Propellant Gauging Experience with Meteosat Second Generation S/C Fleet

Transcription

1 SpaceOps Conferences 28 May - 1 June 2018, 2018, Marseille, France 2018 SpaceOps Conference / Propellant Gauging Experience with Meteosat Second Generation S/C Fleet C. Bihr 1, CLC Space GmbH, Kirchstr. 47, Alsbach-Hähnlein 64665, Germany F. Murolo 2, EUMETSAT, Eumetsat Allee 1, Darmstadt 64295, Germany M. Klinc 3 WGS Workgroup Solutions GmbH, Brüder Knauß Str. 6, Darmstadt 64285, Germany IM. Achkar 4, B. Robert 5 Thales Alenia Space, Cannes 06150, France The measurement results obtained on MSG-1/-2/-3 (now re-named as Meteosat-8/-9/-10) using on-board gauging sensors (GSU for MSG-1/-2, UGS for MSG-3/-4) have demonstrated to be well in line with the values of the Book-Keeping Method (BKM). The UGS measurement results obtained through equivalent UGS measuring campaigns on MSG-4 deviated from the BKM calculations after LEOP. The discrepancy in this case could initially not be justified even considering the worst case conditions for the error budget contributors. Careful investigations and analysis have been made by Thales Alenia Space (TAS) and EUMETSAT to eliminate errors on the input parameters and assumptions for the BKM. In parallel, the sensor outputs and the three stage correction scheme was analyzed by EUMETSAT and TAS. Both parties agreed to test the UGS at different thermal conditions and to compare the calculated values for the remaining propellant in the tanks in order to identify and specify a possible measurement sensitivity to the fluid temperature as function of the tank filling height. For this purpose, extensive calibration campaigns have been performed to fully characterize the actual in flight temperature dependency of the UGS for the different kind of propellant used. This paper describes EUMETSAT s new propellant gauging strategy, resulting from several gauging campaigns conducted on MSG-3 and MSG-4 in 2015/2016. With this innovative strategy the behavior of the sensor could be characterised by the operator in space at different propellant temperatures. The advantage of the operational use of these information leads to a reduction of the uncertainty margins of which the gauging instruments are specified and to a potential release of fuel reserves for the extension of MSG-3/-4 operational lifetime. BKM EOL EUMETSAT GSU LEOP MMH MON MSG MSG-1 NSSK UGS Nomenclature = book keeping method = end of life = European Organisation for the Exploitation of Meteorological Satellites = gauging sensor unit = launch and early orbit phase = monomethyl hydrazine = mixed oxides of nitrogen = Meteosat Second Generation Programme = Meteosat-8, MSG-2 = Meteosat-9, MSG-3 = Meteosat-10, MSG-4 = Meteosat-11 = north south station keeping maneuver (inclination maneuver, out of plane maneuver) = ultrasonic gauging sensor 1 Spacecraft Operations Engineer, CLC Space GmbH c/o EUMETSAT, Flight Operations Division. 2 MTG Operations Preparation Team Leader and MSG-IODC project manager, EUMETSAT, Flight Operations Division. 3 Flight Dynamics Engineer, WGS GmbH c/o EUMETSAT, Flight Operations Division. 4 MSG & MTG Propulsion Engineer, Thales Alenia Space, France 5 MSG Routine Operations Support Manager, Thales Alenia Space, France 1 Copyright 2018 by EUMETSAT. Published by the, Inc., with permission.

