2016-01-07 Chiara.Bracco@cern.ch LHC_MD292: TCDQ-TCT retraction and losses during asynchronous beam dump C. Bracco,R. Bruce and E. Quaranta CERN, Geneva, Switzerland Keywords: asynchronous dump, abort gap, losses. Abstract The protection provided by the TCDQs in case of asynchronous beam dump depends strongly on their correct setup. They have to respect the strict hierarchy of the full collimation system and shield the tertiary collimators in the experimental regions. This MD aimed at performing asynchronous beam dump tests with different configurations, in order to assess the minimum allowed retraction between TCTs and TCDQs and, as a consequence, on the * reach. This is an internal CERN publication and does not necessarily reflect the views of the CERN management.
Contents Introduction... 1 Settings and Procedures... 1 MD History and Measurements... 2 Preliminary Data Analysis... 2 First comparison with simulations... 7 Conclusions... Acknowledgments... References... 11 i
Introduction Asynchronous beam dumps in the LHC are particularly critical at 6.5 TeV due to the large stored energy (~350 MJ) and the potential risk of damaging machine components. A stopper (TCDQ) and a secondary like collimator (TCSP, with embedded BPMs) are installed at 90 phase advance from the extraction kickers (MKD). They are used to intercept miskicked beams and shield the arc aperture and the tungsten tertiary collimators (TCTs) which protect the triplets upstream of the high-luminosity experimental regions. HiRadMat tests and simulations proved that TCTs at 7 TeV can be damaged by the direct impact of a single pilot bunch (5e9 protons); their protection in case of an asynchronous beam dump is thus vital. The retraction between TCTs and TCDQs must be carefully evaluated and a safety margin for optics errors, orbit variations and setup accuracy has to be taken into account. Still, a too large margin affects the lowest achievable * and thus the LHC luminosity reach. The retraction between TCTs and TCDQs is minimum at the end of the squeeze and the present operational retraction in 2015, for a * of 80 cm in IP1 and IP5, is 4.6 (TCDQ/TCSP at 9.1 and TCTs in IR1 andir5 at 13.7 ). In order to evaluate the minimum allowed safe TCT aperture, it was proposed to perform several asynchronous beam dump tests, moving the TCTs in steps closer to the TCDQ aperture. The procedures and proposed settings are presented together with the MD history and some preliminary data analysis. Settings and Procedures The proposed TCT and TCDQ/TCSP settings planned to be used for the different asynchronous beam dump tests are summarised in Table 1. The rest of the collimators should be positioned at their standard physics settings. Table 1: TCT and TCDQ/TCSP settings (in units) to be used during the different tests) Test 1 [ ] Test 2 [ ] Test 3 [ ] Test 4 [ ] Test 5 [ ] TCDQ/TCSP 9.1 9.1 9.1 9.1 9.1 TCT IR1-IR5 12.7 11.7.7 9.7 8.7 In principle only the horizontal TCTs could be closed to the defined settings, since the sweep during the asynchronous beam dump occurs in the horizontal plane. Nevertheless, since there could be particles intercepted by the TCDQ and/or the TCSP (not perfect alignment between TCDQ and TCSP) and scattered in a symmetric 3D cone, also the vertical TCTs were closed. The asynchronous beam dumps have to be performed at 6.5 TeV and with squeezed optics at 80 cm. Both beams are used and one nominal bunch per beam is required in order to establish the nominal LHC reference orbit. A few (1 nominal plus 5 probes) additional bunches per beam could be added in the machine to allow for parallel measurements. One nominal bunch has to be in bucket 1 to minimize the needed time to fill the abort gap with the required intensity (from a few e9 up to 1e protons, to be monitored with the BSRA). To perform the test, several channels have to be masked: BLMs, RF, interlocked BPMs and collimators (position, energy and *). The approved MD procedure consists in establishing the reference orbit at the end of the squeeze and then reduce the intensity of the nominal bunch down to 5e protons by exciting the beam with the ADT and scraping it with the betatron collimators. The TCT position thresholds in IR1 and IR5 has to be relaxed to allow moving the jaws to the required new positions (movements across the limits are blocked even if the interlock is masked). The nominal orbit bump (1.2 mm away from the TCDQ) which 1
