A UK research team has been awarded a £4.9 million ($7.9 million) government grant to design a new generation of far more accurate modeling and simulation tools for multiphase flow systems, such as oil and gas pipelines and complex emulsions and foams used in fast-moving consumer goods (FMCG). The project is being funded by the UK government’s Engineering and Physical Sciences Research Council (EPSRC) and involves leading researchers in multiphase flows from several universities. The research will be conducted as part of the “Multi-scale Examination of MultiPHase physIcs in flowS,” or MEMPHIS, Programme.
Working with leading companies across a breadth of industrial sectors including oil and gas (BP, Chevron and Statoil), FMCG (Procter & Gamble) and fluid modeling (including AspenTech and CD-Adapco), the team will design modeling and simulation tools that better leverage computational fluid dynamics. Omar Matar, professor of fluid dynamics at Imperial College London‘s Department of Chemical Engineering, and the team leader, said:
“We believe our work will result in a paradigm shift in the way we think about the prediction of complex multiphase flows – which are key to the design of virtually every processing and manufacturing technology. Currently, there is an over-reliance on existing models which contain many assumptions and are often used beyond their range of validity. We want to change that culture and ensure researchers in academia and industry rely more on fundamentals. This approach will ultimately lead to more innovative and inventive products for us all, more reliable equipment design, with an associated reduction in emissions and our carbon footprint.”
Matar went on to explain:
“A single-phase flow is one in which a single fluid phase (a gas or a liquid) flows along a channel. In a multiphase flow, there is a mixture of phases (gas, solid or liquid) flowing in the channel. Perhaps the most common multiphase flow is the flow of mixtures of gases and liquids (as in boilers or condensers), but there are also many examples of liquid-liquid, gas-solid and liquid-solid two-phase flows [here the solid phase is present as particles].”
Matar noted that three-phase flows (e.g., flows of oil, natural gas and water in hydrocarbon recovery) are also commonly encountered in industry, and even four-phase flows (liquid-liquid-gas-solid) are seen, particularly in the production of everyday household products.
Current models for multiphase flows can be divided approximately into empirical, phenomenological and computational methodologies. In the empirical models, data are fitted with correlating equations which often are based on traditional treatments of single-phase flows. In the phenomenological models, each of the constituent processes (e.g., droplet deposition and entrainment in annular flows, as shown in the photograph) is modeled and the models are integrated to provide system predictions.
In terms of accuracy of the various models currently used, empirical models have typical standard deviations of 30 to 50 percent but can produce errors of several hundred percent. The phenomenological models produce more accurate predictions and also allow more confident extrapolation outside their validation base. However, phenomenological models are restricted to relatively simple geometries, and the constituent models have unproven generality, Matar said. “To predict multiphase flows in complex geometries, computational fluid dynamics (CFD) tools are the only option, and their predictive power is increasing day by day.”
Matar noted that ever-expanding computational power has enabled greater use of CFD models. “In these models, solutions of the basic equations for the fluid flow (e.g., Navier-Stokes) and interfacial behavior are generated,” he said. “However, generally applicable solutions to these equations have proved elusive.”
He went on:
“It is important to remember that, even with single-phase turbulent flows, there is a restriction to the range of the Reynolds number that can be predicted in a fundamental way (the maximum Reynolds number which can be modeled by direct numerical simulation is of order 104). Most systems of engineering significance operate at much higher Reynolds numbers and, to use CFD for these systems, turbulence needs to be correlated using fairly arbitrary relationships. It is only by harnessing the power of new numerical methods combined with massively-parallel computing and mesh-refinement technology that further progress can be made.”
There are two main areas where improved CFD could be used to increase accuracy in multiphase flow predictions. First, CFD (coupled with well-designed experiments) could be used in the development of better constituent models to be employed in phenomenological modeling; this should lead to better prediction and improved generality. Second, CFD could be used to obtain a better understanding of multiphase flows in complex three-dimensional geometries encountered in industrial applications.
Routine CFD calculations could be carried out by non-specialist engineers and scientists with a modicum of training. The same would apply to the case of developed phenomenological models. However, to make real progress in developing and applying CFD tools for complex multiphase flow situations, a much higher level of skill would be required.
“The tools developed as part of the MEMPHIS Programme could be used in the design and investigation of complex chemical process plants, in the design of hydrocarbon recovery systems and in the design and investigation of nuclear power plants,” Matar said.