2 I. Introduction HE length of the satellite useful lifetime in orbit clearly depends on the accuracy of the determination of the T amount of propellant available on-board. The aim is to be able to make use of the propellant for station keeping maneuvers as long as operationally possible, up to the point of reaching a minimum amount, which is needed for a successful re-orbiting at the end of the mission, according to the ISO standard on space debris mitigation. EUMETSAT s geostationary and spin-stabilized satellites of the Meteosat second generation spacecraft fleet, Meteosat-8/-9/-10/-11, are equipped with gauging sensors (GSU or UGS) for the measurement of the on-board oxidizer (MON) and fuel (MMH) propellants. They are installed in the propellant feed port of each of the four tanks (two MON tanks with GSU/UGS-1/2, two MMH tanks with GSU/UGS -3/4) and are part of the bi-propellant Unified Propulsion System. The gauging sensors were specified to be able to determine the propellant mass available in the tanks with an accuracy of 4 kg. This is a significantly smaller uncertainty, compared to the value estimated for the book keeping method (BKM), which can instead result in errors of up to 14 kg at the end of life of the satellite. While the propellant measurements using GSU on Meteosat-8 and -9 are done on a routine basis due to various operational reasons, several UGS test campaigns were conducted on Meteosat-10 in 2012 with a new method, to extend the proposed warm up period up to one full hour in order to ensure thermal equilibrium between propellant, tank and sensor. A further improvement was observed, by powering the sensors ON at the same time, i.e. in parallel, instead of the proposed in-series philosophy. The idea was to further reduce uncertainties due to possible fuel migration effects during the different gauging periods for each tank. This strategy was introduced and described in previous Space Ops conferences by the authors (Ref. [1]). Thanks to the new method, applied at the second measurement campaign in December 2012, the difference in the total propellant amount between BKM calculations and UGS measurements could be reduced from 0.81% to 0.69% of the estimated mass. When MSG-4 was launched in 2015 the differences between the UGS measured propellant and the BKM predictions after LEOP were larger than initially anticipated. As a result, the BKM calculations tracing the propellant consumption during LEOP were reviewed by Thales Alenia Space. In parallel, the sensor outputs and the three stage correction scheme were analyzed by EUMETSAT and TAS. Both parties agreed to test the UGS at different thermal conditions and to compare the calculated values for the remaining propellant in the tanks in order to identify and specify possible sensor drifts and/or response variation as function of the fluid temperature and tank filling height. II. Sensor Concepts Two different sensor concepts are used for the Meteosat Second Generation Satellites. Meteosat-8 and -9 have gauging sensors installed, the so called GSU s (gauging sensor units) which are based on capacity measurement of the wetted sensor segment (capillarity principle), whereas on Meteosat-10 and -11 ultrasonic gauging sensors are mounted, the so called UGS which are based on ultrasonic signal transmission through the propellant. Each of the four sensors per spacecraft provides a high-frequency output signal which is correlated to the liquid level height inside the oxidizer or fuel propellant tanks. A. Gauging Sensor Units GSU s The gauging sensor units, installed in the propellant tanks of MSG-1 and MSG-2 are described in Ref. [2] in detail. Due to its design, the GSU s are not designed to provide reliable measurements, when the MON tank filling height is between 85 mm and 59 mm (within the so called guard segment of the unit). For MSG-1 the top of the guard segment (i.e. 85 mm filling level) was reached in October 2010 after the execution of the last NSSK maneuver. The last valid GSU measurements in May 2010 have confirmed the BKM values. For example the difference between GSU and BKM results for the MON part per tank was 575 grams less measured with GSU s only. As a comparison for the MON part on MSG-2, the difference between GSU and BKM results at the measurement campaigns in July 2017 showed +750 grams per each tank. Fig. 1 Gauging Sensor Unit Picture courtesy of Thales Alenia Space & Bradford-NL 2

