multiUQ: An intrusive uncertainty quantification tool for gas-liquid multiphase flows
multiUQ: An intrusive uncertainty quantification tool for gas-liquid multiphase flows
复制标题
DOI：
10.1016/j.jcp.2019.108951
复制
发表时间：
2017-11
期刊：
影响因子：
--
通讯作者：
Brian Turnquist;M. Owkes
中科院分区：
文献类型：
--
作者：
Brian Turnquist;M. Owkes
流体流动的不确定性量化(UQ)提供了了解流体性质、边界条件和初始条件变化对模拟结果的影响的能力。在这项工作中，开发了一个名为MultiUQ的开源程序，它使用一种适用于气液多相流的侵入式方法来执行UQ。侵入式方法需要通过加入随机(不确定)变量来修改控制方程。与非侵入性方法(例如蒙特卡罗)相比，这增加了复杂性，但降低了计算成本。到目前为止，关于侵入式UQ的大部分工作都集中在单相流上。我们扩展了这项工作，增加了气液两相流动的功能，其中包括捕捉相界面位置的随机保守水平集方法，计算随机曲率，以及发展随机表面张力。文中给出了几个测试用例，说明了该框架的强度。确定性和随机性两种渠道流情况都收敛到解析结果，并证明了水平集输运的准确性。Zalesak的圆盘和形变测试案例进一步突出了传输方法的能力以及保持水平集轮廓的重新初始化方程的健壮性。确定性和随机振荡液滴测试案例与分析结果相结合，解决了真正的多相流问题，并突出了UQ框架的能力。最后，随机雾化射流的结果表明，在表面张力系数和入射速度不确定的情况下，液滴会发生破碎和合并。
Uncertainty quantification (UQ) of fluid flows offers the ability to understand the impact of variation in fluid properties, boundary conditions, and initial conditions on simulation results. In this work, an open-source program called multiUQ is developed which performs UQ using an intrusive approach applied to gas-liquid multiphase flows. Intrusive methods require modifying the governing equations by incorporating stochastic (uncertain) variables. This adds complexity but reduces computational cost compared to non-intrusive methods (e.g. Monte Carlo). To date, much of the work on intrusive UQ has focused on single phase flows. We extend this work by adding capabilities for gas-liquid flows which include a stochastic conservative level set method to capture the location of the phase interface, computing a stochastic curvature, and development of a stochastic surface tension force. Several test cases are presented which illustrate the strength of the framework. Both deterministic and stochastic channel flow cases converge to analytic results and demonstrate the accuracy of the level set transport. Zalesak's disk and the deformation test cases further highlight the abilities of the transport method as well as the robustness of the reinitialization equation, which maintains the level set profile. Deterministic and stochastic oscillating droplet test cases paired with analytic results, solve a true multiphase flow problem, and highlight the abilities of the UQ framework. Finally, results from a stochastic atomizing jet show droplet breakup and merging for cases of uncertainty about the surface tension coefficient and incoming velocity.