CO2 pipelines and transport

Sofia Reyes had built gas gathering systems and a sales line for Prairie Crest, and her first sketch of the hub's transport system looked familiar: lateral lines from each capture plant to a compressor station, and a trunk line 70 miles to the storage site. The differences emerged as soon as she started the calculations. “Natural gas lines are sized for velocity and compression cost,” she told Ethan. “A CO₂ line also has to stay in one phase from end to end, survive a rupture without the crack running for kilometers, and protect people from a heavy, invisible gas that collects in low ground. Each of those sets a number in the design.”

Conditioning and compression

CO₂ leaves an amine stripper at about 150 to 200 kPa, saturated with water and at about 40 °C. Raising it to pipeline pressure, 2,175 psia for the hub, is done in two parts. A multistage compressor, often an integrally geared machine with four to eight stages and intercoolers, raises it to just above the critical pressure; each intercooler condenses water, and a dehydration unit, usually triethylene glycol (TEG) contact or molecular sieves, between the middle stages brings water down to the specification of Lesson 4. Once the CO₂ is dense, a pump completes the pressure rise at much lower energy, because pumping a liquid-like fluid needs far less work than compressing a gas. Compression and pumping together take about 90 to 110 kWh per metric ton for a low-pressure source, and less for sources that deliver CO₂ at higher pressure, such as the flash stages of a pre-combustion plant.

For the hub, Sofia proposed a compressor at each large source and a central pump station at the head of the trunk line. Smaller sources would send compressed CO₂ through short laterals. Electrical power for compression became one of the hub's largest operating costs: about 30 MW for 3 million metric tons a year. Its emissions, from the Texas grid, are part of the avoided-CO₂ calculation of Lesson 1.

# CO2 trunk line: compression power, line size, pressure profile and booster spacing (dense phase).
import math
SI = False     # True: Canadian hub trunk line, metric; False: US hub trunk line, US units

rate = 3.0e9 / (8760 * 3600)            # kg/s for 3.0 Mt/yr at steady flow
rho, mu, eps = 860.0, 7.5e-5, 4.5e-5    # kg/m3, Pa s, m (dense CO2 at ~10 C and 10-15 MPa; illustrative)
L_km = 110.0 if SI else 113.0           # trunk length
dz = 60.0 if SI else -15.0              # net elevation gain along the line, m
p_in, p_min = 15.0e6, 10.0e6            # Pa: discharge, and the minimum kept to stay well above the critical pressure

def friction(Re, d):                    # Swamee-Jain explicit form of Colebrook
    return 0.25 / math.log10(eps / (3.7 * d) + 5.74 / Re ** 0.9) ** 2

def dp_per_m(d):
    v = rate / rho / (math.pi * d * d / 4); Re = rho * v * d / mu
    return friction(Re, d) * rho * v * v / (2 * d), v, Re

# compression: 5 intercooled stages from 0.15 MPa to 7.5 MPa (ideal gas, k = 1.29), then a dense-phase pump to 15 MPa
k, eta, T1 = 1.29, 0.80, 313.15
ratio = (7.5 / 0.15) ** (1 / 5)
w_stage = k / (k - 1) * 8.314 / 0.04401 * T1 * (ratio ** ((k - 1) / k) - 1) / eta     # J/kg per stage
w_comp = 5 * w_stage / 3.6e6 * 1000                     # kWh/t
w_pump = (15.0e6 - 7.5e6) / (rho * 0.75) / 3.6e6 * 1000
print(f"Compression {w_comp:.0f} kWh/t + pumping {w_pump:.0f} kWh/t = {w_comp + w_pump:.0f} kWh/t; "
      f"{(w_comp + w_pump) * rate * 3.6 / 1000:.1f} MW for {rate:.1f} kg/s")

