IISPH (Masterarbeit Noah)

Project.toml

@@ -18,6 +18,7 @@ JSON = "682c06a0-de6a-54ab-a142-c8b1cf79cde6"
 KernelAbstractions = "63c18a36-062a-441e-b654-da1e3ab1ce7c"
 LinearAlgebra = "37e2e46d-f89d-539d-b4ee-838fcccc9c8e"
 MuladdMacro = "46d2c3a1-f734-5fdb-9937-b9b9aeba4221"
+OrdinaryDiffEq = "1dea7af3-3e70-54e6-95c3-0bf5283fa5ed"
 PointNeighbors = "1c4d5385-0a27-49de-8e2c-43b175c8985c"
 Polyester = "f517fe37-dbe3-4b94-8317-1923a5111588"
 Printf = "de0858da-6303-5e67-8744-51eddeeeb8d7"
@@ -33,7 +34,6 @@ TrixiBase = "9a0f1c46-06d5-4909-a5a3-ce25d3fa3284"
 WriteVTK = "64499a7a-5c06-52f2-abe2-ccb03c286192"
 
 [weakdeps]
-OrdinaryDiffEq = "1dea7af3-3e70-54e6-95c3-0bf5283fa5ed"
 OrdinaryDiffEqCore = "bbf590c4-e513-4bbe-9b18-05decba2e5d8"
 
 [extensions]

docs/src/refs.bib

@@ -361,6 +361,16 @@ @Article{Hormann2001
   publisher = {Elsevier BV},
 }
 
+@article{ihmsen2013,
+  author = {Ihmsen, Markus and Cornelis, Jens and Solenthaler, Barbara and Horvath, Christopher and Teschner, Matthias},
+  year = {2013},
+  month = {07},
+  pages = {1--11},
+  title = {Implicit incompressible SPH},
+  volume = {20},
+  journal = {IEEE transactions on visualization and computer graphics},
+  doi = {10.1109/TVCG.2013.105}
+}
 
 @Article{Jacobson2013,
   author    = {Jacobson, Alec and Kavan, Ladislav and Sorkine-Hornung, Olga},

