Initial Phase AOCS Operation of Infrared Astronomy Satellite “AKARI” Yasuhiro Kawakatsu, Tatsuaki Hashimoto, Nobutaka Bando, Takaji Kato, Takao Nakagawa (JAXA/ISAS),

Ken Maeda (NTSpace), Hiroshi Iida, Masatoshi Matsuoka (NAS), Takafumi Ohnishi (FJT)

Abstract Reported in this paper is the initial phase operation of the attitude and orbit control system (AOCS) of Japanese satellite “AKARI”. AKARI is the first Japanese satellite dedicated to the infrared astronomy. AKARI was successfully launched by M-V rocket from Kagoshima Space center on February 22, 2006. Just after the launch, AKARI faced to a serious problem. It was found that something interferes with the fields of view of two (out of two) sun sensors. The two sensors were out of use in the subsequent AOCS operation. The anomaly prevents us from carrying out the pre-planned initial phase operation, and constrains us to reconfigure the AOCS architecture. The experiences in this recovery operation are mainly reported in the paper. In addi-tion, unexpected continuous orbit rising was observed in the operation. The cause of the phenomenon and its investigation process are also reported.

「あかり」の姿勢軌道制御系初期運用 川勝康弘、橋本樹明、坂東信尚、加藤隆二、中川貴雄（JAXA/ISAS）、 前田健（NT スペース）、飯田浩、松岡正敏（NAS）、大西隆史（FJT）

摘要 日本初の赤外線天文衛星「あかり」は、2006 年 2 月 22 日に M-V ロケット 8 号機により打ち上げられ、現在も順調に観測運用

を続けている。しかし打上直後から、搭載していた２つの太陽センサ（NSAS、TFSS）の視野が隠されるという異常事態に遭遇

し、初期運用を大きく変更する必要があったとともに、搭載プログラムも変更することとなった。本発表ではその状況を報告す

る。また、初期運用中、当初は想定されていなかった定常的な軌道上昇が観測され、軌道制御計画等に影響を与えた。その

推定要因、影響についても報告する。

1. Introduction

AKARI (Fig. 1) is the first Japanese satellite dedicated to the infrared astronomy. It was successfully launched by M-V rocket from Kagoshima Space Center on February 22, 2006 (Japan Standard Time, JST). The mission of AKARI is to make an all-sky survey at infrared wavelength, with much better sensitivity, spatial resolution and wider wavelength coverage than its predecessor, IRAS (Infrared Astronomical Satellite, launched in 1983 by the United Kingdom, the United States, and the Netherlands.) The observation data acquired by AKARI is expected to contribute to various sci-ence areas, especially to understand the formation and evolu-tion of galaxies, and to inquire into the formation process of stars and planetary systems.

Fig. 1 Infrared Astronomy Satellite “AKARI”

AKARI has a telescope with its primary mirror’s diameter of 68.5cm and two focal plane instruments, FIS (Far Infrared Scanner) and IRC (Infrared Scanner). In order to suppress harmful thermal radiation radiated from the telescope itself, the telescope and scientific instruments are stored in the cry-ostat and cooled down to about 6K by liquid Helium and mechanical coolers (Fig. 2). The operational life of AKARI ends when the liquid Helium runs out. The expected lifetime

was estimated to be about 550 days from launch.

Fig. 2 Telescope and Scientific Instruments in Cryostat

The cryostat had an aperture lid which was ejected when the observation got prepared on orbit. The role of the aper-ture lid on the ground was to keep a tight vacuum in the cry-ostat, and the role on orbit was to prevent stray light inflow into the cryostat. And once the aperture lid has been ejected, the light inflow from the sun and the earth into the cryostat is fatal to AKARI, since it leads to the nearly immediate evaporation of the liquid Helium. Fig. 3 depicts the direction in which the light inflow from the sun and the earth is pro-hibited. Basically, the sun and the earth must not be in front of the telescope, which condition strongly restricts the atti-tude of AKARI after the aperture lid was ejected.

87

The earth prohibitedin this direction.

71

73.8deg.

