A Method for Estimating Costs and Benefits Of Space Assembly and Servicing By Astronauts and Robots Lloyd R. Purves NASA Goddard Space Flight Center Greenbelt, MD 20771 301-286-4207 lloyd.r.purves. [email protected] Abstract - One aspect of designing future space missions is when to use either astronauts or robots, and whether there is to determine whether Space Assembly and Servicing (SAS) some productive way to use the two together. is useful and, if so, what combination of robots and astronauts provides the most effective means of This paper presents the results of an initial study to address accomplishing it. Certain aspects of these choices, such as these questions by using a quantitative model of the the societal value of developing the means for humans to associated costs and benefits. Because this initial effort was live in space, do not lend themselves to quantification. limited, it uses simple assumptions. Despite its simplicity, However, other SAS costs and benefits can be quantified in the model appears capable of deriving some important SAS a manner that can help select the most cost-effective SAS guidelines. Should sufficiently precise answers be desired to approach. make credible decisions about specific missions, it appears that a similar approach could be used, but with more Any space facility, whether it is assembled and serviced or complex and realistic representations of costs and benefits. not, entails an eventual replacement cost due to wear and obsolescence. Servicing can reduce this cost by limiting replacement to only failed or obsolete components. 2. BACKGROUND However, servicing systems, such as space robots, have their own logistics cost, and astronauts can have even greater logistics requirements. On the other hand, humans can be To provide some background and basis for the quantitative more capable than robots at performing dexterous and model, this paper will first summarize the: unstructured tasks, which can reduce logistics costs by allowing a reduction in mass of replacement components. Definition of SAS Overall, the cost-effectiveness of astronaut SAS depends on Rationale for SAS its efficiency; and, if astronauts have to be wholly justified Options for accomplishing SAS by their servicing usefulness, then the serviced space facility History of SAS has to be large enough to fully occupy them. Definition of Space Assembly and Servicing TABLE OFCONTENTS 1. INTRODUCTION Conceptually, space assembly and servicing are fairly simple 2. BACKGROUND and quite similar. Assembly consists of attaching together 3. APPROACH in space the various components that make up some space 4. EXAMPLES facility. Servicing largely consists of replacing certain 5. CONCLUSIONS components on an existing space facility. Replacing a given 6. REFERENCES component generally means detaching it and then attaching 7. BIOGRAPHY its replacement. Thus any system for attaching and detaching components in space can potentially provide I.INTRODUCTION space assembly and servicing. Significant challenges have been associated with Space Rationale for Space Assembly and Servicing Assembly and Servicing (SAS) since almost the beginning of the space age. However, these challenges have been The rationale for space assembly and servicing is likewise more associated with the cost effectiveness of SAS than its basically simple. If a space facility is desired that is too technology. With respect to SAS cost effectiveness, some large or too heavy for one launch vehicle then it has to be questions that have long been considered important are Z assembled in space. Potentially, space assembly could also and has been tested as an autonomous robot (i.e. be justified if it were effectively less expensive than some computer controlled), but the risks of this mode are sufficiently complex set of deployment mechanisms. considered greater than its labor saving value, so it not known to have been used operationally. In its tele- Space servicing is potentially important because all space operated mode the RMS has been used extensively on components in time lose value because of failure, many Shuttle missions, particularly the HST servicing degradation, or obsolescence. If, as usually happens, some missions. components on a space facility lose value faster than others, then it can be more economical to replace just these . Combinations of any of the above. SAS on the ISS is a components rather than the whole facility. good example as it employs a combination of options 3, 4 and 5. Options for Space Assembly and Servicing History of Space Assembly and Servicing Space assembly and servicing starts to get complex when one looks at the options, all of which must be considered in In the early years of the space program the main objective order to design the most cost effective space mission. Some was to successfully carry out groundbreaking missions that of the major options include: were quite limited compared to those of today and for which SAS was not applicable. The most important consequence 1, Avoid space assembly and servicing altogether by of the launch of Sputnik in 1957 was the immense boost it making the facility small enough and replacing all of it gave the USSR in its international standing. Given the