Facilities and infrastructure

Hannah's facilities concept is on the critical path, and for good reason: at Spruce Ridge the facilities cost almost as much as the wells, and they decide whether the project can be built and operated safely with sour fluids. Sofia Reyes, who designed Prairie Crest's most recent sour battery, spends two sessions with her. Sofia's view is that the facility design at this stage has three jobs: to define what the battery must do, to size the main equipment well enough to estimate its cost, and to identify the decisions that cannot be changed later without great expense. This lesson follows those three jobs.

What the battery must do

A battery takes the wells' mixed stream of oil, water and gas and separates it into products that can be sold, reused or disposed of. At Spruce Ridge it must also deal with hydrogen sulfide in every stream, supply gas for gas lift, and, in options B and C, inject water. Hannah writes a design basis that states the rates, pressures, compositions and conditions the battery must handle over the life of the field, from the plateau with low water cut to late life with high water cut and gas lift in every well.

The process is conventional. Production from the wells enters an inlet manifold, where any well can be switched to a test separator to measure its oil, water and gas. The main stream goes to a production separator operating at a moderate pressure, where most of the gas leaves the liquid. The liquids go to a free-water knockout, which removes most of the water, and then a heater treater, which heats the remaining emulsion and separates it to sales specification. The treated oil goes to tanks, from which it is pumped into a pipeline. The water goes through a skim tank to remove residual oil and then to the injection pumps. The gas is scrubbed of liquid droplets and compressed, part of it for gas lift and the rest for the pipeline to the gas plant.

Because every stream is sour, the design has features that a sweet battery might not. All vessels and piping are sour-service materials. The oil tanks are connected to a vapor recovery unit, so that the gas that flashes from the oil in the tanks is compressed and recovered rather than released. H₂S detectors throughout the site are connected to an emergency shutdown system that can isolate the wells and depressurize the plant to the flare. The flare is designed to burn sour gas safely in an emergency, with a continuous pilot. The site is fenced, and its location is chosen with the emergency planning zone in mind, as Lesson 11 discusses.

Separator sizing

Separator size is set either by the gas, which must flow slowly enough that liquid droplets settle out rather than being carried over, or by the liquid, which must stay long enough for gas bubbles to escape and for oil and water to separate. The gas capacity is usually calculated with the Souders-Brown relation, which gives the maximum vapor velocity through the vessel in terms of the liquid and gas densities and an empirical coefficient K that depends on the vessel's internals:

The liquid capacity depends on the retention time, a few minutes for two-phase separation and more for three-phase separation of oil and water. Hannah sizes the production separator for the largest gas flow, which occurs late in life when gas lift is in full use, and for the larger of two liquid cases: the plateau with low water cut, and late life with a water cut near 80 percent. A 20 percent design margin is added to both.

# Facility sizing at Spruce Ridge: separator, water injection and the sour gas line to the gas plant.
import math
SI = False     # True: Canadian case (metric), False: US case (oilfield units)
if SI:
    oil, oil_late, wc_late, gor, lift = 320.0, 130.0, 0.80, 150.0, 50400.0   # plateau and late-life oil m3/d, late water cut, m3/m3, gas-lift gas m3/d
    p_sep, T_sep, Z = 1500.0, 313.15, 0.93                         # separator kPa(a), K
    Bo, Bw, res_p, inj_ii, depth = 1.45, 1.03, 30850.0, 1.2, 3090.0 # injectivity m3/d per kPa; depth m
    L_line, p1, p2, gg, T_line = 22.0, 4500.0, 3500.0, 0.82, 288.0  # km, kPa(a), K
    h2s = 0.022
else:
    oil, oil_late, wc_late, gor, lift = 1600.0, 650.0, 0.80, 900.0, 1.32e6  # bbl/d, scf/STB, scf/d
    p_sep, T_sep, Z = 215.0, 560.0, 0.93                           # psia, deg R
    Bo, Bw, res_p, inj_ii, depth = 1.48, 1.03, 6050.0, 9.0, 13000.0  # injectivity bbl/d per psi; depth ft
    L_line, p1, p2, gg, T_line = 13.0, 650.0, 500.0, 0.80, 520.0    # miles, psia, deg R
    h2s = 0.010
margin = 1.2                                     # design margin on flow rates

