Simulation of QUELL

by C.Marinucci (EPFL/CRPP) & CryoSoft

The QUELL Experiment

QUELL (QUench Experiment on Long Length) is a thermohydraulic experiment specifically designed to produce extrapolation and validation data on a scaled-down version of the CICC with cooling hole for ITER (International Thermonuclear Experimental Reactor). The experiment has been described abundantly in literature (for a general description see for instance : A. Anghel, QUELL Experiment: Analysis and Interpretation of the Quench Propagation Results, Cryogenics, 38, 5, pp. 459-466, 1998).
[quell conductor] The photograph shows the cross section of the QUELL conductor The conductor has an outer diameter of 19.4 mm and a critical current of 32 kA at 4.5 K and 12 T. The central cooling hole has an inner diameter of 6 mm.

The QUELL cable is a 1:5 scaled down version of the ITER-Central Solenoid Cable.

The QUELL sample is approximately 100 m long, and is highly instrumented with temperature and voltage sensors, as well as inductive and resistive heaters. The sketch below gives the location of the heaters (in red) and of the main temperature sensors (in blue) along the sample length. The temperature sensors were glued on the Ti-alloy jacket. Helium flow in the experiments was from terminal J into the inner layer, the heater sections, the outer layer and out of terminal K

[quell sample]

During the test period several type of thermo-hydraulic transients were induced and followed in great detail. The transients spanned the whole range of operation of a superconducting coil, from slow pulsed heating and subsequent recooling, to fast stability transients followed either by recovery or by a quench. The experimental results produced during the QUELL experiment is as of today the most complete and wide ranging calibration data-base available for thermo-hydraulic analysis codes, and has been intensively used to validate Gandalf (for the results on quench propagation see: C. Marinucci, L. Bottura, G. Vecsey, R. Zanino, The QUELL Experiment as a Validation Tool for the Numerical Code Gandalf, Cryogenics, 38, 5, pp. 467-477, 1998). Here we concentrate on slow heating transients (heat slug propagation tests) that proved to be among the most difficult to reproduce by simulation. We show in particular how the complete two-channels model implemented in Gandalf 2.0 is capable of reproducing with good accuracy most experimental features.

Model in Gandalf 2.0

Gandalf was originally developed with the ITER application in mind as its primary objective. This required a model of a CICC with at least two helium flow channels.
The cable model as used in Gandalf up to version 1.8 had independent flow conditions (velocity), but was limited to identical thermodynamic state in the two helium channels (i.e. the same temperature and pressure).

The number of degrees-of-freedom per node in the finite element of Gandalf 1.8 was therefore 6 (cable and jacket temperature, hole and bundle velocity, helium temperature and pressure)

[gandalf1.8]
It was clear during the interpretation of the QUELL data that this model is adequate for engineering predictions of quench propagation, but it is not sufficient to describe accurately the slow temperature transients typical of pulsed heating and re-cooling phases in a CICC with cooling hole. We have therefore developed the code upgrading the model to a full description of the two channels.
[gandalf2.0] The model as upgraded in Gandalf 2.0 takes into account for two independent cooling channels in the conductor, with distinct helium flow rate, pressure and temperature.

The number of degrees-of-freedom per node in the finite element has increased correspondingly to 8 in Gandalf 2.0 (the helium pressure and temperature is treated separately for hole and bundle).

In addition we have implemented appropriate correlations for the heat transfer between the two channels (from: A. Long, Transverse Heat Transfer in a Cable-in-Conduit Conductor with Central Cooling Channel, Master Thesis, June 1995, MIT) that are used to achieve the thermal coupling among the flows. It is this upgrade that has formed the basis for Gandalf version 2.0.

Results

As we already remarked, the slow propagation of heat slugs in the QUELL cable turned out to be one of the most difficult experiments to interpret using numerical simulation. A heat slug experiment consisted in:
  • establishing a steady state condition;
  • pulsing one of the heaters with a pre-set energy;
  • recording (among others) the temperature evolution at the thermometers.
[run12] The measured temperature increase at termometers downstream of the heater (crosses) was largely overestimated by Gandalf 1.8 (dashed lines). On the other hand the augmented model of Gandalf 2.0 (solid lines) demonstrates a much better agreement with the measured data.

The reason for this significant improvement is in the separate treatment of the two independent helium streams: a part of the helium flows in the central hole with low hydraulic impedance and large velocity, while a part flows in the intersticial space among the strands with high hydraulic impedance and necessarily low velocity.

The differential of temperature and velocity between the two flows causes a spread of the original heated slug that thus grows in length and lowers in amplitude, as demonstrated by the measurement and the simulation with Gandalf 2.0.

Simulations of heat slug propagation were performed for several runs at different input energy and length (different heaters). Below we report a summary of the results obtained. To ease the comparison we have taken the maximum temperature increase observed at selected thermometers (crosses), and we compare this to the results of simulations performed with Gandalf 1.8 (dashed line) and Gandalf 2.0 (solid line). The results confirm that the full two-channels model presently available in Gandalf 2.0 is indeed a significant improvement with respect to the previous approximation.

[TA5] [TA6]

[TA7] [TA8]

back to Gandalf by CryoSoft, © 2013