sizes = [("NPS 12", 0.3048, 0.0103), ("NPS 16", 0.4064, 0.0127), ("NPS 20", 0.5080, 0.0143)]
print(f"\nTrunk line {L_km:.0f} km ({L_km / 1.609:.0f} mi), net rise {dz:.0f} m")
for name, od, wt in sizes:
    d = od - 2 * wt; g, v, Re = dp_per_m(d)
    loss = g * L_km * 1000 + rho * 9.81 * dz
    reach = (p_in - p_min - rho * 9.81 * dz) / g / 1000 if g > 0 else float('inf')
    if SI: print(f"{name}: v {v:.2f} m/s, Re {Re:,.0f}, drop {g:.1f} kPa/km, outlet {(p_in - loss) / 1000:,.0f} kPa, reach to 10 MPa {reach:,.0f} km")
    else:  print(f"{name}: v {v * 3.281:.1f} ft/s, drop {g * 1609 / 6894.8:.1f} psi/mi, outlet {(p_in - loss) / 6894.8:,.0f} psia, reach to 1,450 psia {reach / 1.609:,.0f} mi")
# booster spacing for a hypothetical extension of the same NPS 12 line to 400 km (250 mi)
d = sizes[0][1] - 2 * sizes[0][2]; g = dp_per_m(d)[0]
spacing = (p_in - p_min) / g / 1000
n_pumps = math.ceil(400 / spacing) - 1
if SI: print(f"\nNPS 12 extended to 400 km: a booster every {spacing:.0f} km, so {n_pumps} intermediate pump stations")
else:  print(f"\nNPS 12 extended to 250 mi: a booster every {spacing / 1.609:.0f} mi, so {n_pumps} intermediate pump stations")
g16 = dp_per_m(0.4064 - 0.0254)[0]
pts = [(x, p_in - g16 * x * 1000 - rho * 9.81 * dz * x / L_km) for x in [L_km * i / 5 for i in range(6)]]
if SI: print("Pressure profile, NPS 16:", ", ".join(f"{x:.0f} km {p / 1e6:.2f} MPa" for x, p in pts))
else:  print("Pressure profile, NPS 16:", ", ".join(f"{x / 1.609:.0f} mi {p / 6894.8:,.0f} psia" for x, p in pts))

Compression 88 kWh/t + pumping 3 kWh/t = 91 kWh/t; 31.2 MW for 95.1 kg/s

Trunk line 113 km (70 mi), net rise -15 m
NPS 12: v 5.7 ft/s, drop 14.4 psi/mi, outlet 1,185 psia, reach to 1,450 psia 52 mi
NPS 16: v 3.2 ft/s, drop 3.2 psi/mi, outlet 1,971 psia, reach to 1,450 psia 234 mi
NPS 20: v 2.0 ft/s, drop 1.0 psi/mi, outlet 2,125 psia, reach to 1,450 psia 760 mi

NPS 12 extended to 250 mi: a booster every 50 mi, so 4 intermediate pump stations
Pressure profile, NPS 16: 0 mi 2,176 psia, 14 mi 2,135 psia, 28 mi 2,094 psia, 42 mi 2,053 psia, 56 mi 2,012 psia, 70 mi 1,971 psia

Hydraulics of dense-phase lines

In the dense phase, CO₂ flows like a liquid and the line is sized with the Darcy-Weisbach equation:

The Colebrook equation (or its explicit Swamee-Jain approximation, used in the code) gives the friction factor from the Reynolds number and relative roughness. Dense CO₂ has a high density and a low viscosity, so Reynolds numbers are in the millions and flow is fully turbulent. Unlike a gas line, the density hardly changes along the line, so velocity is nearly constant, and elevation matters: a 200 ft rise costs about 75 psi of head, as it would for water.

Two pressure limits frame the design. The maximum is the line's maximum operating pressure, set by pipe grade, wall thickness and the design code, and limited in practice by the compressor discharge. The minimum is the pressure needed to stay safely in the dense phase all the way to the delivery point, including the bubble point of the impure stream and a margin; for the hub, 1,450 psia. The allowed pressure drop, 725 psi, divided by the friction gradient gives the maximum distance between pump stations. For the 70-mile trunk line, an 12-inch pipe would drop below the minimum before reaching the hub. An 16-inch line arrives at about 1,970 psia, with enough margin to carry about 5 million metric tons a year later. Larger diameters cost more steel but save pump stations and leave room for growth; for a hub that hopes to attract more sources, Sofia recommended the 16-inch line.

