\tau | \Vert \phi\Vert_{H^1} | Rate | \Vert p\Vert_{L^2} | Rate |
0.007812 | 0.102393 | 0.469438 | ||
0.003906 | 0.0653769 | 0.647262 | 0.282333 | 0.733534 |
0.001953 | 0.0367805 | 0.829842 | 0.143976 | 0.971574 |
0.000976 | 0.0196276 | 0.906053 | 0.0725853 | 0.988076 |
In this paper, we consider the numerical approximations of the Cahn-Hilliard phase field model for two-phase incompressible flows with variable density. First, a temporal semi-discrete numerical scheme is proposed by combining the fractional step method (for the momentum equation) and the convex-splitting method (for the free energy). Second, we prove that the scheme is unconditionally stable in energy. Then, the L2 convergence rates for all variables are demonstrated through a series of rigorous error estimations. Finally, by applying the finite element method for spatial discretization, some numerical simulations were performed to demonstrate the convergence rates and energy dissipations.
Citation: Mingliang Liao, Danxia Wang, Chenhui Zhang, Hongen Jia. The error analysis for the Cahn-Hilliard phase field model of two-phase incompressible flows with variable density[J]. AIMS Mathematics, 2023, 8(12): 31158-31185. doi: 10.3934/math.20231595
[1] | Saulo Orizaga, Ogochukwu Ifeacho, Sampson Owusu . On an efficient numerical procedure for the Functionalized Cahn-Hilliard equation. AIMS Mathematics, 2024, 9(8): 20773-20792. doi: 10.3934/math.20241010 |
[2] | Jean De Dieu Mangoubi, Mayeul Evrard Isseret Goyaud, Daniel Moukoko . Pullback attractor for a nonautonomous parabolic Cahn-Hilliard phase-field system. AIMS Mathematics, 2023, 8(9): 22037-22066. doi: 10.3934/math.20231123 |
[3] | Martin Stoll, Hamdullah Yücel . Symmetric interior penalty Galerkin method for fractional-in-space phase-field equations. AIMS Mathematics, 2018, 3(1): 66-95. doi: 10.3934/Math.2018.1.66 |
[4] | Wei Li, Guangying Lv . A fully-decoupled energy stable scheme for the phase-field model of non-Newtonian two-phase flows. AIMS Mathematics, 2024, 9(7): 19385-19396. doi: 10.3934/math.2024944 |
[5] | Pierluigi Colli, Gianni Gilardi, Jürgen Sprekels . Distributed optimal control of a nonstandard nonlocal phase field system. AIMS Mathematics, 2016, 1(3): 225-260. doi: 10.3934/Math.2016.3.225 |
[6] | Haifeng Zhang, Danxia Wang, Zhili Wang, Hongen Jia . A decoupled finite element method for a modified Cahn-Hilliard-Hele-Shaw system. AIMS Mathematics, 2021, 6(8): 8681-8704. doi: 10.3934/math.2021505 |
[7] | Joseph L. Shomberg . Well-posedness and global attractors for a non-isothermal viscous relaxationof nonlocal Cahn-Hilliard equations. AIMS Mathematics, 2016, 1(2): 102-136. doi: 10.3934/Math.2016.2.102 |
[8] | Hyun Geun Lee . A mass conservative and energy stable scheme for the conservative Allen–Cahn type Ohta–Kawasaki model for diblock copolymers. AIMS Mathematics, 2025, 10(3): 6719-6731. doi: 10.3934/math.2025307 |
[9] | Amber Yousaf Dar, Nadia Saeed, Moustafa Omar Ahmed Abu-Shawiesh, Saman Hanif Shahbaz, Muhammad Qaiser Shahbaz . A new class of ratio type estimators in single- and two-phase sampling. AIMS Mathematics, 2022, 7(8): 14208-14226. doi: 10.3934/math.2022783 |
[10] | Paola F. Antonietti, Benoît Merlet, Morgan Pierre, Marco Verani . Convergence to equilibrium for a second-order time semi-discretization ofthe Cahn-Hilliard equation. AIMS Mathematics, 2016, 1(3): 178-194. doi: 10.3934/Math.2016.3.178 |
In this paper, we consider the numerical approximations of the Cahn-Hilliard phase field model for two-phase incompressible flows with variable density. First, a temporal semi-discrete numerical scheme is proposed by combining the fractional step method (for the momentum equation) and the convex-splitting method (for the free energy). Second, we prove that the scheme is unconditionally stable in energy. Then, the L2 convergence rates for all variables are demonstrated through a series of rigorous error estimations. Finally, by applying the finite element method for spatial discretization, some numerical simulations were performed to demonstrate the convergence rates and energy dissipations.
In this paper, we focus on the following hydrodynamically consistent Cahn-Hilliard phase field model for incompressible two-phase flows with variable density
ρt+∇⋅(ρu)=0, | (1.1a) |
ϕt+u⋅∇ϕ−γΔw=0, | (1.1b) |
w=−Δϕ+1ε2(ϕ3−ϕ), | (1.1c) |
ρ(ut+u⋅∇u)−ηΔu+∇p−λw∇ϕ=0, | (1.1d) |
∇⋅u=0, | (1.1e) |
where ϕ, w, u and p are the phase field function, the chemical potential, the velocity of flow and pressure respectively. Moreover, ρ=12(ρ1+ρ2)+ϕ2(ρ1−ρ2) is the density, where ρ1 and ρ2 are the densities of the two fluids. The parameters γ, ε, η and λ are the mobility parameter related to the relaxation time scale, the interface thickness [1], the viscosity of the field, and the mixing energy density respectively. The system (1.1) is supplemented with appropriate boundary and initial conditions which are given as
ϕ(x,0)=ϕ0(x), ρ(x,0)=ρ0(x), u(x,0)=u0(x),ϕ|∂Ω=0, u|∂Ω=0. |
This model is also called the Cahn-Hilliard-Navier-Stokes model when the density is constant, which has many practical applications in physical and engineering, such as wetting, coating, and painting. By considering the influence of variable density, this model has broad applicability, which include highly stratified flows, interfaces between fluids of different densities and some problems of inertial confinement.
The phase field method which has a wide range of applications is one of the main methods to deal with the fluid interface in two-phase flow modeling; see [2,3,4,5] and the references therein. It was initially developed to simulate solid-liquid phase transitions, where the interface is treated as a thin, smooth transition layer[6,7,8] to remove the singularities between two phases. The basic framework is to use phase field variables to represent the volume fraction of fluid components and then adopt a variational form to derive the model. Recently, it has increasingly attracted researchers' interest, mainly because the phase field method is superior to other available methods in some aspects of two-phase flow [9]. Various material properties or complex interface behaviors can be simulated directly by introducing suitable energy functions. Numerical solutions play a crucial role in their study and applications because the analytic solutions are usually not available.
A few works have been devoted to the design, analysis and implementation of numerical schemes for the phase field model, although this model is very classical and canonical. We briefly review the available methods used in the phase field model. It is worth noting that there are projection/gauge/penalty methods [10,11,12], scalar auxiliary variables [13,14], linear stability [15,16], convex splitting [17,18,19], invariable energy quadratization(IEQ) [20,21], nonlinear quadratic [22], exponential time differencing [23], etc. In practical applications, we often couple the flow-field equation with the phase field equation. Typically two-phase incompressible flow models are coupled to phase-field models. The Cahn-Hilliard model is taken into account in this paper, because it is effective in the following two aspects: (ⅰ) Cahn-Hilliard models can accurately conserve the volume and dynamics; (ⅱ) the equation is one of the most important models in mathematical physics. Because of these reasons, here we use the Cahn-Hilliard equation developed in [24] to couple the incompressible flows with variable density.
There also have been a lot of works on numerical approximates for the Cahn-Hilliard phase field model for incompressible two-phase flows with variable density. Hohenberg and Halperin proposed the model in [25] to simulate two incompressible viscous fluids with constant density. In [26], Gurtin et al. obtained the equal model by using the framework of rational continuum mechanics. A fully adaptive energy stabilization scheme is proposed in [27]. An efficient Picard iteration procedure was designed in [28] to further decouple the model. In the last years, many authors have been concerned with designing incompressible two-phase flow models with variable densities. Several efficient and energy-stable time discretization schemes for the coupled nonlinear Cahn-Hilliard phase field system with variable density are constructed; see [29]. Yang and Dong presented an energy-stable scheme in [30] for the numerical approximation of the two-phase governing equations with variable density and viscosity for the two fluids by introducing a scalar-valued variable related to the total of the kinetic energy and the potential free energy. A second-order accurate, coupled, energy-stable schemeis proposed in [31], where the Crank-Nicolson method and the IEQ method were used. In [32], the conservation scheme of the first-order energy law was established, in which the Cahn-Hilliard solver was used to decouple from the two-phase incompressible flows solver through the use of the fractional step method. Ye et al.[33] have designed a fully-decoupled type scheme to solve the Cahn-Hilliard phase field model for a two-phase incompressible fluid flow system with constant density. They only give a detailed practical implementation method and also prove the solvability. None of the various existing schemes have been subjected to error analysis, where the main difficulty lies in the delicate treatment of a several of nonlinear terms. Rigorous error estimates of models with variable densities, using an optimal order error bound, may seem to be a difficult prospect, but is a very interesting direction for future research.
In this paper, we finally arrive at an unconditionally stable in energy, first-order time-accurate scheme for the incompressible Cahn-Hilliard two phase flows with variable density by coming up with a fractional step method. This method has an advantage over the projection method in that the original boundary conditions of the problem can be implemented in all substeps of the scheme. The popular approach to discretizing the Cahn-Hilliard phase field model (1.1b) and (1.1c) in time is based on the convex-splitting of the free energy functional, i.e., an idea that can be traced back to [32]. In the convex-splitting framework, one treats the contribution from the convex part implicitly and the contribution from the concave part explicitly. This treatment promotes the energy stability of the scheme and this property is unconditional in terms of time steps. We also give a rigorous proof of the convergence results and error estimates in the theoretical analysis. The main contribution of this paper is a rigorous error analysis, particularly under the condition that energy stability is available. To the best of the authors' knowledge, the proof developed in this article is the first to have the description of error estimation. The accuracy and stability are also demonstrated through the simulation of various numerical examples, where the challenge is in creating an efficient and easy to implement numerical scheme that preserves the energy dissipation law.
The rest of this article is organized as follows. In Section 2, we construct an efficient time discrete scheme for variable density and derive unconditional energy stability. In Section 3, the error analysis of the semi-discrete scheme in time is provided. Some numerical experimentations are given in Section 4. Finally, conclusions are drawn in Section 5.
For the sake of simplicity, some notations are needed for the following content. We assume that the domain Ω∈R2 is open, sufficiently smooth and bounded. For any two functions ϕ(x) and ψ(x), their L2 inner product on Ω is denoted by (ϕ,ψ)=∫Ωϕ(x)ψ(x)dx, and the L2 norm of ϕ(x) is denoted by ‖ϕ‖=(ϕ,ϕ)12. Let τ>0 be the time step size and set tn=nτ for 0≤n≤N with T=Nτ. Moreover, we introduce the following spaces,
H={u∈L2(Ω)|divu in Ω, u⋅n=0 on ∂Ω},V=H10, V0={u∈V|divu=0 in Ω},M=L20(Ω)={q∈L2(Ω)|∫Ωqdx=0}, |
where n is the outward normal vector of ∂Ω. Next, we reformulate the system (1.1) as follows:
ρn+1−ρnτ+∇ρn+1⋅un=0, | (2.1a) |
ϕn+1−ϕnτ+˜un+1⋅∇ϕn−γΔwn+1=0, | (2.1b) |
wn+1=1ε2((ϕn+1)3−ϕn)−Δϕn+1, | (2.1c) |
ρn˜un+1−unτ−ηΔ˜un+1+ρn+1(un⋅∇)˜un+1−λwn+1∇ϕn+14ρn+1(∇⋅un)˜un+1=0, | (2.1d) |
ρn+1un+1−˜un+1τ−η(Δun+1−Δ˜un+1)+∇pn+1=0, | (2.1e) |
∇⋅un+1=0 | (2.1f) |
Remark 1. In (2.1d), the term 14ρn+1(∇⋅un)˜un+1 is added to obtain the unconditional stability, as it is 0 if ∇⋅un=0.
