Using the DG2-RANS-k-ε turbluent flow solver (resp. the DG2-RANS laminar flow solver)

The DG2-RANS-k-ε turbulent flow solver (resp. DG2-RANS laminar flow solver) can be set-up and run on LISFLOOD-FP8.0 in the same way as any other uniform-grid solver (see “Input files and their format”).

In addition, a number of extra parameters need to be typed and initialised in LISFLOOD-FP8.0’s *.par file. These extra parameters are described in the Tale below:

Item name input	Description
turbulent	Boolean keyword instructing LISFLOOD-FP to use DG2-RANS-k-ε
laminar	Boolean keyword instructing LISFLOOD-FP to use DG2-RANS
inviscid	Boolean keyword instructing LISFLOOD-FP to use the inviscid flow solver (DG2-SWE). By default, this flow solver is selected if none of laminar, turbulent and inviscid are pointed in *.par file
DtRatioLimit	Keyword followed by a decimal number. It sets the upper bound limit for the turbulent eddy viscosity $D_t$ = DtRatioLimit × $D_m$. The default value is 10,000 which means $D_t$ =0.01
TurbIntensity	Keyword followed by a decimal number. It is the turbulence intensity $I$ and is used in the turbulent inflow boundary condition calculations. It is between 0 to 20%. The default value is 1%. The guidance for its value: Low-turbulence cases: 0%<I≤1% Medium-turbulence cases: 1%<I≤5% High-turbulence cases: 5%<I≤20%
TurbViscousityRatio	Keyword followed by a decimal number. It specifies the turbulent viscosity ratio $D_r$ which is used in the turbulent inflow boundary condition calculation. It is usually $D_r$ < 10. Its default value is 0.1
turboutput	Boolean keyword instructing LISFLOOD-FP to write out ASCII raster files of the turbulent-flow variables $Hk$, H$\varepsilon$ and $D_t$. Their average values will be printed in files with the extensions of *.Hk, *.He and *.Dt. Their slope coefficients will be in files with the extensions of *.Hk1x, *.Hk1y, *.He1x and *.He1y. Without turboutput, there will not be any outputs for the turbulence-flow variables.
startHkHe2d	Boolean keyword instructing LISFLOOD-FP to read initial condition files for Hk and H$\varepsilon$ in ASCII raster format. It will read files with the extensions of *.Hk, *.Hk1x, *.Hk1y, *.He, *.He1x and *.He1y. Without startHkHe2d, $k$ = $k_{tol}$ = $10^{-12}$ and $\varepsilon$ = $\varepsilon_{tol}$ = $10^{-12}$ will be used.

By default, the DG2-RANS-k-ε turbulent flow solver (resp. DG2-RANS laminar flow solver) will be run on multicore CPU. It is also possible to make the run on the GPU by further typing the item cuda in the the *.par file (see “Parameter file (.par)”).

The technicalities underpinning the turbulent flow DG2-RANS-k-ε solver (resp. laminar flow DG2-RANS solver), as well as the simpler (invicid) DG2-SWE solver, are all described in a pre-print Kesserwani et al. 2025 with validation for turbulent and laminar flow experimental benchmarks involving quasi-steady vortical structures. This is the pre-print accepted for publication in Advances in Water Resources.

Videos demonstrating the turbulent flow DG2-RANS-k-ε solver’s (resp. laminar flow DG2-RANS solver’s) performance for the most challenging test cases: a video will start after clicking on a test-case image from the list below.

Test case	Reynolds number	Image
Surface-piercing: Vortex shedding past a conical island (DG2-RANS-k-ε-medium)	5,175
Submerged: Vortex shedding past a conical island (DG2-RANS-k-ε-fine)	6,210
Compoundeddies: Flow past a square block in a diverting T-junction (DG2-RANS-k-ε-fine)	7,432
Laminar wake interactions past cylinders (DG2-RANS-medium)	220

“Recirculation flow in sharp building cavities”

In contrast to all the other experimental benchmarks, this benchmark involves a skewed (non-uniform distribution of) infow velocity components. Therefore, resolution-specific pre-processing of the inflow boudary was applied and can be activated by further typing these two items.

Item name input	Description
cavity_coarse	Boolean keyword instructing LISFLOOD-FP to simulate the “Recirculation flow in sharp building cavities” benchmark test using the coarse grid (0.32 m) -
cavity_fine	Boolean keyword instructing LISFLOOD-FP to simulate the “Recirculation flow in sharp building cavities” benchmark test using the fine grid (0.16 m) -skewed infow is resolution specific

back