Mission Operations for LISA Pathfinder

Transcription

1 Albert Einstein Institute Max Planck Institute for Gravitational Physics and Leibniz Universität Hannover Mission Operations for LISA Pathfinder Martin Hewitson for the LISA Pathfinder Team COSPAR Moscow, Russia August 7th 2014 We gratefully acknowledge support by the European Space Agency (ESA) (22331/09/NL/HB, 16238/10/NL/HB) and the German Aerospace Center (DLR) (50OQ0601, 50OQ1301) and thank the German Research Foundation for funding the Cluster of Excellence QUEST (Centre for Quantum Engineering and Space-Time Research).

2 Aims of operations 2

3 Aims of operations Obtain the best geodesic motion possible quietest differential acceleration of the two TMs 3 x m s -2 / Hz at 1 mhz pm accuracy position measurement of TM-SC, TM-TM optimisation by changing system parameters determine best configuration by experiments 2

4 Aims of operations Obtain the best geodesic motion possible quietest differential acceleration of the two TMs 3 x m s -2 / Hz at 1 mhz pm accuracy position measurement of TM-SC, TM-TM optimisation by changing system parameters determine best configuration by experiments Develop a noise model of the system allows the projection of the performance of technologies to LISA 2

5 Structure of Mission Operations LPF operations comprises many phases Launch IOCR Launch, LEOP, Transfer, Separation, De-spin Commissioning LTP Science Ops DRS Commissioning DRS Operations 60 days 14 days 3 months 10 days 3 months 3

6 Structure of Mission Operations LPF operations comprises many phases Launch IOCR Launch, LEOP, Transfer, Separation, De-spin Commissioning LTP Science Ops DRS Commissioning DRS Operations 60 days 14 days 3 months 10 days 3 months H1 H2 H3 H4 H5 Day 1 Day 2 Day 3 Day 4 Noise Run Sys ID Discharge Working Point Noise Run Discharge Stray Potentials 3

7 Data Flow 4

8 Data Flow Mission Operations Centre 4

9 Data Flow Science and Technology Operations Centre Analysis Client Analysis Client Mission Operations Centre Data retrieval and conversion Data Repository Analysis Client 4

10 Data Flow Science and Technology Operations Centre Analysis Client Analysis Client Mission Operations Centre Data retrieval and conversion Data Repository Analysis Client Analysis Client Data Repository Analysis Client Complementary Data Centre(s) e.g. APC, Paris 4

11 Running Operations 5

12 Running Operations Each day has two data deliveries: early morning (10am Z): Instrument Configuration & Evaluation data (ICE) a few key channels of data to allow 'quick look' late afternoon (6pm Z): Full data set 5

13 Running Operations Each day has two data deliveries: early morning (10am Z): Instrument Configuration & Evaluation data (ICE) a few key channels of data to allow 'quick look' late afternoon (6pm Z): Full data set We have one team on duty to perform the quick-look and planned STOC front-line analysis A team comprises: A senior scientist A scribe 2 data analysts Operations Scientist 5

14 Example shifts Online: Perform quick look and planned front-line analyses Offline: Consolidate analysis and logbook from previous day Travel: Commute to/from STOC Off duty: Time at home Day N N+1 N+2 N+3 N+4 N+5 N+6 N+7 N+8 N+9 Team 1 Online Offline Online Offline Travel Off Duty Off Duty Travel Online Offline Team 2 Online Offline Online Offline Travel Off Duty Off Duty Travel Online Team 3 Online Offline Online Offline Travel Off Duty Team 4 Online Offline Online Offline Travel 6