docs/src/systems/implicit_incompressible_sph.md

@@ -0,0 +1,233 @@
+# [Implicit Incompressible SPH](@id iisph)
+
+Implicit Incompressible SPH as introduced by [M. Ihmsen](@cite ihmsen2013) is a scheme that computes pressure values, by iteratively solving a linear system using a relaxed jacobi system, to resolve the particles density deviation from the reference density. This method solves a linear system $Ax=b$ where the pressure values that are used for the pressure acceleration are the unknown variable $x$.
+It does not use a state equation to generate pressure values like the [weakly compressible SPH scheme](weakly_compressible_sph.md).
+
+
+To derive the formulation for the linear system we start by discretizing the Continuity equation
+
+```math
+\frac{D\rho}{Dt} = - \rho \nabla \cdot v
+```
+
+For a particle $i$ a forward difference method is used to discretizie the left hand side of the equation to get
+
+```math
+\frac{\rho_i(t+ \Delta t - \rho_i(t))}{\Delta t}.
+```
+
+The right-hand side is discretized by the difference formulation of the divergence of the velocity field.
+
+```math
+\nabla \cdot \textbf{v}_i = \frac{1}{\rho_i} \sum_j m_j \textbf{v}_{ij} \nabla W_{ij},
+```
+where $\textbf{v}_{ij}$ is equal to $\textbf{v}_i - \textbf{v}_j$.
+
+So all in all the following discretized version of the continuity equation for a particle $i$ is achieved:
+
+```math
+\frac{\rho_i(t + \Delta t) - \rho_i(t)}{\Delta t} = \sum_j m_j \textbf{v}_{ij}(t+\Delta t) \nabla W_{ij}
+```
+
+In this implicit formulation the unknown velocities $\textbf{v}_{ij}(t + \Delta t)$ are needed.
+
+They are given by adding all pressure and non-pressure accelerations to the current velocity:
+
+```math
+\textbf{v}_i(t + \Delta t) = \textbf{v}_i(t) * \frac{\textbf{F}_i^{adv}(t) + \textbf{F}_i^p(t)}{m_i}
+```
+
+$\textbf{F}_i^{adv}$ are all non-pressure forces such as gravity, viscosiy, surface tension and more, while $\textbf{F}_i^p(t)$ are the unknown pressure forces, which have to be determined.
+
+Note that the IISPH is an incompressible fluid system, which means that the density of the fluid remain constant over time. By assuming a fixed reference density $\rho_0$ for all fluid particle over the whole time of the simulation, the density value at the next time step $\rho_i(t + \Delta t)$ also has to be this rest density. So $\rho_0$ can be plugged in for $\rho_i(t + \Delta t)$ in the formula above.
+
+The goal is to compute the pressure values required to obtain the pressure acceleration that ensures each particle reaches the rest density in the next time step. At the moment these pressure values are unknown in $t$, but all the non-pressure forces are already known.
+Therefore a predicted velocity can be calculated which depends only on the non-pressure forces $\textbf{F}_i^{adv}$:
+
+```math
+\textbf{v}_i^{adv}= \textbf{v}_i(t) + \Delta t \frac{\textbf{F}_i^{adv}(t)}{m_i},
+```
+Using this predicted velocity and the continuity equation a predicted density can also be determined.
+
+```math
+\rho_i^{adv} = \rho_i(t) + \Delta t \sum_j m_j \textbf{v}_{ij}^{adv} \nabla W_{ij}(t)
+```
+
+To achieve the rest density, the unknown pressure forces must counteract the compression caused by the non-pressure forces. In other words, they must resolve the deviation between the predicted density and the reference density.
+
+Therefore following equation needs to be fulfilled:
+```math
+\Delta t ^2 \sum_j m_j \left(  \frac{\textbf{F}\_i^p(t)}{m_i} - \frac{\textbf{F}\_j^p(t)}{m_j} \right) \nabla W_{ij}(t) = \rho_0 - \rho_i^{adv}
+```
+
+This expression is derived by substituting the reference density $\rho_0$ for $\rho_i(t+\Delta t)$ in the discretized continuity equation and inserting the definitions of the predicted velocity $\textbf{v}_{ij}(t+\Delta t)$ and predicted density $\rho_i^{adv}$.
+
+The pressure force is defined as:
+
+```math
+\textbf{F}_i^p(t) = m_i * \nabla p_i = m_i \sum_j m_j \left( \frac{p_i(t)}{\rho_i^2(t)} - \frac{p_j(t)}{\rho_j^2(t)} \right) \nabla W_{ij}
+```
+
+Substituting this definition into the equation, leadsa linear system $\textbf{A}(t) \textbf{p}(t) = \textbf{b}(t)$ with one equation and one unknown pressure value for each particle. So in total it is a system with  n equations and n unknown variables for n particles.
+
+```math
+\sum_j a_{ij} p_i = b_i = \rho_0 - \rho_i^{adv}
+```
+
+This linear system must be solved in order to get the pressure values, which is done by using a relaxed jacobi scheme.
+
+This is a iterative numerical method to solve a linear system $Ax=b$. In each iteration the new values of the variable $x$ get computed by the following formular