is used for asynchronous beam dump tests (in YASP) has to be applied. The abort gap cleaning has to be switched off. The RF cavities are then switched off for ~50-55 seconds or until the abort gap population reaches a few e9 protons. A beam dump is then triggered with the dump switch. MD History and Measurements The proposed MD is highly time consuming. The process from injection (including injection setup) until the end of the squeeze takes about 1.5 hours. An additional 0.5 hours can be counted to set up the collimators, perform all the needed checks, reduce the intensity of the nominal bunch with the ADT and finally dump the de-bunched beam. The time for the magnet ramp down (~0.5 hours) has to be included as well and, in total, each measurement takes ~2.5 hours. For this reason it was proposed to combine this MD with other parallel studies and/or to profit of other MDs for end of fill studies. Despite that, dedicated time was foreseen for MD292 already during the first Block (20/07 25/07) but, due to several problems with the machine, the global program was shortened and the MD postponed. It was proposed to perform two parasitic asynchronous dumps at the end of MD311 and MD346 respectively but, due to the lack of time, this was not possible. The MD was then rescheduled at the beginning of the 2 nd MD block (26/08 31/08). This time the MD was postponed to allow the completion of the VdM scans. A floating time slot of hours was assigned to perform MD292 in the night between October 8 th and 9 th. The MD started at 19:30 (beam injection) but the beam was dumped a few minutes after reaching the flattop due to an RQD.A78 FREE WHEEL DIODE fault. An access was needed to reset the error and the machine was ready for beam at ~2:40. A fault occurred at 2:59 during the BLM MCS checks and the beam was back at 4:50. Since the available time was sufficient to perform only a single test, it was decided to skip Test 1 and Test 2 (Table 1) and directly close the TCTs in IR1 and IR5 to.7 This setting guaranteed a reasonable safety margin to ensure the TCT protection. Moreover the intensity of the nominal bunches was reduced down even to 2e-3.5e protons (to stay compatible with the assumed damage limit where ejection of tungsten fragments occur) in bucket 1 (for Beam 2 and Beam 1) and to 5e protons in bucket 15000 before closing the tertiary collimators. The RF was switched off for ~52 seconds and the beam was dumped by the operator at 07:13:25. Preliminary Data Analysis The total abort gap population between the moment when the RF is switched off and the beam is dumped is shown in Fig.1 for Beam 1 and Beam 2. The leakage from the LHC dump protection system to the TCTs is defined as the ratio between the BLM signals at these elements. Past studies showed that a leakage lower than 1e-3 provides an adequate protection of the tertiary collimators [1]. The analysis of several asynchronous beam dumps, periodically performed to check the status of the system, allowed to define more realistic thresholds for each TCT and those are now implemented in the XPOC. As a preliminary analysis, the MD data were compared to those of an asynchronous beam dump performed with the same optics conditions and the TCTs at their nominal aperture (test performed on September 9 th at 06:42:02). Fig.2 shows the distribution of the particles at the moment of the dump for Beam 1 and Beam 2 respectively. The number of charges was calculated using the logged raw data in ADC counts (LHC.BSRA.US45.B1/B2:ABORT_GAP_SPILL in Timber) and multiplying them by the calibration factor at the dump (LHC.BSRA.US45.B1/B2:ABORT_GAP_CAL_FACT in timber, see Table 2). The XPOC values for the total number of charges at the moment of the dump is also shown in Table 2. 2