cold when, for whatever reasons, its utility drops below some war competition between the US and the USSR, this created threshold. This is in fact how most space missions have a comparably immense challenge for the US. When in 1961 been and are still designed. the USSR also became the first country to put a human, Yuri Gargarin, into space, the geo-political challenge for the US . If possible, achieve the capability of a large space became even greater. facility by flying a set of smaller facilities in formation, as is proposed for some future missions such as the The consequence of these USSR "space spectaculars" was to Terrestrial Planet Finder (TPF), which is envisioned as set the US on the course of the immensely expensive Apollo a set of formation flying telescopes acting together as an program to land the first humans of the moon. Basically interferometer. With this approach, assembly does not Apollo was a throwaway system that was neither assembled require physical attachment; nor does servicing require nor serviced in space. However, space rendezvous and detachment and reattachment, as long as the docking was used two times on each lunar landing mission components to be replaced are defined as the individual to connect and reconnect the command and landing facilities flying in formation. modules. It is interesting to note that serious consideration was given to first assembling a space station in LEO and . Use rendezvous and docking (along with undocking and then using it to assemble a lunar vehicle in orbit, but this departure) to provide the assembly and disassembly approach was eventually rejected as taking too long functions. This was the means for the assembly of certain modules of the International Space Station By the time Apollo's goal was successfully accomplished in (ISS). The Russian Control and Service modules were 1969, the US had surpassed the accomplishments of the joined using automated rendezvous and docking, and USSR in space by such an extent that the space race had the US node was attached to ISS using a manually effectively come to an end. This left the US with a manned controlled Shuttle rendezvous and docking. space infrastructure for which there was no need important enough tojustify its cost. However, the US did not have the . Use Astronauts. A good example is the Hubble Space option of wholly abandoning a human space effort, in part Telescope (HST) on which astronauts have performed because the USSR continued to slowly advance its own the vast majority of servicing tasks. While the HST more modest human space program directed toward a space station in low earth orbit. It should be noted that these servicing tasks were performed using Extra-Vehicular Activity (EVA), astronauts can also use Intra-Vehicular USSR Soyuz space stations used space rendezvous and Activity (IVA) for SAS. docking for assembly, and astronauts for servicing. . Use Robots. Examples of this are more complex to It was at this point that the issue of space assembly and define as the Remote Manipulator System (RMS) on the servicing became important in the US. It was decided that Shuttle is sometimes described as a robot arm and the cost of a US human space program could be better sometimes as a crane. It is probably most accurately justified if astronauts could support other space efforts, such described as tele-operated (i.e. human controlled) as by launching, and possibly assembling and servicing the manipulator of fairly large objects. It can be operated communication and observation satellites used for military, 3 scientific, environmental and commercial applications. 3. APPROACH Consequently, and at considerable cost, the US Space The approach for evaluating the cost of various options for Shuttle was developed to support SAS by providing such Space Assembly and Servicing is based on the following capabilities as rendezvous and docking, an airlock and space simplified assumptions suits for EVA, the RMS to manipulate large objects, and an overall ability to support more than a half dozen astronauts 1. A space facility has an objective of some value that it for over a week of activity in space. The Shuttle was also accomplishes over its mission lifetime. intended to lower the cost of all (SAS and non-SAS) space missions, by being mostly reusable and by providing a large . During the course of mission components fail or payload volume and mass capacity degrade at constant average rate. Also, during the course of the mission technology improves, which Despite the setback of the total loss in 1986 of the Space causes components to become obsolete at a constant Shuttle Challenger and all of its crew, the Shuttle has used average rate. its capabilities over the last decade to successfully support on the order of 10 SAS missions for the Hubble Space 3. After its mission is accomplished the facility is replaced Telescope (HST) and the International Space Station (ISS). by some follow-on facility that has more capabilities Yet, the outcome has been that HST servicing is due to be and commensurately greater objectives. terminated in about 2004 after about 3 more missions, and, outside of the ISS, no other existing or planned space 4. The average cost of space flight equipment is mission