15 Analysis software and infrastructure 7

16 Analysis software and infrastructure LTPDA Toolbox MATLAB toolbox which implements an object-oriented data analysis environment objects track their history so results are traceable and reproducible heavily tested and documented ~700 page user manual ~6000 unit tests running every 3 hours multiple system test campaigns ao('built-in'='retrieve_in_timespan', 'START TIME'= :00: UTC, 'NSECS'=86400, 'NAMES'= 'ADT20112', 'HOSTNAME'='lpsdas01.esac.esa.int', 'DATABASE'='STOC_Sim_2_ _Full', 'VERSION'='Version 1', 'CONN'= [], 'BINARY'=1, 'STOP TIME'= ' :00:00.000', 'TIME RANGE'=[], 'CONSTRAINTS'= '', 'FSTOL'= e-07, 'SORT'=1) index('i'=1, 'J'=[]) split('start_time'=' ao('built-in'='tm_parameters', 08:00:00', 'END_TIME'=' 'PARAMETER NAME'='EOM_TM2_M', 16:00:00') 'VERSION'='Initial version') rdivide(empty-plist) pzmodel('name'='', 'DESCRIPTION'= '', 'GAIN'=1, 'POLES'=[], split('offsets'=[ ]) 'ZEROS'=[], 'IUNITS'='', 'OUNITS'= '', 'DELAY'= ) ao('vals'= , fftfilt('npad'=[], 'FILTER'= 'DY'=[], 'AXIS'='y', 'N'=1, '', 'INITIAL CONDITIONS'={} 'YUNITS'='', 'NAME'='', 'DESCRIPTION'= [0x0]) '') mtimes(empty-plist) setname('name'='s_cmd') split('offsets'=[ ]) ao('built-in'='retrieve_in_timespan', 'START TIME'= :00: UTC, 'NSECS'=86400, 'NAMES'= 'LAT10005', 'HOSTNAME'='lpsdas01.esac.esa.int', 'DATABASE'='STOC_Sim_2_ _Full', 'VERSION'='Version 1', 'CONN'= [], 'BINARY'=1, 'STOP TIME'= ' :00:00.000', 'TIME RANGE'=[], 'CONSTRAINTS'= '', 'FSTOL'= e-07, 'SORT'=1) index('i'=1, 'J'=[]) pzmodel('name'='', 'DESCRIPTION'= split('start_time'=' '', 'GAIN'=1, 'POLES'=[], 08:00:00', 'END_TIME'=' 'ZEROS'=[], 'IUNITS'='', 'OUNITS'= 16:00:00') '', 'DELAY'= ) ao('built-in'='retrieve_in_timespan', 'START TIME'= :00: UTC, 'NSECS'=86400, 'NAMES'= 'LAT10003', 'HOSTNAME'='lpsdas01.esac.esa.int', fftfilt('npad'=[], 'FILTER'= 'DATABASE'='STOC_Sim_2_ _Full', '', 'INITIAL CONDITIONS'={} 'VERSION'='Version 1', 'CONN'= [0x0]) [], 'BINARY'=1, 'STOP TIME'= ' :00:00.000', 'TIME RANGE'=[], 'CONSTRAINTS'= '', 'FSTOL'= e-07, 'SORT'=1) ao('vals'= e-06, diff('method'='3point', 'F0'= 'DY'=[], 'AXIS'='y', 'N'=1, '1/Nsecs', 'ORDER'='ZERO', split('offsets'=[ ]) index('i'=1, 'J'=[]) 'YUNITS'='s^-2', 'NAME'='', 'COEFF'=[]) 'DESCRIPTION'='') diff('method'='3point', 'F0'= split('start_time'=' '1/Nsecs', 'ORDER'='ZERO', mtimes(empty-plist) 08:00:00', 'END_TIME'=' 'COEFF'=[]) 16:00:00') fftfilt('npad'=[], 'FILTER'= setname('name'='diff acc') setname('name'='s_w2') '', 'INITIAL CONDITIONS'={} [0x0]) ao('vals'= e-07, 'DY'=[], 'AXIS'='y', 'N'=1, split('offsets'=[ ]) split('offsets'=[ ]) 'YUNITS'='s^-2', 'NAME'='', 'DESCRIPTION'='') split('offsets'=[ ]) mtimes(empty-plist) minus(empty-plist) split('offsets'=[ ]) setname('name'='s_dw') plus(empty-plist) split('offsets'=[ ]) plus(empty-plist) formal deliveries to ESA with acceptance tests END setname('name'='residual') 7