Figure 1: Abort gap population (instantaneous total number of charges) between RF off and beam dump for Beam 1for Beam 1 (red line) and Beam 2 (blue line). Table 2: Logged calibration factors used to convert the ADC counts in number of charges in the abort gap at the moment of the dump. Nominal MD LHC.BSRA.US45.B1:ABORT_GAP_CAL_FACT 2.9e7 1.4e6 LHC.BSRA.US45.B2:ABORT_GAP_CAL_FACT 2.8e6 9.1e5 XPOC B1 total number of charges in abort gap 4.4e9 1.1e9 XPOC B2 total number of charges in abort gap 1.8e9 7.2e8 The fraction of beam in the abort gap which is intercepted by the TCDQ/TCSP and by the TCDS during the dump is highlighted in Fig.2 and Fig.3. This is defined by calculating the needed MKD kick ( ) to touch the TCDQ/TCSP (~4.6 mm aperture + 1.2 mm orbit = 11.5 ) and TCDS (~16.3 mm) aperture using the formula: H 1 2 sin( ) For each element, H is the aperture, 1 and 2 the -functions at the MKD and at the element and the phase advance from the extraction kickers. Particles with an amplitude smaller than the TCDQ aperture escape and are dumped at the next turn. Particles with an amplitude larger than 40 mm are cleanly extracted. About 99% of the particles intercepted by the TCDQ are also absorbed by it. Particles grazing the TCDQ jaw surface or impacting the TCSP can be only scattered and a part of them can reach the TCTs. The aperture of the horizontal TCTs in IR1 and IR5 are also displayed in Fig. 2 and Fig. 3 for the MD settings (dashed green and black lines). All TCTs but the horizontal TCT in IR1 (TCTH1) for Beam 1 are still in the shadow of the TCDQ/TCSP thanks to the favourable phase advance. 3
Figure 2: Abort gap population at the moment of the MD and nominal dumps for Beam 1 (top) and Beam 2 (bottom). The MKD waveform is also shown and allows to define the range of mis-kicked particles which are intercepted by the TCDQ/TCSP and the TCDS. Two dashed lines indicating the TCTH in IR1 (green) and in IR5 (black) with MD settings are displayed. 4
The absolute losses at the TCDQ, TCSP and TCTs for the nominal and the MD dump are shown in Table 3. The losses during the MD are in general lower due to the smaller number of charges in the abort gap. Only in IR1 for Beam 1 losses appear above the noise level when closing the TCTs to.7. This is expected since this collimator is now slightly closer to the miskicked beam than the TCDQ (see Fig. 1). The ratio between Nominal and MD losses at the TCDQ and TCSP (3.7 for Beam 1 and 2.6 for Beam 2) is almost exactly the ratio between the abort gap populations (4 for Beam 1 and 2.5 for Beam 2). This is expected and confirmed by the abort gap distribution when normalising the nominal dump by this multiplication factor (see Fig. 3). The population profiles overlap almost perfectly in the TCDQ/TCSP region for Beam 2 while a discrepancy is observed or Beam 1 (~% more charges for nominal dump). Table 3: Absolute losses at the TCDQ, TCSP and TCTs for nominal and MD TCT settings. (BLM running sum 1, multiplication factor for electronic filter included). Only values above noise level are shown. Nominal [Gy/s] MD [Gy/s] BLMTL.TCSP.A4R6.B1 1.25e3 3.13e2 BLMTL.TCDQA.A4R6.B1 3.19e3 9.58e2 BLMTI.TCTPH.4L8.B1 6.18e-2 3.16e-2 BLMTI.TCTPV.4L8.B1 2.28e-2 9.14e-3 BLMTI.TCTPH.4L1.B1 8.24e-2 BLMTI.TCTPV.4L1.B1 5.07e-2 BLMTI.TCTPH.4L5.B1 3.62e-4 1.99e-3 BLMTI.TCTPV.4L5.B1 5.43e-4 1.27e-3 BLMTL.TCSP.A4L6.B2 7.16e2 2.48e2 BLMTL.TCDQA.A4L6.B2 2.57e3 1.07e3 BLMTI.TCTPH.4R5.B2 7.50e-1 3.36e-1 BLMTI.TCTPH.4R1.B2 1.47e-2 7.96e-3 BLMTI.TCTPV.4R1.B2 6.70e-3 2.99e-3 As a last check, the ratios between losses at the TCT and at the TCSP/TCDQ, for nominal and MD TCT settings, were analysed and are shown in Table 4. For Beam 1 the loss ratios are almost doubled in IR8, increase by two orders of magnitude in IR1 and are a factor ~-20 higher at the TCTH and TCTV in IR5. For Beam 2 a maximum increase of 60% is recorded at the TCTH in IR1 while losses change by only ~20% at the other tertiary collimators. The 1e-3 limit is respected everywhere but at the TCTPH.4R5.B2; still all the ratios are below the newly defined XPOC thresholds. This preliminary analysis confirms that the TCTs were still adequately protected by the TCDQ/TCSP when closed at.7 as expected according to previous calculations 5
Figure 3: Abort gap population at the moment of the MD and nominal dumps for Beam 1 (Top) and Beam 2 (Bottom). The population for the Nominal dump is scaled with the number of charges for a direct comparison with the MD dump. 6
