOSPSIM and impact the prediction of liquid carry over (LCO) and the fractional gas carry under. The separator volumes occupied by foam and emulsion layers are accounted for in the separator residence time calculations. Using the separator residence time calculations for the individual oil, gas and water phases, and the design basis droplet sizes specified, GOSPSIM calculated the effluent gas, water and oil qualities (water-in-oil and oil-in water) in response to the separator throughput rate, water cut and the fluid physical properties.
Slug Flow Calculations The prediction of slug flow in the multiphase trunkline inlet to the separators, and estimates of the gas and liquid slug size and frequency were key to sizing the Hamaca gas oil separators to ensure adequate volume to safely handle the approaching slugs. Burke and Kashou of Texaco conducted a review of published methods and concluded that the slug volume and frequency predictions that are most used in practice for large pipe diameters are the Brill/Scott correlations. These are contained in EnSys Yocum GOSPSIM Black Oil Model.
Based on a wealth of data collected, Brill’s correlation predicts the gas bubble and liquid slug lengths, their frequencies and the liquid content of the gas bubble and the gas content of the liquid slug. Brill’s correlations are dynamic in that the same information is predicted across a range of probability levels. GOSPSIM performs detailed calculations for the 2% probability and the 50% (mean) probability level for liquid slug and gas bubble lengths and frequencies. Brill’s data extends to flowlines up to 24 inches in diameter- analysis of Brill’s data by Scott indicated a zone of uncertainty when extended to even larger pipe diameters (to include the 36 inch Hamaca trunkline). Based on this consideration, EnSys Yocum multiplied the slug lengths and volumes predicted by Brill by a factor of 1.3 (Scott related the liquid slug lengths to pipeline diameter and the EnSys Yocum calculated factor was arrived at to maintain gas and liquid material balance across slug frequency.
Simulation Results and Recommended Design Basis: The specific EnSys Yocum findings which influenced the final design recommendation as submitted for the Hamaca flow system included:
1) Based on the calculated slug flow characteristics in the approach lines to the separator bank and with the full production rate evenly divided among seven separators. 10 x 60 foot separators in Hamaca could handle the mean and maximum slug flow volumes/frequency with feasible control valve lift reactions. The buildup is higher for the low GOR case because the liquid slugs are less gassy. However, this is offset by the increasing quantities of gas relative to the steady state flow that must be handled as the GOR increases.
2) Data provided by ConocoPhillips for the performance of separator vortex tube internals indicates satisfactory separator performance and the ability to handle sand when operating in conjunction with the installation of perforated de-foaming baffles.
3) The liquid flow surge rate increases by approximately a third relative to the separator steady state flow indicating an increase in the control valve lift. Control valve calculations indicate an increase from a normal 60% lift to approximately 70% for a 12 inch inlet control valve.
4) Relative to the steady state flow, the corresponding gas rate is predicted to increase by approximately a factor of 1.3 to 1.7 across the range of the low to the high GOR cases. This indicates an increase from a normal 60% control valve lift to approximately 75% for a 12 inch gas line control valve.
5) The rapid changes in the exit gas volumes and pressures from varying control valve lifts cause pressure pulsation in the downstream gas and were dampened by a recommended sized surge drum installed before the compressors.
6) The basis for the EnSys Yocum slug flow calculations is the slug unit, which is comprised of the liquid slug and gas bubble and which is in material balance with the steady state flow. The mean and maximum (2 percent probability) size slugs are both in material balance since the Brill correlation compensates for slug and bubble frequencies versus slug and bubble volumes to maintain material balance. Therefore, calculating the effects of the mean and maximum cases is deemed sufficient for determining the effects on separator rates for all combinations (probabilities) of slug sizes, their sequence and frequency variations.
7) Based on EnSys Yocum field experience in the Middle East, the gas residence times may be marginal for the higher GOR cases depending on the degree of crude oil foaminess, with ongoing foam tests being recommended. An adequately sized liquid knockout drum was therefore recommended.
8) The separator gas velocities remain below the recommended maximum 2.1 feet/ second for the separator conditions and below the mesh pad flooding velocity of 5-8 feet/second. This is consistent with the recommended installation of a 4 ft2 mesh pad.
9) Simulation runs indicated the need to install a shell and tube heat exchanger upstream of the separators to raise the fluid temperature to 130 degrees F. The flowlines are assumed to be coated and wrapped for a resulting overall heat transfer coefficient of 0.5 BTU/ ft2 /hour. This reduces the naphtha diluted crude viscosity from approximately 400 to 140 centipoises, with the temperature increase having a significant effect on reducing the GVF exiting the separator. The surge tank following the separator was assumed to have liquid entry at the top to eliminate any gas-carry-under in the surge tank exit liquid.
10) The separator horizontal liquid velocities are above the recommended limit of 0.1 feet/ second and gas entrainment in the liquid will likely be exacerbated by the arrival of liquid slugs. The variation in the liquid control valve lift position will cause variations in the liquid rates and pressures downstream creating pressure waves or surges in the exiting liquid, normally containing a gas fraction (GVF). The second stage atmospheric pressure cone roof surge tank was sized to handle these pressure surges, and thus avoids compromising the tank roof integrity.