1: \begin{abstract}

2:   The advantage of particle Lagrangian methods in computational fluid

3:   dynamics is that advection is accurately modeled. However, the

4:   calculation of space derivatives is complicated. On the other hand,

5:   Eulerian formulations benefit from a fixed mesh, but feature

6:   non-linear advection terms. It seems natural to combine these two,

7:   using particles to advect the information and a fixed mesh to

8:   calculate space derivatives. This idea goes back to Particle-in-Cell

9:   methods, and is here considered within the context of the finite

10:   element (FE) method for the fixed mesh, and the particle-FEM (pFEM)

11:   for the particles.  A ``projection'' method is required to transfer

12:   field values from particles to mesh and vice versa --- in this work

13:   simple interpolation is used. Our results confirm that projection

14:   errors, especially from particles to mesh, cause a slow convergence

15:   of the method if standard, linear, FEs, are employed --- but also

16:   that the convergence rate is restored to values comparable to

17:   Lagrangian simulations if quadratically consistent shape functions

18:   for the particles are used. The same applies to computational

19:   resources, making projection a viable alternative to Lagrangian

20:   simulations. The procedure is validated on Zalesak's disk problem,

21:   the Taylor-Green vortex sheet flow, and the Rayleigh-Taylor

22:   instability.

23: \end{abstract}

24: