The first stage of the project involved verification and history
matching of the OLGA hydraulics flow model, prior to uncertainty evaluation.
Laboratory-based, low-liquid-loading experiments were also undertaken. This
provided a theoretical verification of the transition from low to high holdup,
for low-liquid-loading flow in large-diameter pipes. Laboratory data was also
compared with OLGA simulation data for steady-state and ramp-up conditions. The
two sets of results matched within a reasonable tolerance.
The project team then combined the OLGA simulator with the RMO workflow
to outline key flow assurance risks. Parameters such as production capacity,
liquid content at turndown production, and arrival temperature were defined.
These elements also depended on a significant number of uncertainty parameters,
including pipeline routing, hydraulic roughness, reservoir fluid properties,
ambient temperatures, and arrival pressures, as well as liquid processing and
slug catcher drainage capacities.
Sensitivity studies were then undertaken to examine the effects of each
parameter individually, followed by analyses to examine the overall effects of
Thousands of simulations were performed to determine where liquid
accumulation starts, as well as to compare liquid content and flow rate
distribution—underlining that overall uncertainty is significantly less
for the flow rate than for the liquid content.
Additional simulations were performed to estimate the distribution of
the required trunkline inlet pressure with different combinations of
uncertainty parameter values.
Finally, the team clarified the capacity at maximum pressure drop,
concluding that overall flow rate uncertainty is slightly higher for the
required inlet pressure. This allowed systematic derivation of key
uncertainties relating to pressure drop and capacity at high-gas flow rate,
liquid holdup, and turndown flexibility at minimum flow.
The OLGA simulator allowed Statoil to quantify and visualize flow
assurance uncertainties more systematically and study risks through an
automated workflow, avoiding reliance on assumptions such as linear response
variation or the decoupling of uncertainty contributions. Adopting this
methodology increased confidence in the design, as well as the