intends to employ SAS. proportional to its mass. The associated launch and operations costs are also assumed to be proportional to Basically, and partly as a result of the Challenger disaster, the mass of the space flight hardware. Naturally items current thinking is that the cost and constraints of making like detectors have a much greater cost per unit mass the Shuttle sufficiently safe for astronauts have priced it out than, say, aluminum structure. So what this statement is of the reach of space missions that do not specifically need effectively saying is that the proportion of these astronauts. There also appears to be a general consensus different cost-per-unit-mass items stays relatively constant from one mission to the next. that SAS, despite it promise, is not cost effective, at least outside of the ISS program. . Therefore, whatever approach allows a given mission to The are multiple reasons why the extensive, successful, and be accomplished with the least mass of space flight continuing SAS activities on the ISS program are not hardware will provide the lowest cost solution. considered proof that SAS would be cost-effective on other missions. A primary one was that the ISS program was not based on providing a more cost effective way of 4. EXAMPLES accomplishing things that other missions could do. In fact, like the Apollo moon program, a major impetus for the ISS was political, specifically a perceived need to provide anon- How these assumptions could work out in practice can be military space project for the Russian space establishment, illustrated by using the example of a space telescope similar which was largely unemployed after the collapse of the to HST. For purposes of comparison HST has a mass of USSR and might otherwise have sold it skills to wealthy 12,000 kg, a Low Earth Orbit (LEO), and an expected nations that were antagonistic to the US. Another reason mission life of 20 years. HST is expected to remain was that there was no realistic choice but to use SAS on the completely functional for the approximately 3-year period ISS. Being too large for any single LV the ISS has to be between Shuttle servicing missions. Because HST is assembled. Being too expensive to replace, it also has to be statistically expected to have component failures during serviced. these periods, it has enough redundancy that anticipated failures should not degrade its performance. HST actually This brings us to the current situation where little becomes more capable with servicing when some HST quantitative analysis exists to indicate how different SAS instrumentation is replaced with improved versions. approaches (including not doing SAS) could affect mission costs. It is an initial and simplified approach to the kind of For purposes of illustration, let us consider a HST-like analysis that is presented below. telescope [1] called the Reference Space Telescope (RST) and assume that it has a mass of 10,000 kg and needs an average of one component replaced per year per 1000 kg of mass, due to failure, degradation or upgrading. This comes out to 10component replacements per year. Therefore, over its 20-year lifetime some 200 new components will be supplied. At the end of 20 years, we assume its basic q architecture (for example aperture of the primary mirror) is serviceable RST's will therefore produce the same LSR of so obsolete that total replacement is better than continued 30,000, but will also entail the development of 30,000 kg of servicing. spaceflight hardware, which gives a FOM of 1LSR/kg. The concept that the number of annual component Scenario 2 (SAS using EVA from the STS) replacements is proportional to mass is important and has different ramifications for different SAS approaches. This scenario basically reflects the HST mission. A key Basically what is being assumed is that the number of assumption is that, with astronaut EVA servicing, the components on a space facility is proportional to its mass, average mass of a module containing a RST replacement and that a constant percentage of the total number of component is 25 kg. Therefore, with 200 components components needs to be replaced each year. However, the replaced over 20 years, the RST gets 5000 kg of new mass of the replacement component depends on the modules over its lifetime. This means the RST program has servicing scenario. For example, the HST has had a number to develop 15,000 kg of spaceflight hardware to get the 20- of gyro failures, most of which were due to the failure of a year lifetime and the LSR of 30,000. tiny wire whose mass is measured in grams. However, EVA servicing does not replace the wire. Instead, a module If we assume that the RST is serviced by 6 dedicated containing two whole gyros and having a mass of some tens Shuttle missions (one every three years for 18 years) and of kg is replaced. Conceivably, if the Shuttle had the right that the human space program covers all of the astronaut equipment on board, the astronaut could go back into the related servicing costs in including Shuttle launches, then Shuttle and just replace the wire, thereby essentially the cost for the RST program stays proportional to just the eliminating the need to use any mass for this particular 15,000 kg of hardware. servicing operation. This approach yields a FOM of 2 LSR/kg, which is twice as Let us define the benefit of the RST as having a scientific good as the one for the non-serviceable RST; however the return of 1000 in its first year. For this study the figure of costs to the human space program are ignored. 