One-Dimensional OLGA Modelling

OLGA is one of the multiphase flow simulation codes widely used in the oil and gas industry. OLGA was originally developed for two-phase hydrocarbon flow in pipelines and pipeline networks, with processing equipment included (SPT Group, 2006). Later, a water option was included which treats the water as a separate liquid phase.

Essentially OLGA is a transient one-dimensional (1-D) modified two-fluid model, i.e. separate continuity equations for the gas, liquid bulk and liquid droplets are applied; these may be coupled through interfacial mass transfer. Only two momentum equations are used; one for the continuous liquid phase and one for the combination of gas and possible liquid droplets. The velocity of any entrained liquid droplets in the gas phase is given by a slip relation. One mixture energy equation is applied; both phases are at the same temperature. This yields six conservation equations to be solved: three for mass, two for momentum and one for energy. The continuity equations for bulk water and water droplets are added with the water option on. The bulk water velocity is obtained from a correlation for water velocity relative to the average liquid bulk velocity. The closure laws are based on two main flow regime classifications, namely, separated flow and distributed flow. In the case of separated flows such as stratified flow and annular mist flow, the closure laws are in the form of correlations for each flow-regime dependent parameter. For distributed flows such as bubble flow and slug flow, the closure laws take the form of a ‘slip relations’, relating the phase velocities to one another.

Applications to Severe Slugging

The first attempt at predicting severe slugging was that carried out as part of the OLGA code development. The code predictions were compared against the data of Schmidt et

al. (1980) and from the SINTEF Two-Phase Flow Laboratory (Linga and Østvang,

holdup discontinuities in severe slugging. The pressure cycling characteristics were not predicted by the code reasonably well.

Mazzoni et al. (1993) described the predictions of severe slugging in an offshore field using OLGA. The pressure cycling and liquid accumulation process in the riser was clearly shown. Courbot (1996) reported the use of OLGA to predict the region of potential severe slugging in an offshore pipeline/riser application. Unfortunately there was little recorded data to be compared with the model predictions in both of the above cases.

Kashou (1996) verified OLGA by comparing the simulation results with the experimental data from two riser configurations (an S-shaped riser and a catenary riser). Generally the simulations showed a degree of success in predicting the overall flow regimes, cycle times and slug lengths in the pipeline/riser systems; however, details of the severe slugging characteristics such as peak production rate were not correctly predicted by the code. OLGA experienced difficulty in predicting the slug production period, particularly at high gas velocities.

Yeung et al. (2003) reported a series of simulation results from an S-shaped riser. The results showed that the variations of boundary conditions affected the flow behaviour in the riser significantly and the liquid holdup in the down comer of the riser was over predicted due to the assumption of a horizontal gas/liquid interface (curved interface in reality). The effects of the downstream equipment conditions on the flow behaviour in a catenary riser were demonstrated through OLGA by Yeung et al. (2006). The model predictions agreed with the experimental data reasonably well although there were some differences in detail. It was found that imposing a pressure boundary at the riser outlet to represent the downstream equipment resulted in quite different flow behaviour in the riser. Therefore, they recommended that, in any simulation study on pipeline/riser systems, the downstream equipment and controls need to be included in the model.

Applications to Hydrodynamic Slug Flow

OLGA is the first commercial code to incorporate a non-diffusive slug tracking scheme (Bendiksen et al., 1990; Straume et al., 1992). Two numerical schemes are employed to

solve the model equations. One is an implicit scheme used for separated and bubbly flows and a Lagrangian tracking scheme for slug flows (Straume et al., 1992). The implicit scheme is used for most normal transient calculations. Bendiksen et al. (1991) stated that the implicit numerical schemes were more efficient and stable for pipeline simulations of slow transients and were favoured over explicit schemes in these cases due to the improved computational efficiency. However, for slug flow regime the purely implicit schemes were highly undesirable (Straume et al., 1992). Numerical diffusion inherent in the implicit schemes causes a ‘smoothing out’ of void fraction discontinuities, such as the slug front and tail. To track the propagation of the slugs a Lagrangian slug tracking scheme is adopted by OLGA. The slug tracking scheme traces the movement of a discontinuity. The cell that contains the discontinuity is divided into two regions with separate flow regimes. Then the flow parameters are calculated based on each flow regime.

As summarised by King (1998) the slug tracking model in OLGA had received a lot of attentions. Hustvedt (1993) used OLGA to predict the slug distribution in a Tunisian oil pipeline and compared with experimental data. The experimental results exhibited the typical Log-Normal (Brill et al., 1981) or Inverse Gaussian (Dhulesia et al., 1993) shapes of slug length distributions. OLGA predicted the correct distribution shape, maximum and mean slug length. However, it predicted twice the number of slugs observed in practice. Burke and Kashou (1993) highlighted one of the shortcomings of the OLGA model that the slug frequency within the pipe needed to be specified. Because the slug frequency has not been modelled adequately, a significant limitation is imposed on the application of the slug tracking model within OLGA. This is most limiting in the design of new pipelines where there is no experimental data available to ‘tune’ the simulation model.

Most recently, Nordsveen et al. (2009) reported an investigation on slug flow development in high risers. Based on the analysis of the field data and previous experimental data from SINTEF Two-Phase Flow Laboratory, they found that: (1) the slugs generated in a near horizontal pipeline could develop a long high void (low liquid holdup) zone in front of a low void zone (high liquid holdup) as the slugs moved upwards in a high riser; (2) the two zones were separated by a relatively sharp void front.

The high void zone was confirmed to be liquid continuous and thus a part of the slug observed in their new experiment conducted in the Well Flow Loop at IFE. Different versions of OLGA were applied to predict the high void zone. A new numerical scheme, 2nd order explicit TVD (Total Variation Diminishing) scheme, for the mass equations and the convective transport volume in the volume equation was implemented in OLGA 6.0. The 2nd order explicit TVD is less diffusive than the 1st order upwind implicit (Backward Euler) scheme in the previous versions. It was found that the prediction of the void front with the previous OLGA code failed mainly because the mass gradients were smeared out due to numerical diffusion, while the new numerical scheme was shown to be able to improve the predictions of the sharp void gradient. It was pointed out that OLGA did not predict the slug length distribution properly and the correlations for the gas entrainment and slip in slugs needed to be improved.