3 B. Ultrasonic Gauging Sensors UGS s The last two satellites of the Meteosat Second Generation MSG-3 and MSG-4 are equipped with Ultrasonic Gauging Sensors, a liquid level gauging device based on ultrasonic transit time measurement principle. A piezo emitter and receiver are employed to send an ultrasonic pulse into the propellant and to receive the reflected pulses from the propellant surface. The measured time-of transmission is converted into a frequency signal which is sent to ground for processing. For the propellant calculation a sophisticated three stage correction scheme is applied, by considering the propellant temperatures in order to allow a post processing correction to its measurements. For further details see Ref. [1, 3]. Fig. 2 Ultrasonic Gauging Sensor Picture courtesy of Thales Alenia Space & Bradford-NL III. Analysis of MSG-4 results and new gauging strategy in 2016 After commissioning of the MSG-4 satellite in 2015 several gauging campaigns have been conducted. For the first time on Eumetsat s Meteosat Second Generation Spacecraft fleet, the difference between the BKM and UGS results observed was larger than expected. Although the propellant temperature is already considered and taken into account in the correction scheme provided by the manufacturer for the UGS frequency output, it was deemed useful to perform several gauging measurements at different temperatures to have information on the sensitivity of the results to the temperature variations. Between gauging campaigns on DOY211 and DOY217, 2.58 kg of propellant were used for both drift stop maneuvers. Between DOY217 and DOY337, only 55.6 grams of total propellant were used for a regular EWSKand small spin down maneuver. Hence, the propellant gauging results of the campaigns between DOY217 and DOY337 can be compared, assuming constant propellant volume. Figure 3 summarizes the MSG-4 UGS measurement results from 2015, conducted at random times decoupled from regular on-board temperature patterns and therefore exploit measurements at several different temperatures. Fig. 3 MSG-4 UGS results in

4 At higher tank temperatures higher fluid level and thus more propellant was measured. The results for each UGS are very close, when gauging was conducted at similar tank temperatures - see Fig. 3. For example UGS-2 measurements on DOY217 and DOY280 as well as UGS-4 measurements on DOY217 and DOY266 and, as a very good example, on UGS-1 on DOY335 and DOY337, when the same amount of MON in tank 1 was measured at the same temperature. For all 4 UGS a correlation trend was obtained from the measurement campaigns in 2015 at tank temperatures between 19 C and 25 C - see Fig. 4. Fig. 4 MSG-4 measurement sensitivity to temperature; results of 2015 campaign C. New strategy: Propellant gauging at higher and lower tank temperatures to determine the sensor sensitivity to thermal variations In March 2016 four additional measurement campaigns were conducted on MSG-4 at different tank temperatures in order to identify and specify the temperature related measurement change. Between the last measurement campaign in 2015 (DOY337) and the first in March 2016, when the new strategy applied (DOY060-DOY063) only 60.5 grams propellant were consumed at the execution of a single EWSK maneuver, which still allows to consider the results comparable to the gauging results, obtained in 2015 (after the drift stop maneuvers, i.e. from DOY217). The UGS tests could be performed at lower (19.3 C to 20.0 C) and higher tank temperatures (21.2 C to 22.3 C) exploiting the standard thermal regulation cycles without additional operational effort thanks to the nominal tank heater cycling pattern. These thermal cycles are providing a variation of about two degrees Celsius, which was sufficient to characterize the temperature dependent shift of the measurements. Table 1 summarizes the MSG-4 UGS measurement results from the March 2016 campaigns. Table 1 MSG-4 UGS measurement campaigns in March 2016 at higher and lower Tank temperatures (Lx = tank liquid level height) 4

5 It was shown, that tanks temperature variations of about 2 C (between lower and higher temperature) caused a difference of 4 kg in the measurement of the total propellant. The discrepancy of the propellant in the tanks was kg at higher and kg at lower tank temperatures compared to BKM values. The new strategy in March 2016 was based on gauging the propellant at specific temperatures. In this way the temperature dependant variations, initially observed from the measurements in 2015, were verified and the individual sensor sensitivity trend-lines for 2015 confirmed. These propellant specific sensor characteristics were later also verified at measurement campaigns on Meteosat-10, considering the tank filling level - see next chapter. Fig. 5 MSG-4 measurement results from 2016 (new strategy to gauge at higher and lower temperatures) confirming sensor thermal response observed in 2015 IV. New strategy verification approach with MSG-3 With the different measurement campaigns conducted on MSG-4 by applying the new thermal variation compensation strategy, i.e. gauging the propellant at higher and lower tank temperatures, the individual sensor sensitivity to temperature could be specified and the measurement results derived considering a constant reference temperature. The new strategy was then adopted and used for measurement campaigns on Meteosat-10. TAS and Eumetsat agreed to have two gauging campaigns before and after the NSSK maneuver in June 2016 at higher and lower tank temperatures. The idea was to determine the sensor temperature sensitivity pre- and post NSSK, where approximately 20 kg of total propellant had been used. Table 2 shows the specified individual sensor temperature variations for MSG-3 before and after the NSSK maneuver in June 2016: MSG-3 Gauging Sensor and Tank Drift before NSSK [mm/ C] Drift after NSSK [mm/ C] UGS-1 in MON Tank UGS-2 in MON Tank UGS-3 in MMH Tank UGS-4 in MMH Tank Table 2 Meteosat-10 sensor thermal response pre- and post NSSK in June