1000 is arbitrary, but it could be related or at least proportional to the number of peer-reviewed papers that Scenario 3 (SAS using EVA from a Space Station) result from the first year of RST operations. Let us also assume that 20 years of equipment upgrades leaves it twice Using the Shuttle, it is difficult to separate EVA cost from as capable at the end of its life as it was at the beginning. STS launch costs. To get a clearer picture of what astronaut Assuming it improves at a roughly constant rate due to EVA costs might be, let us assume a Reference Space equipment upgrades, it has an average annual scientific Station (RSS), which is similar to the current ISS, but return of 1500 over its lifetime, which results in a Lifetime optimized for servicing the RST. Scientific Return (LSR) of 30,000. The actual ISS (circa 2001) has an approximate mass of 100 Given that RST cost is proportional its effective mass, and metric tons and three permanent crewmembers. To simplify that its value can be measured in LSR units, then the most calculations we will assume a RSS of 100 metric tons, which cost effective RST mission is the one with the highest is 100,000 kg, and a crew of two, who together require LSR/kg, which will be defined as its Figure Of Merit 20,000 kg of logistics annually. Assuming the RSS is in the (FOM). same orbit as the ISS, this amount of annual logistics comes to about two STS missions per year. Thus, over a 20-year Described below are 12 SAS scenarios selected to illustrate period, the RSS has a total mass requirement of 500,000 kg. major options and major ways of changing the FOM. All are summarized in Table 1. The RSS will be designed so that only one astronaut will be required to maintain the RSS and the other can be dedicated Scenario I (No SAS) to RST servicing or other tasks. The RSS will also include a space tug to bring one or more co-orbiting RST's over for The baseline or reference mission will be one that produces servicing. For this scenario we take the RSS as a provided an LSR of 30,000 without SAS. For purposes of illustration, facility and assume that the RST only pays for astronaut let us assume that the RST can be given enough additional logistics to cover the time spent servicing the RST. redundancy and margin to be reliably expected to last 10 years instead of 3, but at the cost of increasing its mass 50%, We will furthermore assume that each RST module i.e. the non-serviceable RST will have a mass of 15,000 kg. replacement takes an average of one day of RSS astronaut At the end of 10 years it will be replaced by an improved time, which covers not only the EVA time, but also training, RST, which also has a mass of 15,000 kg and a 10-year etc. Therefore, replacing 200 modules represent 200 work design life, but a LSR of 2000 due to technology days, or about l-astronaut work year, assuming they have improvements over the 10-year interval. The two non- weekends off. Therefore, to have 1year of RSS astronaut timefortheRSTwouldrequiretheRSTprogramtopayfor replacements over the 20-year life require 400 hours or 20,000kgofRSSlogisticssuppliesT. hemeantshatthe about a fifth of a year of astronaut time, which at 20,000 kg RSTprogramwouldhavetopayfor10,000kgofRST, of astronaut logistics per year imposes a 4,000 kg mass 5,000kgofreplacememntodulesa,nd20,000kgofRSS requirement on the RST program for the astronaut time. astronaluotgisticsfo,ratotalof35,00k0gtogettheLSRof The total mass requirement for this RST servicing option is 30,000. 15,000 kg (10,000 for the RST, 1,000 for the modules, and 4,000 for the astronaut time), which gives a FOM of 2 TheFOMistherefor0e.86LSR/kgw,hichisworsethanfor LSR/kg non-serviceaRbSleT. This gives as good a FOM as shuttle servicing with free Scenario 4 (RSS IVA servicing of the RST) astronaut EVA. However, it still requires the human space program to cover both the development cost of the 100,000 Let us next look at a SAS scenario where we try to kg RSS and its logistics costs for the period of time that the maximize astronaut efficiency and effectiveness. One RST is not being serviced. promising way of accomplishing this will be to let the astronaut perform IVA rather than EVA, which should result Scenario 5 (SAR servicing of the RST) in more dexterous and faster serving. Let us next look at what could happen if the RST were robot Therefore, in this scenario, the RSS space tug brings the serviced. 