Bent optical waveguide finite element analysis with a 3D envelope Maxwell model

This paper presents a novel numerical methodology using an ultraweak variational formulation of the envelope Maxwell model, discretized by the discontinuous Petrov-Galerkin (DPG) method with specialized perfectly matched layers, to accurately extract optical field losses in 3D circularly coiled waveguides as a boundary value problem, achieving stable convergence for the first time with this specific approach.

The Gist

Imagine you have a garden hose. If you hold it straight, water flows through it smoothly with almost no loss. But if you bend that hose into a tight circle, some water squirts out the side, right?

In the world of high-tech lasers and internet cables (optical fibers), light behaves similarly. When you coil a fiber optic cable tightly, some of the light "leaks" out, causing a loss of signal. This is a big problem for engineers who need to pack these cables into small spaces or use them in powerful lasers.

This paper is about building a super-precise digital simulator to predict exactly how much light leaks out when you bend a fiber optic cable, without having to build and break hundreds of physical cables to test it.

Here is a breakdown of how they did it, using some everyday analogies:

1. The Problem: The "Squishy" Light

Normally, light travels in a straight line inside the glass core of a fiber. But when you bend the fiber, the light wants to keep going straight (like a car trying to take a sharp turn without slowing down). It hits the wall of the bend and tries to escape.

• The Challenge: Calculating this is incredibly hard because light waves wiggle trillions of times per second. Trying to simulate every single wiggle over a long distance is like trying to count every grain of sand on a beach while the tide is coming in. It takes too much computer power.

2. The Solution: The "Envelope" Trick

The authors used a clever math trick called an "Envelope Maxwell Model."

• The Analogy: Imagine a surfer riding a giant wave. The wave itself is huge and moves fast, but the surfer (the "envelope") just rides the top. Instead of tracking every tiny ripple of the ocean (the high-frequency light), the computer only tracks the surfer's path (the "envelope" of the light).
• The Benefit: This allows the computer to simulate the light traveling for miles (or thousands of wavelengths) without getting overwhelmed by the sheer speed of the light waves.

3. The Method: The "Smart Mesh" (DPG)

To solve the math, they used a technique called Discontinuous Petrov–Galerkin (DPG).

• The Analogy: Imagine you are trying to draw a perfect circle on a piece of paper using a grid of squares. If you use big squares, your circle looks blocky and inaccurate. If you use tiny squares, it looks smooth, but you need millions of them.
• The Innovation: Their method is like a smart painter. It starts with big squares. Where the light is simple, it leaves the squares big. But where the light is getting complicated (like near the bend where it's leaking), the computer automatically zooms in and chops those squares into tiny, tiny pieces. This saves massive amounts of computer power while keeping the result super accurate.

4. The "Black Hole" Walls (PMLs)

In a real fiber, light that leaks out just disappears into the air. In a computer simulation, if you just stop the simulation at the edge of the screen, the light hits the wall and bounces back, ruining the math.

• The Analogy: To stop the light from bouncing back, they built Perfectly Matched Layers (PMLs). Think of these as "black hole" walls surrounding your simulation. When the light hits them, it doesn't bounce; it gets absorbed and vanishes, just like it would in real life.
• The Twist: Because the fiber is bent, they had to invent two types of these black hole walls: one to catch light leaking out the side of the bend, and another to catch light leaking out the end of the coil.

5. The Results: Predicting the Leak

They tested their simulator in three ways:

1. The Vacuum Tube: They simulated light in a simple, empty bent tube where they already knew the answer. Their simulator matched the known answer perfectly.
2. The Flat Strip: They simulated a flat, bent glass strip. They compared their results to a known mathematical formula and found they matched almost exactly.
3. The Real Fiber (The Big Win): Finally, they simulated a real 3D coiled fiber optic cable. This is the first time anyone has successfully done this specific type of simulation for a 3D coiled fiber.
   • They showed that as the light travels around the bend, it slowly spreads out of the core and gets absorbed by their "black hole" walls.
   • They calculated exactly how much power was lost, matching theoretical predictions.

Why Does This Matter?

This isn't just about math for math's sake. This technology helps engineers:

• Design better lasers: By knowing exactly how much light is lost when coiling a fiber, they can design lasers that don't overheat or fail.
• Filter signals: They can intentionally bend fibers to "scoop out" bad light signals while keeping the good ones.
• Pack tighter: We can make smaller, more efficient devices because we can trust the math to tell us how tightly we can coil the cables without breaking them.

In short: The authors built a "virtual wind tunnel" for light. Instead of guessing how light behaves when you bend a fiber, they created a digital tool that zooms in exactly where needed and absorbs the "leaks" perfectly, giving engineers a crystal-clear map of how to build better optical devices.

Technical Summary

1. Problem Statement

The paper addresses the challenge of accurately simulating optical field losses in three-dimensional (3D), circularly coiled (bent) optical waveguides (e.g., optical fibers).

• Physical Context: When an optical fiber is bent, the curvature causes the guided optical field to partially escape the core region, leading to confinement (bend) losses. This phenomenon is critical for applications such as filtering higher-order modes (HOMs) to improve beam quality and suppressing thermally induced transverse mode instability (TMI) in high-power fiber lasers.
• Computational Challenge: Simulating this requires resolving wave propagation over thousands of wavelengths in a 3D domain. Standard full-wave Maxwell solvers are computationally prohibitive for such long propagation distances due to the high frequency of the optical field. Furthermore, the geometry involves complex boundary conditions where unguided energy radiates outward, requiring effective absorbing boundary conditions in both the propagation direction and the transverse (radial) directions.