Theorem 1. (Stabilityofρ) For any τ>0 and any sequence {un}n=0,…,N satisfying the boundary condition un⋅n=0 on ∂Ω, the solution {ρn}n=1,…,N to (2.1a) satisfies
‖ρN‖2+N−1∑n=1‖ρn+1−ρN‖2=‖ρ0‖2. | (2.2) |
Proof. Testing (2.1a) with 2τρn+1 gives
‖ρn+1‖2−‖ρn‖2+‖ρn+1−ρn‖2+2τ∫Ω(un⋅∇ρn+1+12ρn+1∇⋅un)ρn+1dx=0. | (2.3) |
Owing to the boundary conditions on un, we note that
∫Ω(un⋅∇ρn+1+12ρn+1∇⋅un)ρn+1dx=12∫Ω∇⋅(|ρn+1|2un)dx=12∫∂Ω|ρn+1|2un⋅ndx=0. | (2.4) |
Thus, we get
‖ρn+1‖2+‖ρn+1−ρn‖2=‖ρn‖2. |
Summing all indices n ranging 0 to N−1, the proof is completed.
Next, the stability of (2.1b)–(2.1e) will be proved in the theorem below. Moreover, since the kinetic energy of the fluid is 12‖√ρnun‖2, it is more suitable to establish a bound based on ‖√ρnun‖2 than on the velocity itself. For simplicity, let us say that σn=√ρn for all 1≤n≤N and σ0=√ρ0.
Theorem 2. (Stabilityofenergy) For any τ>0, (2.1b)–(2.1e) satisfy the the conditions of following energy estimates:
‖σNuN‖2+λ‖∇ϕN‖2−λε2‖ϕN‖2+λ2ε2‖ϕN‖4L4+2τγλN−1∑n=0‖∇wn+1‖2+λN−1∑n=0(‖∇(ϕn+1−ϕn)‖2+12ε2‖ϕn+1−ϕn‖2)+λ2ε2N−1∑n=0(‖ϕn+1‖2−‖ϕn‖)2+N−1∑n=0(‖σn(˜un+1−un)‖2+‖σn+1(un+1−˜un+1)‖2)+ητN−1∑n=0(‖∇un+1‖2+‖∇˜un+1‖2+‖∇(un+1−˜un+1)‖2)+λε2N−1∑n=0(‖ϕn+1(ϕn+1−ϕn)‖2)=‖σ0u0‖2+λ‖∇ϕ0‖2−λε2‖ϕ0‖2+λ2ε2‖ϕ0‖4L4. |
Proof. Multiplying (2.1b) by 2τλwn+1 and integrating over Ω, we get
2λ(ϕn+1−ϕn,wn+1)+2τλγ‖∇wn+1‖2+2τλ∫Ω˜un+1∇ϕnwn+1dx=0. | (2.5) |
Multiplying (2.1c) by 2λ(ϕn+1−ϕn) yields
2λ(ϕn+1−ϕn,wn+1)=λ2ε2(‖ϕn+1‖4L4−‖ϕn‖4L4)+λ2ε2(‖ϕn+1‖2−‖ϕn‖2)2+λε2‖ϕn+1(ϕn+1−ϕn)‖2+λε2(‖ϕn‖2−‖ϕn+1‖2+‖ϕn+1−ϕn‖2)+λ(‖∇ϕn+1‖2−‖∇ϕn‖2+‖∇(ϕn+1−ϕn)‖2), | (2.6) |
where we use the following identity:
2a(a−b)=a2−b2+(a−b)2,a3(a−b)=14(a4−b4+(a2−b2)2+2a2(a−b)2). | (2.7) |
Testing (2.1a) with τ|˜un+1|2 leads to
‖σn+1˜un+1‖2−‖σn˜un+1‖2+τ∫Ω∇ρn+1⋅un|˜un+1|2dx+τ2∫Ωρn+1(∇⋅un)|˜un+1|2dx=0. | (2.8) |
By taking the inner product of (2.1d) with 2τ˜un+1, we have
‖σn˜un+1‖2−‖σnun‖2+‖σn(˜un+1−un)‖2−2τλ∫Ωwn+1∇ϕn˜un+1dx+2τη‖∇˜un+1‖2+τ∫Ωρn+1(un⋅∇)|˜un+1|2dx+τ2∫Ωρn+1(∇⋅un)|˜un+1|2dx=0. | (2.9) |
Testing (2.1e) with 2τun+1 yields
‖σn+1un+1‖2−‖σn+1˜un+1‖2+‖σn+1(un+1−˜un+1)‖2+ητ‖∇un+1‖2+ητ∣∇˜un+1‖2+ητ‖∇(un+1−˜un+1)‖2=0. | (2.10) |
Taking into account the boundary condition on un and using integration by parts, we have
∫Ω∇ρn+1⋅un|˜un+1|2dx+∫Ωρn+1(∇⋅un)|˜un+1|2dx+∫Ωρn+1un⋅∇|˜un+1|2dx=∫Ω∇⋅(ρn+1un|˜un+1|2)dx=∫Ωρn+1un|˜un+1|2⋅ndx=0. | (2.11) |
Summing the above inequality, we arrive at
2τλγ‖∇wn+1‖2+λ2ε2(‖ϕn+1‖4L4−‖ϕn‖4L4‖)+λ2ε2(‖ϕn+1‖2−‖ϕn‖2)2+λε2‖ϕn+1(ϕn+1−ϕn)‖2+λε2(‖ϕn‖2−‖ϕn+1‖2+‖ϕn+1−ϕn‖2)+λ(‖∇ϕn+1‖2−‖∇ϕn‖2+‖∇(ϕn+1−ϕn)‖2)+‖σn+1un+1‖2−‖σnun‖2+‖σn(˜un+1−un)‖2+‖σn+1(un+1−˜un+1)‖2+ητ‖∇un+1‖2+ητ‖∇˜un+1‖2+ητ‖∇(un+1−˜un+1)‖2=0. | (2.12) |
Adding up the above inequality from n=0 to N−1, we obtain Theorem 2.3.
The bound on the pressure p is proved in the following theorem.
Theorem 3. (Stabilityofp) For any τ>0, the solution pn+1 to (2.1e) satisfies the following inequality:
τ2N−1∑n=0‖pn+1‖2≤C(‖σ0u0‖2+λ‖∇ϕ0‖2−λε2‖ϕ0‖2+λ2ε2‖ϕ0‖4L4)(‖ρ0‖+1). | (2.13) |
Proof. Under the inf-sup condition, there exists a positive constant β such that
β‖pn+1‖≤supv∈V,v≠0(∇⋅v,pn+1)‖∇v‖. | (2.14) |
Testing (2.14) with all v∈V leads to
(∇⋅v,pn+1)=η(∇(un+1−˜un+1),∇v)+τ−1(ρn+1(un+1−˜un+1),v)≤η(‖∇un+1‖+‖∇˜un+1‖)‖∇v‖+τ−1‖σn+1(un+1−˜un+1)‖L2‖σn+1‖L3‖v‖L6. | (2.15) |
Given that ‖ρn‖≤‖ρ0‖ for all 1≤n≤N, and by using Hölder's inequality, we have
‖σn+1‖L3=(∫‖pn+1‖32)13≤C‖ρn+1‖12≤C‖ρ0‖12. | (2.16) |
Then, by the Sobolev embedding inequality ‖v‖L6≤C‖∇v‖L2 for any v∈V, we get
(∇⋅v,pn+1)≤η(‖∇un+1‖+‖∇˜un+1‖)‖∇v‖+Cτ−1‖σn+1(un+1−˜un+1)‖‖ρ0‖12‖∇v‖. | (2.17) |
Substituting the above inequalities into Eq (2.14), we obtain
β‖pn+1‖≤η(‖∇un+1‖+‖∇˜un+1‖)+Cτ−1‖σn+1(un+1−˜un+1)‖‖ρ0‖12. | (2.18) |
And by using Theorem 2.2, we get the desired result.
In this section, we will give the time error estimates and show that the scheme has a first-order convergence rate. Although we verified that the scheme (2.1) is unconditionally stable in the previous chapter, we need to make the following assumptions[34,35] when conducting temporal error analysis:
{ρn}n=0,⋯,N is uniformly bounded in L∞, | (3.1) |
for all n=0,⋯,N, it holds that ρn≥χ a.e. in Ω, | (3.2) |
where χ is a number in (0,ρmin0].
We assume that the exact solution (ρ,u,ϕ,w,p) is sufficiently smooth. To be more precise,
ρ∈H2(0,T;L2(Ω))∩L2(0,T;W1,∞(Ω)),ϕ∈L∞(0,T;H3(Ω))∩W1,∞(0,T;H2(Ω))∩W2,∞(0,T;H1(Ω))∩W3,∞(0,T;L2(Ω)),w∈L∞(0,T;H2(Ω))∩W1,∞(0,T;L2(Ω)),u∈H2(0,T;L2(Ω))∩L∞(0,T;V∩H2(Ω)),p∈W2,∞(0,T;H1(Ω)). | (3.3) |
We denote
enϕ=ϕ(tn)−ϕn,enw=w(tn)−wn,enρ=ρ(tn)−ρn,enu=u(tn)−un,˜enu=u(tn)−˜un,enp=p(tn)−pn. |
In (1.1), taking t=tn+1 and subtracting from (2.1), we get the following error equations:
en+1ρ−enρτ+∇en+1ρ⋅u(tn+1)+∇ρn+1⋅(u(tn+1)−u(tn))+∇ρn+1⋅enu=Rn+1ρ, | (3.4a) |
en+1ϕ−enϕτ−γ△en+1w+u(tn+1)∇ϕ(tn+1)−˜un+1∇ϕn=Rn+1ϕ, | (3.4b) |
en+1w+△en+1ϕ=1ε2(ϕ(tn+1)3−(ϕn+1)3−ϕ(tn+1)+ϕn), | (3.4c) |
ρn˜en+1u−enuτ−η△˜en+1u+∇p(tn+1)+ρ(tn+1)(u(tn+1)⋅∇)u(tn+1)−ρn+1(un⋅∇)˜un+1−λw(tn+1)∇ϕ(tn+1)+λwn+1∇ϕn=Rn+1u, | (3.4d) |
ρn+1en+1u−˜en+1uτ−η△(en+1u−˜en+1u)−∇pn+1=0, | (3.4e) |
∇⋅en+1u=0, | (3.4f) |
where
Rn+1ϕ=ϕ(tn+1)−ϕ(tn)τ−ϕt(tn+1),Rn+1ρ=ρ(tn+1)−ρ(tn)τ−ρt(tn+1),Rn+1u=ρnu(tn+1)−u(tn)τ−ρ(tn+1)ut(tn+1). |
If the exact solution is sufficiently smooth, it is easy to establish the following estimate of the truncation error.
Lemma 1. Under the regularity assumptions given by (3.3), the truncation errors satisfy:
‖Rn+1ρ‖2≤Cτ∫tn+1tn‖ρtt(t)‖2dt≤Cτ2,‖Rn+1ϕ‖2≤Cτ∫tn+1tn‖ϕtt(t)‖2dt≤Cτ2,‖Rn+1u‖2≤Cτ∫tn+1tn(‖utt(t)‖2+‖ρt(t)‖2)dt+C‖enρ‖2≤Cτ2+C‖enρ‖2, |
for all 0≤n≤N−1.
