Resonance in rarefaction and shock curves: Local analysis and numerics of the continuation method

Amaury Álvarez, Guilherme Tegoni Goedert, D. Marchesin

TL;DR

This paper introduces a robust numerical algorithm to solve the Riemann problem in systems of conservation laws where classical assumptions—like distinct wave speeds and predictable nonlinearity—break down. By using "Generalized Jordan Chains" and lifting calculations into higher-dimensional spaces, the authors provide a mathematical bridge to continue wave curves through singular points that previously crashed standard solvers.

What Problem Does This Paper Address?

In the study of fluid dynamics and physics, we often use systems of conservation laws to describe how quantities like mass or momentum move. To solve these, researchers typically rely on the "Riemann problem": predicting what happens when two different states of a medium (like two different pressures in a pipe) meet at a single point.

For decades, the gold standard for solving these was Lax’s theory, which relies on two big assumptions:

  1. Strict Hyperbolicity: Every wave family moves at a unique, distinct speed.
  2. Genuine Nonlinearity: Waves either sharpen into shocks or spread out into rarefactions in a predictable way.

The catch? Real-world models—especially those involving multi-phase flow in porous media (like oil recovery or CO2 sequestration)—frequently violate these rules. Wave speeds can coincide (resonance), and nonlinearity can vanish (inflection points). When these "singularities" occur, standard numerical integrators get lost, unable to decide which direction the wave curve should go.

Methods

The authors propose a continuation method designed to "sail through" these mathematical storms. Their approach combines several sophisticated tools:

  • Generalized Jordan Chains: When wave speeds coincide, the underlying matrices become singular. The authors use Jordan chains to regularize these points, effectively providing the missing directional information needed to keep the algorithm moving.
  • Manifold Lifting: Instead of looking at the wave curves only in the "state space" (the physical variables), they lift the problem into an extended space that includes the wave speeds ($\lambda$) as a coordinate. This turns a messy intersection of curves into a smooth, navigable surface.
  • Versal Deformation Theory: They use this to create a local coordinate system near the coincidence points, allowing the solver to maintain accuracy even when the "discriminant" (the math that tells speeds apart) hits zero.
  • Event Detection: The algorithm uses specific stopping criteria to identify exactly when a curve hits a coincidence or inflection locus, triggering a switch to the specialized continuation logic.

Datasets

As a theoretical and algorithmic paper, the "datasets" are specific physical models used to stress-test the new method:

  • The ICDOW Model: A system of three conservation laws describing complex fluid flow (water saturation and chemical composition).
  • Quadratic Corey Permeability Model: A classic 2x2 system used in petroleum engineering that features an "umbilic point" where wave speeds coincide.
  • RPN C++ Library: The algorithms were implemented and validated using this specialized library alongside Matlab.

Key Insights and Ideas

  • The Geometry of Resonance: The paper treats the "coincidence locus" (where speeds match) not as a dead end, but as a submanifold of codimension one. By understanding its geometry, they can predict how a rarefaction wave might "transform" into a different family or a shock wave upon crossing it.
  • Removing Singularities: One of the "aha" moments is the realization that singularities in the composite wave field can be removed by scaling the eigenvectors correctly. This allows the use of standard ODE solvers in regions that were previously considered numerically unstable.
  • Composite Waves are Mandatory: When a rarefaction curve hits an inflection point, it can't just stop. The authors prove that a "composite wave"—a hybrid of a rarefaction and a shock—must exist to continue the solution, even in non-strictly hyperbolic settings.

Key Results

The authors successfully demonstrated that their algorithm can construct complete Riemann solutions for non-strictly hyperbolic systems.

  • Numerical Stability: They showed that by using the lifted coordinate system, they avoided the "chatter" and error accumulation that happens when solvers try to integrate curves that are nearly tangent to each other.
  • Theoretical Proofs: They provided new proofs for the existence and uniqueness of composite wave curves in "anomalous" cases where the wave traverses an inflection locus.
  • Visualization: The paper provides clear phase-space diagrams (like the triangular geometry in the Corey model) showing wave curves successfully navigating through the umbilic point and inflection lines to connect the left and right states.

Limitations and Future Work

While the method is a significant step forward, the authors note a few hurdles:

  • Parameter Sensitivity: The method relies on a small parameter ($\epsilon$) to "jump" past the singularity. Choosing an $\epsilon$ that is small enough for accuracy but large enough to avoid machine precision issues remains a delicate balancing act for the user.
  • Higher-Order Singularities: The current work focuses on "codimension one" coincidences (where two speeds match). Systems where three or more speeds coincide simultaneously (higher codimension) are much more complex and are left for future research.
  • Automation: While the framework is robust, integrating it into a fully "black-box" solver for any arbitrary system of equations still requires significant manual setup of the model's derivatives and Hessians.

Summary updated

Abstract

We describe certain crucial steps in the development of an algorithm for finding the Riemann solution to systems of conservation laws. We relax the classical hypotheses of strict hyperbolicity and genuine nonlinearity due to Lax. First, we present a procedure for continuing wave curves beyond points where characteristic speeds coincide, i.e. at wave curve points of maximal co-dimensionality. This procedure requires strict hyperbolicity on both sides of the coincidence locus. Loss of strict hyperbolicity is regularized by means of a Generalized Jordan Chain, which serves to construct a four-fold sub-manifold structure on which wave curves can be continued. Second, we analyze the loss of genuine nonlinearity. We prove a new result: the existence of composite wave curves when the composite wave traverses either the inflection locus or an anomalous part of the non-local composite wave curve. In this sense, we find conditions under which the composite field is well defined and its singularities can be removed, allowing use of our continuation method. Finally, we present numerical examples for a non-strictly hyperbolic system of conservation laws.
BibTeX
@article{alvarez2020resonance,
  author = {Amaury Álvarez and Guilherme Tegoni Goedert and D. Marchesin},
  title = {{Resonance in rarefaction and shock curves: Local analysis and numerics of the continuation method}},
  journal = {Journal of Hyperbolic Differential Equations},
  year = {2020},
  doi = {10.1142/s0219891620500198},
}
Figure