CreepLaws
The following viscous creep laws are implemented:
GeoParams.MaterialParameters.ConstitutiveRelationships.LinearViscous Type
LinearViscous(η=1e20Pa*s)Defines a linear viscous creeplaw
The (isotopic) linear viscous rheology is given by
or
where
GeoParams.MaterialParameters.ConstitutiveRelationships.PowerlawViscous Type
PowerlawViscous(η0=1e18Pa*s, n=2.0NoUnits, ε0=1e-15/s)Defines a power law viscous creeplaw as:
where
GeoParams.MaterialParameters.ConstitutiveRelationships.DislocationCreep Type
DislocationCreep(n = 1.0NoUnits, r = 0.0NoUnits, A = 1.5MPa/s, E = 476.0kJ/mol, V = 6e-6m^3/mol, apparatus = AxialCompression )Defines the flow law parameter of a dislocation creep law.
The (isotropic) dislocation creep law, as used by experimentalists, is given by
where
is the power law exponent is the exponent of fugacity dependence is a pre-exponential factor (if manually defined, and must be either pre-defined or substituted) is the activation energy is the activation volume is the strain rate is the differential stress which are converted into second invariants using the Apparatusvariable that can be
either AxialCompression, SimpleShear or Invariant. If the flow law parameters are already given as a function of second invariants, choose Apparatus=Invariant.
Example
julia> x2 = DislocationCreep(n=3)
DislocationCreep: Name = , n=3.0, r=0.0, A=1.5e6 Pa⁻³·⁰ s⁻¹·⁰, E=476000.0 J mol⁻¹·⁰, V=6.0e-6 m³·⁰ mol⁻¹·⁰, FT=1.7320508075688772, FE=1.1547005383792517, Apparatus=1GeoParams.MaterialParameters.ConstitutiveRelationships.ViscosityPartialMelt_Costa_etal_2009 Type
ViscosityPartialMelt_Costa_etal_2009(LinearMeltViscosity())The viscosity of a partially molten rock depends on the melt viscosity, melt fraction and strainrate.
This implements a parameterisation of Costa et al. [2009].
Reference
Costa, A., Caricchi, L., Bagdassarov, N., 2009. A model for the rheology of particle‐bearing suspensions and partially molten rocks. Geochem Geophys Geosyst 10, 2008GC002138. https://doi.org/10.1029/2008GC002138
sourceGeoParams.MaterialParameters.ConstitutiveRelationships.LinearMeltViscosity Type
LinearMeltViscosity(η=1e20Pa*s)Defines a simple temperature-dependent melt viscosity, given by
or
where
and \eta_0 is the scaling viscosity, A and B are constants, and T_0 is a reference temperature, and T is the temperature [in K].
Typical parameters for basalt (default) are: A = -9.6012, B = 1.3374e+04K, T_0 = 307.8043K and \eta_0 = 1Pas.
Typical parameters for rhyolite are: A = -8.1590, B = 2.4050e+04K, T_0 = -430.9606K and \eta_0 = 1Pas.
GeoParams.MaterialParameters.ConstitutiveRelationships.GiordanoMeltViscosity Type
GiordanoMeltViscosity(; oxd_wt = oxd_wt, η0=1Pas)Defines the melt viscosity model after Giordano et al. (2008) given by
or
where
Parameters
oxd_wt::NTuple{9,T}: Melt composition as 9-element Tuple containing concentrations in [wt%] of the following oxides ordered in the exact sequence(SiO2 TiO2 Al2O3 FeO MgO CaO Na2O K2O H2O) Default values are for a hydrous N-MORB melt.
compute_εII/compute_τII (and their derivatives) also accept an mH2O keyword (melt water content, mass fraction) that overrides the struct's own oxd_wt[9] for that call, recomputing only the water-dependent aggregates. mH2O is accepted as given: GeoParams does not check it for consistency with any Solubility or other closure's dissolved-water output — that is the caller's/solver's responsibility.