17 Analysis software and infrastructure LTPDA Toolbox MATLAB toolbox which implements an object-oriented data analysis environment objects track their history so results are traceable and reproducible heavily tested and documented ~700 page user manual ~6000 unit tests running every 3 hours multiple system test campaigns formal deliveries to ESA with acceptance tests ao('built-in'='retrieve_in_timespan', 'START TIME'= :00: UTC, 'NSECS'=86400, 'NAMES'= 'ADT20112', 'HOSTNAME'='lpsdas01.esac.esa.int', 'DATABASE'='STOC_Sim_2_ _Full', 'VERSION'='Version 1', 'CONN'= [], 'BINARY'=1, 'STOP TIME'= ' :00:00.000', 'TIME RANGE'=[], 'CONSTRAINTS'= '', 'FSTOL'= e-07, 'SORT'=1) index('i'=1, 'J'=[]) split('start_time'=' ao('built-in'='tm_parameters', 08:00:00', 'END_TIME'=' 'PARAMETER NAME'='EOM_TM2_M', 16:00:00') 'VERSION'='Initial version') rdivide(empty-plist) pzmodel('name'='', 'DESCRIPTION'= '', 'GAIN'=1, 'POLES'=[], split('offsets'=[ ]) 'ZEROS'=[], 'IUNITS'='', 'OUNITS'= '', 'DELAY'= ) ao('vals'= , fftfilt('npad'=[], 'FILTER'= 'DY'=[], 'AXIS'='y', 'N'=1, '', 'INITIAL CONDITIONS'={} 'YUNITS'='', 'NAME'='', 'DESCRIPTION'= [0x0]) '') mtimes(empty-plist) setname('name'='s_cmd') split('offsets'=[ ]) ao('built-in'='retrieve_in_timespan', 'START TIME'= :00: UTC, 'NSECS'=86400, 'NAMES'= 'LAT10005', 'HOSTNAME'='lpsdas01.esac.esa.int', 'DATABASE'='STOC_Sim_2_ _Full', 'VERSION'='Version 1', 'CONN'= [], 'BINARY'=1, 'STOP TIME'= ' :00:00.000', 'TIME RANGE'=[], 'CONSTRAINTS'= '', 'FSTOL'= e-07, 'SORT'=1) index('i'=1, 'J'=[]) pzmodel('name'='', 'DESCRIPTION'= split('start_time'=' '', 'GAIN'=1, 'POLES'=[], 08:00:00', 'END_TIME'=' 'ZEROS'=[], 'IUNITS'='', 'OUNITS'= 16:00:00') '', 'DELAY'= ) ao('built-in'='retrieve_in_timespan', 'START TIME'= :00: UTC, 'NSECS'=86400, 'NAMES'= 'LAT10003', 'HOSTNAME'='lpsdas01.esac.esa.int', fftfilt('npad'=[], 'FILTER'= 'DATABASE'='STOC_Sim_2_ _Full', '', 'INITIAL CONDITIONS'={} 'VERSION'='Version 1', 'CONN'= [0x0]) [], 'BINARY'=1, 'STOP TIME'= ' :00:00.000', 'TIME RANGE'=[], 'CONSTRAINTS'= '', 'FSTOL'= e-07, 'SORT'=1) ao('vals'= e-06, diff('method'='3point', 'F0'= 'DY'=[], 'AXIS'='y', 'N'=1, '1/Nsecs', 'ORDER'='ZERO', split('offsets'=[ ]) index('i'=1, 'J'=[]) 'YUNITS'='s^-2', 'NAME'='', 'COEFF'=[]) 'DESCRIPTION'='') diff('method'='3point', 'F0'= split('start_time'=' '1/Nsecs', 'ORDER'='ZERO', mtimes(empty-plist) 08:00:00', 'END_TIME'=' 'COEFF'=[]) 16:00:00') fftfilt('npad'=[], 'FILTER'= setname('name'='diff acc') setname('name'='s_w2') '', 'INITIAL CONDITIONS'={} [0x0]) ao('vals'= e-07, 'DY'=[], 'AXIS'='y', 'N'=1, split('offsets'=[ ]) split('offsets'=[ ]) 'YUNITS'='s^-2', 'NAME'='', 'DESCRIPTION'='') split('offsets'=[ ]) mtimes(empty-plist) minus(empty-plist) split('offsets'=[ ]) setname('name'='s_dw') plus(empty-plist) split('offsets'=[ ]) plus(empty-plist) setname('name'='residual') END LTPDA Repository provides a centralised database structure with web interface for administration and searching interface to LTPDA toolbox directly from within MATLAB to submit and retrieve objects core client/server system to be used by ESA for LPF mission also in heavy daily use in various labs 7

18 Operational Constraints 8

19 Operational Constraints Investigations are time-line driven no real-time / 'joystick' control 8

20 Operational Constraints Investigations are time-line driven no real-time / 'joystick' control Investigations are packed into 24 hour groups called Payload Operation Requests 8

21 Operational Constraints Investigations are time-line driven no real-time / 'joystick' control Investigations are packed into 24 hour groups called Payload Operation Requests 6 days of time-line are on-board LPF at all times 8