6 The measured drifts for the UGS on MSG-3 confirmed the measurements sensitivity to temperature observed on MSG-4. Furthermore after the NSSK on MSG-3 it was possible to conclude, that the propellant measurement results depend not only as expected on the tank filling level and on the density of the fluid but are also influenced by the temperature at which the measurement is acquired. Furthermore, the static residuals on the oxidizer side depends strongly on the temperature due to the vapor pressure, which increases significantly with higher temperatures. Figure 6 shows the MSG-3 measurement temperature variation, identified before and after the NSSK manoeuvre in June 2016 as trend-lines. Fig. 6 MSG-3 UGS temperature sensitivity before and after NSSK D. UGS measurement results comparison at corrected temperatures Thanks to the characterization of the individual sensor sensitivity to temperature for the UGS on MSG-3 and MSG-4 in 2016, a comparison to the BKM values was possible, taking different propellant temperatures into account. Table 3 summarizes the MSG-3 and MSG-4 UGS measured total propellant considering the specified individual sensor characteristics and corrected temperatures for comparison to the BKM values in July The temperatures listed (i.e. 19 C to 23 C = Tcorrected) are the equivalent reference temperatures to which the propellant gauging measurements were normalised in consideration of the individual temperature sensitivity effect of each gauging sensor. As can simply be seen from Table 3 and Fig. 7, the higher the propellant temperature, the more propellant was gauged. The BKM values for MSG-3 were very well in line with the corrected UGS measurements at 22 C, whereas on MSG-4 a larger discrepancy between BKM vs. UGS results was observed Table 3 MSG-3 and MSG-4 UGS results at for operational tank temperatures between 19 C to 23 C. different temperatures vs. BKM values in 2016 At e.g. 22 C the difference was of about 10.8 kg less total propellant gauged than predicted by BKM for MSG-4. Eumetsat and TAS agreed to compare the more conservative results, obtained at 20 C with the BKM values. One of the reasons for this safety conclusion was, that the sensors were characterised on ground by the manufacturer at 20 C. By taking the results obtained at the corrected temperature of 20 C, MSG-3 had approximated 2.6 kg less and MSG-4 about 15 kg less total propellant compared to the BKM calculations. Both parties agreed further, that the measurement results at 20 C might be too conservative. 6

7 E. In-flight calibration of the MSG-3 UGS utilizing the NSSK maneuver propellant consumption Since the difference between planned and determined maneuver performance was 0.1 % only, the BKM values for the propellant consumed during the NSSK maneuver in June 2016 were used as a reference to recalculate the UGS measurements. The BKM results for the used propellant were as follows: MON= kg, MMH=7.702 kg, total= kg. With consideration of the measurement dependency from temperature as determined before and after NSSK, the following calculations have been made (example at Tcorrected = 20 C): The gauged propellant prior NSSK maneuver was corrected by using the individual sensor temperature sensitivity at 20.0 C. For example for UGS-1 in tank MON-1, the gauged MON-1 propellant at C was kg and kg at C. The resulting sensor drift pre NSSK was therefore 1.51 mm/c - see Table 2. With the sensor drift information, the MON-1 propellant was corrected to T = 20 C and resulted to kg. Similar calculations were done for the other sensors (i.e. UGS-2 in MON tank 2, UGS-3 in MMH tank 1 and UGS-4 in MMH tank 2), considering their individual pre NSSK sensor variation and by adjusting the gauging results to the corrected temperature of 20 C. As can be seen from Table 4 at 20 C, kg MON and kg MMH summarizing up to a total propellant of kg pre NSSK. The true post NSSK maneuver gauged MON-1 propellant for UGS-1 in tank MON-1 at C was kg and kg at C. Hence, the UGS-1 sensor drift post NSSK was 1.14 mm/c. With this information, the gauged MON-1 propellant was corrected to T = 20 C, and resulted to kg. Again, the same calculations were done for the other sensors, considering their true measured propellant at different tank temperatures and their individual post NSSK temperature sensitivity by adjusting the gauging results to 20 C. The post NSSK gauging results, corrected to 20 C were as follows: kg MON and kg MMH resulting in a total