88.3deg.The sun prohibitedin this direction.

+z

Ranges are symmetricaround z axis.

Fig. 3 Restriction as to the Light Inflow into the Telescope

Fig. 4 shows the definition of satellite body coordinate system of AKARI. z axis is in the direction of the bore sight of the telescope, y axis is in the direction where the solar panel is attached, and x axis is perpendicular to z and y axes.

Fig. 4 Definition of Satellite Body Coordinate System

The observation orbit of AKARI is a sun synchronous or-bit. The altitude and the inclination are approximately 700km and 98.2 deg. respectively. AKARI goes around the Earth above the twilight zone, in other words, the direction of the sun is approximately perpendicular to the orbit plane. Fig. 5 shows the usual (i.e. not in the special operation such as an orbit maneuver or pointing observation) attitude of AKARI. Basically, +y direction always points to the sun, and +z direc-tion, i.e. the telescope bore sight, points to not exactly but approximately in the opposite direction from the earth. On this attitude, the telescope scans the sky along a great circle in an orbit plane, and combined with the orbit perturbation due to the earth’s oblateness, an all-sky survey is completed by a half year observation.

Fig. 5 On Orbit Attitude of AKARI

Just after the launch, AKARI faced to a serious problem. The first phenomenon detected was the data anomaly of NSAS (Non-spin-type Solar Aspect Sensor). It was found just 15 minutes after the launch, and naturally it was sup-posed as a failure of NSAS itself. However, in the following

operation, anomalies are found one after another in the in-struments loaded on +y side panel. The anomalies found are,

- No input of the sun light on TFSS (Two-dimensional Fine Sun Sensor).

- Power generation of SAP (Solar Array Panel) approxi-mately 150W lower than predicted.

- Temperature difference between two symmetrically lo-cated SHNT (Shunt regulators).

- Temperature of some instruments loaded on +y side panel higher than predicted.

- Beam pattern of SANT (S-band Antenna) different from that measured on the ground.

No longer, it was not supposed as failures of each instru-ments, but supposed as a systematic anomaly affecting many instruments on +y side panel. And the phenomena are called “+y side panel problem” (Fig. 6). Later, the phenomena were analyzed in detail, and it has been concluded that a part of +y side panel is covered with something. However, the object that covers the panel is not specified in spite of elaborate investigation.

Fig. 6 Instruments Affected by “+y Side Panel Problem”

As to AOCS (Attitude and Orbit Control System), two (out of two) sun sensors are affected by the problem and they are out of use in the subsequent AOCS operation (Fig. 7). As other many satellites, sun sensors were planned to play im-portant roles in AKARI’s AOCS operation. Firstly, they were to be used for sun acquisition at the very initial phase. Sec-ondly, as mentioned earlier, to prevent the sun light inflow to the telescope, the attitude of AKARI is strongly restricted after the aperture lid was ejected. Sun sensors were to be used to watch the sun direction to keep the sun in appropriate direction even when the satellite falls into the safe mode. The absence of the two sun sensors prevents us from carrying out the pre-planned operation. However, the difficulty had to be overcome to achieve the mission of AKARI.

Fig. 6 Instruments Affected by “+y Side Panel Problem”

Reported in this paper is the AOCS operation of AKARI under the circumstance mentioned above. It covers the

AOCS operation from the launch to the end of the initial phase, i.e. the aperture lid ejection. The topics included are, the operation on the day of launch and the orbit rising prob-lem (another problem arose in the initial phase operation).

2. Operation on the Day of Launch

AKARI was launched by M-V rocket from Kagoshima Space center at 6:28 a.m. on February 22, 2006 (JST). Re-ported in this section is the operation on the day of launch. Fig. 8 shows the visible paths from the ground stations on the day of launch. The main ground station for the operation of AKARI is Uchinoura space center. Add to that, four more stations (Santiago, Perth, Kirna, and Svalbard) were also assigned for the critical operation on the day of launch. The basic role of the four additional stations was to watch the status of AKARI, and no nominal command operations were planned for these stations. The first command operation was planned to be performed at the Uchinoura path, which was planned at about 11 hours from the launch. Therefore, the critical operations in very initial phase were planned to be performed automatically onboard. A feature to be pointed out about the paths schedule is that, there were no visible paths for about 5 hours after the second Santiago path (SNT #2). Therefore, it was planned to complete the SAP deployment and three axes attitude establishment before SNT #2, so that we could release AKARI in stable condition at SNT #2. In particular, the completions of SAP deployment and the sun acquisition were necessary to secure sufficient power supply to survive the succeeding invisible 5 hours.