Proof. By using the integral residual of the Taylor formula, we have
Rn+1ρ=1τ∫tn+1tn(t−tn)ρtt(t)dt. | (3.5) |
By Hölder's inequality, we can derive
‖Rn+1ρ‖2=∫Ω(1τ∫tn+1tn(t−tn)ρtt(t)dt)2dx≤1τ2∫Ω∫tn+1tn(t−tn)2dt12⋅2∫tn+1tnρtt(t)2dt12⋅2dx≤1τ2(τ33∫tn+1tn∫Ωρtt(t)2dxdt)≤Cτ∫tn+1tn‖ρtt(t)‖2dt≤Cτ2. | (3.6) |
Similarly, we can prove the inequality of Rn+1ϕ. For Rn+1u, we can rewrite
Rn+1u=ρn(u(tn+1)−u(tn)τ−ut(tn+1))−(enρ+∫tn+1tnρt(t)dt)ut(tn+1)=ρnτ∫tn+1tn(t−tn)utt(t)dt−(enρ+∫tn+1tnρt(t)dt)ut(tn+1). | (3.7) |
Using Rn+1ρ estimation and Hölder's inequality can yield the result for Rn+1u.
We introduce the following Gronwall's inequality, which will frequently be used in error estimates.
Lemma 2. Let ak, bk, ck and γk, for integers k≥0, be the nonnegative numbers such that
an+τn∑k=0bk≤τn∑k=0γkak+τn∑k=0ck+B for ≥0. |
Suppose that τγk<1, for all k, and set σk=(1−τγk)−1. Then,
an+τn∑k=0bk≤exp(τn∑k=0γkσk)(τn∑k=0ck+B) for ≥0. |
We verify Lemma 5 by the following lemma.
Lemma 3. Define
Gn+1c=(ϕ(tn+1))3−(ϕn+1)3=3(ϕ(tn+1))2en+1ϕ−3ϕ(tn+1)(en+1ϕ)2+(en+1ϕ)3. | (3.8) |
Then, for n<Tτ−1, we have
‖Gn+1c‖≤C‖en+1ϕ‖H1,‖en+1ϕ‖H2≤C(τ+‖en+1w‖+‖en+1ϕ‖H1+‖enϕ‖). | (3.9) |
Proof. For ‖Gn+1c‖, we use (3.8) to conclude that
‖Gn+1c‖≤C(‖en+1ϕ‖‖ϕ(tn+1)‖2L∞+‖en+1ϕ‖2L4‖ϕ(tn+1)‖L∞+‖en+1ϕ‖3L6)≤C(‖en+1ϕ‖‖ϕ(tn+1)‖2L∞+‖en+1ϕ‖2H1‖ϕ(tn+1)‖L∞+‖en+1ϕ‖3H1)≤C‖en+1ϕ‖H1, | (3.10) |
where we have used the a priori bound ‖ϕn‖H1≤C implied by the stability result given by Theorem 2. Using the H2 regularity results for elliptic equations, we conclude that
‖en+1ϕ‖H2≤c(‖en+1ϕ‖L2+‖△en+1ϕ‖L2); |
from (3.4c), we know that
△en+1ϕ=−en+1w+1ε2Gn+1c−1ε2enϕ−∫tn+1tnϕt(t)dt; |
thus,
‖en+1ϕ‖H2≤C(‖en+1ϕ‖+‖en+1w‖+‖Gn+1c‖+‖enϕ‖+τ)≤C(τ+‖en+1w‖+‖en+1ϕ‖H1+‖enϕ‖). |
The error estimate for the discrete density ρn+1 is derived in the following lemma.
Lemma 4. Suppose that the solution to (1.1) satisfies the regularity assumptions given by (3.3), and suppose that (3.1)–(3.2) hold. Then, we have
‖en+1ρ‖2+2n∑m=0‖em+1ρ−emρ‖2≤C(τ2+τn∑m=0‖σmemu‖2) | (3.11) |
for all 0≤n≤N−1.
Proof. Multiplying (3.4a) by 2τen+1ρ and integrating over Ω, we have
‖en+1ρ‖2−‖enρ‖2+‖en+1ρ−enρ‖2=−2τ(∇en+1ρ⋅u(tn+1),en+1ρ)+2τ(Rn+1ρ,en+1ρ)−2τ(∇ρn+1⋅(u(tn+1)−u(tn)),en+1ρ)−2τ(∇ρn+1⋅enu,en+1ρ)=4∑i=1Ki. | (3.12) |
Using ∇⋅u(tn+1)=0 in Ω and u(tn+1)=0 on ∂Ω, we have
K1=12∫∂Ω|en+1ρ|2u(tn+1)⋅nds=0. | (3.13) |
By using Young's inequality, the Cauchy-Schwarz inequality and Lemma 1, we have
K2≤2τ‖Rn+1ρ‖‖en+1ρ‖≤Cτ‖Rn+1ρ‖2+ετ‖en+1ρ‖2≤Cτ3+ετ‖en+1ρ‖2, |
where ε>0 is a sufficiently small constant.
Then, by the Sobolev inequality and Young inequality, we get
K3=−2τ(∇ρ(tn+1)⋅(u(tn+1)−u(tn)),en+1ρ)+2τ(∇en+1ρ⋅(u(tn+1)−u(tn)),en+1ρ)=−2τ(∇ρ(tn+1)⋅(u(tn+1)−u(tn)),en+1ρ)≤2τ‖∇ρ(tn+1)‖L∞‖∫tn+1tnut(t)dt‖‖en+1ρ‖≤ετ‖en+1ρ‖2+Cτ2∫tn+1tn‖ut(t)‖2dt≤Cτ3+ετ‖en+1ρ‖2. | (3.14) |
Similarly, the last term can be estimated as follows:
K4≤2τ‖∇ρ(tn+1)‖L∞‖1σn‖L∞‖‖σnenu‖‖en+1ρ‖≤ετ‖en+1ρ‖2+Cτ‖σnenu‖2. | (3.15) |
If ε is sufficiently small such that ετ≤16, substituting the estimates of Ki (1≤i≤4) into (3.12), we have
‖en+1ρ‖2−‖enρ‖2+2‖en+1ρ−enρ‖2≤Cτ3+Cτ‖enρ‖2+Cτ‖σnenu‖2. | (3.16) |
Using the discrete Gronwall inequality, we obtain the desired result.
Lemma 5. Suppose that the solution to (1.1) satisfies the regularity assumptions given by (3.3), and suppose that (3.1)–(3.2) are valid. For sufficiently small τ, there are the following error estimates:
τγN−1∑n=0(λ‖∇en+1w‖2+‖en+1w‖2)+‖eNϕ‖2+λ‖∇eNϕ‖2+λ2ε2‖eNϕ‖4L4+‖σNeNu‖2+N−1∑n=0(12‖σn(˜en+1u−enu)‖2+‖σn+1(en+1u−˜en+1u)‖2)+τηN−1∑n=0(‖∇en+1u‖2+12‖∇˜en+1u‖2+‖∇(en+1u−˜en+1u)‖2)≤Cτ. | (3.17) |
Proof. Let us multiply (3.4a) by τ|˜en+1u|2, (3.4b) by 2τen+1ϕ and 2τλen+1w, (3.4c) by 2τγen+1w and −2λ(en+1ϕ−enϕ), (3.4d) by 2τ˜en+1u and (3.4e) by 2τen+1u. Summing up all of the above equations, we have
2τγλ‖∇en+1w‖2+‖en+1ϕ‖2−‖enϕ‖2+‖en+1ϕ−enϕ‖2+2τγ‖en+1w‖2+λ‖∇en+1ϕ‖2−λ‖∇enϕ‖2+λ‖∇en+1ϕ−enϕ‖2‖σn+1en+1u‖2−‖σnenu‖2+‖σn(˜en+1u−enu)‖2+‖σn+1(en+1u−˜en+1u)‖2+τη‖∇en+1u‖2+τη‖∇˜en+1u‖2+τη‖∇(en+1u−˜en+1u)‖2=−2τ(ρ(tn+1)(u(tn+1)⋅∇)u(tn+1),˜en+1u)+2τ(ρn+1(un⋅∇)˜un+1,˜en+1u)−τ(∇ρn+1⋅un,|en+1u|2)−2τ(∇p(tn+1),˜en+1u)+2τ(Rn+1u,˜en+1u)+2τλ(w(tn+1)∇ϕ(tn+1)−wn+1∇ϕn,˜en+1u)−2τλ(u(tn+1)∇ϕ(tn+1)−˜un+1∇ϕn,en+1w)+2τλ(Rn+1ϕ,en+1w)−2τ(u(tn+1)∇ϕ(tn+1)−˜un+1∇ϕn,en+1ϕ)+2τ(Rn+1ϕ,en+1ϕ)+2τγε2(ϕ(tn+1)3−(ϕn+1)3,en+1w)−2τγε2(ϕ(tn+1)−ϕn,en+1w)−2λε2(ϕ(tn+1)3−(ϕn+1)3,en+1ϕ−enϕ)+2λε2(ϕ(tn+1)−ϕn,en+1ϕ−enϕ)=i=14∑i=1Ai. | (3.18) |
Thanks to ∇⋅un=0 in Ω, we have
A2+A3=2τ(ρn+1(un⋅∇)˜un+1,˜en+1u)−τ(∇ρn+1⋅un,|˜en+1u|2)=2τ(ρn+1(un⋅∇)u(tn+1),˜en+1u)−τ(∇⋅(unρn+1|˜en+1u|2),1)=2τ(ρn+1(un⋅∇)u(tn+1),˜en+1u). | (3.19) |
Therefore, we get
A1+A2+A3 | (3.20) |
=2τ(ρn+1(un⋅∇)u(tn+1)−ρ(tn+1)(u(tn+1)⋅∇)u(tn+1),˜en+1u)=−2τ(ρn+1(enu⋅∇)u(tn+1),˜en+1u)−2τ(en+1ρ(u(tn)⋅∇)u(tn+1),˜en+1u)−2τ(ρ(tn+1)((u(tn+1)−u(tn))⋅∇)u(tn+1),˜en+1u)≤Cτ‖ρn+1‖L∞‖1σn‖L∞‖σnenu‖‖∇u(tn+1)‖L3‖˜en+1u‖L6 | (3.21) |
+Cτ‖en+1ρ‖‖u(tn)‖L∞‖∇u(tn+1)‖L3‖˜en+1u‖L6+Cτ‖ρ(tn+1)‖L∞‖∫tn+1tnut(t)dt‖‖∇u(tn+1)‖L3‖˜en+1u‖L6≤Cτ(‖σnenu‖+‖en+1ρ‖+τ∫tn+1tn‖ut(t)‖dt)‖∇˜en+1u‖≤ητ12‖∇˜en+1u‖2+Cτ(‖σnenu‖2+‖en+1ρ‖2+τ2)≤Cτ3+ητ12‖∇˜en+1u‖2+Cτ‖σnenu‖2+Cτ2n∑m=0‖σmemu‖2, | (3.22) |
where the Sobolev embedding inequality ‖v‖L6≤C‖∇v‖L2 for any v∈V is used.