# 1. Gas capacity of a vertical separator (Souders-Brown) and liquid retention
if SI:
    gas_std = (oil * gor + lift) * margin        # m3/d at 101.325 kPa, 15 C
    gas_act = gas_std * (101.325 / p_sep) * (T_sep / 288.15) * Z / 86400   # m3/s at separator conditions
    rho_g = p_sep * 28.97 * gg / (Z * 8.314 * T_sep)   # kg/m3
    rho_l, K = 820.0, 0.107                       # m/s, with a mesh pad
    v = K * math.sqrt((rho_l - rho_g) / rho_g)
    d_gas = math.sqrt(4 * gas_act / (math.pi * v))
    liq = max(oil / 0.9, oil_late / (1 - wc_late)) * margin   # larger of plateau (10 percent cut) and late-life liquid, m3/d
    vol = liq / 1440 * 5.0                        # five minutes of retention, m3
    print(f"design gas {gas_std:,.0f} m3/d ({gas_act:.3f} m3/s at {p_sep:,.0f} kPa); gas density {rho_g:.1f} kg/m3; allowable velocity {v:.2f} m/s")
    print(f"minimum diameter for gas {d_gas:.2f} m; liquid design {liq:,.0f} m3/d needs {vol:.1f} m3 of liquid volume (5 min)")
else:
    gas_std = (oil * gor + lift) * margin        # scf/d
    gas_act = gas_std * (14.7 / p_sep) * (T_sep / 520) * Z / 86400          # ft3/s
    rho_g = p_sep * 28.97 * gg / (Z * 10.73 * T_sep)   # lb/ft3
    rho_l, K = 51.2, 0.35                         # ft/s, with a mesh pad
    v = K * math.sqrt((rho_l - rho_g) / rho_g)
    d_gas = math.sqrt(4 * gas_act / (math.pi * v)) * 12
    liq = max(oil / 0.9, oil_late / (1 - wc_late)) * margin
    vol = liq * 5.6146 / 1440 * 5.0               # ft3
    print(f"design gas {gas_std / 1e6:.2f} MMscf/d ({gas_act:.2f} ft3/s at {p_sep:,.0f} psia); gas density {rho_g:.2f} lb/ft3; allowable velocity {v:.2f} ft/s")
    print(f"minimum diameter for gas {d_gas:.0f} in; liquid design {liq:,.0f} bbl/d needs {vol:.0f} ft3 of liquid volume (5 min)")

# 2. Water injection for voidage replacement late in life
water = oil_late * wc_late / (1 - wc_late)
inj = oil_late * Bo + water * Bw                      # reservoir voidage per day, injected as water (Bw)
inj /= Bw
bhp = res_p + inj / 2 / inj_ii                   # two injectors share the water
grad = 10.1 if SI else 0.446                     # injected water gradient, kPa/m or psi/ft
whp = bhp - grad * depth + (1500.0 if SI else 220.0)   # plus tubing friction
pump_kw = inj * (whp * (1 if SI else 6.894757)) / 86400 / 0.70 if SI else inj * 0.158987 * whp * 6.894757 / 86400 / 0.70
print(f"late-life produced water {water:,.0f} {'m3/d' if SI else 'bbl/d'}; injection for full voidage replacement {inj:,.0f} {'m3/d' if SI else 'bbl/d'}")
print(f"bottomhole injection pressure {bhp:,.0f}, wellhead {whp:,.0f} {'kPa' if SI else 'psi'}; pump power at 70 percent efficiency {pump_kw:,.0f} kW")

# 3. Sour gas line to the gas plant: Weymouth equation (oilfield form), smallest standard size that carries the flow
sales = oil * gor * margin * (35.3147 if SI else 1.0)   # scf/d; gas-lift gas is recycled at the battery
P1, P2 = (p1 / 6.894757, p2 / 6.894757) if SI else (p1, p2)
Lmi = L_line / 1.609344 if SI else L_line
Tl = T_line * 1.8 if SI else T_line
for d_in, nps in ((2.067, "2 in"), (3.068, "3 in"), (4.026, "4 in"), (6.065, "6 in")):
    q = 433.5 * (520 / 14.7) * ((P1 ** 2 - P2 ** 2) / (gg * Tl * Lmi * Z)) ** 0.5 * d_in ** 2.667
    if q >= sales: break
print(f"plant gas {sales / 1e6:.2f} MMscf/d" + (f" ({sales / 35.3147:,.0f} m3/d)" if SI else "") + f": {nps} line (inside diameter {d_in} in) carries {q / 1e6:.2f} MMscf/d")
s_t = sales / 35.3147 * h2s / 23.645 * 32.06 / 1000 if SI else sales * h2s / 379.5 * 32.06 / 2204.6
print(f"sulfur in the plant feed: {s_t:.2f} metric tons per day" if SI else f"sulfur in the plant feed: {s_t:.2f} metric tons per day")

design gas 3.31 MMscf/d (2.62 ft3/s at 215 psia); gas density 0.89 lb/ft3; allowable velocity 2.63 ft/s
minimum diameter for gas 14 in; liquid design 3,900 bbl/d needs 76 ft3 of liquid volume (5 min)
late-life produced water 2,600 bbl/d; injection for full voidage replacement 3,534 bbl/d
bottomhole injection pressure 6,246, wellhead 668 psi; pump power at 70 percent efficiency 43 kW
plant gas 1.73 MMscf/d: 3 in line (inside diameter 3.068 in) carries 1.79 MMscf/d
sulfur in the plant feed: 0.66 metric tons per day