USC #1 USC #2

SNT #1 SNT #2 SNT #3

PRT #1 PRT #2 PRT #3

KRN #1 KRN #2 KRN #3

SVD #1 SVD #2 SVD #3 SVD #4 SVD #5Svalbard

Kirna

Perth

Santiago

Uchinoura

Fig. 8 Visible Paths from the Ground Stations

Listed on Table 1 are the planned operation items to be performed up to three axes establishment. AKARI was going to be separated from M-V rocket with the spin rate of 7.5 rpm around z axis. At the first apogee passage, a perigee up maneuver of 25m/s was to be performed. The maneuver was set to cope with the case that the perigee altitude of the injec-tion orbit was extremely low. Therefore, as long as the injec-tion by the rocket was performed properly, the maneuver was not necessary. After the maneuver, the spin rate was to be suppressed, and the SAP was to be deployed. Then the three axes attitude was to be established by firstly searching the sun by NSAS and point +y direction to the sun, and secondly searching the earth by CES (Conical scan Earth Sensor) and point –z direction to the earth. Before the launch, these op-eration items were planned to be completed before SNT #2.

Table 1 Operations up to Three Axes Attitude Establishment 1. Separation 2. Perigee up maneuver 3. Spin down and rate dump 4. SAP deployment 5. Sun acquisition 6. Earth acquisition 7. Attitude maintenance

In actual operation, because of the “+y side panel problem”, NSAS didn’t work properly, and the sequence didn’t proceed as it had been planned. Depicted in the followings are what had been occurred in actual operation.

The first opportunity after the launch that we could know the status of NSAS was the deployment of the fairing of the rocket. It was before the separation and the event was visible from the launch site. The moment that the fairing was de-ployed, the sun light was input to NSAS and the telemetry shows that NSAS output proper predicted signals. The first visible path after the separation, that is the first Perth path (PRT #1), was about 15 minutes after the launch. The data anomaly of NSAS was firstly detected at PRT #1. The anom-aly was not the kind of “perfect blind”, but that of “disor-dered data”. The next visible path was the first Santiago path (SNT #1), about 45 minutes after the launch. Originally, the perigee up maneuver was planned to be performed at the beginning of this path, and it was predicted to take time to establish the communication link at this path. Additionally, the information about the anomaly was limited at this mo-ment. Then, we decided to monitor the status of AKARI, and didn’t uplink any emergency command at this path. As a result of monitoring at SNT #1, two facts became clear. Firstly, the perigee up maneuver was interrupted at 6 seconds, which was originally planned to continue for 300 seconds. It was supposed to be automatically aborted due to NSAS data anomaly. Secondly, the spin rate was suppressed to the ap-propriate level, but not in nominal manner. The spin rate was originally planned to be measured by detecting the zero-cross of NSAS data, however, it was supposed to be automatically switched to the measurement by GAS (Geomagnetic Aspect Sensor) data. Based on these facts, it was concluded that NSAS was out of use. And, it was expected that SAP de-ployment would be performed as planned since the spin rate was suppressed sufficiently, however, it was expected that the sun acquisition would not complete by the next visible path, SNT #2, since NSAS was out of use. As mentioned before, the completion of the sun acquisition at SNT #2 was necessary to secure sufficient power supply for the survival.