Table 4: Ratio between losses at the TCT and at the TCSP/TCDQ, for nominal and MD TCT settings. The increase factor when closing the TCTs to.7 is also shown. Beam 1 TCT/TCSP Nominal TCT/TCSP MD Increase factor TCT/TCDQ Nominal TCT/TCDQ MD Increase factor BLMTI.TCTPH.4L8.B1 4.93E-05 1.01E-04 2.1 1.94E-05 3.30E-05 1.7 BLMTI.TCTPV.4L8.B1 1.82E-05 2.92E-05 1.6 7.15E-06 9.55E-06 1.3 BLMTI.TCTPH.4L1.B1 2.63E-04 8.62E-05 BLMTI.TCTPV.4L1.B1 1.62E-04 5.30E-05 BLMTI.TCTPH.4L5.B1 2.88E-07 6.35E-06 22.1 1.13E-07 2.08E-06 18.4 BLMTI.TCTPV.4L5.B1 4.33E-07 4.04E-06 9.3 1.70E-07 1.32E-06 7.8 Beam 2 TCT/TCSP Nominal TCT/TCSP MD Increase factor TCT/TCDQ Nominal TCT/TCDQ MD Increase factor BLMTI.TCTPH.4R5.B2 1.05E-03 1.35E-03 1.3 2.91E-04 3.15E-04 1.1 BLMTI.TCTPV.4R5.B2 2.37E-04 2.75E-04 1.2 6.60E-05 6.40E-05 0.97 BLMTI.TCTPH.4R1.B2 2.05E-05 3.20E-05 1.6 5.71E-06 7.46E-06 1.3 First comparison with simulations The simulations of asynchronous beam dump were performed using a special setup of the SixTrack code [2]. SixTrack is a multiturn six-dimensional symplectic tracking code optimized to track single particles in high-energy rings. The code was later extended with a new built-in Monte Carlo routine [3] in order to simulate the particle-matter interaction inside collimators and machine elements hit by escaping particles. This version of the software became the standard tool for collimation studies at CERN. The reliability of the SixTrack simulation results has been confirmed by the very good agreement with loss measurements using the LHC beam, as shown in [4]. In order to study fast failure scenarios in the LHC, SixTrack was adapted to simulate the simultaneous misfiring of all the beam dump kicker magnets in the extraction line [5]. A train of 6.5 TeV protons with 25 ns spacing between consecutive bunches was simulated for both beams. Each bunch receives a different kick from each MKD according to the estimated rise of the magnetic field and the retriggering time of adjacent kickers using the nominal MKD waveform [6]. Only bunches belonging to the dangerous windows of kicker angles (grey band in Figure 4), i.e. those bunches with higher probability to be miskicked and driven towards the TCTs, were taken into account. 7
charges 1 9 8 8 6 8 4 8 2 8 B1 B2 0 0 50 0 150 200 kick Σ Figure 4: Abort gap population, measured by the BSRA during the asynchronous dump test for both beams, given as a function of total kick in summed over all MKDs. The gray band indicates the region where particles could end up at the TCTs. A Gaussian bunch profile of macro-particles is tracked for 3 turns: the particles pass the MKDs at the second turn and they receive an intermediate kick, which sweep them across the machine aperture. At the third turn, the MKDs have reached their full field and all remaining particles are extracted. The full LHC collimation system was deployed in the simulations: the TCTs in IP1 and IP5 were tightened to.7, as it was done during the MD, and the presence of standard bumps for asynchronous dump in IP6 was taken into account by retracting the TCDQs and TCSP settings of additional 2.4. The other collimators were set at the standard 2015 configuration [7]. Figure 5 shows the particle loss distribution in the ring as simulated by SixTrack. As expected, the main loss location is in IP6, but high loss peaks also stick out from the betatron cleaning collimators in IP7 as well as the TCTs in IP1 and IP5. Figure 5: Simulated loss map distribution around the ring. The losses at collimators and magnets were summed over all bunches considered in the SixTrack simulation. Note that, to better compare these results with the measurements, the number of particles lost at the TCDQ was increased manually by a factor 7: this takes into account the contribution of the bunches not simulated, but which however contribute to the BLM signals. We can reasonably assume they are all intercepted by the TCDQs. 8
The signals recorded by the BLMs when the beam dump was triggered during the MD are collected in Figure 6. A quantitative comparison between the simulated and measured loss maps cannot be done using only Figures 5 and 6, because SixTrack does not account for the shower of particles generated by the interaction of the primary beams with the matter, which cause the actual BLM signals. However, an overall good qualitative agreement can be stated: the level of normalized losses foreseen by simulations in IP6 and IP7 is very well reproduced by the measurement as well as the higher contribution to the losses in IP7 due to Beam 2 (right side of the insertion) with respect to Beam 1. The same trend is visible in IP3, although the discrepancy between simulations and measurements is higher than the other IPs. Losses lower of a factor 3 to 4 in Figure 5 for IP1, IP5 and IP8 are a good achievement if we consider that the shower is not accounted for in SixTrack. The only major anomaly is the blue spike in