+
+```math
+x_i^{(k+1)} = (1-\omega) x_i^{(k)} + \omega \left( \frac{1}{a_{ii}} \left( b_i - \sum_{j \neq i} a_{ij} x_j^{(k)} \right)\right)
+```
+
+In the context of the linear system for the pressure values the formula is
+
+```math
+p_i^{l+1} = (1-\omega) p_i^l + \omega \frac{\rho_0 - \rho_i^{adv} \sum_{j \neq i} a_{ij}p_j^l}{a{ii}}
+```
+
+Therefore the diagonal elements $a_{ii}$ and the sum $\sum_{j \neq i} a_{ij}p_j^l$ need to be determined.
+This can be done efficiently by separating the pressure force acceleration into two components: One that describes the influence of particle i's own pressure on its displacement, and one that describes the displacement due to the pressure values of the neighboring particles $j$.
+
+The pressure acceleration is given by:
+```math
+\Delta t^2 \frac{\textbf{F}_i^p}{m_i} = -\Delta t^2 \sum_j m_j \left( \frac{p_i}{\rho_i^2} + \frac{p_j}{\rho_j^2} \right)\nabla W_{ij} = \left( - \Delta t^2 \sum_j \frac{m_j}{\rho_i^2} \nabla W_{ij} \right) p_i + \sum_j - \Delta t^2 \frac{m_j}{\rho_j^2} \nabla W_{ij}p_j
+```
+
+The $d_{ii}p_i$ value describes the displacement of particle $i$ because of the particle $i$# and $d_{ij}p_j$ describes the influence from the neighboring particles $j$.
+Using this new values the linear system can be rewritten as
+
+```math
+\rho_0 - \rho_i^{adv} = \sum_j m_j \left( d_{ii}p_i + \sum_j d_{ij}p_j - d_{jj}p_j - \sum_k d_{jk}p_k \right) \nabla W_{ij}
+```
+
+where $k$ stands for the neighbor particles of the neighbor particle $j$ from $i$.
+So the sum over the neighboring pressure values $p_j$ also includes the pressure values $p_i$, since $i$ is a neighbor of $j$
+To separate this sum it can be written as
+
+```math
+\sum_k d_{jk} p_k = \sum_{k \neq i} d_{jk} p_k + d_{ji} p_i
+```
+
+With this separation the equation for the linear system can again be rewritten as
+
+```math
+\rho_0 - \rho_i^{adv} = p_i \sum_j m_j ( d_{ii} - d_{ij})\nabla W_{ij}  + \sum_j m_j \left ( \sum_j d_{ij} p_j - d_{jj} p_j - \sum_{k \neq i} d_{jk}p_k \right) \nabla W_{ij}
+```
+
+In this formulation all coefficients that are getting multiplied with the pressure value $p_i$ are separated from the other. The diagonal elements $a_{ii}$ can therefore be defined as:
+
+```math
+a_{ii} = \sum_j m_j ( d_{ii} - d_{ij})\nabla W_{ij}
+```
+
+The remaining part of the equation represents the influence of the other pressure values $p_j$
+​
+ . Hence, the final relaxed Jacobi iteration takes the form:
+
+```math
+p_i^{l+1} = (1 - \omega) p_i^{l} + \omega \frac{1}{a_{ii}} \left( \rho_0 -\rho_i^{adv} \sum_j m_j \left( \sum_j d_{ij} p_j^l - d_{jj} p_j^l - \sum_{k \neq i} d_{jk} p_k^l \right) \nabla W_{ij} \right).
+```
+
+Note that the pressure value of a particle $i$ depends only on its own current pressure value, the pressure values of his neighboring particles, and the neighbor of those neighboring particle. That means that the most entries of the matrix are zero (i.e the system is sparse).
+
+The diagonal elements $a_ii$ get computed and stored at the beginning of the simulation step and remain unchanged throughout the relaxed Jacobi iterations. The same applies for the $d_{ii}$. The coefficients $d_{ij}$ are computed, but not stored, as part of the calculation of $a_{ii}$.
+During the jacobi iterations, two loops over all particles are performed:
+The first updates the values of $\sum_j d_{ij}$ and the second is to compute the updated pressure values $p_i^{l+1}$.
+
+The final pressure update follows the given formula of the relaxed jacobi scheme, with two exceptions:
+#### 1. Pressure clamping
+To avoid negative pressure values, one can enable pressure clamping. In this case, the pressure update is given by:
+
+```math
+\max(0, p_i^{l+1}) = (1-\omega) p_i^l + \omega \frac{\rho_0 - \rho_i{adv} \sum_{j \neq i} a_{ij}p_j^l}{a{ii}}
+```
+
+#### 2. Small diagonal elements
+If the diagonal element $a_{ii}$ becomes too small or even zero, the simulation may become unstable. This can occur, for example, if a particle is isolated and receives little or no influence from neighboring particles, or due to numerical cancellations. In such cases, the updated pressure value is set to zero.
+
+There are also other options, like setting $a_{ii}$ to the threshold value is its beneath and then updare with the known formula, or just don't update the pressure value at all, and continue with the old value. By setting the pressure value to zero, the numerical error through this can not be so big to mess up a whole simulation, as long as it doesn't happens for too many particles.
+
+
+## Boundary Handling
+