propellant of kg. By subtracting the post NSSK results from the pre NSSK results (e.g. at 20 C), the following deltas were calculated: MON = kg, MMH = kg, Total = kg. These deltas did not match the BKM values for the NSSK maneuver consumption (MON= kg, MMH=7.702kg, total= kg). The calibration procedure was then repeated at Tcorrected = 20.1 C, 20.2 C, 20.3 C, etc. At 22.7 C the deltas for MON, MMH and the overall propellant were well matching the BKM values for the propellant consumed during the NSSK see Table 4. From the calibration exercise it can be concluded, that for the corrected temperatures between 20 C and 22 C (as shown as examples in the table below), the MSG-3 UGS calculated values for MON and MMH differ from the BKM values for the propellant used during NSSK (i.e. pre- and post NSSK delta values) and also for the total propellant, whereas at corrected 22.7 C the UGS calculated difference matches the BKM values. The 22.7 C is the equivalent sensing temperature (not the actual propellant temperature) at which the individual sensor temperature sensitivity is compensated for all 4 UGS and minimizes the deviation with respect to the measured consumption during NSSK for both components (i.e kg MON and kg MMH) and the total propellant. Table 4 MSG-3 UGS results [kg] corrected at 20.0 C, 22.0 C, 22.6 C, 22.7 C and 22.8 C considering the individual sensor drift before/after NSSK and BKM propellant consumption during NSSK. 7

8 F. Total propellant considering sensor drift and corrected temperatures for comparison to BKM values At the equivalent sensing temperature of 22.7 C the UGS results (110.5 kg) are about 1 kg higher only compared to the BKM results (109.5 kg). These data may confirm, that the assumption of taking measurement results at 20 C for comparison to BKM values was too conservative. In particular for the UGS, the first measurement campaign conducted on Meteosat-10 in 2012 showed that the difference in the total propellant amount between BKM calculations and UGS measurements with respect to the BOL propellant mass was of about 0.13 %. At the measurement campaigns in 2016, the UGS temperature variation was considered and showed that the difference compared to the BKM may be reduced even further to less than 0.08 % (measurement results from October 2016, corrected to T=22.7 C) with respect to the BOL propellant mass loaded. Since the first MSG-4 NSSK maneuver is planned in mid-2018, the opportunity to verify the UGS measurement results with the re-calibration method utilizing the propellant consumption of a larger maneuver also on that satellite will be performed later this year. As Fig. 7 shows (red lines), there was a larger difference between BKM vs. UGS results for MSG-4 for the operational tank temperatures between 19 C to 23 C identified. For measurements at higher temperature between 21.3 C and 22.3 C (Table 1) the difference between UGS and BKM results was 1.23 %, whereas for measurements at lower temperatures (19.3 C to 20 C) the difference was 1.65 % in comparison to the amount of oxidizer and fuel at BOL. V. Re-assessment of the MSG-4 propellant consumption during LEOP The measurements performed on MSG- 4, gave results which differed from those of the book keeping and were consistently repeated in the successive measurement tests. The MSG unified propulsion system consumes the majority of its propellant during the large orbit raising maneuvers performed in LEOP. During the first days after launch, a pair of dedicated 400 N apogee motor engines is fired during four apogee passes, to achieve an opportune near-synchronous orbit, before the start of routine phase operations. The performance factors for the large Apogee Engine Firing (AEF) manoeuvres determined for MSG4 were very atypical and significantly higher than the ones seen during the LEOP of all the previous satellites of the MSG series, see Fig. 8. During the routine phase in orbit, Fig. 7 Comparison total propellant of BKM vs. UGS corrected results for MSG-3 and MSG-4 in 2016 Fig. 8 Determined apogee engine firing thrust calibration factors of MSG-fleet Meteosat-11 continued to apparently exhibit higher thrust levels than expected in periodic orbit maintenance maneuvers. At the same time, a larger discrepancy for the estimated propellant mass on board, between the bookkeeping values and the first results, derived from the propellant gauging sensor, became evident. 8