At AOS (Acquisition Of Signal) of SNT #2, as had been expected, AKARI continued searching the sun by spinning around z axis. The objective of the operation of SNT #2 was to stop the spin so that +y direction pointed at the appropriate direction, that is the direction of the sun. Instead of NSAS, which was out of use, mainly monitored was the output cur-rent (equivalent to the power supply) of SAP. The output current changes sinusoidally as AKARI spins around z axis, and when the current was at the peak, a real time command was sent to stop the spin. Of course, it was rather course to measure the angle between +y direction and the sun based on the power supply, however, it was clear that sufficient power supply was secured. And that’s enough in this situation. AKARI was saved.

Though it was not relied on, the data of NSAS was also monitored as a reference. It seemed that the sun is approach-ing to the +y direction, that is the signal of the sun seemed to approach the center of NSAS view. However, the signal of the sun suddenly disappeared just before the arrival at the target, and NSAS didn’t trigger to stop the spin. Fig. 9 shows the reproduced data of NSAS output at the sun acquisition. The plots indicate the signal of the sun in the view of NSAS. The center of the view, that is the point (1024, 1024), coin-cides with the +y direction, the horizontal and vertical axes

express the direction around z and x axes respectively. The data indicate that the left half of the view seems to generate correct data, since the signal of the sun moved from the left to the right as AKARI span around z axis, and additionally, the signal of the sun seems to approach the center of the view, that is +y direction. On the other hand, the right half of the view seems to generate confused data, and the point of the signal seems not to express the correct direction of the sun. And the point is that, the very center of the view, where the direction of the sun should converge, is in “confused data region”. After all, NSAS caught the signal of the sun in the left half of its view, and the attitude was controlled correctly to point +y direction to the sun. However, the signal of the sun was lost just before the arrival at the target, and the search process was restarted. It was what repeated many times, and it was what observed in SNT #2.

Fig. 9 NSAS Data at Sun Acquisition

As a result of the “manual” sun acquisition procedure, the minimum necessary condition at the end of SNT #2 was satisfied. At LOS (Loss of Signal) of SNT #2, +y direction was approximately pointing the sun, and AKARI was spin-ning around y axis in the orbit rate. Only the body rate was controlled based on IRU (Inertial Reference Unit). The power supply, the thermal input, and the communication link to the ground were all in stable conditions.

At the next visible path, the first Kirna path (KRN #1), a real time command was sent to start the earth acquisition. It was performed normally, and AKARI established the normal three axes attitude (see Fig. 5). After the monitoring at sev-eral paths at Kirna and Svalbard, the main operation on the day of launch was performed at the first and the second Uchinoura path (USC #1 and USC #2). AOCS RAM (Ran-dom Access Memory) program was loaded and run, RW (Reaction Wheel) was run up, and the attitude was started to be controlled by RW instead of thrusters.

At the end of the operation on the day of launch, +y direc-tion was approximately pointing the sun and the body rate around x and z axes are controlled to be zero based on IRU. The body rate around y axis is controlled to be the orbit rate based on CES and IRU. The attitude was controlled by RW. The power supply, the thermal input, and the communication link to the ground were all in stable conditions.

3. Orbit Rising Problem

From right after the launch, it was observed that the semi major axis of the orbit of AKARI gradually increased day by

day. The rate of increase was approximately 190m per day (Fig. 10).

7080

7085

7090

Mar.09 Mar.19 Mar.29 Apr.08 Apr.18

Fig. 10 Increase of the Semi Major Axis of AKARI (before aperture lid ejection)

The phenomenon didn’t influence the operation instantly. However, in the long term, it was expected to cause the orbit to deviate from the sun synchronous condition, and lead to the prolongation of the time of eclipse to unacceptable level. Fig. 11 shows the prediction of the deviation from the sun synchronous condition (which is expressed as the transition of the angle between the right ascension of the sun and the ascending node of the orbit) and the time of eclipse, assum-ing that the semi major axis of the orbit would increase in the same rate. Additionally, the increase of the semi major axis (in other words, the increase of the orbit energy) is unusual phenomenon, which suggests that AKARI was ejecting something. Therefore, the phenomenon was called “orbit rising problem” and regarded as the problem to be solved by the ejection of the aperture lid.

