Automate tests for gas leak on RPC detectors of the ATLAS experiment

Author(s):
R. De Asmundis - ISTITUTO NAZIONALE DI FISICA NUCLEARE

Industry:
University/Education

Products:
LabVIEW, Data Acquisition

The Challenge:
Perform automatic gas leak test at the atmospheric pressure range on RPC Detectors installed in the ATLAS Experiment. Detectors are in number of about 1200 units, grouped in 12-24 units on a single gas supply line: this generates 128 gas supply lines to be tested, each of them being 90 to 120 liters as testing volume.

The Solution:
A PID controller has been implemented to stabilize the testing pressure to two moderate values (standard testing pressures are of 5 and 2 hPa), by setting an appropriate (and tipically small) gas flow. Once the Process Variable (the pressure) is stable, in general after few minutes, the flow required to stabilize it is measured and integrated in the time to get a direct measure of the gas loss.

Short Summary

Although classically the test is performed by putting a moderate pressure into the chambers (about 5 to 6 hPa) and looking at the pressure decay in the time, here a different approach has been adopted: a low flow, high performance mass flow controller (MFC) is used to send a Nitrogen flow to the testing volume in such a way that the testing pressure is kept at the nominal values during test time. The MFC is controlled via an applied voltage in the 0-5 V range, sent through the NI DAQCard-6024E Analogue Output Channel, and its value is calculated by the PID process where the Process Variable is the pressure. At this point a flow is established to keep conditions stable and such a required flow is integrated in the time. Finally the ratio between the integrated flow and the total integration time is calculated, obtaining a measure of the leak flow directly in n-ml/s (normal-millilitres per second) or, if needed, normalised to the total volume (n-ml/(s * l)). Note that the MFC is able to measure the real flow independently from the setting point sent to it by the control VI: this ensures that the integration is performed on the effective gas flowing into the line and not on the theoretical one.

Article

The Resistive Plate Chamber, usually called RPC, [ ] is an interesting ionization detector for Particle Physics characterized by an high detection efficiency (97-98 %) a very high time resolution (1.5-2 ns), a good spatial resolution (~ 1 cm), a good reliability proved in previous applications[ ]. All these characteristics, an industrial-supported production and low cost [ ], make the RPC an ideal instrument for fast response application like Muon Trigger in big detectors or wide Cosmic Rays detectors arrays (see also [ , ]). RPC are currently under installation as Muon trigger detectors for the barrel region of the Muon Spectrometer in the ATLAS (“A Toroidal LHC Apparatus”) experiment at LHC Accelerator (Large Hadron Collider) at CERN. See the Figure 1 for a particular of the ATLAS installation: almost 1200 RPC modules are used, covering a total sensitive surface of more than 7000 square meters. The ATLAS Experiment [ ], one of the four Collaborations at LHC, is a very complex detection system, dedicated to new discoveries in proton-proton collision physics. Beside annexed electronics and services, it consists of four major sub detectors: 1) an inner detector (semiconductor pixels and strips detectors plus 2) straw tubes as a sensitive part of a Transition Radiation Detector) gives tracks for charged particles; 3) calorimeters follow to get the total energy for both hadrons and electromagnetic radiation; 4) the outer part is a precision muon spectrometer, which measures muons momentum by observing track deviation in a toroidal magnetic field.

Since RPC are gaseous detector in which the quality of the gas mixture together with the engineering of the gas supplying system play an essential role in the correct functioning of them, a great care must be devoted to any possible gas leak that the detectors can present: in fact any gas leak is always accompanied by an equivalent in air that comes into the detector, making the gas mixture (Tetrafluorethane, Isobuthane and Sulfur Exafluoride) polluted by air and causing unwanted dispersions of gas in the experimental area. Even if the mixture is not toxic nor inflammable, dispersions means higher managing costs due to the needs of producing and distributing more mixture. Since a certain level of gas leak is unavoidable, the only solution is, once all actions to limit the leaks have been carried on, to measure the gas leak level with a reference gas (in our case Nitrogen).