22 Operational Constraints Investigations are time-line driven no real-time / 'joystick' control Investigations are packed into 24 hour groups called Payload Operation Requests 6 days of time-line are on-board LPF at all times Changing a POR has a 3-5 day lead-time 8

23 Operational Constraints Investigations are time-line driven no real-time / 'joystick' control Investigations are packed into 24 hour groups called Payload Operation Requests 6 days of time-line are on-board LPF at all times Changing a POR has a 3-5 day lead-time Mid- and long-term plans will be generated before launch 8

24 Philosophy of Science Operations 1. Start out with low-risk, gentle probing of the system first to gain experience and to understand the state of the system 2. Move on to more invasive investigations and begin tuning the system 3. Higher risk investigations are planned to be later in the operations 9

25 Week 1: Gentle Probing The first two weeks are all about gathering information and gaining experience This is our first interaction with the system Focus on: noise runs first tests of signal injection (system identification) getting a handle on the charge rate and discharging 10

26 Week 1: Gentle Probing The first two weeks are all about gathering information and gaining experience This is our first interaction with the system Focus on: Hour noise runs first tests of signal injection (system identification) 1 Noise run in Sci CE1 CE2 Noise run in Sci 1.2 getting a handle on the charge rate and discharging 3 CE1 CE2 Sys ID (low amp) Noise run in Sci CE1 CE2 Working point scan (x,y,z), both TMs 5 CE1 CE2 Cross-talk investigations, low amplitude 6 CE1 CE2 Noise run in Sci Station Keeping Transition Acc3 -> Sci 1.2 FD1 FD2 CE1 CE2 FD1 FD2 Charge estimate TM1 Charge estimate TM2 Fast Discharge TM1 Fast Discharge TM2 10

27 What is a noise run? 11

28 What is a noise run? Enter nominal science mode (DFACS mode Sci. 1.2) SC following TM1 TM2 following TM1 11

29 What is a noise run? Enter nominal science mode (DFACS mode Sci. 1.2) SC following TM1 TM2 following TM1 Put the system in the 'best' state we know discharged TMs optimal dc compensation voltages best test-mass working point for OMS and GRS 11

30 What is a noise run? Enter nominal science mode (DFACS mode Sci. 1.2) SC following TM1 TM2 following TM1 Put the system in the 'best' state we know discharged TMs optimal dc compensation voltages best test-mass working point for OMS and GRS Take data for, e.g., 10 hours 11

31 Estimating Residual Differential Acceleration 12

32 Estimating Residual Differential Acceleration Understanding the purity of the free-fall we achieve, and what limits it, requires us to assess the residual forces acting on the TMs what s left when we subtract the forces we can account for 12

33 Estimating Residual Differential Acceleration Understanding the purity of the free-fall we achieve, and what limits it, requires us to assess the residual forces acting on the TMs what s left when we subtract the forces we can account for We compute the relative acceleration of the two TMs based on the observed relative position 12

34 Estimating Residual Differential Acceleration Understanding the purity of the free-fall we achieve, and what limits it, requires us to assess the residual forces acting on the TMs what s left when we subtract the forces we can account for We compute the relative acceleration of the two TMs based on the observed relative position Try to account for the contributions of g_res that we know applied control forces couplings due to force gradients 12

35 Analysing the data 1. Download the time-series 2. Assemble the current best estimate of the required system parameters actuator gains, delays, stiffnesses, 3. Form linear combination of the time-series with delays, and filtering as necessary 4. Take spectrum of the residuals 13

36 The contributions sqrt(psd(in loop Acceleration)) sqrt(psd(applied Force)) sqrt(psd(tm2 Stiffness Coupling)) sqrt(psd(sc Jitter)) sqrt(psd(residual Diff. Acc.)) [m s 2 Hz 1/2 ] Frequency [Hz] 14

37 System Identification 15

38 System Identification Estimating our residual acceleration requires knowledge of certain system parameters How do we gain that knowledge? 15

39 System Identification Estimating our residual acceleration requires knowledge of certain system parameters How do we gain that knowledge? At the beginning of operations, this comes from ground measurements system modelling results of industrial commissioning campaign 15

40 System Identification Estimating our residual acceleration requires knowledge of certain system parameters How do we gain that knowledge? At the beginning of operations, this comes from ground measurements system modelling results of industrial commissioning campaign How do we improve and update that knowledge? through dedicated investigations 15