10,000 kg RST to the RSS for servicing, and an RSS mounted robot will perform the EVA activity of removing Because robots themselves are complex subsystems that can and replacing the RST modules. After removal, the module fail we will add two identical Supervised Autonomy Robots would be transferred inside the RSS through an airlock (SAR's) to the RST, each of which is designed to be where IVA astronaut activity would be used to replace just serviced by the other. Let us assume that this robot the failed component within the module. Afterwards the servicing system and a docking port increase the RST mass repaired module would be passed back through the RSS by 50%, i.e. from 10,000 to 15,000 kg. We assume that the airlock and be replaced on the RST by the RSS robot. launch vehicle performs the rendezvous and docking. Consistent with this assumption of improving astronaut We still assume 1component failure per year per 1,000 kg, efficiency, the robot will be a Supervised Autonomy Robot so the robot serviceable RST will need 300 new components (SAR). This means that it will basically be pre-programmed over 20 years. However, as described above, a SAR is less but will have a variety of sensors to detect anomalies. In the dexterous than an EVA astronaut; therefore, the average event of any detected problem the SAR will automatically mass of a replaced module will increase to 50 kg. This stop and wait for a ground operator to decide what to do means that the 15,000 kg robot serviceable RST will need about the situation. Even with no detected anomalies, the 15,000 kg of replacement modules over its 20 year life. SAR would also stop at appropriate intervals to allow Assuming the same 6 servicing missions as for the STS ground operators to check and confirm status. Should the serviced RST (one every three years), each servicing LV robot itself need servicing, the support astronaut who is will bring up filly 50-kg modules, or 2500 kg of modules. responsible for RSS operation would perform an EVA. Based on the above assumptions, a robot serviceable RST We will make the not unreasonable assumption that the RSS requires the same mass of HW (30,000 kg) as Scenario 1 EVA robot is like the SRMS on the ISS and therefore not as (two non-serviceable RST's) and produces the same amount dexterous as an EVA astronaut. The consequence will be of science. If a more detailed and realistic cost-benefit that the average mass of the robot replaceable module analysis came to the same conclusion, the expected decision containing the failed or obsolete components will be 50 kg, would be to stay with the better understood non-serviceable i.e. twice the mass of an EVA replaced modular component. approach. However this mass is not relevant here because we are assuming that, with IVA servicing, the average mass of the Scenario 6 (DTR servicing of the RST) component to be replaced within the 50-kg module is only 5 kg. Finally, let us assume that effective use can be made of the technology for a Dexterous Tele-operated Robot (DTR). Therefore the 200 components the RST needs over it 20- DTR's, or their technological equivalents, are now being year life have a total mass of 1000 kg. Let us also assume used successfully to allow surgery to be performed by a that the astronaut is 4 times more efficient doing IVA rather remotely located surgeon, so their dexterity is equivalent to than EVA and can replace a component in two hours rather that of an IVA astronaut. than one 8-hour workday. Now the 200 component Thetimedelayforperforminsguchsurgerhyastobeno Scenario 7 (1VA Servicing of lO0 RST's) morethanabou0t.1sec,whichstillallowsadoctortobe locatedhalfwayaroundtheworld,i.e.20,000kmdistant. A first look at the previous scenarios would indicate that ThisapproacchouldbeaccomplisheindLEOusinga there is no mass dependency, since all of the replacement groundoperatosrinceLEOisonlyafewhundrekdilometers rates and servicing costs are defined to vary linearly with abougtroundw,hichisnotsignificanctomparetdo20,000 mass. However, some other effects come into play, which km. Howeverth, eGEOTDRSSlinkcouldnotbeused serve to create important mass dependencies. sinceitisintroducemsorethan70,000kmtothelengthof thesignaplath. Tokeepacontinuouasndlowtimedelay Looking first at the effect of reducing RST mass, there is an communicatiolinnskbetweeangroundoperatoarndaLEO obvious limit when the RST mass begins to approach the DTRwouldrequireeithearlargenumbeorfgroundstations oruseofaLEOcommunicatiosnastellitceonstellatiolinke mass of a replacement module. While more detailed study IRIDIUM. would be needed to determine the real cross over point, let us assume that the limit is 100 times the average mass of a replacement module. What this implies are minimum RST Toputsomenumberbsehindthisoptionl,etusassumtheat masses of: theDTRresidesin itsownRST-equivalent-mfaacsislity, 5000 kg for robot servicing, whichwillalsoneeditsownduasl ervicinrgobotssinceitis 2500 kg for astronaut EVA servicing acomplefxacilitythatwillhavetohavemodulereplacement 500 kg for IVA servicing duetoequipmenfatiluresandupgradesT.hereforew,e assigtnheDTRfacilityamassequivalentottheScenar5io SARserviceRdST.Thismeanisthasamassof15,00k0g If we further assumed that half of the RST mass is made up of non-serviceable items such as structure and cable andneeds15,00k0gofreplacememnotduleosvera20-year harnesses, this suggests there should be a minimum of about timeframet;hereforiet represenats30,000kgcost.We 50 modules to make a serviceable RST worthwhile. alsoassumtheeDTRfacilityisabletorendezvoaunsddock withtheRST. As the RST increases in mass one consequence that comes into play is that it is able to carry a greater portion of the Inthisscenarioth,eservicingrobotpasseasmodulefrom total cost of the RSS. Recall that