openswmm::fv Namespace Reference

Namespaces
namespace	implicit_wet
namespace	kernels
namespace	secgeom

Classes
struct	AcceptedWaterVolumes
class	ExplicitFvSolver
struct	FvClosure
	One section's closure. Fixed-size POD: inline arrays, no pointers. More...
struct	FvGeometry
	Cross-section closure for one conduit's cell chain. More...
struct	FvOptions
	Knobs for FLOW_ROUTING FV. More...
struct	FvStepForcing
class	ImplicitNetworkSolver
class	ImplicitReachSolver
class	INetworkSolver
	Abstract time integrator for the explicit FV 1D network. More...
struct	MeshBuildReport
class	NativeGmres
struct	NetworkMeshData
	SoA mesh geometry and topology for the FV network solver. More...
struct	NetworkStateData
	Mutable solver state — the conserved variables and the node volumes. More...
class	NonlinearPressureStorage
class	PressurizedHeadSolver
struct	PressurizedView

Enumerations
enum	: uint8_t {   kClosureTabulated = 0 ,   kClosurePolynomial = 1 }
	The closure's representation class. More...
enum class	RiemannSolver : int {   HLL = 0 ,   HLLC = 1 }
enum class	Limiter : int {   MINMOD = 0 ,   VANLEER = 1 ,   SUPERBEE = 2 }
	Slope limiter used by the second-order (MUSCL) reconstruction. More...
enum class	ScalarScheme : int {   UPWIND = 0 ,   MUSCL = 1 ,   QUICKEST_ULTIMATE = 2 }
	Reconstruction used for the advected scalar field. More...
enum class	TimeIntegration : int {   EULER = 0 ,   RK2 = 1 }
	Temporal integrator. More...
enum class	StructureCoupling : int {   SUBSTEP = 0 ,   ROUTING_STEP = 1 }
	When pump/orifice/weir/outlet structure equations are re-evaluated. More...
enum class	Backend : int {   CPU = 0 ,   AUTO = 1 ,   OMP = 2 ,   CUDA = 3 ,   HIP = 4 ,   SYCL = 5 }
	Backend selection for the solver kernels. More...
enum	: uint8_t {   kNodeJunction = 0 ,   kNodeVirtual = 1 ,   kNodeStorage = 2 ,   kNodeOutfall = 3 }
	Node kind codes stored in NetworkMeshData::node_kind. More...

Functions
double	advanceNetworkChecked (INetworkSolver &solver, double start, double target, const FvStepForcing &forcing)
	Advance an FV backend and verify it reached the full requested time window.
void	buildGeometry (const XSectParams &xs, bool is_open, double slot_celerity, FvGeometry &g, int barrels=1)
	Construct an aggregate-barrel FV section and its tapered slot/closure.
MeshBuildReport	buildNetworkMesh (SimulationContext &ctx, const FvOptions &opts, NetworkMeshData &mesh)
	Build the cell/face mesh and the per-conduit geometry closures.
std::unique_ptr< INetworkSolver >	makeNetworkSolver (const FvOptions &opts, std::string *chosen, int n_cells)
	Construct an uninitialized CPU or compatible plugin network solver.
double	limitSlope (double a, double b, Limiter lim)

Variables
constexpr int	kClosurePanels = 128
	Panels of the closure table, uniform in depth on [0, y_full].
constexpr int	kMaxLtsTiers = 8
constexpr int	kNodeVolSamples = 129
	Samples in a STORAGE node's flattened depth→volume table.

Enumeration Type Documentation

anonymous enum

anonymous enum : uint8_t

The closure's representation class.

Enumerator
kClosureTabulated
kClosurePolynomial

anonymous enum

anonymous enum : uint8_t

Node kind codes stored in NetworkMeshData::node_kind.

Enumerator
kNodeJunction
kNodeVirtual
kNodeStorage
kNodeOutfall

Backend

enum class openswmm::fv::Backend : int

strong

Backend selection for the solver kernels.

Enumerator
CPU
AUTO	default: try plugins above the parallel gate, else CPU
OMP
CUDA
HIP
SYCL

Limiter

enum class openswmm::fv::Limiter : int

strong

Slope limiter used by the second-order (MUSCL) reconstruction.

Enumerator
MINMOD	most diffusive, most robust (default)
VANLEER
SUPERBEE	sharpest; can artificially steepen smooth profiles

RiemannSolver

enum class openswmm::fv::RiemannSolver : int

strong

Face flux function. HLLC is the default and HLL exists only as a debugging/comparison baseline — see plan §3.2: the contact wave HLL averages away is exactly the wave that carries an advected scalar, so HLL is not a legitimate production choice once transport is on the same mesh.

Enumerator
HLL	baseline / debug only
HLLC	default

ScalarScheme

enum class openswmm::fv::ScalarScheme : int

strong

Reconstruction used for the advected scalar field.

Enumerator
UPWIND	1st-order upwind — monotone, diffusive
MUSCL	2nd-order TVD, one-cell stencil (default)
QUICKEST_ULTIMATE	3rd-order, two-cell upstream stencil

StructureCoupling

enum class openswmm::fv::StructureCoupling : int

strong

When pump/orifice/weir/outlet structure equations are re-evaluated.

Enumerator
SUBSTEP	every explicit substep (default; physically exact)
ROUTING_STEP	once per routing step (matches DW-scale control cadence)

TimeIntegration

enum class openswmm::fv::TimeIntegration : int

strong

Temporal integrator.