+The previously introduced formulation only considered interactions between fluid particles and neglected interactions between fluid and boundary particles. To account for boundary interactions, a few modifications to the previous equations are required.
+
+First, the discretized form of the continuity equation must be adapted for the case in which a neighboring particle is a boundary particle. From now on, we distinguish between neighboring fluid particles (indexed by $j$) and neighboring boundary particles (indexed by $b$).
+
+The updated discretized continuity equation becomes:
+
+```math
+\frac{\rho_i(t + \Delta t) - \rho_i(t)}{\Delta t} = \sum_j m_j \textbf{v}_{ij}(t+\Delta t) \nabla W_{ij} + \sum_b m_b \textbf{v}_{ib}(t+\Delta t) \nabla W_{ib}
+```
+
+Since boundary particles do not have an own velocity the difference between the fluid particles velocity and the boundary particles velocity simplifies to just the fluid particles velocity $\textbf{v}_{ib}(t+\Delta t) = \textbf{v}_{i}(t+\Delta t)$
+Accordingly, the predicted density $\rho^{adv} also includes contributions from boundary particles:
+
+```math
+\rho_i^{adv} = \rho_i (t) + \Delta t \sum_j m_j \textbf{v}_{ij}^{adv} \nabla W_{ij}(t) + \Delta t \sum_b m_b \textbf{v}_{i}^{adv} \nabla W_{ib}(t)
+```
+
+This leads to the following updated formulation of the linear system:
+
+```math
+\Delta t^2 \sum_j m_j \left(  \frac{\textbf{F}_i^p(t)}{m_i} - \frac{\textbf{F}_j^p(t)}{m_j} \right) \nabla W_{ij} + \Delta t^2 \sum_b m_b \frac{\textbf{F}_i^p(t)}{m_i} \nabla W_{ib} = \rho_0 - \rho_i^{adv}
+```
+
+It is important to note that, because boundary particles have no velocity, the pressure acceleration of a fluid particle does not directly depend on the pressure forces of boundary particles. However, the pressure forces of both the particle itself and its neighboring fluid particles depend on the pressure values of all neighboring particles—this includes boundary particles. Therefore, the pressure of a fluid particle is indirectly influenced by the pressure values of nearby boundary particles.
+
+The pressure force acting on a fluid particle is computed as:
+
+```math
+\textbf{F}_i^p(t) = -m_i \sum_j \left( \frac{p_i(t)}{\rho_i^2(t)} + \frac{p_j(t)}{\rho_j^2(t)} \right) \nabla W_{ij}(t) -m_i \sum_b \left( \frac{p_i(t)}{\rho_i^2(t)} + \frac{p_b(t)}{\rho_j^2(t)} \right) \nabla W_{ib}(t)
+````
+
+From this point forward, the computation of the coefficients required for the Jacobi scheme (such as $d_{ii}$, $d_{ij}$ etc.) depends on the specific boundary density evaluation method used in the chosen boundary model.
+
+
+
+### Pressure Mirroring
+When using pressure mirroring, the pressure value $p_b$ of a boundary particle is defined to be equal to the pressure value of the corresponding fluid particle $p_i$.
+ In other words, the boundary particle "mirrors" the pressure of the fluid particle interacting with it.
+As a result, the coefficient that describes the influence of a particle's own pressure value $p_i$
+​must also include contributions from boundary particles. Therefore, the formula for calculating the coefficient $d_{ii}$ must be adjusted as follows:
+
+```math
+d_{ii} = -\Delta t^2 \sum_j \frac{m_j}{\rho_i^2} \nabla W_{ij} - \Delta t^2 \sum_b \frac{m_b}{\rho_i^2} \nabla W_{ib}
+```
+
+The corresponding relaxed Jacobi iteration for pressure mirroring then becomes:
+
+```math
+p_i^{l+1} = (1 - \omega) p_i^l + \omega \frac{1}{a_{ii}} \left( \rho_0 - \rho_i^{adv} - \sum_j m_j \left( \sum_j d_{ij} p_j^l - d_{jj}p_j^l - \sum_{k \neq i} d_{jk} p_k^l \right) \nabla W_{ij} - \sum_b m_b \sum_j d_{ij} p_j^l \nabla W_{ij} \right)
+```
+
+### Pressure Zeroing
+If pressure zeroing is used instead, the pressure value of a boundary particle $p_b$
+​is assumed to be zero. Consequently, boundary particles do not contribute to the pressure forces acting on fluid particles.
+In this case, the computation of the coefficient $d_{ii}$ remains unchanged and is given by:
+
+```math
+d_{ii} = -\Delta t^2 \sum_j \frac{m_j}{\rho_i^2} \nabla W_{ij}
+```
+
+The formula for the relaxed Jacobi iteration remains the same as in the pressure mirroring approach. However, the contribution from boundary particles vanishes due to their zero pressure:
+
+```math
+p_i^{l+1} = (1 - \omega) p_i^l + \omega \frac{1}{a_{ii}} \left( \rho_0 - \rho_i^{adv} - \sum_j m_j \left( \sum_j d_{ij} p_j^l - d_{jj}p_j^l - \sum_{k \neq i} d_{jk} p_k^l \right) \nabla W_{ij} - \sum_b m_b \sum_j d_{ij} p_j^l \nabla W_{ij} \right)
+```
+
+
+
+
+
+
+
+
+
+
+@autodocs
+Modules = [TrixiParticles]
+Pages = [joinpath("schemes", "fluid", "implicit_incompressible_sph", "system.jl")]