Reference
- Giordano D, Russell JK, & Dingwell DB (2008). Viscosity of Magmatic Liquids: A Model. Earth & Planetary Science Letters, 271, 123-134. (https://dx.doi.org/10.1016/j.epsl.2008.03.038)
GeoParams.MaterialParameters.ConstitutiveRelationships.MeltWeakening Type
MeltWeakening(law; α)Creep law law weakened by a small melt fraction
so the viscosity at a given stress drops by
α is dimensionless and must be non-negative. The melt fraction is passed at runtime, e.g. compute_εII(law, τII; T, ϕ = 0.05); without it law. All other keyword arguments are passed on to law.
Example
julia> law0 = SetDiffusionCreep(GeoParams.Diffusion.wet_olivine_Mei_2000a);
julia> law = MeltWeakening(law0; α = 26);
julia> args = (; T = 1473.0, P = 300e6, d = 15e-6, f = 300e6);
julia> compute_εII(law, 1e6; args..., ϕ = 0.05) / compute_εII(law0, 1e6; args...)
3.6692966676192444References
Hirth, G., & Kohlstedt, D. L. (1995a). Experimental constraints on the dynamics of the partially molten upper mantle: Deformation in the diffusion creep regime. Journal of Geophysical Research, 100(B2), 1981–2001. https://doi.org/10.1029/94JB02128
Hirth, G., & Kohlstedt, D. L. (1995b). Experimental constraints on the dynamics of the partially molten upper mantle: 2. Deformation in the dislocation creep regime. Journal of Geophysical Research, 100(B8), 15441–15449. https://doi.org/10.1029/95JB01292
Mei, S., Bai, W., Hiraga, T., & Kohlstedt, D. L. (2002). Influence of melt on the creep behavior of olivine–basalt aggregates under hydrous conditions. Earth and Planetary Science Letters, 201, 491–507. https://doi.org/10.1016/S0012-821X(02)00745-8
GeoParams.MaterialParameters.ConstitutiveRelationships.DiffusionCreep Type
DiffusionCreep(r = 0NoUnits, p = A = 1.5MPa/s, E = 476.0kJ/mol, V = 6e-6m^3/mol, apparatus = AxialCompression )Defines the flow law parameter of a dislocation creep law.
The (isotropic) diffusion creep law, as used by experimentalists, is given by
where
is the grain size (default 5 mm), stored in the law is the exponent of fugacity dependence is the exponent of grain size is a pre-exponential factor (if manually defined, and must be either pre-defined or substituted) is the activation energy is the activation volume is the strain rate is the differential stress
The experimental parameters are converted into second invariants using the Apparatus variable that can be either AxialCompression, SimpleShear or Invariant. If the flow law parameters are already given as a function of second invariants, choose Apparatus=Invariant.
The grain size used in the computations is the stored field d, unless a grain size is passed at runtime as keyword (or in the args NamedTuple), e.g. compute_εII(law, τII; T, P, d = 1e-4), in which case the runtime value (scalar or array) overrides the stored one. In the Unitful methods d must be a length (e.g. d = 100μm).
Example
julia> x2 = DiffusionCreep(Name="test")
DiffusionCreep: Name = test, n=1.0, r=0.0, p=-3.0, A=1.5e6 m³·⁰ Pa⁻¹·⁰ s⁻¹·⁰, E=500000.0 J mol⁻¹·⁰, V=2.4e-5 m³·⁰ mol⁻¹·⁰, FT=1.7320508075688772, FE=1.1547005383792517GeoParams.MaterialParameters.ConstitutiveRelationships.SetDiffusionCreep Function
SetDiffusionCreep["Name of Diffusion Creep"]This is a dictionary with pre-defined creep laws
sourceGeoParams.MaterialParameters.ConstitutiveRelationships.GrainBoundarySliding Type
GrainBoundarySliding(; d = 1mm, n = 3.5, p = -2, A = 6.5e-30Pa^(-3.5) * m^2 / s, E = 400kJ/mol, V = 18e-6m^3/mol, Apparatus = AxialCompression)Defines the flow law parameters of grain boundary sliding (GBS).
The grain boundary sliding, as used by experimentalists, is given by
where
is the grain size (default 1 mm), stored in the law is the exponent of grain size is a pre-exponential factor (if manually defined, must be either pre-defined or substituted) is the activation energy is the activation volume is the strain rate is the differential stress
The experimental parameters are converted into second invariants using the Apparatus variable that can be either AxialCompression, SimpleShear or Invariant. If the flow law parameters are already given as a function of second invariants, choose Apparatus=Invariant.
