Main dipole circuit simulations Behavior and performance analysis PSpice models Simulation results Comparison with QPS data Ongoing activities Emmanuele Ravaioli LHC-CM
Main dipole circuit simulations Emmanuele Ravaioli LHC-CM Main dipole circuit Components Circuit behavior PSpice model Main parameters Results Means for reducing voltage oscillations Quench Protection System Conclusions and further work 2
LHC main dipole circuit Emmanuele Ravaioli LHC-CM Power converterFilterSwitch1 Switch2 77 Magnets Crowbar77 Magnets
Main dipole circuit – Charging of the circuit Emmanuele Ravaioli LHC-CM A variation of the voltage across the capacitors of the filter causes an oscillation to occur. The frequency of the oscillations depends on the inductance and capacitance of the filter, L_filter and C_filter. The damping of the oscillations depends on the resistance of the filter R_filter.
Main dipole circuit – Switch-off of the power converter Emmanuele Ravaioli LHC-CM
Main dipole circuit – Fast Power Abort (Switch opening) Emmanuele Ravaioli LHC-CM
Main dipole circuit – Distinct voltage transients Emmanuele Ravaioli LHC-CM Voltage waves due to the filter ringing They occur every time the voltage across the capacitance of the filter changes: strong effect when the power converter is shutting down; weak effect when the thyrirstors of the crowbar are already conducting. Their frequency depends on the inductance and capacitance of the filter, L_filter and C_filter. Their damping depends on the resistance of the filter R_filter. 2.Voltage waves due to the switch opening They occur when the switches are opened, due to the sudden change of the voltage across the switches; the magnet string behaves as a lumped transmission line. Their frequency depends on the magnet inductance L_magnet and on the capacitance to ground C_ground. Their damping depends on the characteristics of the magnet chain.
Simulated circuit – Complete model Emmanuele Ravaioli LHC-CM Power converter Filter Switch1 Switch2 Crowbar Earthing point 77 Magnets
Simulated circuit - Power converter with output filter Emmanuele RavaioliLHC-CM Power Converter + 2 Thyristors Grounding point Filter Capacitors PC composed of two parallel units 6x Crowbars to allow by-pass of the PC at the shut-down (Thyristor model needed) Filter at the output of the PC PC grounded in the positive and negative branches through capacitors Grounding point Filter Inductors Power Converter + 2 Thyristors 6x Crowbars with Thyristors
Simulated circuit – Old dipole model Emmanuele RavaioliLHC-CM From Methods and results of modeling and transmission-line calculations of the superconducting dipole chains of CERN’s LHC collider, F. Bourgeois and K. Dahlerup-Petersen
Simulated circuit – New dipole model Emmanuele RavaioliLHC-CM Model of an aperture (refined for particular dipoles) Standard parameters F_bypass = 0.75 R_bypass = 10 Model of a magnet 19 components 7 components: 1 hour 20 minutes of simulation time Physically explainable by the effects of the eddy currents The distribution of unbalanced dipoles in each sector can be simulated assigning a different value to the R_bypass parameter ( and eventually f_bypass2 and R_bypass2 ) of each magnet
Simulated circuit – Switch model Emmanuele RavaioliLHC-CM Each switch is modeled by four switches in series to model the different phases of the switch opening.
PSpice simulation – Main parameters Emmanuele Ravaioli LHC-CM Number of dipoles154 Inductance Lmag of each magnet98 mH Capacitance to ground Cg of each magnet300 nF Parallel resistance R// of each magnet100 Ohm Capacitance C of the power-converter filter110 mF Inductance L of the power-converter filter284 uH Resistors R in the filter branches (8x in parallel)27 mOhm Resistance R_EE of the extraction resistor147 mOhm
Main dipole circuit simulations Emmanuele Ravaioli LHC-CM Main dipole circuit Results I_max=6 kA; dI/dt=10 A/s; No switch opening I_max=6 kA; dI/dt=10 A/s; Delay_s1=0 ms; Delay_s2=0 ms I_max=6 kA; dI/dt=0 A/s; Delay_s1=0 ms; Delay_s2=0 ms I_max=6 kA; dI/dt=10 A/s; Delay_s1=400 ms; Delay_s2=400 ms I_max=6 kA; dI/dt=10 A/s; Delay_s1=400 ms; Delay_s2=560 ms New model of a dipole aperture Means for reducing voltage oscillations Quench Protection System Conclusions and further work 14
I_max = 6 kA; dI/dt = 10 A/s; No switch opening Magnet 001 Blue Magnet 154 Red Max oscillation ≈ 9 V Min voltage ≈ -5 V Simulation results – Typical configuration Emmanuele Ravaioli LHC-CM