For A4, we have
A4≤2τ‖∇p(tn+1)‖‖1σn‖L∞‖σn(˜en+1u−enu)‖≤Cτ2+12‖σn(˜en+1u−enu)‖2. | (3.23) |
Using the Poincarˊe inequality and Cauchy-Schwarz inequality, we can deal with A6, A7 and A9:
A6=2τλ(w(tn+1)∇ϕ(tn+1)−wn+1∇ϕn,˜en+1u)=2τλ(w(tn+1)(∇ϕ(tn+1)−∇ϕ(tn)),˜en+1u)+2τλ(w(tn+1)∇enϕ,˜en+1u)+2τλ(∇ϕnen+1w,˜en+1u)≤2τλ‖w(tn+1)‖L∞‖∫tn+1tn∇ϕtdt‖‖˜en+1u‖+2τλ‖w(tn+1)‖L∞‖∇enϕ‖‖˜en+1u‖+2τλ(∇ϕnen+1w,˜en+1u)≤Cτ3+ητ12‖∇˜en+1u‖2+Cτ‖∇enϕ‖2+2τλ(∇ϕnen+1w,˜en+1u),A7=−2τλ(u(tn+1)∇ϕ(tn+1)−˜un+1∇ϕn,en+1w)=−2τλ(u(tn+1)(∇ϕ(tn+1)−∇ϕ(tn))+u(tn+1)∇enϕ+∇ϕn˜en+1u,en+1w)≤2τλ‖u(tn+1)‖L∞‖∫tn+1tn∇ϕtdt‖‖en+1w‖+2τλ‖u(tn+1)‖L∞‖∇enϕ‖‖en+1w‖−2τλ(∇ϕn˜en+1u,en+1w)≤Cτ3+τγ3‖en+1w‖2+Cτ‖∇enϕ‖2−2τλ(∇ϕn˜en+1u,en+1w), | (3.24) |
and
A9=−2τ(u(tn+1)∇ϕ(tn+1)−˜un+1∇ϕn,en+1ϕ)≤2τ‖u(tn+1)‖L∞‖∫tn+1tn∇ϕtdt‖‖en+1ϕ‖+2τ‖u(tn+1)‖L∞‖∇enϕ‖‖en+1ϕ‖+2τ‖∇ϕn‖‖˜en+1u‖L6‖en+1ϕ‖L6≤Cτ3+Cτ‖∇en+1ϕ‖2+Cτ‖∇enϕ‖2+ητ12‖∇˜en+1u‖2. | (3.25) |
From Lemma 1, we have
A5+A8+A10≤2τ‖Rn+1u‖‖˜en+1u‖+2τλ‖Rn+1ϕ‖‖en+1w‖+2τ‖Rn+1ϕ‖‖en+1ϕ‖≤Cτ3+Cτ2n∑m=0‖σmemu‖2+ητ12‖∇˜en+1u‖2+τγ3‖en+1w‖2+12‖en+1ϕ‖2. | (3.26) |
From Lemma 3, we obtain
A11+A12=2τγε2(Gn+1c−enϕ−∫tn+1tnϕt(t)dt,en+1w)≤2τγε2(‖enϕ‖+‖Gn+1c‖+‖∫tn+1tnϕt(t)dt‖)‖en+1w‖≤Cτ3+Cτ(‖enϕ‖2+‖en+1ϕ‖2+‖∇en+1ϕ‖2)+τγ3‖en+1w‖2, | (3.27) |
and
A13+A14=−2λε2(Gn+1c−enϕ−∫tn+1tnϕt(t)dt,en+1ϕ−enϕ)=−λ2ε2(‖en+1ϕ‖4L4−‖enϕ‖4L4+‖(en+1ϕ)2−(enϕ)2‖2+2‖en+1ϕ(en+1ϕ−enϕ)‖2)−2λε2‖en+1ϕ−enϕ‖2−2λε2(3(ϕ(tn+1))2en+1ϕ−3ϕ(tn+1)(en+1ϕ)2−en+1ϕ,en+1ϕ−enϕ)+2λε2(∫tn+1tnϕt(t)dt,en+1ϕ−enϕ), | (3.28) |
where we use this identity
a3(a−b)=14(a4−b4+(a2−b2)2+2a2(a−b)2). | (3.29) |
We denote
˜Gn+1=−2λε2(3(ϕ(tn+1))2en+1ϕ−3ϕ(tn+1)(en+1ϕ)2−en+1ϕ). | (3.30) |
Similar to the method estimated from Lemma 3, we can obtain
‖˜Gn+1‖≤C‖en+1ϕ‖H1. | (3.31) |
Taking the gradient of ˜Gn+1, we get
∇˜Gn+1=−2λε2[(3(ϕ(tn+1))2−1)∇en+1ϕ+6ϕ(tn+1)en+1ϕ∇ϕ(tn+1)−3(en+1ϕ)2∇ϕ(tn+1)−6ϕ(tn+1)en+1ϕ∇en+1ϕ]. | (3.32) |
Since H2(Ω)⊂L∞(Ω) and by utilizing the bound of ‖∇en+1ϕ‖L2 implied by Theorem 2, we conclude that
‖en+1ϕ∇en+1ϕ‖L2≤‖en+1ϕ‖L∞‖∇en+1ϕ‖L2≤C‖en+1ϕ‖H2. | (3.33) |
In view of (3.10) and the bound ‖ϕn‖H1<C, we have
‖∇˜Gn+1‖≤C[(‖ϕ(tn+1)‖2L∞+1)‖∇en+1ϕ‖L2+‖ϕ(tn+1)‖L∞‖∇ϕ(tn+1)‖L3‖en+1ϕ‖L6+‖∇ϕ(tn+1)‖L6‖en+1ϕ‖2L6+‖ϕ(tn+1)‖L∞‖en+1ϕ∇en+1ϕ‖L2]≤C(‖∇en+1ϕ‖L2+‖en+1ϕ‖H1+‖en+1ϕ‖2H1+‖en+1ϕ∇en+1ϕ‖L2)≤C(τ+‖en+1w‖+‖en+1ϕ‖+‖∇en+1ϕ‖+‖enϕ‖). | (3.34) |
Dealing with the penultimate term of (3.28) and in view of (3.31) and (3.33), we get
(˜Gn+1,en+1ϕ−enϕ)=τ(˜Gn+1,γΔen+1w−u(tn+1)∇enϕ−˜en+1u∇ϕn)+τ(˜Gn+1,−u(tn+1)∫tn+1tn∇ϕt(t)dt+Rn+1ϕ)≤γτ‖∇en+1w‖‖∇˜Gn+1‖+τ‖∇ϕn‖‖˜en+1u‖H1‖˜Gn+1‖H1+τ‖˜Gn+1‖(‖Rn+1ϕ‖+‖u(tn+1)‖L∞‖∇enϕ‖+‖u(tn+1)‖L∞‖∫tn+1tn∇ϕt(t)dt‖)≤Cτ3+γλτ2‖∇en+1w‖2+ητ12‖∇˜en+1u‖2+Cτ(‖en+1w‖2+‖en+1ϕ‖2+‖∇en+1ϕ‖2+‖enϕ‖2+‖∇enϕ‖2). | (3.35) |
Then, we estimate the last term of (3.28),
2λε2(∫tn+1tnϕt(t)dt,en+1ϕ−enϕ)=2λτε2(∫tn+1tnϕt(t)dt,γΔen+1w−u(tn+1)∇enϕ−˜en+1u∇ϕn)+2λτε2(∫tn+1tnϕt(t)dt,−u(tn+1)∫tn+1tn∇ϕt(t)dt+Rn+1ϕ)≤2λγτε2‖∫tn+1tn∇ϕt(t)dt‖‖∇en+1w‖+2λτε2‖∇ϕn‖‖˜en+1u‖H1‖∫tn+1tnϕt(t)dt‖H1+2λτε2‖∫tn+1tnϕt(t)dt‖(‖Rn+1ϕ‖+‖u(tn+1)‖L∞‖∇enϕ‖)+2λτε2‖∫tn+1tnϕt(t)dt‖‖u(tn+1)‖L∞‖∫tn+1tn∇ϕt(t)dt‖≤Cτ3+γλτ2‖∇en+1w‖2+ητ12‖∇˜en+1u‖2+Cτ‖∇enϕ‖2. | (3.36) |
Combining (3.35) and (3.36), we can obtain
A13+A14≤−λ2ε2(‖en+1ϕ‖4L4−‖enϕ‖4L4+‖(en+1ϕ)2−(enϕ)2‖2+2‖en+1ϕ(en+1ϕ−enϕ)‖2)−2λε2‖en+1ϕ−enϕ‖2+Cτ3+γλτ‖∇en+1w‖2+ητ6‖∇˜en+1u‖2+Cτ(‖∇enϕ‖2+‖en+1w‖2+‖en+1ϕ‖2+‖∇en+1ϕ‖2+‖enϕ‖2+‖∇enϕ‖2). | (3.37) |
Substituting the above estimates into (3.18), we have
τγλ‖∇en+1w‖2+12‖en+1ϕ‖2−‖enϕ‖2+(1+2λε2)‖en+1ϕ−enϕ‖2+τγ‖en+1w‖2+λ‖∇en+1ϕ‖2−λ‖∇enϕ‖2+λ‖∇en+1ϕ−enϕ‖2+‖σn+1en+1u‖2−‖σnenu‖2+12‖σn(˜en+1u−enu)‖2+‖σn+1(en+1u−˜en+1u)‖2+τη‖∇en+1u‖2+τη2‖∇˜en+1u‖2+τη‖∇(en+1u−˜en+1u)‖2+λ2ε2(‖en+1ϕ‖4L4−‖enϕ‖4L4+‖(en+1ϕ)2−(enϕ)2‖2+2‖en+1ϕ(en+1ϕ−enϕ)‖2)≤Cτ2+Cτ‖σnenu‖2+Cτ2n∑m=0‖σmemu‖2+Cτ(‖∇enϕ‖2+‖en+1w‖2+‖en+1ϕ‖2+‖∇en+1ϕ‖2+‖enϕ‖2+‖∇enϕ‖2). | (3.38) |
Adding up from 0 to N−1, and applying Gronwall's inequality, we infer that
τγN−1∑n=0(λ‖∇en+1w‖2+‖en+1w‖2)+‖eNϕ‖2+λ‖∇eNϕ‖2+λ2ε2‖eNϕ‖4L4+‖σNeNu‖2+N−1∑n=0(12‖σn(˜en+1u−enu)‖2+‖σn+1(en+1u−˜en+1u)‖2)+τηN−1∑n=0(‖∇en+1u‖2+12‖∇˜en+1u‖2+‖∇(en+1u−˜en+1u)‖2)≤Cτ. |
Lemma 5 shows that the fractional scheme converges at a rate of O(τ12). However, since we use the first-order backward Euler method of time discretization, it is not optimal from the perspective of theoretical analysis. Next, we will improve the convergence speed to the first order.