The grain size used in the computations is the stored field d, unless a grain size is passed at runtime as keyword (or in the args NamedTuple), e.g. compute_εII(gbs, τII; T, P, d = 1e-4), in which case the runtime value (scalar or array) overrides the stored one. In the Unitful methods d must be a length (e.g. d = 100μm).
Example
julia> x2 = GrainBoundarySliding(Name="test")
GrainBoundarySliding: Name = test, d=0.001 m, n=3.5, p=-2.0, A=6.5e-30 m²·⁰ Pa⁻³·⁵ s⁻¹·⁰, E=400000.0 J mol⁻¹·⁰, V=1.8e-5 m³·⁰ mol⁻¹·⁰, FT=1.7320508075688772, FE=1.1547005383792517GeoParams.MaterialParameters.ConstitutiveRelationships.SetGrainBoundarySliding Function
SetGrainBoundarySliding["Name of GBS"]This is a dictionary with pre-defined creep laws
sourceGeoParams.MaterialParameters.ConstitutiveRelationships.VolumeBoundaryDiffusionCreep Type
VolumeBoundaryDiffusionCreep(; Name = "", d = 1mm, D0v = 9.1e-4m^2/s, Qv = 59.4kJ/mol, D0b = 6.0e-4m^2/s,
Qb = 49kJ/mol, δ = 9.04e-10m, Vm = 1.97e-5m^3/mol, Apparatus = AxialCompression)Diffusion creep by volume (Nabarro-Herring) plus grain-boundary (Coble) diffusion, in the form of eq. (4) of Goldsby & Kohlstedt (2001); the defaults are their values for ice,
where
is the differential stress and the axial strain rate of the experiments is the grain size (default 1 mm), stored in the law , are the pre-exponential factor and activation energy of volume diffusion , are those of grain-boundary diffusion, and is the grain-boundary width is the molar volume.
The prefactor scales with DiffusionCreep cannot express (see #52). The law is linear in stress. Apparatus converts it to second invariants as for the other creep laws; with AxialCompression (the default, matching the uniaxial experiments)
where
The grain size used in the computations is the stored field d, unless a grain size is passed at runtime, e.g. compute_εII(law, τII; T, d = 1e-4), which overrides it.
GeoParams.MaterialParameters.ConstitutiveRelationships.SetVolumeBoundaryDiffusionCreep Function
SetVolumeBoundaryDiffusionCreep(name; d, D0v, Qv, D0b, Qb, δ, Vm)
SetVolumeBoundaryDiffusionCreep(name, CharDim; ...)Returns the pre-defined volume and grain-boundary diffusion creep law name (e.g. GeoParams.VolumeBoundaryDiffusion.ice_Goldsby_2001) in SI units, nondimensionalized if CharDim is given. Keywords override the stored parameters.
Computational routines for creep laws
Grain-size sensitive laws (DiffusionCreep, GrainBoundarySliding, VolumeBoundaryDiffusionCreep) store a grain size d (by default 5 mm for DiffusionCreep and 1 mm for the others; change it with e.g. SetDiffusionCreep(law; d = 100μm)). It is used unless a grain size is passed at runtime, e.g. compute_εII(law, τII; T, P, d = 1.0e-4) or compute_εII(law, τII, (; T, P, d)), in which case the runtime value (a scalar, or an array in the in-place ! routines) overrides the stored one.
Once a creep rheology is defined, we can use the following routines to perform computations within the solvers
GeoParams.MaterialParameters.ConstitutiveRelationships.compute_εII Function
compute_εII(a::DiffusionCreep, TauII; T, P = 0, f = 1, d = a.d, kwargs...)Returns diffusion creep strainrate as a function of 2nd invariant of the stress tensor
The grain size d defaults to the value stored in a; a runtime d (scalar) overrides it.
compute_εII(a::GrainBoundarySliding, TauII; T, P = 0, d = a.d, kwargs...)Returns grain boundary sliding strainrate as a function of 2nd invariant of the stress tensor
The grain size d defaults to the value stored in a; a runtime d (scalar) overrides it.