Running fractures and pipe toughness

A rupture in a dense-phase CO₂ line behaves differently from one in a gas line. When the pipe opens, a decompression wave runs along the line at the speed of sound in the fluid. As the pressure falls to the vapor pressure, the CO₂ starts to boil, and the pressure stops falling for a while: the decompression curve has a long plateau near the saturation pressure, higher than for natural gas at the same initial pressure. A crack in the pipe wall travels at a speed set by the steel's toughness and the pressure at the crack tip. If the crack is faster than the pressure falls, it keeps running, potentially for long distances. Preventing that requires steel with high fracture toughness, measured by Charpy impact energy, a lower operating stress (thicker wall), or mechanical crack arrestors, sleeves fitted at intervals that stop a running crack. Impurities make it worse, because they raise the saturation plateau. The Battelle two-curve method, developed for gas lines and adapted for CO₂ with full-scale burst tests, is the standard way to check the design.

Codes and regulation

In the United States, pipelines carrying CO₂ as a supercritical fluid are regulated by the Pipeline and Hazardous Materials Safety Administration (PHMSA) under the hazardous liquid rules of 49 CFR Part 195; gas-phase and liquid CO₂ lines have not been explicitly covered. About 5,000 miles of CO₂ pipeline operate in the US, mostly serving enhanced recovery in the Permian Basin and the Rockies. Codes set design factors, material requirements, valve spacing, leak detection, integrity management and emergency planning.

The 2020 rupture of a CO₂ pipeline near Satartia, Mississippi, changed the debate. Heavy rain caused a landslide that broke a line operated by Denbury. The plume of cold CO₂ flowed into the village, about 200 people were evacuated and at least 45 were treated in hospital; vehicle engines stalled for lack of oxygen, slowing the response. Emergency responders had not been told what the pipeline carried or how to respond. PHMSA's investigation led to a notice to operators about geohazards and, in January 2025, a proposed rule covering CO₂ in all phases, dispersion modeling, emergency planning and notification of local responders. The proposal was withdrawn under the new administration's regulatory freeze shortly afterward, so operators like Prairie Crest work to the lessons voluntarily, and several states have added requirements of their own.

Dispersion and routing

CO₂ is not flammable, but it is an asphyxiant and, at high concentration, toxic: above about 5 percent in air it causes headaches and breathing difficulty, and above about 10 percent loss of consciousness within minutes. Because it is heavier than air, a large release flows along the ground and collects in valleys, ditches and basements, especially in calm weather. Dispersion modeling for a pipeline route therefore considers terrain, not just distance, and identifies low areas where a plume could pool. Routing avoids running a line uphill of communities where possible, places block valves to limit the inventory that can escape from any section, and adds detection and notification. CO₂ contains no odorant; adding one has been studied but is not standard. The hub's route ran across flat coastal prairie, with one river crossing and a stretch of low ground near the Colorado River that Sofia flagged for detailed modeling.

Ships, trucks and rail

Not every source is near a pipeline. Ships carry CO₂ as a cold liquid, at about −50 °C and 100 psia in low-pressure designs or −30 °C and 220 to 290 psia in medium-pressure ones, and have long served the food and beverage industry on a small scale. The Northern Lights project in Norway, which began injecting in 2025, receives CO₂ by ship at a coastal terminal and pipes it offshore to a saline formation, and several European hubs plan to collect CO₂ from many coastal emitters the same way. Ship transport has lower capital cost and more flexibility than a pipeline but higher operating cost per metric ton, and it requires liquefaction, buffer storage and terminals. Trucks and rail suit small volumes, such as an ethanol plant's fermentation CO₂ before a pipeline is available. For the hub, all four sources were within pipeline distance, but Lena wanted the storage site to keep land for a future truck or rail unloading facility.