Lemma 6. Suppose that the solution to (1.1) satisfies the regularity assumptions given by (3.3), and suppose that (3.1)–(3.2) are valid. For sufficiently small τ, there are the following error estimates:
τγN−1∑n=0(λ‖∇en+1w‖2+‖en+1w‖2)+‖eNϕ‖2+λ‖∇eNϕ‖2+λ2ε2‖eNϕ‖4L4+‖σNeNu‖2+τηN−1∑n=0‖∇en+1u‖2≤Cτ2. | (3.39) |
Proof. Taking the sum of (3.4d) and (3.4e), we get
ρnen+1u−enuτ+ρn+1−ρnτ(en+1u−˜en+1n)−η△en+1u+ρ(tn+1)(u(tn+1)⋅∇)u(tn+1)−ρn+1(un⋅∇)˜un+1+Δen+1p−λw(tn+1)∇ϕ(tn+1)+λwn+1∇ϕn=Rn+1u. | (3.40) |
Let us multiply (3.4a) by τ˜en+1u, (3.4b) by 2τen+1ϕ and 2τλen+1w, (3.4c) by 2τγen+1w and −2λ(en+1ϕ−enϕ) and (3.40) by 2τen+1u. Summing up all of the above equations, we obtain
‖σn+1en+1u‖2−‖σnenu‖2+‖σn(en+1u−enu)‖2+2ητ‖∇en+1u‖2+2τγλ‖∇en+1w‖2+‖en+1ϕ‖2−‖enϕ‖2+‖en+1ϕ−enϕ‖2+2τγ‖en+1w‖2+λ‖∇en+1ϕ‖2−λ‖∇enϕ‖2+λ‖∇en+1ϕ−∇enϕ‖2=−2τ(∇ρn+1⋅un,˜en+1u⋅en+1u)+τ(∇ρn+1⋅un,|en+1u|2)−2τ(ρ(tn+1)(u(tn+1)⋅∇)u(tn+1)−ρn+1(un⋅∇)˜un+1,en+1u)+2τλ(w(tn+1)∇ϕ(tn+1)−wn+1∇ϕn,en+1u)+2τ(Rn+1u,en+1u)−2τλ(u(tn+1)∇ϕ(tn+1)−˜un+1∇ϕn,en+1w)+2τλ(Rn+1ϕ,en+1w)−2τ(u(tn+1)∇ϕ(tn+1)−˜un+1∇ϕn,en+1ϕ)+2τ(Rn+1ϕ,en+1ϕ)+2τrε2(ϕ(tn+1)3−(ϕn+1)3,en+1w)−2τrε2(ϕ(tn+1)−ϕn,en+1w)−2λε2(ϕ(tn+1)3−(ϕn+1)3,en+1ϕ−enϕ)+2λε2(ϕ(tn+1)−ϕn,en+1ϕ−enϕ)=13∑i=1Li. | (3.41) |
Using integration by parts, we have
L1+L2=2τ(ρn+1(un⋅∇)˜en+1u,en+1u)+2τ(ρn+1(enu⋅∇)en+1u,en+1u−˜en+1u)−2τ(ρn+1(u(tn)⋅∇)en+1u,en+1u−˜en+1u); | (3.42) |
we rewrite the L3 as
L3=−2τ(ρn+1(un⋅∇)˜en+1u,en+1u)−2τ(ρn+1((u(tn+1)−u(tn))⋅∇)u(tn+1),en+1u)−2τ(ρn+1(enu⋅∇)u(tn+1),en+1u)−2τ(en+1ρ(u(tn+1)⋅∇)u(tn+1),en+1u); | (3.43) |
then,
L1+L2+L3=2τ(ρn+1(enu⋅∇)en+1u,en+1u−˜en+1u)−2τ(en+1ρ(u(tn+1)⋅∇)u(tn+1),en+1u)−2τ(ρn+1(u(tn)⋅∇)en+1u,en+1u−˜en+1u)−2τ(ρn+1((u(tn+1)−u(tn))⋅∇)u(tn+1),en+1u)−2τ(ρn+1(enu⋅∇)u(tn+1),en+1u)=5∑i=1Ji. | (3.44) |
Lemma 5 shows that
‖∇(en+1u−˜en+1u)‖≤C. | (3.45) |
According to Young's inequality and the Cauchy-Schwarz inequality, we can deduce that
J1≤ητ10‖∇en+1u‖2+Cτ‖∇enu‖2‖∇(en+1u−˜en+1u)‖‖σn+1(en+1u−˜en+1u)‖≤ητ10‖∇en+1u‖2+Cτ32‖∇enu‖2,J2≤2τ‖en+1ρ‖‖u(tn+1)‖L∞‖∇u(tn+1)‖L3‖en+1u‖L6≤ητ10‖∇en+1u‖2+Cτ‖en+1ρ‖2≤Cτ3+ητ10‖∇en+1u‖2+Cτ2n∑m=0‖σmemu‖2,J3≤Cτ‖∇en+1u‖‖u(tn)‖L∞‖σn+1(en+1u−˜en+1u)‖≤Cτ3+ητ10‖∇en+1u‖2+Cτ‖σn+1(en+1u−˜en+1u)‖2,J4≤2τ‖ρn+1‖L∞‖∇u(tn+1)‖L∞‖∫tn+1tnutdt‖‖en+1u‖≤Cτ3+ητ10‖∇en+1u‖2,J5≤2τ‖ρn+1‖L∞‖∇u(tn+1)‖L∞‖1σn‖L∞‖σnenu‖‖en+1u‖≤Cτ‖σnenu‖2+ητ10‖∇en+1u‖2. | (3.46) |
Therefore, we have
M1+M2+M3≤Cτ3+Cτ32‖∇enu‖2+ητ2‖∇en+1u‖2+Cτ‖σnenu‖2+Cτ2n∑m=0‖σmemu‖2+Cτ‖σn+1(en+1u−˜en+1u)‖2. | (3.47) |
Using the embedding inequality and Cauchy-Schwarz inequality, we can deal with M4–M9
M4=2τλ(w(tn+1)∇ϕ(tn+1)−wn+1∇ϕn,en+1u)=2τλ(w(tn+1)(∇ϕ(tn+1)−∇ϕ(tn))+w(tn+1)∇enϕ+∇ϕnen+1w,en+1u)≤2τλ‖w(tn+1)‖L∞‖∫tn+1tn∇ϕt(t)dt‖‖en+1u‖+2τλ‖w(tn+1)‖L∞‖∇enϕ‖‖en+1u‖+2τλ‖∇ϕn‖L2‖en+1w‖L3‖en+1u‖L6≤Cτ3+τη‖∇en+1u‖2+Cτ‖∇enϕ‖2+Cτ‖∇en+1w‖2, | (3.48) |
M5+M7+M9=2τ(Rn+1u,en+1u)+2τλ(Rn+1ϕ,en+1w)+2τ(Rn+1ϕ,en+1ϕ)≤2τ‖Rn+1u‖‖en+1u‖+2τλ‖Rn+1ϕ‖‖en+1w‖+2τ‖Rn+1ϕ‖‖en+1ϕ‖≤Cτ3+Cτ2n∑m=0‖σmemu‖2+τλγ4‖∇en+1w‖2+14‖en+1ϕ‖2, | (3.49) |
M6=−2τλ(u(tn+1)∇ϕ(tn+1)−˜un+1∇ϕn,en+1w)=−2τλ((˜en+1u−en+1u)∇ϕ(tn+1),en+1w)−2τλ(en+1u∇ϕ(tn+1),en+1w)−2τλ(˜un+1∫tn+1tn∇ϕt(t)dt,en+1w)−2τλ(˜un+1∇enϕ,en+1w)≤2τλ‖1σn+1‖L∞‖σn+1(˜en+1u−en+1u)‖‖∇ϕ(tn+1)‖L∞‖en+1w‖+2τλ‖1σn+1‖L∞‖σn+1en+1u‖‖∇ϕ(tn+1)‖L∞‖en+1w‖+2τλ‖∫tn+1tn∇ϕt(t)dt‖L2‖˜un+1‖L3‖en+1w‖L6+2τλ‖∇enϕ‖‖˜un+1‖L3‖en+1w‖L6≤Cτ3+τλ2‖en+1w‖2+Cτ‖σn+1en+1u‖2+τλγ4‖∇en+1w‖2+Cτ‖∇enϕ‖2+Cτ‖σn+1(en+1u−˜en+1u)‖2, | (3.50) |
and
M8=−2τ(u(tn+1)∇ϕ(tn+1)−˜un+1∇ϕn,en+1ϕ)≤Cτ3+14‖en+1ϕ‖2+Cτ‖σn+1en+1u‖2+λ2‖∇en+1ϕ‖2+Cτ‖∇enϕ‖2. | (3.51) |
Next, we estimate M10+M11 and M12+M13 as follows:
M10+M11≤Cτ3+Cτ(‖enϕ‖2+‖en+1ϕ‖2+‖∇en+1ϕ‖2)+τγ2‖en+1w‖2. | (3.52) |
\begin{align} &\left(\tilde{G}^{n+1}, e_{\phi}^{n+1}-e_{\phi}^{n}\right)\\ = &\tau\left(\tilde{G}^{n+1}, \gamma \triangle e_w^{n+1}-u(t_{n+1} ) \nabla \phi(t _{n+1})+\tilde{\boldsymbol{u}}^{n+1} \nabla \phi^{n}+R_{\phi}^{n+1}\right)\\ = &\tau\left(\tilde{G}^{n+1}, \gamma \triangle e_w^{n+1}-\left(\tilde{e}_{u}^{n+1}-e_{u}^{n+1}\right) \nabla \phi\left(t_{n+1}\right)-e_{u}^{n+1} \nabla \phi(t _{n+1})-\tilde{\boldsymbol{u}}^{n+1} \int_{t _n}^{t _{n+1}} \nabla \phi_t(t) d t\right)\\ &-\tau\left(\tilde{G}^{n+1}, \tilde{\boldsymbol{u}} \nabla e_{\phi}^{n}-R_{\phi}^{n+1}\right) \\ \leq &C \tau^{3}+C\tau\left\|\tilde{G}^{n+1}\right\|^{2}+\frac{\tau\gamma \lambda}{4}\left\|\nabla e_{w}^{n+1}\right\|^{2}+C\tau\left\|\nabla \tilde{G}^{n+1}\right\|^{2}+\frac{1}{4}\left\|\sigma^{n+1} e_u^{n+1}\right\|^{2}+C \tau\left\|\nabla e_{\phi}^{n}\right\|^{2}\\ &+C\tau\left\|\sigma^{n+1}\left(e_{u}^{n+1}- \tilde{e}_{u}^{n+1} \right)\right\|^2\\ \leq& C\tau^{3}+\frac{\tau \gamma \lambda}{4}\left\|\nabla e_{w}^{n+1}\right\|^{2}+\frac{1}{4}\left\|\sigma^{n+1} e_{u}^{n+1}\right\|^{2}+C\tau\left\|\sigma^{n+1}\left(e_{u}^{n+1}- \tilde{e}_{u}^{n+1} \right)\right\|^2\\ &+C\tau\left(\left\|e_{w}^{n+1}\right\|^{2}+\left\|e_{\phi}^{n+1}\right\|^{2}+\left\|\nabla e_{\phi}^{n+1}\right\|^{2}+\left\|e_{\phi}^{n}\right\|^{2}+\left\|\nabla e_{\phi}^{n}\right\|^{2}\right), \end{align} | (3.53) |
and
\begin{align} \left(\int_{t_{n}}^{t_{n+1}} \phi_{t}(t) d t, e_{\phi}^{n+1}-e_{\phi}^{n}\right)\leq C \tau^{3}+\frac{\tau \gamma \lambda}{4}\left\|\nabla e_{w}^{n+1}\right\|^{2}+\frac{1}{4}\left\|\sigma^{n+1} e^{n+1}\right\|^{2}+C \tau\|\nabla e _{\phi}^{n}\|^{2}. \end{align} |
From (3.29), combining the two inequalities above, we deduce that
\begin{equation} \begin{split} M_{12}+M_{13} \leq&-\frac{\lambda}{2 \varepsilon^{2}}\left(\left\|e_{\phi}^{n+1}\right\|_{L^{4}}^{4}-\left\|e_{\phi}^{n}\right\|_{L^{4}}^{4}+\left\|(e_{\phi}^{n+1})^{2}-\left(e_{\phi}^{n}\right)^{2}\right\|^{2}\right.\\ &\left.+2\left\|e_{\phi}^{n+1}\left(e_{\phi}^{n+1}-e _{\phi}^{n}\right)\right\|^{2}\right)-\frac{2 \lambda}{\varepsilon^{2}}\left\|e_{\phi}^{n+1}-e _{\phi}^{n}\right\|^{2}\\ &+C \tau^{3}+\frac{\tau\gamma \lambda}{2}\left\|\nabla e_{w}^{n+1}\right\|^{2}+\frac{1}{2}\left\|\sigma^{n+1} e_u^{n+1}\right\|^{2} \\ &+C\tau\left(\left\|e_{w}^{n+1}\right\|^{2}+\left\|e_{\phi}^{n+1}\right\|^{2}+\left\|\nabla e_{\phi}^{n+1}\right\|^{2}+\left\|e_{\phi}^{n}\right\|^{2}+\left\|\nabla e_{\phi}^{n}\right\|^{2}\right).\\ \end{split} \end{equation} | (3.54) |
Plugging the above inequality into (3.41) gives
\begin{align} &\frac{1}{2}\left(\left\|\sigma^{n+1} e_{u}^{n+1}\right\|^{2}-\left\|\sigma^{n} e_{u}^{n}\right\|^{2}+\left\|\sigma^{n}\left(e_{u}^{n+1}-e_{u}^{n}\right)\right\|^{2}\right)+ \eta\tau\left\|\nabla e_{u}^{n+1}\right\|^{2}+ \tau \gamma \lambda\left\|\nabla e_{w}^{n+1}\right\|^{2} \\ &+\frac{1}{2}\left(\left\|e_{\phi}^{n+1}\right\|^{2}-\left\|e_{\phi}^{n}\right\|^{2}+\left\|e_{\phi}^{n+1}-e_{\phi}^{n}\right\|^{2}\right)+ \tau \gamma\left\|e_{w}^{n+1}\right\|^{2}+\frac{\lambda}{2}\left\|\nabla e_{\phi}^{n+1}\right\|^{2}-\lambda\left\|\nabla e_{\phi}^{n}\right\|^{2}\\ &+\lambda\left\|\nabla e _{\phi}^{n+1}-\nabla e_{\phi}^{n} \right\|^{2}+\frac{\lambda}{2 \varepsilon^{2}}\left(\left\|e_{\phi}^{n+1}\right\|_{L^{4}}^{4}-\left\|e_{\phi}^{n}\right\|_{L^{4}}^{4}+\left\|(e_{\phi}^{n+1})^{2}-\left(e_{\phi}^{n}\right)^{2}\right\|^{2}\right.\\ &\left.+2\left\|e_{\phi}^{n+1}\left(e_{\phi}^{n+1}-e _{\phi}^{n}\right)\right\|^{2}\right)+\frac{2 \lambda}{\varepsilon^{2}}\left\|e_{\phi}^{n+1}-e _{\phi}^{n}\right\|^{2}\\ \leq& C \tau^{3}+C \tau^{2} \sum\limits_{m = 0}^{n}\left\|\sigma^{m} e_{u}^{m}\right\|^{2}+C\tau\left\|\sigma^{n+1}\left(e_{u}^{n+1}- \tilde{e}_{u}^{n+1} \right)\right\|^2\\ &+C\tau\left(\left\|\sigma^{n} e_{u}^{n}\right\|^{2}+\left\|e_{\phi}^{n}\right\|^{2}+\left\|\nabla e_{\phi}^{n}\right\|^{2}+\left\|\nabla e_{w}^{n+1}\right\|^{2} +\left\|e_{w}^{n+1}\right\|^{2}+\left\|e_{\phi}^{n+1}\right\|^{2}+\left\|\nabla e_{\phi}^{n+1}\right\|^{2}\right){;} \end{align} | (3.55) |
for sufficiently small \tau , taking the sum of (3.55) from 0 to N -1 and using the discrete Gronwall inequality, we can obtain Lemma 6.