compute_εII(a::VolumeBoundaryDiffusionCreep, TauII; T, d = a.d, kwargs...)Returns the strain rate of volume and grain-boundary diffusion creep for the second invariant of the deviatoric stress TauII; see VolumeBoundaryDiffusionCreep.
compute_εII(s::ConstantElasticity{_T}, τII; τII_old, dt)Computes elastic strainrate given the deviatoric stress at the current (τII) and old timestep (τII_old), for a timestep dt:
Note that we here solve the scalar equation, which is sufficient for isotropic cases. In tensor form, it would be
here
compute_εII(p::DruckerPrager{_T,U,U1}, λdot::_T, τII::_T, P)This computes plastic strain rate invariant for a given
compute_εII(p::DruckerPrager_regularised{_T,U,U1}, λdot::_T, τII::_T, P)This computes plastic strain rate invariant for a given
compute_εII(v::Parallel{T,N}, τII, args; tol=1e-6, verbose=false, n=1)Computing εII as a function of τII for a Parallel elements is (usually) a nonlinear problem
compute_εII(v::CompositeRheology{T,N}, τII, args; tol=1e-6, verbose=false, n=1)Computing εII as a function of τII for a composite element is the sum of the individual contributions
compute_εII(v::AbstractPlasticity, τII::_T, args; tol=1e-6, verbose=false, max_iter=100)Performs local iterations to compute the plastic strainrate. Note that the non-plastic strainrate, ε_np, should be part of args. Throws an error if the tolerance is not reached within max_iter iterations.
GeoParams.MaterialParameters.ConstitutiveRelationships.compute_εII! Function
compute_εII!(EpsII::AbstractArray{_T,N}, a::DiffusionCreep, TauII::AbstractArray{_T,N}; T, P, f, d, kwargs...)Computes strainrate as a function of stress. T, P, f and d can be scalars or arrays; if d is not given, the grain size stored in a is used.
compute_εII!(EpsII::AbstractArray{_T,N}, a, TauII::AbstractArray{_T,N}; T, P, f,d,kwargs...)Computes strainrate as a function of stress
sourcecompute_εII!(EpsII::AbstractArray{_T,N}, s::LinearMeltViscosity, TauII::AbstractArray{_T,N}; T, kwargs...)compute_εII!(EpsII::AbstractArray{_T,N}, s::ViscosityPartialMelt_Costa_etal_2009, TauII::AbstractArray{_T,N}; T, kwargs...)compute_εII!(EpsII::AbstractArray{_T,N}, s::GiordanoMeltViscosity, TauII::AbstractArray{_T,N}; T, kwargs...)compute_εII!(EpsII::AbstractArray{_T, N}, a::HerschelBulkley, TauII::AbstractArray; T = one(_T), kwargs...)In-place function for the second invariant of the strain rate for Herschel-Bulkley rheology.
T may be a scalar, applied to every element, or an array indexed alongside TauII.
compute_εII!(EpsII::AbstractArray, s::LinearViscous, TauII::AbstractArray)compute_εII!(EpsII::AbstractArray{_T,N}, s::ArrheniusType, TauII::AbstractArray{_T,N})compute_εII!(ε_el::AbstractArray{_T,N}, s::ConstantElasticity{_T}; τII::AbstractArray{_T,N}, τII_old::AbstractArray{_T,N}, dt::_T, kwargs...)In-place computation of the elastic shear strainrate for given deviatoric stress invariants at the previous (τII_old) and new (τII) timestep, as well as the timestep dt
GeoParams.MaterialParameters.ConstitutiveRelationships.compute_τII Function
compute_τII(a::DislocationCreep, EpsII; P, T, f, args...)Computes the stress for a Dislocation creep law given a certain strain rate
sourcecompute_τII(a::DiffusionCreep, EpsII; T, P = 0, f = 1, d = a.d, kwargs...)Returns diffusion creep stress as a function of 2nd invariant of the strain rate
sourcecompute_τII(a::GrainBoundarySliding, EpsII; T, P = 0, d = a.d, kwargs...)Returns grain boundary sliding stress as a function of 2nd invariant of the strain rate The grain size d defaults to the value stored in a; a runtime d (scalar) overrides it.