Magnet 001 Blue Magnet 154 Red I_max = 6 kA; dI/dt = 0 A/s; Delay_s1 = 0 ms; Delay_s2 = 0 ms Max oscillation ≈ 9 V Min voltage ≈ V Simulation results – Typical configuration Emmanuele Ravaioli LHC-CM
Magnet 001 Blue Magnet 154 Red I_max = 6 kA; dI/dt = 10 A/s; Delay_s1 = 400 ms; Delay_s2 = 400 ms Max oscillation ≈ 9 V Min voltage ≈ V Simulation results – Typical configuration Emmanuele Ravaioli LHC-CM
Magnet 001 Blue Magnet 154 Red I_max = 6 kA; dI/dt = 10 A/s; Delay_s1 = 350 ms; Delay_s2 = 600 ms Simulation results – Typical configuration Emmanuele Ravaioli LHC-CM Max oscillation ≈ 9 V Min voltage ≈ V
Magnet 001 Blue Magnet 154 Red I_max = 6 kA; dI/dt = 10 A/s; Delay_s1 = 350 ms; Delay_s2 = 600 ms Max oscillation ≈ 9 V Min voltage ≈ V Simulation results – Typical configuration – New model Emmanuele Ravaioli LHC-CM
I_max = 2 kA; dI/dt = 10 A/s; Delay_s1 = 350 ms; Delay_s2 = 600 ms Simulation results – nQPS signals – Comparison Emmanuele Ravaioli LHC-CM nQPS MeasurementSimulation
I_max = 6 kA; dI/dt = 10 A/s; Delay_s1 = 350 ms; Delay_s2 = 600 ms Emmanuele Ravaioli LHC-CM Voltage waves along the magnet chain - Animation
Main dipole circuit simulations Emmanuele Ravaioli LHC-CM Main dipole circuit Results Means for reducing voltage oscillations Different switch opening delays Snubber capacitors (13.3 mF) across each switch Additional resistors (27mOhm 81mOhm) in the PC filter branches Inversion between the filter and the thyristor branches Quench Protection System Conclusions and further work 22
Magnet 001 Blue Magnet 154 Red I_max = 6 kA; dI/dt = 10 A/s; Delay_s1 = 350 ms; Delay_s2 = 600 ms Max oscillation ≈ 9 V Min voltage ≈ -15 V Simulation results – Snubber capacitors Emmanuele Ravaioli LHC-CM
Magnet 001 Blue Magnet 154 Red I_max = 6 kA; dI/dt = 10 A/s; Delay_s1 = 350 ms; Delay_s2 = 600 ms Max oscillation ≈ 7.5 V Min voltage ≈ -950 V Simulation results – Additional resistors in the PC filter Emmanuele Ravaioli LHC-CM
Magnet 001 Blue Magnet 154 Red I_max = 6 kA; dI/dt = 10 A/s; Delay_s1 = 350 ms; Delay_s2 = 600 ms Max oscillation ≈ 3V 17V Min voltage ≈ -850 V Simulation results – Inversion between filter & thyristors Emmanuele Ravaioli LHC-CM
Main dipole circuit simulations Emmanuele Ravaioli LHC-CM Main dipole circuit Results Means for reducing voltage oscillations Quench Protection System nQPS oQPS Conclusions and further work 26
Magnet 001 Blue Magnet 154 Red I_max = 2 kA; dI/dt = 10 A/s; Delay_s1 = 350 ms; Delay_s2 = 600 ms Simulation results – nQPS signals Emmanuele Ravaioli LHC-CM
Magnet 001 Blue Magnet 154 Red I_max = 2 kA; dI/dt = 10 A/s; Delay_s1 = 350 ms; Delay_s2 = 600 ms Simulation results – oQPS signals – All balanced dipoles Emmanuele Ravaioli LHC-CM
Magnet 001 Blue Magnet 154 Red I_max = 2 kA; dI/dt = 10 A/s; Delay_s1 = 350 ms; Delay_s2 = 600 ms Simulation results – oQPS signals – Unbalanced dipoles Emmanuele Ravaioli LHC-CM
Magnet 001 Blue Magnet 154 Red I_max = 2 kA; dI/dt = 10 A/s; Delay_s1 = 350 ms; Delay_s2 = 600 ms Simulation results – oQPS signals – Comparison Emmanuele Ravaioli LHC-CM Magnet 001 Blue Magnet 154 Red QSO MeasurementSimulation
I_max = 6 kA; dI/dt = 10 A/s; Delay_s1 = 350 ms; Delay_s2 = 600 ms Simulation results – oQPS signals – Outlier dipole Emmanuele Ravaioli LHC-CM
I_max = 2 kA; dI/dt = 10 A/s; Delay_s1 = 350 ms; Delay_s2 = 600 ms Emmanuele Ravaioli LHC-CM nQPS and oQPS Simulations - Animation
Main dipole circuit simulations Emmanuele Ravaioli LHC-CM Main dipole circuit Results Means for reducing voltage oscillations Quench Protection System Conclusions and ongoing activities 33
The analysis of the voltage transients in the RB circuit after the switch-off of the power converter and during a fast power abort (power converter switch-off + switch opening) has been carried out by means of a complete PSpice model. The model comprises the power converter and its filter, the dipole chain and its capacitance to ground, the switches and extraction resistors, the paths to ground. A new model of a dipole aperture has been presented: the model is simpler than the previous one but more accurate in predicting the behavior of the circuit. The behavior of the unbalanced dipoles, which are oversensitive to any voltage transient, has been successfully reproduced by assigning a different value to one parameter each aperture model, based on the real behavior observed by the QPS. A slightly more refined model of an aperture has been developed in order to simulate the behavior of the so-called outlier dipoles, whose apertures undergo a strange transient after the opening of the switches. The simulation results are in very good agreement with the data measured by the nQPS (magnet across each dipole) and by the oQPS (voltage difference between the two apertures of each dipole). Emmanuele Ravaioli LHC-CM Conclusions-1