Theorem 4. Suppose that the solution to (1.1) satisfies the regularity assumptions given by (3.3), and suppose that (3.1)–(3.2) are valid. For sufficiently small \tau , there are the following error estimates:
\begin{equation} \begin{aligned} &\left\|\sigma\left(t_{n}\right)\boldsymbol{u}\left(t_{n}\right)-\sigma^{n}\boldsymbol{u}^{n}\right\|^{2}+\left\|\rho\left(t_{n}\right)-\rho^{n}\right\|^{2}+\left\|\nabla\left(w\left(t_{n}\right)-w^{n}\right)\right\|^{2} \\ &+\left\|\nabla\left(\phi\left(t_{n}\right)-\phi^{n}\right)\right\|^{2}+\eta \tau \sum\limits_{m = 1}^{n}\left\|\nabla\left(\boldsymbol{u}\left(t_{m}\right)-\boldsymbol{u}^{m}\right)\right\|^{2} \leq C \tau^{2} \end{aligned} \end{equation} | (3.56) |
for all 1 \leq n \leq N .
Proof. From Lemmas 5 and 6, we obtain
\begin{equation} \left\|\rho\left(t_{n}\right)-\rho^{n}\right\|^{2}+\left\|\nabla\left(w\left(t_{n}\right)-w^{n}\right)\right\|^{2}+\left\|\nabla\left(\phi\left(t_{n}\right)-\phi^{n}\right)\right\|^{2} +\eta \tau \sum\limits_{m = 1}^{n}\left\|\nabla\left(\boldsymbol{u}\left(t_{m}\right)-\boldsymbol{u}^{m}\right)\right\|^{2} \leq C \tau^{2}. \end{equation} | (3.57) |
Thus, we only prove that
\begin{equation} \left\|\sigma\left(t_{n}\right)\boldsymbol{u}\left(t_{n}\right)-\sigma^{n}\boldsymbol{u}^{n}\right\|^{2} \leq C \tau^{2} . \end{equation} | (3.58) |
In fact, we have
\begin{equation} \begin{aligned} \left\|\sigma\left(t_{n}\right)\boldsymbol{u}\left(t_{n}\right)-\sigma^{n}\boldsymbol{u}^{n}\right\|^{2} & \leq C\left\|\left(\sigma\left(t_{n}\right)-\sigma^{n}\right)\boldsymbol{u}\left(t_{n}\right)\right\|^{2}+\left\|\sigma^{n} e_{u}^{n}\right\|^{2} \\ & \leq C\left(\left\|e_{\rho}^{n}\right\|^{2}+\left\|\sigma^{n} e_{u}^{n}\right\|^{2}\right) \leq C \tau^{2}. \end{aligned} \end{equation} | (3.59) |
Theorem 3.6 states that both \sigma^{n}\boldsymbol{u}^{n}, \; \rho^{n}, \; \boldsymbol{u}^{n} , \phi^{n} and w^{n} are order 1 approximations to \sigma\boldsymbol{u}, \; \rho, \; \boldsymbol{u} , \phi and w in l^{\infty}\left(L^{2}(\Omega)\right) , l^{\infty}\left(L^{2}(\Omega)\right) , l^{2}\left(H_{0}^{1}(\Omega)\right) , l^{\infty}\left(H_0^{1}(\Omega)\right) and l^{\infty}\left(H_0^{1}(\Omega)\right) , respectively. Finally, we can obtain order \frac{1}{2} error estimates for p approximation in l^{\infty}\left(L^{2}(\Omega)\right) .
Theorem 5. Under the assumptions in Theorem 4, the following holds true:
\begin{equation} \tau\sum\limits_{m = 1}^{N}\left\|p\left(t_{m}\right)-p^{m}\right\| \leq C \tau. \end{equation} | (3.60) |
Proof. Let us rewrite (3.40) as
\begin{equation} \begin{split} -\nabla e_{p}^{n+1} = &\rho^{n} \frac{e_{u}^{n+1}-e_{u}^{n}}{\tau}+\frac{\rho^{n+1}-\rho^{n}}{\tau}\left(e_{u}^{n+1}-\widetilde{e}_{u}^{n+1}\right)-\eta \triangle e_{u}^{n+1}-R_{u}^{n+1}\\ &+\rho\left(t_{n+1}\right)\left(\boldsymbol{u}\left(t_{n+1}\right) \cdot \nabla\right) \boldsymbol{u}\left(t_{n+1}\right)-\rho^{n+1}\left(\boldsymbol{u}^{n} \cdot \nabla\right) \tilde{\boldsymbol{u}}^{n+1}\\ &- \lambda w\left(t_{n+1}\right) \nabla \phi\left(t_{n+1}\right)+ \lambda w^{n+1} \nabla \phi^{n}.\\ \end{split} \end{equation} | (3.61) |
To prove the theorem, we introduce the discrete inf-sup condition, i.e.,
\begin{equation} \beta\left\|e_{p}^{n+1}\right\| \leq \frac{\left(\nabla e_{p}^{n+1}, v\right)}{\|\nabla v\|}. \end{equation} | (3.62) |
Then, we can constrain the products of the right-hand side of (3.61) with an arbitrary v \in V as follows:
\begin{equation*} \begin{aligned} \frac{1}{\tau}\left(\rho^{n}\left(e_{u}^{n+1}-e_{u}^{n}\right), v\right) &\leq C \tau^{-1}\left\|\sigma^{n}\left(e_{u}^{n+1}-e_{u}^{n}\right)\right\| \| \nabla v \|, \\ -\eta\left(\Delta e_{u}^{n+1}, v\right)-\left(R_{u}^{n+1}, v\right) &\leq C\left(\tau^{2}+\left\|\nabla e_{u}^{n+1}\right\|\right) \| \nabla v \|, \end{aligned} \end{equation*} |
and
\begin{equation} \begin{split} &- \lambda\left(w\left(t_{n+1}\right) \nabla \phi\left(t_{n+1}\right)-w^{n+1} \nabla \phi^{n}, v\right)\\ = & -\lambda\left(w\left(t_{n+1}\right)\left(\nabla \phi\left(t_{n+1}\right)-\nabla \phi\left(t_{n}\right)\right)+w\left(t_{n+1}\right) \nabla e_{\phi}^{n}+\nabla \phi^{n} e_{w}^{n+1}, v\right)\\ \leq& \lambda \left\| w\left(t_{n+1}\right)\right\|_{L^{\infty}}\left\| \int_{t_{n}}^{t_{n+1}} \nabla \phi_{t}(t) d t\right\|\| v\|+ \lambda\left\| w\left(t_{n+1}\right)\right\|_{L^{\infty}}\left\|\nabla e_{\phi}^{n}\right\|\left\|v\right\|\\ &+ \lambda\left\|\nabla \phi^{n}\right\|_{L^2}\left\|e_w^{n+1}\right\|_{L^{3}}\left\|v\right\|_{L^{6}}\\ \leq& C\left( \tau+\left\|\nabla e_{\phi}^{n}\right\|+\left\|\nabla e_{w}^{n+1}\right\|\right)\left\| \nabla v \right\|.\\ \end{split} \end{equation} | (3.63) |
For the second term on the right-hand side, using \|v\|_{L^{3}} \leq\|v\|^{\frac{1}{2}}\|\nabla v\|^{\frac{1}{2}} , we have
\begin{equation} \begin{aligned} &\left(\frac{\rho^{n+1}-\rho^{n}}{\tau}\left(e_{u}^{n+1}-\widetilde{e}_{u}^{n+1}\right), v\right) \\ &\leq C \tau^{-1}\left\|\rho^{n+1}-\rho^{n}\right\|\left\|e_{u}^{n+1}-\widetilde{e}_{u}^{n+1}\right\|_{L^{3}}\|v\|_{L^{6}} \\ &\leq C \tau^{-1}\left\|\rho^{n+1}-\rho^{n}\right\|\left\|e_{u}^{n+1}-\widetilde{e}_{u}^{n+1}\right\|^{\frac{1}{2}}\left\|\nabla\left(e_{u}^{n+1}-\widetilde{e}_{u}^{n+1}\right)\right\|^{\frac{1}{2}}\|\nabla v\| \\ &\leq C\left\|\sigma^{n+1}\left(e_{u}^{n+1}-\widetilde{e}_{u}^{n+1}\right)\right\|^{\frac{1}{2}}\left\|\nabla\left(e_{u}^{n+1}-\widetilde{e}_{u}^{n+1}\right)\right\|^{\frac{1}{2}}\|\nabla v\|. \end{aligned} \end{equation} | (3.64) |
By the split method, we arrive at
\begin{equation} \begin{aligned} &\left(\rho\left(t_{n+1}\right)\left(\boldsymbol{u}\left(t_{n+1}\right) \cdot \nabla\right) \boldsymbol{u}\left(t_{n+1}\right)-\rho^{n+1}\left(\boldsymbol{u}^{n} \cdot \nabla\right) \widetilde{\boldsymbol{u}}^{n+1}, v\right) \\ & = \left(\rho^{n+1}\left(\left(\boldsymbol{u}\left(t_{n+1}\right)-\boldsymbol{u}\left(t_{n}\right)\right) \cdot \nabla\right) \boldsymbol{u}\left(t_{n+1}\right), v\right)+\left(\rho^{n+1}\left(\boldsymbol{u}^{n} \cdot \nabla\right) \tilde{e}_{u}^{n+1}, v\right) \\ &\quad+\left(\rho^{n+1}\left(e_{u}^{n} \cdot \nabla\right) \boldsymbol{u}\left(t_{n+1}\right), v\right)+\left(e_{\rho}^{n+1}\left(\boldsymbol{u}\left(t_{n+1}\right) \nabla\right) \boldsymbol{u}\left(t_{n+1}\right), v\right). \end{aligned} \end{equation} | (3.65) |
Utilizing H \ddot{o} lder's inequality and Young's inequality, we derive
\begin{equation} \begin{split} &\left(\rho\left(t_{n+1}\right)\left(\boldsymbol{u}\left(t_{n+1}\right) \cdot \nabla\right) \boldsymbol{u}\left(t_{n+1}\right)-\rho^{n+1}\left(\boldsymbol{u}^{n} \cdot \nabla\right) \widetilde{\boldsymbol{u}}^{n+1}, v\right) \\ \leq& C\left(\tau+\left\| e_{\rho}^{n+1}\right\|+\left\|\nabla e_{u}^{n}\right\|+\left\|\nabla \tilde{e}_{u}^{n+1}\right\|\right) \| \nabla v\|.\\ \end{split} \end{equation} | (3.66) |
By adding up the inequalities and incorporating (3.62), we get the desired result.