The design gas rate is 3.31 MMscf/d, which at 215 psia is 2.62 ft³/s of actual gas. With a mesh-pad coefficient of 0.35 ft/s, the allowable velocity is 2.63 ft/s, and the minimum diameter for gas is 14 inches. The liquid case governs: 3,900 bbl/d of total liquid needs 76 ft³ of liquid volume for five minutes' retention. Hannah selects a horizontal three-phase separator of 60 inches diameter and 16 ft length, which gives that volume with room for the oil and water layers, and notes that the gas capacity has a large margin, useful if the gas-lift demand turns out higher than estimated.

Water injection

In options B and C, the battery must inject enough water to replace the reservoir voidage. Late in life most of that water is produced water, recycled; early in life, when little water is produced, it needs a source of make-up water, which at Spruce Ridge will be a water source well completed in a deeper saline aquifer that is not used for drinking water. Using saline water rather than fresh water for injection is a strong expectation of regulators in both jurisdictions and an obvious point for the community.

Late in life, with 650 bbl/d of oil at an 80 percent water cut, the battery handles 2,600 bbl/d of produced water and must inject 3,534 bbl/d to replace the voidage completely. The reservoir is close to hydrostatic pressure, so the column of water in the injection wells does most of the work: the required bottomhole pressure of 6,246 psi needs a wellhead pressure of only about 670 psi, and the injection pumps need about 43 kW. Injection at Spruce Ridge is cheap to run. The expensive parts are the injection wells, the water source well and the water treatment needed to keep the formation from plugging.

The low wellhead pressure also has a safety advantage: the injection system can be designed to a modest pressure rating, and the injection pressure stays far below the pressure that would fracture the reservoir or its seal. Hannah checks this against the fracture gradient from the discovery well's leak-off test, with a large margin.

Getting the products to market

The oil is light and sweet once the H₂S has been stripped out in the treater, and in the Bakken region it is commonly gathered by crude oil gathering systems owned by midstream companies. Until a gathering connection is built, the oil will be trucked to a terminal about 25 miles away, which at the plateau rate means about ten truckloads a day, with the traffic, road wear, spill risk and emissions that implies. Prairie Crest has a letter of interest from the midstream company that operates the gas gathering in the area to connect Spruce Ridge to its crude and gas systems in return for a fee per barrel and per Mcf. The team assumes the connection in the base case and treats trucking as a temporary measure for the first months.

The sour gas must go to a plant that can sweeten it. The Sheep Creek gas plant, owned by the midstream company, is 13 miles away and has spare capacity for sour inlet gas. The code sizes the line: the plant gas, excluding the gas-lift gas, which circulates at the battery, is 1.73 MMscf/d with the design margin. With the battery compressor delivering 650 psia and an allowed arrival pressure of 500 psia, a 3-inch line just carries 1.79 MMscf/d, and Hannah recommends a 4-inch line for margin and for gas from any future wells. The line will be a regulated gathering line under PHMSA's rules in 49 CFR Part 192 and North Dakota's pipeline requirements, and it will carry about 0.66 metric tons of sulfur a day, which the plant recovers as elemental sulfur.

Cost estimate and its accuracy

Hannah estimates the facility costs by scaling from Prairie Crest's recent sour battery and pipeline projects, adjusted for size with the usual capacity exponent of about 0.6, and adds allowances for items not yet designed. An estimate made this way at the select stage is a class 4 estimate in the classification of AACE International, with an expected accuracy of roughly −15 to −30 percent on the low side and +20 to +50 percent on the high side. That range is large, and it is the reason the economics in Lesson 10 test a 25 percent capital overrun. The define stage will narrow it with a class 3 estimate based on real equipment quotations.

Decisions that are hard to change later

Sofia's last piece of advice concerns the decisions that lock in the future. The battery site, the pipeline routes and the size of the main lines are expensive to change once built. Building the battery for option C to be ready for injection, with space, foundations and tie-in points for the water plant and pumps, costs about a million dollars more than a battery for depletion only, but it makes adding injection in phase 2 quick and cheap. The same logic applies to the gas line: a 4-inch line costs little more than a 3-inch line, and it leaves room for future wells. These are small insurance premiums that protect the value of the flexibility the team is trying to buy.