Characterisation of the Thermoflow due to the Dry Nitrogen Flushing Scheme in the ATLAS Inner Tracker using Computational Fluid Dynamics

This paper utilizes computational fluid dynamics to characterize and optimize the dry nitrogen flushing scheme for the ATLAS Inner Tracker's High Luminosity upgrade, ensuring that moisture ingress is effectively managed to prevent condensation and protect detector electronics from damage at extreme low temperatures.

The Gist

Imagine the ATLAS detector at CERN as a giant, incredibly sensitive camera sitting inside a massive particle accelerator. This camera is about to get a huge upgrade to take even more detailed photos of the universe. However, this new camera (called the ITk) is built with delicate electronics that are terrified of one thing: moisture.

If even a tiny bit of water vapor condenses on the electronics, it could freeze or rust, ruining the camera. To keep the camera safe, the engineers need to keep the air inside it as dry as desert sand, with a "dew point" (the temperature at which water starts to condense) colder than -60°C.

The Problem: The "Dry Blanket" vs. The "Leaky Roof"

To keep the camera dry, the team pumps in dry nitrogen gas (like a giant, invisible, dry blanket) to push out any humid air. However, there are two big worries:

1. Leaks: Air from the outside (which is humid) might sneak in through tiny cracks or holes.
2. Dead Zones: The dry nitrogen might not reach every corner of the camera. If the gas gets stuck in one spot, it leaves "dead zones" where humid air can hide and condense.

The engineers needed to know: Is our dry nitrogen blanket thick enough to cover the whole camera, even if there are leaks? And where should we put our "smoke detectors" (humidity sensors) to catch a leak before it causes damage?

The Solution: A Digital Wind Tunnel

Instead of building a full-size model and blowing air through it (which would be expensive and slow), the authors used Computational Fluid Dynamics (CFD). Think of this as a super-advanced video game simulation of the air inside the camera.

They built a digital 3D model of the detector's interior and simulated how the dry nitrogen flows, how it mixes with humid air, and how the temperature changes.

What They Discovered

The simulation acted like a magnifying glass, revealing two major issues with their original design:

1. The "Short-Circuit" Problem: In the original plan, the dry nitrogen was entering and leaving the room too close to each other. It was like opening a window and a door right next to each other; the fresh air would just rush straight out without cleaning the rest of the room. This left the far corners of the detector "stale" and humid.
2. The "Hot and Cold" Layers: Because warm air rises and cold air sinks (like in a room with a heater), the simulation showed that the dry nitrogen was floating on top, leaving the bottom of the detector humid and warm. This was dangerous for the electronics.

The Fix: Rearranging the Pipes

The engineers used the simulation to test a new layout. They moved the dry nitrogen pipes so they were closer to the exit, but arranged in a way that forced the gas to sweep through the entire volume before leaving.

The results of this new design were promising:

• No More Dead Zones: The dry nitrogen now flowed evenly throughout the detector, reaching the bottom corners that were previously neglected.
• The Leak Test: They simulated two scenarios: a "small leak" and a "big leak."
  • Big Leak (0.1 liters/second): The dry nitrogen couldn't keep up. The humidity rose, and the dew point got too warm (above -60°C). This was a fail.
  • Small Leak (0.02 liters/second): The dry nitrogen successfully pushed out the moisture. The dew point stayed safely below -60°C. This was a pass.

The Bottom Line

This paper is essentially a "safety check" for the ATLAS detector's upgrade. By using a computer simulation to visualize the invisible flow of air, the team proved that:

1. They needed to move their pipes to ensure the dry nitrogen covers the whole detector.
2. The system can handle small leaks (up to 0.02 liters per second) without the electronics getting wet, but it would fail if the leaks were five times bigger.

This ensures that when the real detector is built, it will stay dry and functional, ready to capture the secrets of the universe without short-circuiting from a drop of water.

Technical Summary

Technical Summary: Characterisation of the Thermoflow due to the Dry Nitrogen Flushing Scheme in the ATLAS Inner Tracker

Problem Statement
The High Luminosity Large Hadron Collider (HL-LHC) upgrade at CERN necessitates a complete replacement of the ATLAS Inner Detector with a new all-silicon Inner Tracker (ITk). A critical operational constraint for the ITk is the prevention of moisture condensation on detector electronics, which could lead to corrosion or ice formation. To mitigate this, the system must maintain a dew point temperature at or below −60°C (corresponding to a water content of 10.5 ppm). The design employs a dry nitrogen (N₂) flushing scheme to maintain a positive overpressure and remove moisture. However, there is a risk of humid air ingress through leaks or diffusion through service feed-throughs. The primary challenge addressed in this work is the potential for non-uniform N₂ distribution, leading to "dead zones" with low atmospheric renewal rates, and the need to determine the optimal placement of humidity sensors and the maximum permissible leak rate to ensure the entire volume remains within the specified dew point range.

