On the rate of convergence to steady state in a linear chromatography model

Technical Summary: On the rate of convergence to steady state in a linear chromatography model

Problem Statement
The paper investigates the rate of convergence to the steady state in the True Moving Bed (TMB) model of linear chromatography. This model is a system of eight linear partial differential equations of hyperbolic type, describing the transport of solute concentrations in liquid ($c$) and solid ($q$) phases across four distinct zones. The system is coupled through the equations themselves and through specific boundary conditions representing injection and extraction ports. While previous work by the authors established the existence of a unique steady state and provided asymptotic approximations for low mass transfer resistance, this study focuses on the precise rate of convergence to this steady state and the asymptotic profile of the solutions for large times. The convergence rate is directly related to the duration of transient states in industrial Simulated Moving Bed (SMB) processes.

Methodology
The authors employ a combination of functional analysis, spectral theory, and numerical techniques:

1. Abstract Operator Formulation: The evolution problem is cast in the abstract form $(c_t, q_t)^T = A(c, q)^T$ on a Hilbert space $X = L^2 \times L^2$. The operator $A$ is defined on a domain $D(A)$ incorporating the continuity and jump conditions at the zone interfaces.
2. Characteristic Equation Construction: The eigenvalue problem $A(c, q)^T = \lambda(c, q)^T$ is reduced to a single scalar characteristic equation $\Delta(\lambda) = 0$. This is achieved via a "Return Map" (or Monodromy map), a $2 \times 2$ matrix $C(\lambda)$ constructed by multiplying transfer matrices across the four zones. The eigenvalues correspond to the roots of $\Delta(\lambda) = -\det(C(\lambda) - I) = 0$.
3. Functional Analysis and Positivity: Due to the complexity of the characteristic equation, the authors utilize the theory of Positive Operators and the Krein-Rutman Theorem. They prove that the resolvent operator $(-A + sI)^{-1}$ is compact and positive for $s > 0$. This allows them to establish the existence of a dominant real eigenvalue without relying on the Spectral Mapping Theorem, which they note cannot be proven to hold for this specific system due to the lack of compactness of the evolution operator $e^{tA}$.
4. Sensitivity Analysis: Using the Adjoint Method, the authors derive explicit formulas for the partial derivatives of the dominant eigenvalue $\lambda_0$ with respect to the six model parameters (four zone velocities $v_i$, and kinetic/thermodynamic parameters $R$ and $P$).
5. Numerical Implementation: The theoretical results are validated and illustrated through a case study involving the separation of omeprazole enantiomers. Numerical methods, including a volume-of-fluid approach combined with operator splitting, are used to simulate the time evolution and verify the asymptotic profile.

Key Contributions and Results

• Existence of a Dominant Eigenvalue: The paper proves that the operator $A$ possesses a unique largest real eigenvalue $\lambda_0 < 0$ (under strict inequality conditions for zone velocities). This eigenvalue is simple and has a strictly positive eigenfunction.
• Rate of Convergence: It is established that $\lambda_0$ determines the optimal decay rate of the system. Specifically, for any bounded initial condition, the solution satisfies an inequality of the form $|(c(x,t), q(x,t))| \leq K e^{\lambda_0 t} |(c_0(x), q_0(x))|$. This confirms that the spectral bound of $A$ governs the convergence speed to the steady state.
• Asymptotic Profile Conjecture: While a rigorous analytical proof is not provided, the authors conjecture (supported by numerical experiments) that the long-time profile of the solution converges to the shape of the dominant eigenfunction $(c_0, q_0)^T$.
• Explicit Limit Case: In the special limit case where all zone velocities are equal ($v_1=v_2=v_3=v_4$), the authors derive an explicit formula for the entire spectrum of eigenvalues, showing that the spectrum lies in a vertical strip and does not allow for the compactness of the evolution operator.
• Sensitivity Formulas: The paper provides explicit integral formulas for $\frac{\partial \lambda_0}{\partial v_i}$, $\frac{\partial \lambda_0}{\partial R}$, and $\frac{\partial \lambda_0}{\partial P}$, enabling the quantification of how parameter variations affect the convergence rate.
• Case Study Application: Applying these results to the separation of omeprazole enantiomers, the authors calculate a dominant eigenvalue of $\lambda_0 \approx -0.11$, corresponding to a time constant of approximately 13.6 minutes. They demonstrate that the transient concentration profiles align with the dominant eigenfunction.

Significance and Claims
The paper claims to provide the first rigorous proof that the rate of convergence to the steady state in the TMB model is governed by a dominant eigenvalue, utilizing the Krein-Rutman Theorem to overcome the difficulty of proving the compactness of the evolution semigroup. The authors emphasize that their approach offers explicit quantitative information regarding the convergence rate, which is crucial for designing and optimizing SMB processes.

The work is presented as a theoretical advancement that bridges the gap between abstract spectral theory and practical chromatographic engineering. The authors modestly note that while they have proven the existence of the dominant eigenvalue and its role in the decay rate, the proof that the solution profile asymptotically approaches the eigenfunction shape remains a conjecture supported by numerical evidence, leaving it as a direction for future work. The paper also highlights that the characteristic equation derived is complex and requires a mix of analytical and numerical methods to solve, except in the specific limit case of equal velocities where a complete explicit picture of the spectrum is possible.