Enumerator
EULER	forward Euler (default)
RK2	SSP-RK2 (Heun) — strong-stability-preserving.

Function Documentation

advanceNetworkChecked()

double openswmm::fv::advanceNetworkChecked ( INetworkSolver & solver, double start, double target, const FvStepForcing & forcing )

inline

Advance an FV backend and verify it reached the full requested time window.

Throws invalid_argument for unordered/nonfinite times, runtime_error for nonfinite/backward/partial reached time; backend exceptions propagate. Does not roll back backend state or publish Router means/receipts.

Parameters

[in,out]	solver	Borrowed initialized backend; advance may mutate it even if reached-time validation later fails.
[in]	start	Finite accepted elapsed start seconds.
[in]	target	Finite target elapsed seconds>=start.
[in]	forcing	Borrowed boundary/source forcing valid for the requested window; lifetime covers advance.

Returns

Backend reached seconds after checking target agreement within16*epsilon*max(1,abs(start),abs(target)).

Here is the call graph for this function:

buildGeometry()

void openswmm::fv::buildGeometry	(	const XSectParams &	xs,
		bool	is_open,
		double	slot_celerity,
		FvGeometry &	g,
		int	barrels = 1 )

Construct an aggregate-barrel FV section and its tapered slot/closure.

Depth/radius remain per-barrel, area/width multiply max(1,barrels). Closed slot uses g*A/c^2, caps at5% full width (fallback width1ft) and floors nonpositive result1e-6ft; crown uses SLOT_CROWN_CUTOFF. Calls buildClosure to replace provisional crown area/integral. No complete finite/physical validation or ownership transfer.

Parameters

[in]	xs	Borrowed initialized native scalar section; optional table pointers copy as non-owning references and must remain live.
[in]	is_open	True extends the open section above full depth with vertical-wall width; false creates a pressurization slot.
[in]	slot_celerity	Design speed ft/s; closed-section calculation floors nonpositive/<=1/NaN to1 through comparison.
[in,out]	g	Borrowed destination overwritten with copied geometry, barrel-scaled area/width and constructed closure; unrelated fields not assigned here retain prior values.
[in]	barrels	Parallel barrel count; <=0 becomes1.

Here is the caller graph for this function:

buildNetworkMesh()

MeshBuildReport openswmm::fv::buildNetworkMesh	(	SimulationContext &	ctx,
		const FvOptions &	opts,
		NetworkMeshData &	mesh )

Build the cell/face mesh and the per-conduit geometry closures.

Parameters

ctx	Model context — links, nodes and conduit side-table must be fully resolved, and Router::init must already have computed mod_length/rough_factor (the mesh reuses the Courant lengthening as its Δx floor).
opts	FV options (cell length, min cells, slot celerity).
mesh	[out] Cleared and filled.

Returns

Build report; a non-empty errors list means the mesh is unusable.

Here is the call graph for this function:

Here is the caller graph for this function:

limitSlope()

double openswmm::transport::fvkernels::limitSlope ( double a, double b, Limiter lim )

inline

limitSlope moved to transport/fvkernels/SpeciesTransportKernels.hpp (phase E0) — shared between the species kernels and the hydrodynamic second-order reconstruction below.

makeNetworkSolver()

std::unique_ptr< INetworkSolver > openswmm::fv::makeNetworkSolver	(	const FvOptions &	opts,
		std::string *	chosen,
		int	n_cells )

Construct an uninitialized CPU or compatible plugin network solver.

OPENSWMM_FV_BACKEND overrides opts.backend after lowercasing; unknown names fall back to CPU. pressurized_implicit forces CPU even for an explicit plugin request. AUTO tries cuda, hip, sycl, omp in order unless below min_parallel_cells (overridden by atol(OPENSWMM_FV_MIN_PARALLEL_CELLS)). Plugin absence, load/probe/ABI/device/factory failures fall through to CPU. Filesystem/string/allocation errors and plugin-thrown exceptions can propagate; this is not a no-throw factory. Loaded modules remain mapped for process lifetime.

Parameters

[in]	opts	Options used synchronously for backend, minimum cell count and CPU-only pressurized_implicit selection; not retained.
[out]	chosen	Optional writable backend label, or nullptr; receives a CPU label or backend/device label before successful return.
[in]	n_cells	Mesh cell count for AUTO gating; <=0 disables that gate. Explicit backend requests bypass it.

Returns

An owning non-null solver on normal return; caller initializes it against the actual mesh/state/options.

Variable Documentation

kClosurePanels

int openswmm::fv::kClosurePanels = 128

inlineconstexpr

Panels of the closure table, uniform in depth on [0, y_full].

kMaxLtsTiers

int openswmm::fv::kMaxLtsTiers = 8

inlineconstexpr

Hard ceiling on LTS tiers. Tier k advances at 2^k·dt₀, so 8 tiers already span a 128× stiffness ratio — beyond that the coarse tier's lag behind a moving front stops being a bounded integration-path difference. (A 14-tier cap was trialled for the MIN_SURFAREA-junction stiffness on macdonald-long-sub and did not help: the stiff set there is every junction, and the scheduler's base-step count is the cost floor. The remedy is the algebraic junction interface, which removes the bound.)

kNodeVolSamples

int openswmm::fv::kNodeVolSamples = 129

inlineconstexpr

Samples in a STORAGE node's flattened depth→volume table.