Methodology
The study utilizes Computational Fluid Dynamics (CFD) to model the thermoflow, temperature, humidity, and dew point distributions within the ITk Strips region.

• Geometry and Mesh: A simplified geometric model was developed, idealizing detector petals as discs and staves as rectangular plates. The model leverages symmetry to simulate a quarter of the ITk volume (reducing computational load). The geometry includes the Outer Service Volume (OSV), Strips Endcaps, and Strips Barrel, separated by structural components like the bulkhead, stiffener disc, and polymoderator. The mesh consists of approximately 8.8 million cells, utilizing a mix of structured hexahedral and unstructured tetrahedral cells.
• Physics Models: The simulation employs a steady-state, incompressible flow solver based on the Reynolds-Averaged Navier-Stokes (RANS) equations. Turbulence is modeled using the realizable $k-\epsilon$ model. Species transport equations are solved for nitrogen, oxygen, and water vapor to track humidity. Heat transfer is modeled with constant temperature boundaries for detector surfaces (−25°C) and cooling pipes, while structural components like the stiffener disc are modeled with 3D conduction.
• Boundary Conditions: The simulation investigates three scenarios: (1) a baseline with no leaks, (2) a high leak rate of 0.1 l/s (representing 10% of the nominal flush rate), and (3) a lower leak rate of 0.02 l/s (2% of the flush rate). Humid air leaks are introduced at 12 specific points on the OSV. Two distinct piping configurations were tested: an original design and a revised design where inlet manifolds were repositioned closer to the outlets to mitigate stratification.
• Dew Point Calculation: The dew point is calculated using the Bögel modification of the Magnus formula.

Key Contributions and Results

1. Optimization of Inlet Positioning: Initial simulations using the original piping design revealed significant non-uniformity in flow distribution, characterized by temperature and humidity stratification (warmer, more humid air accumulating at the bottom of the volume) and potential short-circuiting of flow. By repositioning the inlet manifolds closer to the outlets, the revised design achieved a more uniform circulation of N₂, improved coverage in the lower regions of the Endcaps and Barrel, and eliminated the short-circuiting issue.
2. Leak Rate Assessment:
   • 0.1 l/s Leak Rate: Under the revised piping design, a leak rate of 0.1 l/s resulted in a volume-averaged dew point of −41.2°C. This exceeds the design specification of ≤−60°C, indicating that this leak rate is unacceptable.
   • 0.02 l/s Leak Rate: A reduced leak rate of 0.02 l/s resulted in a volume-averaged dew point of −69.7°C. This value satisfies the design requirement of ≤−60°C and maintains the relative humidity below 10% throughout the volume.
3. Structural Integrity: The CFD model calculated the temperature gradient across the stiffener disc, a critical structural component. The gradient was found to be 0.01°C, well below the design specification limit of 4°C, confirming that thermal deformation is not a concern under the modeled conditions.
4. Sensor Coverage and Dead Zones: The study identified that the original manifold design led to attenuation of N₂ flow along the length of the Strips, creating potential dead zones. The revised design improves coverage, though the authors note that flow attenuation still occurs downstream, suggesting that further manifold optimization (e.g., varying nozzle sizes) is required for perfect uniformity.

Significance and Claims
The paper claims that the presented CFD model provides a quantitative and qualitative understanding of the internal fluid environment of the ATLAS ITk under various operational and failure conditions. The primary significance lies in its ability to inform engineering design changes, specifically the repositioning of inlet manifolds, to ensure the ITk remains dry. The study establishes that a maximum permissible leak rate of 0.02 l/s is required to maintain the dew point within the acceptable design specification.

The authors modestly frame the work as an initial estimate and a tool for the "Conceive, Design, Implement and Operate" (CDIO) engineering paradigm. They acknowledge that the current model uses simplified geometries (discs/plates instead of petals/staves) and assumes uniform temperature boundaries. Consequently, the model predictions are intended to guide design iterations, with final validation against experimental data scheduled for post-commissioning of the ITk Upgrade (2028–2030). The paper does not claim to have solved all flow distribution issues, noting that future work will focus on optimizing manifold nozzle sizes and incorporating more realistic geometric details.