In this section, we will present some numerical experiments to prove the validity and accuracy of our method. In the following simulation, for phase field \phi , chemical potential w and pressure p , we take the P1 finite element space (continuous piecewise linear), and for fluid velocity \boldsymbol{u} , we take the P2 finite element space. All experiments were conducted in Freefem++. We fixed \eta = 0.8, \lambda = 0.7, \gamma = 0.0006, \varepsilon = 0.1, T = 0.1, \rho_1 = 1 and \rho_2 = 3. The computational domain and the initial conditions were taken as
\begin{equation*} \begin{split} &\Omega = \{(x, y)\in R^2:x^2+y^2 < 1\}, \\ &\phi_0 = cos(\pi x)cos(\pi y), \\ &\boldsymbol{u}_0 = ( \pi cos(\pi x)sin(\pi y), -\pi sin(\pi x)cos(\pi y)).\\ \end{split} \end{equation*} |
Tables 1 and 2 verify that (3.4) is the first-order convergence rate O(\tau) of (\phi, \sigma\boldsymbol{u}, \boldsymbol{u}, \rho) , which is consistent with the conclusion obtained from theoretical analysis. It is only proved in Theorem 3.2 that the pressure is the half-order convergence rate O(\tau^{\frac{1}{2}}) because of technical reasons. However, the numerical results on p in Table 1 still reach the first-order optimal convergence rate O(\tau) . Tables 3 and 4 show the convergence rate with another set of parameters.
\tau | \Vert \phi\Vert_{H^1} | Rate | \Vert p\Vert_{L^2} | Rate |
0.007812 | 0.102393 | 0.469438 | ||
0.003906 | 0.0653769 | 0.647262 | 0.282333 | 0.733534 |
0.001953 | 0.0367805 | 0.829842 | 0.143976 | 0.971574 |
0.000976 | 0.0196276 | 0.906053 | 0.0725853 | 0.988076 |
\tau | \Vert \sigma \boldsymbol{u}\Vert_{L^2} | Rate | \Vert \boldsymbol{u}\Vert_{H^1} | Rate | \Vert \rho\Vert_{L^2} | Rate |
0.007812 | 0.114448 | 0.554732 | 0.0779992 | |||
0.003906 | 0.0610184 | 0.90737 | 0.2956 | 0.90814 | 0.0463154 | 0.751968 |
0.001953 | 0.02533 | 1.2684 | 0.132451 | 1.15819 | 0.022386 | 1.04889 |
0.000976 | 0.0127883 | 0.98602 | 0.071371 | 0.89204 | 0.0107557 | 1.05749 |
\tau | \Vert \phi\Vert_{H^1} | Rate | \Vert p\Vert_{L^2} | Rate |
0.007812 | 0.041787 | 0.0794722 | ||
0.003906 | 0.0215011 | 0.958642 | 0.0435045 | 0.869285 |
0.001953 | 0.0106192 | 1.01774 | 0.0210772 | 1.04548 |
0.000976 | 0.0054229 | 0.96953 | 0.0106356 | 0.986787 |
\tau | \Vert \sigma \boldsymbol{u}\Vert_{L^2} | Rate | \Vert \boldsymbol{u}\Vert_{H^1} | Rate | \Vert \rho\Vert_{L^2} | Rate |
0.007812 | 0.0308723 | 0.387768 | 0.0585448 | |||
0.003906 | 0.0161585 | 0.934018 | 0.201806 | 0.942225 | 0.0300705 | 0.961192 |
0.001953 | 0.00587585 | 1.45943 | 0.0765909 | 1.39772 | 0.00932566 | 1.68907 |
0.000976 | 0.00302534 | 0.957701 | 0.0398452 | 0.942768 | 0.00468607 | 0.992826 |
Figure 1 shows the evolution of the total energy at \tau = 0.02 . The downward trend of the energy curve confirms that our scheme is unconditionally energy stable. We also see a downward trend in energy when using different parameters. The energy curves for different time steps are shown in Figure 2 as a result of keeping the other parameters unchanged. It can been found that the curves are very similar, which means that the scheme is robust against different time steps.
To solve the Cahn-Hilliard phase field model for two-phase incompressible flows with variable density, we have designed a novel time marching scheme, which can significantly improve the calculation efficiency. The method is efficient because we decoupled the pressure from the velocity and phase field. We have also proved the unconditional energy stability, presented the error analysis and provided various numerical examples to demonstrate the stability and accuracy of the scheme. In addition, the decoupling method developed in this paper is universally applicable, and this method is always applicable for the generation of an effective fully decoupling scheme.
The authors declare they have not used Artificial Intelligence tools (AI) in the creation of this article.
This work was supported by the Research Project Supported by Shanxi Scholarship Council of China (No. 2021-029), Shanxi Provincial International Cooperation Base and Platform Project (202104041101019), Shanxi Province Natural Science Research (202203021211129), Shanxi Province Natural Science Research (No. 202203021212249) and Special/Youth Foundation of Taiyuan University of Technology (No. 2022QN101).
All authors declare that they have no competing interests in this paper.
[1] |
P. Yue, J. J Feng, C. Liu, J. Shen, A diffuse-interface method for simulating two-phase flows of complex fluids, J. Fluid Mech., 515 (2004), 293–317. http://dx.doi.org/10.1017/S0022112004000370 doi: 10.1017/S0022112004000370
![]() |
[2] |
H. Ding, P. D. M. Spelt, C. Shu, Diffuse interface model for incompressible two-phase flows with large density ratios, J. Comput. Phys., 226 (2007), 2078–2095. http://dx.doi.org/10.1016/j.jcp.2007.06.028 doi: 10.1016/j.jcp.2007.06.028
![]() |
[3] |
D. Jacqmin, Calculation of two-phase Navier-Stokes flows using phase-field modeling, J. Comput. Phys., 155 (1999), 96–127. http://dx.doi.org/10.1006/jcph.1999.6332 doi: 10.1006/jcph.1999.6332
![]() |
[4] |
J. Lowengrub, L. Truskinovsky, Quasi-incompressible Cahn-Hilliard fluids and topological transitions, Proc. R. Soc. Lond. A., 454 (1998), 2617–2654. http://dx.doi.org/10.1098/rspa.1998.0273 doi: 10.1098/rspa.1998.0273
![]() |
[5] |
C. Liu, J. Shen, A phase field model for the mixture of two incompressible fluids and its approximation by a Fourier-spectral method, Physica D, 179 (2003), 211–228. http://dx.doi.org/10.1016/S0167-2789(03)00030-7 doi: 10.1016/S0167-2789(03)00030-7
![]() |
[6] |
L. Rayleigh, On the theory of surface forces. Ⅱ. compressible fluids, The London, Edinburgh, and Dublin Philosophical Magazine and Journal of Science, 33 (1892), 209–220. http://dx.doi.org/10.1080/14786449208621456 doi: 10.1080/14786449208621456
![]() |
[7] |
J. S. Rowlinson, Translation of J. D. van der Waals' "The thermodynamik theory of capillarity under the hypothesis of a continuous variation of density", J. Stat. Phys., 20 (1979), 197–200. http://dx.doi.org/10.1007/BF01011513 doi: 10.1007/BF01011513
![]() |
[8] |
J. W. Cahn, J. E. Hilliard, Free energy of a nonuniform system. Ⅰ. interfacial free energy, J. Chem. Phys., 28 (1958), 258–267. http://dx.doi.org/10.1063/1.1744102 doi: 10.1063/1.1744102
![]() |
[9] |
J. Shen, X. Yang, Decoupled, energy stable schemes for phase-field models of two-phase incompressible flows, SIAM J. Numer. Anal., 53 (2015), 279–296. http://dx.doi.org/10.1137/140971154 doi: 10.1137/140971154
![]() |
[10] |
J. Shen, X. Yang, A phase-field model and its numerical approximation for two-phase incompressible flows with different densities and viscosities, SIAM J. Sci. Comput., 32 (2010), 1159–1179. http://dx.doi.org/10.1137/09075860X doi: 10.1137/09075860X
![]() |
[11] |
R. H. Nochetto, J. H. Pyo, The gauge-uzawa finite element method. part Ⅰ: The Navier-Stokes equations, SIAM J. Numer. Anal., 43 (2005), 1043–1068. http://dx.doi.org/10.1137/040609756 doi: 10.1137/040609756
![]() |
[12] |
J. L. Guermond, L. Quartapelle, A projection FEM for variable density incompressible flows, J. Comput. Phys., 165 (2000), 167–188. http://dx.doi.org/10.1006/jcph.2000.6609 doi: 10.1006/jcph.2000.6609
![]() |
[13] |
H. Li, L. Ju, C. Zhang, Q. Peng, Unconditionally energy stable linear schemes for the diffuse interface model with Peng-Robinson equation of state, J. Sci. Comput., 75 (2018), 993–1015. http://dx.doi.org/10.1007/s10915-017-0576-7 doi: 10.1007/s10915-017-0576-7
![]() |
[14] |
Z. Yang, S. Dong, An unconditionally energy-stable scheme based on an implicit auxiliary energy variable for incompressible two-phase flows with different densities involving only precomputable coefficient matrices, J. Comput. Phys., 393 (2019), 229–257. http://dx.doi.org/10.1016/j.jcp.2019.05.018 doi: 10.1016/j.jcp.2019.05.018
![]() |
[15] |
J. Shen, X. Yang, Numerical approximations of Allen-Cahn and Cahn-Hilliard equations, Discrete Cont. Dyn. A, 28 (2010), 1669–1691. http://dx.doi.org/10.3934/dcds.2010.28.1669 doi: 10.3934/dcds.2010.28.1669
![]() |
[16] |
J. Shen, X. Yang, Decoupled energy stable schemes for phase-field models of two-phase complex fluids, SIAM J. Sci. Comput., 36 (2014), 122–145. http://dx.doi.org/10.1137/130921593 doi: 10.1137/130921593