Simulations with different delay of the two switch openings have been performed; in particular, the adopted delay of 400 ms and 560 ms has been investigated in order to assess the advantages of this solution. The analysis of the circuit highlighted two different kinds of voltage transients occur after a FPA, caused by different phenomena and characterized by different frequency, maximum value and damping. Oscillations due to power converter switch-off: They happen due to the ringing of the PC filter, thus their frequency is determined by the filter parameters. Oscillations due to switch opening: They present a much larger peak value (up to ≈1000 V), but since the current decays faster they are damped more quickly; their frequency depends mostly on the characteristics of the magnet chain (inductance and capacitance to ground of the apertures). Emmanuele Ravaioli LHC-CM Conclusions-2
A set of simulations has been conducted in order to study the proposed (and partly implemented) modifications to the circuit: snubber capacitors across the switches of the extraction system; additional resistors in the PC filter branches; inversion between the PC filter and the thyristor branches. Delay of the switch openings: The simulations show that delaying the switch opening with respect to the power converter switch-off effectively separates the events, and decreases the voltage differences between electrically-close magnets. Snubber capacitors across the switches of the extraction system: With this configuration, the maximum voltage observed across the magnets decreases dramatically (≈1000 V ≈15 V). Additional resistors in the PC filter branches: This modification leads to a quicker damping of the voltage waves, and to a decrease of the oscillation maximum amplitude of about 20%. Inversion between the PC filter and the thyristor branches: This modification significantly decreases the voltage oscillations due to the power-converter ringing; nevertheless, it does not influence the ringing due to the switch opening, which remains the same with respect to the maximum peak and to the damping. Emmanuele Ravaioli LHC-CM Conclusions-3
Aperture model: Understanding the cause of the unbalanced behavior of a number of dipoles (hypothesis: eddy currents). A set of tests is foreseen in SM18 in order to obtain information about the frequency transfer function of a few dipoles at different current levels and to verify the initial hypothesis. Switch model: Refining is required, in particular for smoothing the extremely sharp rise of the switch resistance during the last phase of the opening. Power converter model: Understanding the reasons why the measured voltage across the PC oscillates at a frequency smaller than the nominal one (28.5 Hz instead of 31.8 Hz) and damps faster. The present model has been corrected according to the measured data. Quadrupole circuit: Comparing the results of the performed simulations with measured data. Emmanuele Ravaioli LHC-CM Ongoing activities
Emmanuele Ravaioli LHC-CM
Annex 39 Emmanuele Ravaioli LHC-CM
Unbalanced dipoles – Measured data (QSO signal) 40 Emmanuele Ravaioli LHC-CM Same event : FPA at 2 10 A/s (S67 20/05/ ) ONLY BALANCED MAGNETS ONLY UNBALANCED MAGNETS The amplitude of the voltage difference between the two apertures of the unbalanced dipoles is ~5-6 times larger than that of the balanced dipoles, and in some cases exceeds the threshold (100 mV) Dipoles oversensitive to any voltage transient The phenomenon peaks around 2 kA and scales up linearly with the current ramp-rate % of the dipoles in every sector affected The distribution of unbalanced dipoles is not dependent on the electrical or physical position, or on the manufacturer of the magnets and their cables, or on the date of installation
Unbalanced dipoles – Modelling 41 Emmanuele Ravaioli LHC-CM FPA at 2 10 A/s (S67 20/05/ ) The behavior of the unbalanced dipoles has been successfully simulated by means of a new simplified model of a dipole aperture The distribution of unbalanced dipoles in each sector is simulated assigning a different value to the R_bypass parameter of each magnet Possible physical explanation: Eddy currents ( see Possible cause of quench in B30R7, where U_QSO exceeds 100 mV during fast decay from 7000 A, Arjan Verweij, 2008 ) Standard parameters F_bypass = 0.75 R_bypass = 10