2. Methodology

The authors propose a novel numerical framework combining a specialized Envelope Maxwell model with the Discontinuous Petrov–Galerkin (DPG) finite element method.

A. Envelope Maxwell Model

• Full Envelope Ansatz: Instead of solving for the rapidly oscillating full electromagnetic fields ($E, H$), the model introduces an ansatz that shifts the wavenumber. The fields are expressed as $E = e^{-ikr_0\theta}\mathcal{E}$ and $H = e^{-ikr_0\theta}\mathcal{H}$, where $k$ is an envelope wavenumber and $r_0$ is the bend radius.
• Benefit: This transformation allows the solver to compute low-frequency envelope fields ($\mathcal{E}, \mathcal{H}$) rather than the full high-frequency waves, drastically reducing the required mesh resolution and computational cost while maintaining accuracy over long propagation distances.
• Coordinate Transformation: The model adapts the vectorial envelope formulation from straight Cartesian coordinates to curvilinear toroidal coordinates to account for the bending geometry.

B. Ultraweak Variational Formulation & DPG

• Ultraweak Formulation: The authors derive an ultraweak variational formulation of the boundary value problem (BVP). This approach reduces numerical dispersion (pollution) errors common in wave propagation problems.
• DPG Method: The problem is discretized using the Discontinuous Petrov–Galerkin (DPG) method.
  • Stability: DPG guarantees stability by using optimal test functions.
  • Adaptivity: The method provides a built-in residual-driven error indicator. This allows for automated $hp$-adaptivity (refining mesh size $h$ and polynomial order $p$) specifically where the solution requires higher resolution (e.g., near the core-cladding interface or where radiation escapes).

C. Perfectly Matched Layers (PML)

A unique contribution is the construction of PMLs for the curved geometry to absorb outgoing radiation without reflection:

1. Longitudinal PML ($\theta$): Absorbs waves exiting the end of the waveguide segment.
2. Transverse PML ($r$ or $\rho$): Absorbs radiation caused by the curvature that escapes the waveguide radially.

• The authors implement complex coordinate stretching in both the longitudinal and transverse directions, deriving the necessary Jacobian transformations for both cylindrical and toroidal geometries to pull the Maxwell equations back to physical coordinates.

3. Key Contributions

1. Extension to Bent Geometries: Successfully extended the envelope Maxwell formulation from straight fibers to bent/toroidal waveguides, handling the transformation from Cartesian to toroidal coordinates.
2. Novel PML Construction: Developed a specific PML construction for circularly coiled waveguides that absorbs radiation in both the propagation direction and the transverse escape direction.
3. First 3D Adaptive DPG Simulation: Presented the first example of an adaptive finite element method for the 3D Maxwell simulation of bent optical fibers using the DPG framework.
4. Verification Strategy: Validated the 3D model against semi-analytical solutions for 2D bent slab waveguides, demonstrating that the 3D full-field simulation converges to the same loss values as the eigen-solution methods.

4. Numerical Results

The authors validated their methodology through three numerical experiments:

• Experiment 1 (Vacuum Waveguide): A 2D-like vacuum waveguide with Perfect Electric Conductor (PEC) walls.
  • Result: Demonstrated expected convergence rates ($O(h^{p/2})$) for the DPG residual and relative error, verifying the discretization scheme.
• Experiment 2 (Step-Index Slab Waveguide): A 2D bent slab waveguide with a radiation condition.
  • Result: The numerically computed loss factors (imaginary part of the propagation constant) converged to the semi-analytical values derived from eigen-solutions. The method accurately captured bend losses for various modes and bend radii.
• Experiment 3 (3D Bent Optical Fiber): A full 3D simulation of a step-index fiber coiled into a torus.
  • Setup: Excited with the $LP_{02}$ mode of a straight fiber and propagated through a bent domain.
  • Observations:
    • The adaptive mesh successfully refined the solution in the transverse directions where the field spread out.
    • Visualizations showed the optical field profile distorting and spreading from the core into the cladding as it propagated.
    • Power Loss: The simulation accurately captured the exponential decay of optical power. As expected, a tighter bend radius ($r_0=1300$) resulted in significantly higher power loss compared to a larger radius ($r_0=2600$).
  • Scale: The simulation utilized up to 24.4 million degrees of freedom on the Frontera supercomputer.

5. Significance and Future Work

• Scientific Impact: This work provides a high-fidelity, computationally efficient tool for analyzing bend losses in 3D optical fibers, a problem previously difficult to solve with full-wave 3D methods due to scale.
• Application: The methodology is a critical step toward modeling Thermally Induced Mode Instability (TMI) in bent fiber amplifiers, where accurate loss calculation of higher-order modes is essential for predicting system stability.
• Future Directions: The authors plan to:
  • Simulate the onset of TMI in bent fibers.
  • Extend the stability analysis to domains with impedance boundary conditions.
  • Incorporate elasto-optical effects (stress-induced anisotropy) caused by bending and tension into the refractive index model.

In summary, the paper presents a robust, adaptive, and highly accurate numerical framework for simulating light propagation in bent optical fibers, bridging the gap between theoretical semi-analytical models and complex 3D physical realities.