examples/fluid/dam_break_2d.jl

@@ -108,7 +108,7 @@ stepsize_callback = StepsizeCallback(cfl=0.9)
 
 callbacks = CallbackSet(info_callback, saving_callback, stepsize_callback, extra_callback,
                         density_reinit_cb, saving_paper)
-
-sol = solve(ode, CarpenterKennedy2N54(williamson_condition=false),
+solver_function = CarpenterKennedy2N54(williamson_condition=false)
+sol = solve(ode, solver_function,
             dt=1.0, # This is overwritten by the stepsize callback
             save_everystep=false, callback=callbacks);

examples/fluid/dam_break_2d_IISPH.jl

@@ -0,0 +1,36 @@
+# 2D dam break simulation using an ImplicitIncompressible SPH system
+using TrixiParticles
+
+# Load setup from dam break example
+trixi_include(@__MODULE__,
+              joinpath(examples_dir(), "fluid", "dam_break_2d.jl"),
+              sol=nothing, ode=nothing)
+
+# Change smoohing kernel and length
+smoothing_length = 1.25 * fluid_particle_spacing
+smoothing_kernel = GaussianKernel{2}()
+
+# Calculate nu for the viscosity model
+nu = 0.02 * smoothing_length * sound_speed/8
+
+# Use IISPH as fluid system
+time_step=0.001
+min_iterations=10
+max_iterations=30
+IISPH_system = ImplicitIncompressibleSPHSystem(tank.fluid, smoothing_kernel,
+                                               smoothing_length, fluid_density,
+                                               viscosity=ViscosityAdami(nu=nu),
+                                               acceleration=(0.0, -gravity),
+                                               min_iterations=min_iterations,
+                                               max_iterations=max_iterations,
+                                               time_step=time_step)
+
+# Run the dam break simulation with these changes
+trixi_include(@__MODULE__,
+              joinpath(examples_dir(), "fluid", "dam_break_2d.jl"),
+              viscosity=ViscosityAdami(nu=nu),
+              fluid_system=IISPH_system,
+              boundary_density_calculator=PressureZeroing(),
+              state_equation=nothing,
+              callbacks=CallbackSet(info_callback, saving_callback),
+              solver_function=SymplecticEuler(), dt=time_step)