optimum IVA use of the theRSTovertoa DTR,whichisprobablyenclosetdo RSS astronauts required only one fifth of a year of work prevenltossof objectsthatmightcomeloose.Wewill over 20 years for RST servicing, in other words only 1% of assumteheDTRis onlyhalfasdexterouassanIVA their time. Thus, if RST servicing were the only task, a two- astronauItn.otherwordso,ntheaveragitereplaceas10kg astronaut crew could service one hundred 10,000-kg RST's, componennto,ta5kgcomponenTt.hereforthee200RST a single massive RST with a mass of 1,000,000 kg, or any componentotsbereplaceodver20yearsamountot 2,000 combination in between that came to one million kg. kg. Let us look at the consequences of having 100 RST's and Wealsoassumtheateachcomponernetplacemetanktesthe having the RST program carry the full cost of the RSS. At DTR4 hoursratherthantwohoursneedebdythe IVA the reference replacement rate of 1component per 1000 kg astronaut, so a total of 800 hours of work is needed over 20 per year, 20 years of use of 1,000,000 kg of RST's will years. There are 8760 hours in a year, and even though a require 20,000 module replacements. ground controlled DTR can presumably work around the clock, we will assume that about half of the time it is Using the 5 kg average module mass for IVA servicing, the involved in its own servicing so only 4000 hours a year of RST program will be responsible for 100,000 kg of are available for tasks like RST servicing. Therefore over modules, which, added to the 1,000,000 kg of RST and 20 years it would have 80,000 hours for such tasks, and the 500,000 kg for the RSS, comes to a total of 1,600,000 kg. 800 hours needed by the RST represent 1% of its total capability. Therefore the RST program is responsible for It would be expected that the 100 RST's would have a paying for 300 kg, which is 1% of the 30,000 kg cost of the variety of designs so that groups will be optimized for DTR system. different wavelength ranges. A given group of RST's could be used to look at multiple targets simultaneously or to Thus, this solution requires the RST program to pay for a function as a formation flying interferometer. However, we 10,000 kg RST, 2,000 kg of replaceable modules, and 300 assume that each RST still has an average LSR of 30,000 so kg of DTR facility needs, which is a total of 12,300 kg of that their overall scientific value increases with their mass mass. This approach gives a FOM of almost 2.44 LSR/kg, and they will produce a total LSR of 3,000,000. which is not only the best so far, but also has the RST program covering its full share of the DTR costs. Using these assumptions, their FOM of merit is still only 1.88; however, now the RST program is paying the full cost of the RSS development and logistics. Scenario 9 (SAR assembly and servicing of the Scenario 8 (1VA Assembly and Servicing of a MST) Massive Space Telescope) If the assembly is to be done by less dexterous supervised- For this scenario, let us assume that instead of 100 RST's, autonomy robots (SAR), then we will assume a more crude their 1,000,000 kg of mass is used to create a single larger construction technique that requires larger and less mass telescope, which will be designated the Massive Space efficient assembly pieces, with the result that the mass of the Telescope (MST). We will also assume that the mass of a MST doubles to 2 million kg. We now need 40,000 telescope is proportional to the area of its primary mirror. replacement modules of 50 kg, which adds another 2 million Thus the mass of 100 RST's will be the same as a MST with kg, for atotal mass requirement of 4 million kg. 100 times the mirror area, or equivalently 10 times the mirror diameter. The LSR is still 30,000,000 so the FOM is 7.5 MSR/kg, which is significantly worse than using the combination of Although not directly relevant to this cost-benefit analysis, it astronaut IVA and DTR EVA. is interesting to note the result of trying to assign reasonable numbers to RST and MST PM diameters. Current space Scenario 10 (DTR assembly and servicing of the telescope technology is represented by the Next Generation MST) Space Telescope (NGST) [2], which has the objective of providing at least 25 sq. meter of primary mirror (PM) area The only difference between this and Scenario 8 is that with a total observatory mass of about 5,000 kg. Keeping a ground operators will now control the DTR's so that we can constant ratio of PM area to mass means that a 10,000 kg avoid all astronaut related costs. Since a DTR in LEO could RST using NGST technology might has have a 50 sq m PM. be just as effective using ground operators as astronauts, the To keep the numbers simple let us assume that foreseeable total assembled mass of MST will still be 1,000,000 kg, but technology improvements increase the ratio of PM area to the total mass of the replacement modules doubles to 200,00 mass by 50% so that that a 10,000 kg RST would have a 75 kg, because the DTR will do the module replacement rather sq. m PM, which means about a 10 m diameter PM. At this than the more dexterous IVA astronaut. technology level, the mass of 100 RST's could alternatively yield a MST with a PM diameter of 100 m. Recalling that the DTR facility only needed 1% of its time to service a 10,000 kg