41 x-axis system identification: part 1 Goal is to measure the key parameters needed for estimating the residual differential acceleration can be done by 16

42 x-axis system identification: part 1 Goal is to measure the key parameters needed for estimating the residual differential acceleration can be done by Guidance Input 16

43 x-axis system identification: part 1 Goal is to measure the key parameters needed for estimating the residual differential acceleration can be done by Force on SC Guidance Input 16

44 x-axis system identification: part 1 Goal is to measure the key parameters needed for estimating the residual differential acceleration can be done by Force on SC SC X Guidance Input 16

45 x-axis system identification: part 1 Goal is to measure the key parameters needed for estimating the residual differential acceleration can be done by Force on SC Diff. X SC X Guidance Input 16

46 x-axis system identification: part 2 17

47 x-axis system identification: part 2 Guidance Input 17

48 x-axis system identification: part 2 Guidance Input Force on TM2 17

49 x-axis system identification: part 2 Diff. X Guidance Input Force on TM2 17

50 x-axis system identification: part 2 Diff. X Guidance Input SC X Force on TM2 17

51 What do we learn from that? Athruster Drag-free Controller Fsc guidance injection SC Dynamics [m/n] X1 IFO dt x1 x1 K1 kg s -2 TM1 Dynamics [m/n] X12 IFO K2 TM2 Dynamics [m/n] dt x12 kg s -2 Ftm2 guidance injection Asus Suspension Controller 18

52 What do we learn from that? Athruster Drag-free Controller Fsc SC Dynamics [m/n] Thruster Response X1 IFO dt x1 guidance injection x1 K1 kg s -2 TM1 Dynamics [m/n] X12 IFO K2 TM2 Dynamics [m/n] dt x12 kg s -2 Ftm2 guidance injection Cap. Act. Response Asus Suspension Controller 18

53 What do we learn from that? Athruster Drag-free Controller Fsc SC Dynamics [m/n] Thruster Response X1 IFO dt x1 guidance injection x1 Spring-like couplings of TMs to SC K1 kg s -2 TM1 Dynamics [m/n] X12 IFO K2 TM2 Dynamics [m/n] dt x12 kg s -2 Ftm2 guidance injection Cap. Act. Response Asus Suspension Controller 18

54 What do we learn from that? Athruster Drag-free Controller Fsc SC Dynamics [m/n] Thruster Response X1 IFO dt x1 guidance injection x1 Spring-like couplings of TMs to SC K1 kg s -2 TM1 Dynamics [m/n] X12 IFO Sensing Delays K2 TM2 Dynamics [m/n] dt x12 kg s -2 Ftm2 guidance injection Cap. Act. Response Asus Suspension Controller 18

55 The data 1.5 x SC Injection X1 IFO X12 IFO Amplitude [m] TM2 Injection Origin: :00: Time [m] 19

56 Analysis Follows the same form as for estimating residual differential acceleration But now the coefficients in the model are fit so that the linear combination of terms fit the observation When a good fit is found, the residuals contain no trace of the injected signals Fit to 20

57 Residuals 21

58 Residuals 10 9 sqrt(psd(in loop Acceleration)) [m s 2 Hz 1/2 ] Frequency [Hz] 21

59 Residuals sqrt(psd(in loop Acceleration)) sqrt(psd(applied Force)) sqrt(psd(residuals)) [m s 2 Hz 1/2 ] Frequency [Hz] 21

60 Residuals sqrt(psd(in loop Acceleration)) sqrt(psd(residuals)) sqrt(psd(tm2 Stiffness Coupling)) sqrt(psd(new Residuals)) [m s 2 Hz 1/2 ] Frequency [Hz] 21

61 Residuals sqrt(psd(in loop Acceleration)) sqrt(psd(residuals)) sqrt(psd(sc Jitter)) sqrt(psd(new Residuals)) [m s 2 Hz 1/2 ] Frequency [Hz] 21

62 A general scheme Balancing forces: improves physical modelling and interpretation simplifies the analysis a great deal This 'acceleration scheme can be used for other contributions cross-talk thermal magnetic free-flight experiments 22

63 Noise Budget How does our observed residual differential acceleration differ from what we expect? Why does it differ? this drives the next activities to be performed h Spectral Density fm s 2 / p i Hz Requirements Noise Breakdown OMS Angular Total Direct Forces Star Tracker Thrusters OMS Sensing Electrostatic Actuation Frequency [mhz] i h fm s 2 / p Hz Spectral Density Direct Forces Star Tracker Current Best Estimate Requirements Total OMS Sensing Electrostatic Actuation Thrusters Frequency [mhz] 23