The Testing System implemented
The testing system consists in a PID process in which the Process Variable is the chamber pressure read from the incoming gas line, and the Output Variable is the Nitrogen flow setting point. See the Figure 2 for a complete scheme of the leak test measurement system. In the Figure 3 you can see the main panel of the VI implemented. The Human Interface consists of a part (on the left) dedicated to the control of the run: specifications for the sector under test and on the duration of the run can be indicated here. The user can also indicate if the specified time for the run is to be intended as a “total time” or “integration” time: in the second case, the elapsing of the time is calculated only if the PID indicates a stable condition and the flow integration is in progress. This solution has been preferred since it allows a good standardization on the “effective” (i.e. integration time) test time.
The right side of the Instrument consists of a two page Tab Control containing “PID” and “Field & Integration” controls and indicators, both shown in the Figure 3. A detailed description follows:

The PID tab contains all information concerning the strict Control Process: user can manage the Set Point pressure (here at 5 hPa) of the process and the three PID gains. To be noted that the PID controlling VI chosen is the one containing the automatic tuning process subVI, so that an “on field” tuning of the PID gains parameters can be performed. Indicators of the Output Variable values are also present here: the main value produced by the PID control is in ml/min of Nitrogen flow, then converted into the physical value to be sent to the DAC channel (a voltage in the 0 to 5 V range) and the equivalent in percent of full scale, useful for a quick understanding of the working point of the MFC valve.
The behaviour of the PID controller is visible by observing the time evolution of the parameters as shown in the picture.
- The “Field & Integration” tab contains real time information on the data acquired from the field: the testing pressure (i.e. the Process Variable) and the actual flow measured by the MFC, the latter reported in liters per hour too, a preferred unit in the case of very low flow. A waveform chart reporting the time evolution of the two parameters and, in the third area, the “stability flag”: this flag indicates when the PID is stable enough and the integration process can proceed. The tab contains also the flow subject to the integration, that is basically a subset of the actual MFC flow array, extracted when the “stability flag” is TRUE. Note that the timescale reported on the integration chart is the effective relative time scale accumulated during the integration time: its maximum value is considered as the total integration time and is used for comparison to stop the entire VI if required by the user (see the RUN control, “Run End” switch setting).
The results of the integration is indicated in the upper right Cluster in which the total flow (in ml) is calculated together with the integration time and their ratio (giving the average gas loss of the whole testing volume in ml/s). Finally, aritmetic division by the testing volume (in this case 110 liters) is performed in order to obtain the specific gas loss (the mean exchange between gas mixture and air to which every liter of gas contained into the detector undergoes). Typical total integration time used are of half an hour to one hour (in the Figure few seconds are reported for simplicity) gining a total testing time of 35-40 to 70-80 minutes for each gas line.

Saving data
Data are filed in two different ways: a new binary Stream File is generated for each line under test (one of the 128 gas channels), containing the “field” results (so the actual pressure and flow value measured at the MFC) and the “stability flag” as a binary number (0-1) recorder at a time step of 0.5 seconds. These stream files are useful for future references to look at the evolution of the test and to compare the different behaviors on the various lines.
Secondarily a single LabVIEW Log file is updated at the end of the test, containing only the results of the integration and obviousely the test date/time and the name of the Sector under test (supplied by the user in the Run control part of the VI).

Analyzing data
Some rapid Viewing VI have been implemented to quickly look at the data: a reader for the LOG file to get information on the final results obtained online by the test; a 3-D viewer showing the evolution of the testing pressure and of the gas flow and a converter to LabVIEW Measurement file (binary format) in order to be compatible with DIAdem software. The first two VIs are shown in the Figure 4.
We intend to further develop these VIs in order to deeply automate the analysis process: once the stream files are converted into .lvm format, script and reports must be prepared in NI DIAdem to accommodate and present data. We plan to develop a converting VI from LabVIEW LOG file to MS Access database using the NI LabVIEW Database Connectivity Toolkit.

Browse All Case Studies »