9 G. Re-evaluation of the MSG-4 BKM results After the first year in orbit, the decision was made to re-evaluate the bookkeeping results computed during the MSG-4 LEOP by making use of the UGS measurement results. During LEOP the method used to calibrate the maneuver performance was to attribute over- or underperformance equally to the deviation from nominal thrust and mass flow rate of the Liquid Apogee Engines (LAE). In fact in absence of in-situ measurements, these two sources of error cannot be discriminated and correctly weighted. Therefore the most reasonable way to optimise the error modelling is to attribute to each of these sources of error the same weight. The approach for the propellant re-assessment was instead to set the thrust factor to a fixed value of (i.e. nominal performance), and to determine only adjusted mass flow-rate factors that would then match the observed delta-v during the LEOP maneuvers, see Table 5. Manoeuvre Delta-V Isp Consumption [AEF] [m/s] [s] [kg] Table 5 MSG-4 GTO to GEO apogee maneuver history The new consumption figures computed in this way were then applied to reset the starting point for the BKM determining the propellant mass value used during Meteosat-11 routine operations. It was therefore agreed to reduce the BKM values for the overall propellant mass by 7.7 kg. Figure 9 shows the effect of the propellant mass update, visible in the determined manoeuvre calibration data during the routine phase. A clear discontinuity at the beginning of the second year in orbit shows how the reduced total mass estimated for the satellite caused then as a consequence the subsequent maneuvers to exhibit calibration factors closer to the expected values (like for Meteosat-10). A difference however still remains, suggesting that the amount of propellant mass on board continues to be most probably still over-estimated, also when using the updated bookkeeping results. A possible reason for this result can be the assumed thrust factor of for the LEOP consumption reassessment, is still too optimistic. This hypothesis can be further Fig. 9 Routine phase manoeuvre calibration data for Meteosat-11 with propellant mass update 500 days after launch (Meteosat-10 for comparison) investigated with some sensitivity analysis on the propellant consumption calculated using different thrust factors, once the first NSSK maneuver on MSG-4 will be performed. Indeed for all previous MSG satellites during LEOP the apogee firing thrust factors were all determined to lay near a value closer to

10 H. Comparison of MSG-4 UGS measurement results with the new BKM values in 2017 In 2017 several measurement campaigns were conducted on MSG-4, applying the new strategy (i.e. to gauge the propellant at higher and lower tank temperatures in order to better determine the sensor characteristics). The UGS results from August 2017 for example could be presented as function of the propellant temperature. For T = 20 C, the difference between UGS corrected measurements and revised BKM results was 7 kg less, whereas at 22 C the difference decreased further to about 3 kg less UGS measured versus estimated propellant using bookkeeping see Fig.10. Hence, the difference with respect to the BOL propellant mass loaded was further reduced to 0.33 % (T = 22 C). The exact temperature, at which the UGS measurements needs to be corrected to, will be derived from recalibrating the results against the propellant consumption at the first MSG-4 NSSK maneuver, planned for June Fig. 10 MSG-4 UGS results in August 2017 in comparison to updated BKM results VI. Summary In addition to the BKM the innovative gauging sensors on the MSG-fleet provide an information about the remaining propellant stored in the four tanks of the unified propulsion system. The GSU/UGS measurement results, obtained on Meteosat-8/-9/-10 were always in line with the values of the BKM considering the accuracy of both gauging methods. The gauging results on Meteosat-11, instead, deviated significantly from the BKM values, which led to detailed analyses and investigation of the propellant consumption during LEOP and in particular of the assumptions done in the BKM and of the sensor behaviour, in particular of its three-stage correction scheme. Although the correction algorithm considers already the temperature of the fluid, it was observed that the measured amount of oxidizer and fuel varied at different tank temperatures. In 2016 a new strategy was applied, to gauge the propellant at the higher and lower part of the tank temperature cycling pattern. In this way, the individual sensor sensitivity to the temperature could be estimated and the measurement results corrected to certain temperatures for comparison to the BKM values in other words: the measured amount of propellant was a function of the propellant temperature and the higher the temperature the more propellant was gauged. The following step was to identify the correct gauging temperature, which the BKM results should have compared to. The first assessment, to take the results at 20 C as a baseline for comparison purposes, was chosen for safety operational reasons, even though it was expected it would not have been the conclusive one. On MSG-3 the same new strategy was applied in Furthermore, the consumed propellant during the yearly NSSK maneuver was used to re-calibrate the sensor gauging results. For this new technique propellant gauging campaigns were conducted before and after the maneuver at higher and lower tank temperatures in order to determine the temperature variation of each sensor. 10