RST, the MST can be serviced with However, another effect comes into play here, and it is that 100% of its time. Therefore the total mass requirement of increasing the diameter of a filled aperture telescope the servicing system is 30,000 kg. This leads to a total mass provides certain unique advantages in resolution, FOV, requirement for this scenario of 1,230,000 kg, which for the sensitivity and speed over a number of smaller aperture LSR of 30,000,000 gives a FOM of 24.39 LSR/kg. telescopes that have the same total PM area. Let us say that this value increase is proportional to the diameter of single Scenario 11 (SAR servicing of MST m L2 orbit) telescope relative to the diameter of each of an equivalent mass of separate telescopes. Thus the MST is assumed to Let us next look at how the cost models are affected by have a LSR of 30,000,000 changes in observatory orbit. An orbit around the Sun-Earth L2 Point is attractive for space telescopes. This L2 point is However, a MST is obviously way too large and too heavy 1.5 million km from Earth (four times as far as the Moon) in to fit in single LV, so some form of space assembly becomes the anti-sun direction. Due to the approximate balancing of necessary. Astronaut EVA could be used to construct a 100 gravitational and centrifugal forces in this region, a satellite m diameter space telescope; however, as stated above, this is can orbit around this L2 point not the most effective way of using astronauts. Therefore we will assume that astronaut IVA is used to control DTR's, A good case can be made that a space observatory in orbit which (along with SAR's where appropriate) perform the about the Sun-Earth L2 point will have at least twice the actual EVA assembly of the MST. scientific productivity of a same mass telescope in LEO. This is because: For simplification, we ignore the time that the actual - The view of many astronomical targets will be construction takes and just look at the 20-year operation blocked in LEO by the earth for half of the time, benefit and servicing cost. The mass for which the RST - Temperature variations will be harder to control in program is responsible is the same 1,600,000 kg as was LEO required for the 100 RST's in Scenario 7, but with the LSR - A LEO observatory will have to operate off stored increasing by 10, the FOM goes up by the same factor to power when sunlight is blocked by the earth 18.75. - An L2 observatory can be fairly easily shaded from Sun and Earth light, making it much easier to cool - which is needed for IR observations, q next decade, and it can only send about 8,000 kg to L2. Its On the basis of the current ratios of space hardware cost to capacity would have to be about doubled just to put the launch cost, choosing L2 over LEO looks cost effective if 15,000 kg non-serviceable (Scenario 1) variant of the RST science return is doubled. A representative overall into L2 orbit. development cost for spacecraft is $100K per kg, and launch costs to LEO are around $10K per kg. An LV can only Because an L2 observatory will be about 1.5 million km launch 1/3 of the mass to L2 that it can to LEO, so launch from earth, there will be about a 10 second round trip time costs to L2 are effectively $30K per kg. Thus 10,000 kg in delay for getting feedback from any ground generated LEO effectively costs $1.1B, and the same mass costs $1.3B command due to the finite speed of light. This effectively in L2. Because this is a relatively modest cost increase, we eliminates the ground operated DTR option described in will for purposes of simplification ignore it for now and Scenario 10, which needs at least about a 10 Hz control simply assume that going to L2 doubles the LSR without loop. Thus, one effect of distant orbits is to make astronauts affecting the cost per unit mass. more valuable. This is because any situation in a distant orbit requiring quick human feedback, such as dexterous For the sake of simplicity we will only look at the effect of manipulation or possibly contingency operations, can only putting the MST into L2 orbit because it already has the best be accomplished by having an astronaut in close proximity. FOM. Another consideration that makes the MST appropriate for L2 is that it will be launched in pieces by However, a SAR can still be effectively used in L2 as it multiple LV's and then assembled in orbit. This avoids the basically autonomous, except for anomalies. Assuming that limit on mass that available and even projected launch anomalies do not significantly affect the overall mission, a vehicles can place into more distant orbits. The Delta IV SAR assembled and serviced MST in LW will double the Heavy [3], which has not even yet flown, is the most capable Scenario 9 FOM to a value of 15LRS/kg. launch vehicle that can be expected to be available over the Table 1 - Summary of Figures of Merit for Space Assembly and Servicing Scenarios E A O I- E _ ,'_ o E c o _= O O "6 t_ E O3 O_C0 L7.._ 1 Non-serviceable RST 15,000 15,000 1 15,000 0 30,000 30,000 1.0I 2 STS Astronaut EVA at no 10,000 25 200 5,000 0 15,000 30,000! 