src/TrixiParticles.jl

@@ -61,7 +61,8 @@ include("preprocessing/preprocessing.jl")
 
 export Semidiscretization, semidiscretize, restart_with!
 export InitialCondition
-export WeaklyCompressibleSPHSystem, EntropicallyDampedSPHSystem, TotalLagrangianSPHSystem,
+export ImplicitIncompressibleSPHSystem, WeaklyCompressibleSPHSystem,
+       EntropicallyDampedSPHSystem, TotalLagrangianSPHSystem,
        BoundarySPHSystem, DEMSystem, BoundaryDEMSystem, OpenBoundarySPHSystem
 export BoundaryZone, InFlow, OutFlow, BidirectionalFlow
 export InfoCallback, SolutionSavingCallback, DensityReinitializationCallback,

src/general/semidiscretization.jl

@@ -502,7 +502,7 @@ function update_systems_and_nhs(v_ode, u_ode, semi, t; update_from_callback=fals
     end
 
     # Update NHS
-    @trixi_timeit timer() "update nhs" update_nhs!(semi, u_ode)
+    update_nhs!(semi, u_ode)
 
     # Second update step.
     # This is used to calculate density and pressure of the fluid systems
@@ -528,7 +528,8 @@ function update_systems_and_nhs(v_ode, u_ode, semi, t; update_from_callback=fals
         v = wrap_v(v_ode, system, semi)
         u = wrap_u(u_ode, system, semi)
 
-        update_final!(system, v, u, v_ode, u_ode, semi, t; update_from_callback)
+        update_final!(system, v, u, v_ode, u_ode, semi,
+                      t; update_from_callback)
     end
 end
 
@@ -564,6 +565,7 @@ end
 
 @inline source_terms(system) = nothing
 @inline source_terms(system::Union{FluidSystem, SolidSystem}) = system.source_terms
+@inline source_terms(system::ImplicitIncompressibleSPHSystem) = nothing
 
 @inline add_acceleration!(dv, particle, system) = dv
 
@@ -925,6 +927,15 @@ function check_configuration(system::TotalLagrangianSPHSystem, systems, nhs)
     end
 end
 
+function check_configuration(system::ImplicitIncompressibleSPHSystem, systems, nhs)
+    foreach_system(systems) do neighbor
+        if neighbor isa WeaklyCompressibleSPHSystem
+            throw(ArgumentError("`ImplicitIncompressibleSPHSystem` cannot be used together with
+            `WeaklyCompressibleSPHSystem`"))
+        end
+    end
+end
+
 function check_configuration(system::OpenBoundarySPHSystem, systems,
                              neighborhood_search::PointNeighbors.AbstractNeighborhoodSearch)
     (; boundary_model, boundary_zone) = system

src/io/write_vtk.jl

@@ -321,7 +321,6 @@ function write2vtk!(vtk, v, u, t, system::FluidSystem; write_meta_data=true)
         vtk["smoothing_kernel"] = type2string(system.smoothing_kernel)
         vtk["smoothing_length_factor"] = system.cache.smoothing_length_factor
         vtk["density_calculator"] = type2string(system.density_calculator)
-
         if system isa WeaklyCompressibleSPHSystem
             vtk["solver"] = "WCSPH"
 
@@ -343,6 +342,8 @@ function write2vtk!(vtk, v, u, t, system::FluidSystem; write_meta_data=true)
             vtk["state_equation_pa"] = system.state_equation.background_pressure
             vtk["state_equation_c"] = system.state_equation.sound_speed
             vtk["solver"] = "WCSPH"
+        elseif system isa ImplicitIncompressibleSPHSystem
+            vtk["solver"] = "IISPH"
         else
             vtk["solver"] = "EDAC"
             vtk["sound_speed"] = system.sound_speed

src/schemes/fluid/implicit_incompressible_sph/implicit_incompressible_sph.jl

@@ -0,0 +1 @@
+include("system.jl")