compute_τII(a::VolumeBoundaryDiffusionCreep, EpsII; T, d = a.d, kwargs...)Returns the second invariant of the deviatoric stress of volume and grain-boundary diffusion creep for the strain rate EpsII.
compute_τII(a::PeierlsCreep, EpsII; P, T, f, args...)Computes the stress for a peierls creep law given a certain strain rate.
sourcecompute_τII(s::LinearMeltViscosity, EpsII; kwargs...)Returns second invariant of the stress tensor given a 2nd invariant of strain rate tensor
sourcecompute_τII(s::ViscosityPartialMelt_Costa_etal_2009, EpsII; kwargs...)Returns second invariant of the stress tensor given a 2nd invariant of strain rate tensor
sourcecompute_τII(s::GiordanoMeltViscosity, EpsII; kwargs...)Returns second invariant of the stress tensor given a 2nd invariant of strain rate tensor
sourcecompute_τII(a::HerschelBulkley, EpsII; T = one(precision(a)), kwargs...)compute_τII(s::LinearViscous, EpsII; kwargs...)Returns second invariant of the stress tensor given a 2nd invariant of strain rate tensor
sourcecompute_τII(s::ArrheniusType, EpsII; kwargs...)Returns second invariant of the stress tensor given a 2nd invariant of strain rate tensor
sourcecompute_τII(s::PowerlawViscous, EpsII; kwargs...)Returns second invariant of the stress tensor given a 2nd invariant of strain rate tensor
sourceτII = compute_τII(v::CompositeRheology{T,N}, εII, args; tol=1e-6, verbose=false)GeoParams.MaterialParameters.ConstitutiveRelationships.compute_τII! Function
compute_τII!(TauII::AbstractArray{_T,N}, a::DislocationCreep, EpsII::AbstractArray;
P = zero(_T),
T = one(_T),
f = one(_T))Computes the deviatoric stress invariant for a dislocation creep law.
Each of P, T, and f may be a scalar, applied to every element, or an array indexed alongside EpsII.
compute_τII!(TauII::AbstractArray{_T,N}, a::PeierlsCreep, EpsII::AbstractArray;
T = one(_T), args...)Computes the deviatoric stress invariant for a peierls creep law.
T may be a scalar, applied to every element, or an array indexed alongside EpsII.
compute_τII!(TauII::AbstractArray{_T, N}, a::HerschelBulkley, EpsII::AbstractArray; T = one(_T), kwargs...)In-place function for the second invariant of the stress for Herschel-Bulkley rheology.
T may be a scalar, applied to every element, or an array indexed alongside EpsII.
compute_τII!(τII::AbstractArray{_T,N}, s::ConstantElasticity{_T}. ε_el::AbstractArray{_T,N}; τII_old::AbstractArray{_T,N}, dt::_T, kwargs...)In-place update of the elastic stress for given deviatoric strainrate invariants and stres invariant at the old (τII_old) timestep, as well as the timestep dt
Water fugacity and water content
Wet flow laws take the water term through the argument f (default 1), as in compute_εII(law, τII; T, P, f). For laws written with the water fugacity f is always passed in Pa (for example from compute_water_fugacity); the database stores A in f = C_OH in H/10⁶ Si; at present these are Dislocation.wet_olivine1_Hirth_2003 and Diffusion.wet_olivine1_Hirth_2003 (Hirth & Kohlstedt 2003, constant f sets it to 1 (1 Pa or 1 H/10⁶ Si), which for a wet law is rarely intended.
Refitted database entries
Database entries whose name ends in _refit hold a prefactor A refitted so that the law reproduces the paper's own data where the printed equation does not; the printed law is kept under the original name, and both comments give the evidence. At present these are Dislocation.dry_olivine_Hirth_2003_refit (A = 2.8e5 instead of 1.1e5 Dislocation.wet_olivine_Karato_2003_refit (10^2.7 instead of 10^2.9 Dislocation.wet_quartzite_Lu_2019_refit (1.0e-15 instead of 6.0e-15 Dislocation.felsic_granulite_Wen_2021_refit (10^-4.7 instead of 10^-4.4 Dislocation.ice_dislocation_cold_Goldsby_2001_refit (5.4e4 instead of 4.0e5