265

270

275

280

285

0

5

10

15

20

2006 Mar. 2006 Jun. 2006 Sep. 2006 Dec. 2007 Mar. 2007 Jun. 2007 Sep.

[deg.]

[min.]

time of eclipse

sun– orbitSun : right ascension of the sunOrbit : right ascension of the ascending node of the orbit

Fig. 11 Prediction of Deviation from the Sun Synchronous Condition and Time of Eclipse (assuming 190m/day for in-

creasing rate of semi major axis)

To investigate the phenomenon, firstly, the acceleration acted on AKARI was estimated by way of the orbit determi-nation. The existence of the continuous body fixed small acceleration was assumed, and the acceleration was esti-mated along with the orbit from the ground observation data. Fig. 12 shows the O-C profile (from March 21 to 23) of the several cases of the determined orbits. Fig. 12(a) shows the

results of the two cases. “Cross” indicate the case where no acceleration was estimated, and “solid circle” indicate the case where acceleration was estimated. In the latter case, the existence of x, y, z components of the acceleration (in the satellite body coordinate system) are assumed and estimated. It is apparent that O-C of the latter case is much smaller than that of the former case, which indicates that continuous body fixed acceleration is likely to be acting on AKARI. Next, the compositions of the acceleration are investigated in detail. Fig. 12(b) shows the results of the two cases. “Cross” indi-cate the case where the existence of x, y, z components of the acceleration are assumed and estimated, and “solid circle” indicate the case where the existence of x, y components of the acceleration are only assumed and estimated. The result shows that O-C of the latter case is smaller than that of the former case, which indicates that the acceleration in xy plane (in the satellite body coordinate system) is likely to be acting on AKARI. It is true that, ideally, even in case that z compo-nent of the acceleration doesn’t exist, the former case simply estimates z component to be zero, and two cases yields es-sentially the same results. However, in practice, z component (i.e. radial component) of the acceleration is inherently in-sensitive to the orbit, and hard to be estimated accurately by way of the orbit determination. Fig. 13 shows the estimated x, y components of the acceleration. As a result of the investi-gation, the acceleration acted on AKARI was estimated at approximately 1.0×10-6m/s2.

-50

-40

-30

-20

-10

0

10

20

30

40

50

Mar. 21 Mar. 22 Mar. 23 Mar. 24

O -

C [k

m]

No acceleration estimated 3 components of acceleration estimated

(a) Estimate “No acceleration” and “3 components of acceleration”

-10

-8

-6

-4

-2

0

2

4

6

8

10

Mar. 21 Mar. 22 Mar. 23 Mar. 24

(b) Estimate “3 components” and “2 components” of acceleration

3 components of acceleration estimated 2 components of acceleration estimated

O -

C [k

m]

Fig. 12 Comparison of Orbit Determination Accuracy (O-C)

0.0E+00

5.0E-07

1.0E-06

1.5E-06

2.0E-06

Mar. 21 Mar. 22 Mar. 23 Mar. 24

Acc

eler

atio

n [m

/s2 ]

Fig. 13 Estimated Acceleration in x, y Direction

Fortunately, the phenomenon observed was the orbit “ris-ing” and not “falling”, and the candidates of its cause were limited. The phenomenon indicates that AKARI ejects something in the accelerating direction, and the candidates of the cause are RCS and the cryostat which ejects vaporized Helium. Immediately, from the telemetry of RCS, it was confirmed that there is no unexpected leak of the propellant, and RCS was deleted from the candidates list. As a result of heat exchange, the liquid Helium onboard is gradually va-porized and the gas Helium is ejected from two outlets. The two outlets are attached on the –x/–y side of AKARI, point to the opposite directions (Fig. 14). Therefore, the momentum of the ejected gas Helium is cancelled right after it is ejected. However, it is supposed that the ejected gas Helium is ther-mally exploded, and the thermal energy of the gas Helium is converted to the kinetic energy of Helium molecules. Al-though the expansion is isotropic, the molecules which move to x/y direction would collide with the satellite wall and push the satellite.