Figure 6 from BLMBI.11R2.BOT20.MBB.LECL sticking out from IR2, which can probably be considered as noise: a high signal from the same BLM is indeed recorded in a time interval close to the moment of the dump also when there were not other losses in the machine. Local cleaning inefficiency 1-1 -2-3 -4 Collimator Warm Cold Roman Pot 2015--9 07:13:25-5 -6-7 0 5000 000 15000 20000 25000 s [m] Figure 6: Measured raw loss pattern around the ring when the asynchronous dump test was triggered. An integration time of 1.3 s was used for the BLM signals, which have been then normalized to the highest signal. A more detailed comparison of the TCT losses in measurement and simulations losses has also been performed. As shown in Figure 7, the fraction of the total abort gap population impacting on the horizontal tertiary collimators in IP1 and IP5 measured during the asynchronous dump test is compared with SixTrack simulation results. Please note that, in order to have a meaningful comparison, the measured TCT BLM signals were converted to the estimated number of protons impacting the TCT in the same way as in Ref. [8]. It should be noted that the simulated bunches were all equally populated, while the abort gap population was not fully homogenous during the measurement (Figure 4). Therefore, each simulated bunch was normalized to the measured population profile of the abort gap over the corresponding 25 ns interval and in the end the losses at the TCTs were summed over all bunches. 9
Figure 7: Fraction of the total abort gap population impacting on the TCTs in IR1 and IR5 during the asynchronous dump test. The BLM signals, converted to protons using the conversion factor from Ref. [8], are compared with SixTrack results, which have been normalized by the measured abort gap profile of Figure 4. By looking at the ratio for each TCTs, we can say that the overall accuracy of the simulations with respect to the measurements is fairly good: At the two TCTs with highest losses, the measurements agree with the measurements within a factor 3. At TCTPH.4L5.B1 the discrepancy is a factor 6, while a larger difference of a factor 20 is found for TCTH.4R1.B2. A similar result was already reported during the recent MD performed with ß*=40 cm (see Ref. 8) but the reason is not fully understood yet. Conclusions A preliminary analysis of the data for the asynchronous beam dump performed with the TCTs in IR1 and IR 5 at.7 (instead of the nominal 13.7 ) is presented. All TCTs, but TCTH in IR1, were still in the shadow of the dump protection collimators thanks to the favourable phase advance with respect to the MKD kickers. Still the leakage level was everywhere below the XPOC thresholds. Based on a comparative study, an overall good agreement between the beam loss measurements during the MD and the prediction from simulations can be found, especially if we consider that the contribution to the BLM signal due to secondary particles is not taken into account by SixTrack simulations. A more detailed analysis of the loss ratio for selected TCTs shows reasonable results, however large discrepancies are found for TCTH.4R1.B2 as in Ref. [8]. Further studies are required to evaluate whether more measurements are needed and to conclude on the final range of allowed TCT settings. Acknowledgments Special thanks to: ABT and MPP colleagues, OP and MD crew, collimation team and in particular S. Redaelli and B. M. Salvachua Ferrando.
References [1] C. Bracco et al., Leakage from LHC Dump Protection System, Proceedings of HB20 Workshop, Morschach, Switzerlan, 20. [2] F. Schmidt. SixTrack. User's Reference Manual. CERN/SL/94-56-AP, 1994. [3] G. Robert-Demolaize et al., A new version of SixTrack with collimation and aperture interface. Proc. of the Particle Accelerator Conf. 2005, Knoxville, page 4084, 2005. [4] R. Bruce et al., Simulations and measurements of beam loss patterns at the CERN Large Hadron Collider. Phys. Rev. ST Accel. Beams 17, 0804 2014. [5] R. Bruce et al., Calculations of safe collimation settings and ß* ate the CERN Large Hadron Collider. Phys. Rev. ST Accel. Beams 18, 0601 2015. [6] B. Goddard, private communication [7] R. Bruce et al., Baseline LHC machine parameters and configuration of the 2015 proton run, Proceedings of the LHC Performance Workshop, Chamonix, France (2014) [8] R. Bruce et al., Collimation with tighter TCTs at ß*=40 cm, CERN-ATS-Note, 2015 11