2.0( cost to RST 3 Astronaut EVA with RST 10,000 25 200 5,000 20,000 35,000 30,000 0.86 support of RSS logistics 4 Astronaut IVA with RST 10,000 5 200 1,000 4,000 15,000 30,00C 2.00 support of RSS logistics 5 Supervised Autonomous 15,000 50 300 15,000 0 30,000 30,00C 1.00 Robot (SAR) 6 Dexterous Tele-operated 10,000 10 200 2,000 300 12,300 30,000 2.44 Robot (DTR) 7 RSS IVA servicing of 100 1,000,000 5 20,000 100,000 500,000 1,600,00(] 3,000,00C 1.88 RST's 8 RSS IVA assembly and 1,000,000 5 20,000 100,000 500,000 1,600,000 30,000,00C 18.75 servicing of MST 9 SAR assembly and 2,000,00(] 50 40,000 2,000,000 0 4,000,000 30,000,000 7.50 servicing of MST 10 DTR assembly and 1,000,00(] 10 20,000 200,000 30,00(] 1,230,000 30,000,000 24.39 servicing of MST 11 SAR assembly and 2,000,00(] 50 40,000 2,000,000 0 4,000,000 60,000,000 15.00 servicing of MST at L2 12 RSS IVA assembly and 1,000,000 5 20,000 100,000 500,00C 1,600,00(] 60,000,000 37.50 servicing of MST at L2 [ Scenario 12 (RSS IVA servicing of MST at L2 a. Perform planned astronaut efforts in a well Like the previous SAR scenario, this simply doubles the equipped IVA environment where work can be equivalent (Scenario 8) LEO FOM from 18.75 to 37.5 done quickly and with dexterity LSR/kg. b. Use astronaut controlled DTR's for EVA work requiring dexterity Thus, the end result is that the best FOM comes from putting c. Use SAR's for EVA work that does not require an MST in L2 orbit and using astronaut IVA to service dexterity modules, white the EVA for assembly and module exchange d. Use astronauts in remote orbits where the ground is performed by some optimum mix of SAR's and DTR's. controlled alternative become less effective due to communication delays e. Provide astronaut EVA capability, but limit its use 5. CONCLUSIONS to handling anomalies that cannot be otherwise dealt with f. Minimizing the size, staffing, and logistics of the As noted, the above cost-benefit model is the result of an space station which provides the astronaut support initial and limited investigation. It is, therefore, very for SAS simplified and not useful in its present form for determining g. The program requiring the SAS (the RST program the most cost effective way to carry out any specific mission. in this paper) can save money by having other However, its results appear consistent with actual programs, such as the human space program, defray experience in SAS, and so this model seems to provide a astronaut costs. framework for using sufficiently complex and realistic representations of SAS costs and benefits to get useful 5. Any of the following will contribute to increasing the results for specific missions. cost effectiveness of robot SAS However, even in its current simplified form the results a. Use robot EVA so the astronaut can work more appear realistic enough to derive useful guidelines to follow efficiently in an IVA environment. in the conceptual design of future missions. Below are some b. Have astronauts or ground operators sufficiently guidelines for each of the five distinct SAS options defined near the robot to perform dexterous tasks using at the beginning of the paper. high bandwidth teleoperated control (i.e. DTR's). c. To minimize human workload, use SAWs when . A non-serviceable space facility looks fairly cost- dexterity is not needed effective relative to any form of SAS, until the target d. Use SAR's in distant orbits when establishing facility gets too large for a single launch vehicle. Then astronauts there is not cost effective. An example some form of assembly becomes necessary, and the could be a 30 m diameter telescope in L2, which assembly system can be used to provide the servicing might be too small to cost justify a L2 astronaut capability at little additional cost. facility, but too large for a single LV. . If a set of smaller (and single launch vehicle 6. REFERENCES compatible) space facilities can provide approximately the same mission benefits of a larger, equivalent-mass [1] http://hubble'nasa'g°v/servicing-missi°ns/ facility that would have to be assembled, then the set of [2] http://ngst.gsfc.nasa.gov/ smaller facilities should be more cost effective [3]http://www.boeing.com/defense- space/space/delta/guides.htm . A rendezvous and docking capability is needed for any form of SAS, but it alone does not seem sufficient for 7. BIOGRAPHY assembling and servicing space observatories, or other space facilities that have such a specialized geometry Lloyd Purves is an engineer at NASA's Goddard Space and design. However, rendezvous and docking does Flight Center (GSFC). His work at GSFC has been in the seem to be effective for the initial stages of building areas of satellite data analysis; computer aided design, something like a space station, which can consist of a manufacturing, modeling and simulation; and robotics. He number of similar large modules with similar headed the project to develop the Robot Operated Materials connections. Processing System (ROMPS), a GAS-Hitchhiker payload, which flew successfully on STS-64 in 1995. He is currently 4. Astronauts are expensive and so they must be used as supporting the Next Generation Space Telescope (NGST) effectively as possible to maximize their cost Project. He has a BSEE from Union College and has done effectiveness. Any combination of the following will graduate work in finite element analysis. increase astronaut effectiveness for SAS