64 From requirements to expectations... h fm s 2 / p Hz i Spectral Density Direct Forces Star Tracker Requirements Total Thrusters Frequency [mhz] Electrostatic Actuation OMS Sensing Residual Differential Test-mass Acceleration 24

65 Free-flight experiment h fm s 2 / p Hz i Spectral Density Star Tracker Requirements Total Direct Forces Thrusters Frequency [mhz] OMS Sensing Electrostatic Actuation Residual Differential Test-mass Acceleration 25

66 In context... i fm s 2 / p Hz LPF Mission Requirements elisa Mission Goal LPF Performance (Subsystem Requirements) LPF Performance (Best Estimate) LPF Performance (Free-Flight) Spectral Density h Frequency [mhz] 26

67 Week 2: More Detailed Probing Focus on: detailed system identification state of TMs (stray potentials, charge) Noise run in Sci 1.2 CE1 CE2 9 Stray Potentials: Vscan (z) Stray Potentials: Vscan (y) Stray Potentials: Vscan (x) Noise run in Sci CE1 CE2 Noise run in Sci 1.2 Sys ID 1 Sys ID 2 11 Match stiffness Sys ID 3 Sys ID 4 Sys ID 5 Sys ID 6 Noise run in Sci Long charge estimate TM2 CE1 Noise run in Sci Noise run in Sci 1.2 Cross-talk investigations, nominal amplitudes 14 Station Keeping Transition Acc3 -> Sci 1.2 FD1 FD2 27

68 Week 3: Exploration of the system Long noise run to look at low frequencies Alternative DFACS operation modes More detailed cross-talk investigations Noise run in Sci Noise run in Sci CE1 CE2 Transition from Sci1.2 to Sci2.0 Noise run in Sci Sys ID 1 (Sci2) Sys ID 2 (Sci2) Match stiffness Sys ID 3 (Sci2) Sys ID 4 (Sci2) Sys ID 5 (Sci2) 19 Sys ID 6 (Sci2) CE1 CE2 Noise run in Sci Cross-talk investigations, nominal amplitudes CE1 CE2 21 Station Keeping Transition Acc3 -> Sci 1.2 FD1 FD2 28

69 Longer term plan Main Activities Early Phase Middle Phase Sys ID Electrostatics Charging DFACS Modes OMS Optimisation IS Optimisation Thermal excitations working point experiments Magnetic excitations Thruster characterisation Late Phase Sensor characterisation Swapped test-mass operation Free-flight experiments Discharge tests 29

70 Elements and Capabilities 30

71 Elements and Capabilities Successful science operations requires many elements Ground segment infrastructure Data analysis tools Investigation designs Analysis pipelines 30

72 Elements and Capabilities Successful science operations requires many elements Ground segment infrastructure Data analysis tools Investigation designs Analysis pipelines We also need many capabilities of the system DFACS configuration: modes, gains, offsets, actuation algorithms Sensor configuration: TM bias levels, GRM modes, laser temperature, OMS heterodyne frequency, OMS loop states, alignment Actuator configuration: gains, biases Signal injections: guidances, forces, torques, electrode voltages, OMS loop set points Environment: TM discharge, UV lamp control, CMS configuration, thermal excitation, magnetic field excitation 30

73 Where are we? 31

74 Where are we? Components Ground segment infrastructure mostly all in place, some testing remains Data Analysis Tools LTPDA Toolbox is mature Investigation design we have generated and tested a large number of investigations Analysis Pipelines many exist already some to be developed over the next year 31

75 Where are we? Components Ground segment infrastructure mostly all in place, some testing remains Data Analysis Tools LTPDA Toolbox is mature Investigation design we have generated and tested a large number of investigations Analysis Pipelines Training many exist already some to be developed over the next year We ve had 4 dedicated training sessions We ran 4 large-scale mission simulations tested many of the investigation designs and analysis pipelines on synthesised data in realistic operational environment Participation in industrial hardware test campaigns exposure to real data 31

76 Concluding thoughts LPF Operations is a complex and demanding schedule Packed full of detailed investigations which will allow us to: optimise the system to achieve best possible TM free-fall develop a detailed physical model of the system This all paves the way for commissioning of a LISAlike mission we must design in the necessary capabilities from the start! 32

77 Thank you! 33