The facilities

As a first step in actually predictive simulation, the numerical methods and tools must be validated. The code UNSCYFL3D, developed at MFLab, was used in all simulations shown here, and has been validated in previous publications. Although the experiments were carried out at low mass loading, interparticle collisions are accounted for, as in previous investigations they were found to protect the pipe surface in a non-negligible way. The solution procedure is summarized as follows: first, the single-phase flow solution is calculated by the solution of the Reynolds-averaged Navier-Stokes equations. Then, the computed solution is used for tracking the particles in the computational domain. During this process, the momentum exchange between the phases is calculated in each timestep. As the particles collide against the pipe wall, data such as impact angle, impact velocity, impact frequency and penetration rate are time-averaged at each wall face.
The erosion depth at one of the extrados of the standard elbow is compared to the simulation results. The good match adds confidence to the models.

Below is a comparison involving the standard and two twisted pipes. As can be concluded, the 4-spiral one more effectively decreased the erosion depth. This is because the particles gain swirl upstream of the bend, which in turn redistributes them across the pipe cross section. The outcome is an extended elbow lifetime.

By means of validated numerical simulations, new concepts and/or geometry modifications for mitigating solid-particle erosion can be investigated at low cost and with fast turnaround times. Bearing in mind that experiments are of paramount importance to the confirmation of predicted trends, the experimental campaign can be reduced to a minimum. Numerical models are also useful for explaining complex mechanisms, such as the effects of interparticle collisions, which are otherwise impossible to assess.
As a next step towards more accurate, realistic simulations, the effects of geometry modification due to the erosive wear are to be accounted for by calculations involving dynamic mesh. Particularly in situations where the dimension of the eroded region is comparable to that of the entire geometry, the fluid flow can be affected.