![]() |
[17] |
S. M. Wise, C. Wang, J. S. Lowengrub, An energy-stable and convergent finite-difference scheme for the phase field crystal equation, SIAM J. Numer. Anal., 47 (2009), 2269–2288. http://dx.doi.org/10.1137/080738143 doi: 10.1137/080738143
![]() |
[18] |
D. Han, X. Wang, A second order in time, uniquely solvable, unconditionally stable numerical scheme for Cahn-Hilliard-Navier-Stokes equation, J. Comput. Phys., 290 (2015), 139–156. http://dx.doi.org/10.1016/j.jcp.2015.02.046 doi: 10.1016/j.jcp.2015.02.046
![]() |
[19] |
Y. Gao, D. Han, X. He, U. Rüde, Unconditionally stable numerical methods for Cahn-Hilliard-Navier-Stokes-Darcy system with different densities and viscosities, J. Comput. Phys., 454 (2022), 110968. http://dx.doi.org/10.1016/j.jcp.2022.110968 doi: 10.1016/j.jcp.2022.110968
![]() |
[20] |
C. Chen, X. Yang, Efficient numerical scheme for a dendritic solidification phase field model with melt convection, J. Comput. Phys., 388 (2019), 41–62. http://dx.doi.org/10.1016/j.jcp.2019.03.017 doi: 10.1016/j.jcp.2019.03.017
![]() |
[21] |
X. Yang, H. Yu, Efficient second order unconditionally stable schemes for a phase field moving contact line model using an invariant energy quadratization approach, SIAM J. Sci. Comput., 40 (2018), 889–914. http://dx.doi.org/10.1137/17M1125005 doi: 10.1137/17M1125005
![]() |
[22] |
Z. Yang, S. Dong, An unconditionally energy-stable scheme based on an implicit auxiliary energy variable for incompressible two-phase flows with different densities involving only precomputable coefficient matrices, J. Comput. Phys., 393 (2019), 229–257. http://dx.doi.org/10.1016/j.jcp.2019.05.018 doi: 10.1016/j.jcp.2019.05.018
![]() |
[23] |
X. Wang, L. Ju, Q. Du, Efficient and stable exponential time differencing Runge-Kutta methods for phase field elastic bending energy models, J. Comput. Phys., 316 (2016), 21–38. http://dx.doi.org/10.1016/j.jcp.2016.04.004 doi: 10.1016/j.jcp.2016.04.004
![]() |
[24] |
Y. Yan, W. Chen, C. Wang, S. M. Wise, A second-order energy stable bdf numerical scheme for the Cahn-Hilliard equation, Commun. Comput. Phys., 23 (2018), 572–602. http://dx.doi.org/10.4208/cicp.OA-2016-0197 doi: 10.4208/cicp.OA-2016-0197
![]() |
[25] |
P. C. Hohenberg, B. I. Halperin, Theory of dynamic critical phenomena, Rev. Mod. Phys., 49 (1977), 435. http://dx.doi.org/10.1103/RevModPhys.49.435 doi: 10.1103/RevModPhys.49.435
![]() |
[26] |
M. E. Gurtin, D. Polignone, J. Vinals, Two-phase binary fluids and immiscible fluids described by an order parameter, Math. Mod. Meth. Appl. S., 6 (1996), 815–831. http://dx.doi.org/10.1142/S0218202596000341 doi: 10.1142/S0218202596000341
![]() |
[27] |
Y. Chen, J. Shen, Efficient, adaptive energy stable schemes for the incompressible Cahn-Hilliard Navier-Stokes phase-field models, J. Comput. Phys., 308 (2016), 40–56. http://dx.doi.org/10.1016/j.jcp.2015.12.006 doi: 10.1016/j.jcp.2015.12.006
![]() |
[28] |
D. Han, X. Wang, A second order in time, uniquely solvable, unconditionally stable numerical scheme for Cahn-Hilliard-Navier-Stokes equation, J. Comput. Phys., 290 (2015), 139–156. http://dx.doi.org/10.1016/j.jcp.2015.02.046 doi: 10.1016/j.jcp.2015.02.046
![]() |
[29] |
J. Shen, X. Yang, Energy stable schemes for Cahn-Hilliard phase-field model of two-phase incompressible flows, Chin. Ann. Math. Ser. B, 31 (2010), 743–758. http://dx.doi.org/10.1007/s11401-010-0599-y doi: 10.1007/s11401-010-0599-y
![]() |
[30] |
Z. Yang, S. Dong, An unconditionally energy-stable scheme based on an implicit auxiliary energy variable for incompressible two-phase flows with different densities involving only precomputable coefficient matrices, J. Comput. Phys., 393 (2019), 229–257. http://dx.doi.org/10.1016/j.jcp.2019.05.018 doi: 10.1016/j.jcp.2019.05.018
![]() |
[31] |
Y. Gong, J. Zhao, X. Yang, Q. Wang, Fully discrete second-order linear schemes for hydrodynamic phase field models of binary viscous fluid flows with variable densities, SIAM J. Sci. Comput., 40 (2018), 138–167. http://dx.doi.org/10.1137/17M1111759 doi: 10.1137/17M1111759
![]() |
[32] |
F. Guillén-González, G. Tierra, Splitting schemes for a Navier-Stokes-Cahn-Hilliard model for two fluids with different densities, J. Comp. Math., 32 (2014), 643–664. http://dx.doi.org/10.4208/jcm.1405-m4410 doi: 10.4208/jcm.1405-m4410
![]() |
[33] |
Q. Ye, Z, Ouyang, C, Chen, X. Yang, Efficient decoupled second-order numerical scheme for the flow-coupled Cahn-Hilliard phase-field model of two-phase flows, J. Comput. Appl. Math., 405 (2022), 113875. http://dx.doi.org/10.1016/j.cam.2021.113875 doi: 10.1016/j.cam.2021.113875
![]() |
[34] |
R. An, Error analysis of a new fractional-step method for the incompressible Navier-Stokes equations with variable density, J. Sci. Comput., 84 (2020), 3. http://dx.doi.org/10.1007/s10915-020-01253-6 doi: 10.1007/s10915-020-01253-6
![]() |
[35] |
J. L. Guermond, A. Salgado, A splitting method for incompressible flows with variable density based on a pressure Poisson equation, J. Comput. Phys., 228 (2009), 2834–2846. http://dx.doi.org/10.1016/j.jcp.2008.12.036 doi: 10.1016/j.jcp.2008.12.036
![]() |
1. | Jaeyong Choi, Seokjun Ham, Soobin Kwak, Youngjin Hwang, Junseok Kim, Stability analysis of an explicit numerical scheme for the Allen-Cahn equation with high-order polynomial potentials, 2024, 9, 2473-6988, 19332, 10.3934/math.2024941 | |
2. | Soobin Kwak, Seokjun Ham, Jian Wang, Hyundong Kim, Junseok Kim, Effective perpendicular boundary conditions in phase-field models using Dirichlet boundary conditions, 2025, 0177-0667, 10.1007/s00366-025-02109-z |
\tau | \Vert \phi\Vert_{H^1} | Rate | \Vert p\Vert_{L^2} | Rate |
0.007812 | 0.102393 | 0.469438 | ||
0.003906 | 0.0653769 | 0.647262 | 0.282333 | 0.733534 |
0.001953 | 0.0367805 | 0.829842 | 0.143976 | 0.971574 |
0.000976 | 0.0196276 | 0.906053 | 0.0725853 | 0.988076 |
\tau | \Vert \sigma \boldsymbol{u}\Vert_{L^2} | Rate | \Vert \boldsymbol{u}\Vert_{H^1} | Rate | \Vert \rho\Vert_{L^2} | Rate |
0.007812 | 0.114448 | 0.554732 | 0.0779992 | |||
0.003906 | 0.0610184 | 0.90737 | 0.2956 | 0.90814 | 0.0463154 | 0.751968 |
0.001953 | 0.02533 | 1.2684 | 0.132451 | 1.15819 | 0.022386 | 1.04889 |
0.000976 | 0.0127883 | 0.98602 | 0.071371 | 0.89204 | 0.0107557 | 1.05749 |
\tau | \Vert \phi\Vert_{H^1} | Rate | \Vert p\Vert_{L^2} | Rate |
0.007812 | 0.041787 | 0.0794722 | ||
0.003906 | 0.0215011 | 0.958642 | 0.0435045 | 0.869285 |
0.001953 | 0.0106192 | 1.01774 | 0.0210772 | 1.04548 |
0.000976 | 0.0054229 | 0.96953 | 0.0106356 | 0.986787 |
\tau | \Vert \sigma \boldsymbol{u}\Vert_{L^2} | Rate | \Vert \boldsymbol{u}\Vert_{H^1} | Rate | \Vert \rho\Vert_{L^2} | Rate |
0.007812 | 0.0308723 | 0.387768 | 0.0585448 | |||
0.003906 | 0.0161585 | 0.934018 | 0.201806 | 0.942225 | 0.0300705 | 0.961192 |
0.001953 | 0.00587585 | 1.45943 | 0.0765909 | 1.39772 | 0.00932566 | 1.68907 |
0.000976 | 0.00302534 | 0.957701 | 0.0398452 | 0.942768 | 0.00468607 | 0.992826 |
\tau | \Vert \phi\Vert_{H^1} | Rate | \Vert p\Vert_{L^2} | Rate |
0.007812 | 0.102393 | 0.469438 | ||
0.003906 | 0.0653769 | 0.647262 | 0.282333 | 0.733534 |
0.001953 | 0.0367805 | 0.829842 | 0.143976 | 0.971574 |
0.000976 | 0.0196276 | 0.906053 | 0.0725853 | 0.988076 |
\tau | \Vert \sigma \boldsymbol{u}\Vert_{L^2} | Rate | \Vert \boldsymbol{u}\Vert_{H^1} | Rate | \Vert \rho\Vert_{L^2} | Rate |
0.007812 | 0.114448 | 0.554732 | 0.0779992 | |||
0.003906 | 0.0610184 | 0.90737 | 0.2956 | 0.90814 | 0.0463154 | 0.751968 |
0.001953 | 0.02533 | 1.2684 | 0.132451 | 1.15819 | 0.022386 | 1.04889 |
0.000976 | 0.0127883 | 0.98602 | 0.071371 | 0.89204 | 0.0107557 | 1.05749 |
\tau | \Vert \phi\Vert_{H^1} | Rate | \Vert p\Vert_{L^2} | Rate |
0.007812 | 0.041787 | 0.0794722 | ||
0.003906 | 0.0215011 | 0.958642 | 0.0435045 | 0.869285 |
0.001953 | 0.0106192 | 1.01774 | 0.0210772 | 1.04548 |
0.000976 | 0.0054229 | 0.96953 | 0.0106356 | 0.986787 |
\tau | \Vert \sigma \boldsymbol{u}\Vert_{L^2} | Rate | \Vert \boldsymbol{u}\Vert_{H^1} | Rate | \Vert \rho\Vert_{L^2} | Rate |
0.007812 | 0.0308723 | 0.387768 | 0.0585448 | |||
0.003906 | 0.0161585 | 0.934018 | 0.201806 | 0.942225 | 0.0300705 | 0.961192 |
0.001953 | 0.00587585 | 1.45943 | 0.0765909 | 1.39772 | 0.00932566 | 1.68907 |
0.000976 | 0.00302534 | 0.957701 | 0.0398452 | 0.942768 | 0.00468607 | 0.992826 |