STAR-CD helps CERN to turn back time
TS/CV/DC CFD Team, CERN, CH

Introduction

CERN is the European Organization for Nuclear Research, the world’s largest particle physics laboratory and the birthplace of the World Wide Web. Its primary objective is to provide the scientific community with facilities to study sub-nuclear particles and the forces of matter. Most of the activities at CERN are currently directed towards building a new particle accelerator and collider, the Large Hadron Collider (LHC) and the detector experiments for it. The final aim is to look back in time and recreate the environment present at the origin of the Universe to understand what matter is made of and what forces hold it together. Construction of these experiments requires an extraordinary engineering effort and STAR-CD has been used for numerical simulations of thermal-fluid related problems, particularly during the development, design and construction phases of the LHC experiments. This article presents, therefore, the study performed in one of the five experiments currently being built to run on the collider.

The problem

ALICE (A Large Ion Collider Experiment) is a general-purpose heavyion experiment designed to study the physics of strongly interacting matter and the quark–gluon plasma in nucleus–nucleus collisions at the LHC. It consists of a variety of tracing devices, called subdetectors, enclosed in a large solenoid magnet known as L3, see figure 1. Inside the L3 envelope there are heat sources, in the form of cables and electronic circuits, dissipating heat into the surrounding air. To remove this heat a dedicated ventilation system is in place and the internal surfaces of the magnet are lined with a thermal screen. However, to maintain the quality of the particle measurements, stringent temperature requirements must be satisfied inside the L3 volume. In this system, natural convection plays an important role in transferring this heat away and the resulting gradient of temperatures in the enclosed environment cannot be neglected. As a result, STAR-CD was employed to improve the ventilation system inside the L3 magnet.

The CFD model

The geometry was built in STAR-Design solid modeler and meshed with the automatic mesh generation module, pro-STAR which uses trimmed cell technology, see figure 2. The mesh was locally refined in regions where larger gradients of temperature and velocity are expected to occur, such as close the walls of the internal subdetectors. Since we were only interested in the air region enclosed in the L3magnet, no solid parts were included in the model. The final mesh arrangement contains approximately one million cells. Inlet and outlet boundaries were prescribed at inflow and outflow regions and constant heat flux conditions at the surfaces corresponding to the boundaries of the internal sub-detectors. Additionally, a constant temperature value was defined at the magnet thermal screen. Air was modeled as non-isothermal, incompressible (with density dependent on temperature) and turbulent gas. The widely applied high Reynolds k-? turbulence model with default wall functions was believed to provide good representation of the enclosed flow. In order to accurately predict the effects of buoyancy, a transient solution was employed. The overall model predicts the air flow and temperature fields inside the L3 volume under different ventilation configurations. The CFD computations were performed on a parallel, double CPU Itanium Linux cluster, Openlab, available at CERN.

Results and discussion

The results indicate that without forced ventilation the air temperature in the half upper region of the L3 volume considerably exceeds the maximum temperature limit that ensures adequate running conditions of the sub-detectors, see figure 3. As expected, the hot air accumulates in the volume in a stratified manner with a maximum temperature difference along the height surpassing the 20K.

On its own, the L3 thermal screen is insufficient for adequate removal of the heat generated by the internal sub-detectors so that a ventilation strategy is required to lower down and even out this temperature.

As a consequence, a number of ventilation configurations were considered and the effects of varying the flow rate, quantity, location and orientation of the inlet/outlet ducts investigated. It was found that a combination of 2 inlets placed on the floor and 2 others at the middle level with an outlet positioned at the top would provide good mixing and consequently adequate temperature uniformity of the air enclosed in the L3 volume. Temperature differences along the height were reduced to around 6K and the maximum temperatures registered at the top are now well within the acceptable working limits of the sub-detectors, see figure 4. Natural convection is still the main mode of heat transfer since the velocity magnitudes in the ventilation system and comparable to that of buoyancy driven flows.

Conclusions

At CERN, STAR-CD has been proven to be a useful tool in assisting the development of cooling systems for high energy particle detectors. Because of the detectors large dimensions and tight project timescales, experimental work and prototype modeling is difficult and CFD models become the practical alternative. The CFD study presented allowed improvements to the ventilation system and consequently enhancement of the detector’s performance. More details on this and other similar studies can be found at http://cern.ch/cfd, http://aliceinfo.cern.ch or http://cern.ch/openlab

Figures
01: The ALICE Detector
02: Geometry and Mesh of the ALICE Detector
03: Temperature Field in the L3 volume without Forced Ventilation
04: Temperature Field in the L3 volume with Forced Ventilation