11 At corrected 22.7 C the UGS calculated difference (deltas) for pre- and post NSSK propellant was the one best matching the BKM values. This was the equivalent gauging temperature at which the individual sensor variations were compensated for all 4 UGS and the measured consumption during NSSK for both components (i.e kg MON and 7.7 kg MMH), was best in line with the BKM values. These findings may have confirmed, that the assumption for taking measurement results at 20 C for comparison to BKM values was too conservative. By taking the corrected results at 22.7 C into account, a further measurement campaign in October 2016 on MSG-3 showed, that the difference compared to the BKM value was reduced even further to less than 0.08 % with respect to the BOL propellant mass loaded. For MSG-4, which had no NSSK maneuver so far (status 2017) further analysis of the LEOP propellant consumption was made. With some ad hoc simulations TAS could demonstrate, that by assuming nominal thrust performance for the LEOP maneuvers the overall propellant could be revised and the BKM values for the overall propellant mass should be reduced by 7.7 kg as of September The new strategy applied in 2016 that considers the temperature variation in the measurements from each sensor, allowed to have closer results of the UGS measurements to the BKM values in comparison to the deviation observed in 2015 at the first measurement campaigns in GEO. Therefore, the difference in the total propellant amount between BKM calculations and UGS measurements in 2017 with respect to the BOL propellant mass loaded could be reduced further from 1.23 % to 0.33 %. VII. Conclusion In order to specify the gauging sensor temperature dependency as function of the tank filling height, measurement campaigns have been conducted shortly before and after the yearly NSSK maneuver, which typically require the consumption of a large amount of propellant and therefore make it possible to have available measurements over a wide propellant mass range in a short time. Additionally extensive calibration campaigns have been performed to fully characterize the actual in flight temperature dependency of the UGS for the oxidizer and fuel used. The combination of several UGS measurement campaigns allowed to obtain higher confidence on the residual propellant calculation and to improve the knowledge on the propellant budget during the mission and therefore on the minimum mass of MON and MMH required at EOL to ensure a safe re-orbiting. With the new measurement strategy the sensor behaviour at different temperatures could be characterised in space by the operator. By using this novel method, the understanding of the gauging sensors characteristics on-board has now improved, and the characterization of the non-modelled and not yet compensated temperature dependencies has now been fully achieved. A better consistency of the results between BKM and the UGS has been obtained as outcome of these analyses, which further confirms the applicability of the method. As a consequence the BKM can be more reliably used on other satellites of the same series where different gauging instruments are available (e.g. Meteosat-8/-9) and where specific known limitations in the operational range (guard-region) are present. The effort spent in characterizing the behaviour of the UGS at different thermal conditions allowed furthermore to gain additional confidence that, when taking those thermal effects correctly into account, the obtained measurement accuracy of the sensor can actually be significantly better than the specification value. This innovative characterization of the sensor could only be performed in orbit and is therefore beyond the normal limitations of the qualification tests that the manufacturers can conduct on ground (e.g. non vacuum condition, use of the safe IPA instead of MON and MMH, on-ground calibration performed at ambient temperature close to 20 C). The advantage of the operational use of these instruments allows a more refined characterization of their behaviour across different temperature levels and in turn the operational results acquired lead to a reduction of the uncertainty margins on the measurements and to a potential release of fuel reserves for the extension of MSG-3/-4 operational lifetime. References 1 Murolo, F. and Bihr, C., Pili, P., Klinc, M., Brandt, R. The Ultrasonic Gauging Sensors: Results of an Innovative Spacecraft Propellant Measurement Method, SpaceOps 2014 Conference, Pasadena, California (U.S.A.), (AIAA ) 2 Hufenbach, B., et al. Comparative Assessment of Gauging Systems and Description of a Liquid Level Gauging Concept for a Spin Stabilized Spacecraft, Proceedings of the Second European Spacecraft Propulsion Conference,, May 1997, Noordwijk, the Netherlands.ESA SP-398. Paris: European Space Agency, 1997, pp Moog Inc. Ultrasonic Gauging Sensor, Brochure 2014, Moog Bradford, De Wijper 26, 4726 TG Heerle (NB), The Netherlands, 11