Fig. 14 Ejected Gas Helium Causes Acceleration

Assuming the following four conditions, - The proportion η (less than 1) of the ejected Helium

contribute to this effect. - The temperature of Helium at ejection is 240K. - The expansion of Helium is isotropic. - The Helium molecules collide with the satellite wall as

“perfectly elastic collision”. the acceleration in +x (or +y) direction caused by this effect is estimated as

[ ] [ ] (1) smsmg7.16.0

101.1 26⎟⎟⎠

⎞⎜⎜⎝

⎛⎟⎠⎞

⎜⎝⎛×= − ma η

where is the ejection rate of the gas Helium. (1) indi-cates that, assuming 0.6 for η and 1.7mg/s for (that is the actual ejection rate observed onboard), then, the accelera-tion caused by this effect is estimated as 1.1×10-6m/s2, which approximately coincides with the value estimated by way of the orbit determination. The value 0.6 for η seems somewhat high, however, it was concluded that the accelera-tion was caused by this effect.

mm

It was expected that, after the aperture lid is ejected, the ejection rate of the gas Helium was expected to be reduced to

less than one third of the value above (i.e. the value when the aperture lid was still attached). This means that, after the aperture lid ejection, the rate of semi major axis increase, as well as the acceleration, is expected to be reduced to less than one third of the value observed then. Fig. 15 shows the prediction of the deviation from the sun synchronous condi-tion (which is expressed as the transition of the angle be-tween the right ascension of the sun and the ascending node of the orbit) and the time of eclipse, assuming that the semi major axis of the orbit would increase by the rate 45m/day. The time of eclipse in June 2007 shortens compared with the case of the rate 190m/day in Fig. 11, and it is regarded as acceptable level. As a result, it was concluded that the orbit rising problem would not be the obstacle for the aperture lid ejection operation.

265

270

275

280

285

0

5

10

15

20

2006 Mar. 2006 Jun. 2006 Sep. 2006 Dec. 2007 Mar. 2007 Jun. 2007 Sep.

Fig. 15 Prediction of Deviation from the Sun Synchronous Condition and Time of Eclipse (assuming 45m/day for in-

creasing rate of semi major axis)

After the aperture lid ejection, it was observed that the rate of semi major axis increase was reduced to the level we had predicted, and it was proved that the investigation was cor-rect (Fig. 16).

7080

7085

7090

7095

7100

Mar.09 Mar.29 Apr.18 May.08 May.28 Jun.17 Jul.07 Jul.27

Fig. 16 Profile of Semi Major Axis after Aperture Lid Ejection

It was fortunate that the observed phenomenon was the orbit “rising”. If the observed phenomenon had been the orbit “falling”, it would have been more difficult to specify

the cause of the phenomenon. There would have been more candidates of the cause, which include the unexpected large air drag related with “+y side panel problem”.

4. Conclusion

AKARI is the first Japanese satellite dedicated to the in-frared astronomy, which was successfully launched by M-V rocket from Kagoshima Space center on February 22, 2006. Reported in this paper is the initial phase AOCS operation of AKARI. Just after the launch, AKARI faced to a serious problem, “+y side panel problem”. The fields of view of two (out of two) sun sensors are interfered by something and they were out of use in the subsequent AOCS operation. As other many satellites, the sun sensors has been planned to play important roles in AKARI’s AOCS operation. Firstly, they were to be used for sun acquisition at the very initial phase. The anomaly prevents us from carrying out the pre-planned sequence. However, the crisis was overcome with the appro-priate ground support operation. Secondly, the sun sensors were to be used to watch the sun direction to keep the sun in appropriate direction even when the satellite falls into the safe mode. In order to achieve this function in the absence of the sun sensors, AOCS architecture was reconfigured so that the equivalent function is achieved by the remaining sensors. In addition, unexpected continuous orbit rising was observed in the operation. As a result of the investigation, the cause of the phenomenon was identified as the ejected gas Helium. Consequently, after one month’s delay from the preplanned schedule, the aperture lid of the cryostat has been success-fully ejected on orbit and the scientific observation has been going on.