back to page 1

Selection Criteria Approach


Having set the initial goals we want to achieve, we now want to start developing a design that best meets them, and evaluate how it compares to existing systems. One part of this is turning the vague term "best" into an objective measurement method. Complex systems such as these will have many parameters which can be used to compare alternatives. These parameters will have different units of measurement. Since different features of a design, like cost and performance, cannot be directly compared, we convert them to a unified numerical scale, and can then choose whichever option has a higher score on that scale. The conversion formulas represent a mathematical model of what is desirable to the Customer. In other words, it is an objective model of "best". This model will evolve through the conceptual design process. At this point we can only start to work on it.

Subjectivity of Criteria


The modeling and evaluation is an objective process. The choice of what parameters are to be measured and their relative importance, however, is subjective. It derives from human needs and desires which are not set objectively. In addition, humans often do not know what they would want because they are uninformed on a subject. In our present program, the ultimate customer is civilization as a whole. Most people do not have enough interest or information to know what parameters they would value. Therefore the immediate customer must act as a proxy who expresses their preferences for them. The immediate customers would be the ones in direct contact with the designers. For now that would be contributors to a research foundation, and the program designers themselves. At this stage they cannot educate and poll all of civilization to determine what they would want. Instead they must estimate what they would want if they were well informed and asked their opinion. As an example of this proxy method, the ultimate customers for a smartphone are the people who will eventually use them. Since you don't even know who those people will be, you cannot ask them what their preferences are. Therefore the management, marketing department, and engineering department act as a proxy for those customers and make their best effort to determine what their preferences would be.

Types of Criteria


Some criteria are a clear yes/no or pass/fail type. If the design fails to meet such a criterion, it has no value in the eyes of the customer. This type are not modeled or scored, since the score if not met would be zero. Instead they are included as well defined program requirements that must be met, and other parameters allowed to vary as needed to meet the "hard" (fixed and unchanging) requirements. Usually this type represents fundamental program goals and objectives. Other parameters, such as cost, have an incremental level of desirability - each increment of lower cost is more desirable. Such parameters can be scored on a sliding scale from desirable to undesirable.

Number of Criteria


Any parameter which has value to the end user/customer can potentially be used for design selection. Thus they can cover the full range of goals, requirements, and design features. In practice, some are more important than others, or only are relevant at certain levels of the design. A very large set of criteria is difficult to apply because you have to evaluate all of them for every design option. So the selection criteria are usually limited to the more important ones. Whichever criteria are chosen, there should be a clear description showing the relationship to customer desires and how the scoring formula is derived. We will give a couple of examples of cost criteria to show how this is done. At this point they are not final selection parameters, merely examples.

Example of Cost Criteria


We have assumed that civilization as a whole is who would be paying for the program development as a whole and benefit from the results. Thus they are the ultimate customer from the Systems Engineering standpoint. One major reason for choosing a new design over current ones is cost, which is a measure of the resources input to the program. So we will use development and operating cost as our example criteria. In both cases, lower is better.

  • Affordable Development

Existing space programs are already quite substantial, indicating they are valued by society. Worldwide, government space agency budgets total US $33 billion as of 2011. The Satellite Industry Association estimates global space industry in 2011 of US $290 billion. Industry commonly sells to government programs, so to avoid double counting we only include 30% of government budgets as unique. This leads to a total of $300 billion/year for all space-related activity. Present value converts an annual flow of funds to an equivalent single amount. The current Price/Earnings ratio of the S&P 500 stock index is 12, implying a present value for current space programs of US $3.6 trillion.

A new or modified program should not greatly exceed what people apparently are willing to spend. Therefore we take the net development cost of a new combined system divided by the present value of current space programs as a measure of affordability. Small values are good, indicating not much extra cost. Negative values are even better, indicating cost savings. Some net cost is acceptable if the benefits exceed the cost. Net development cost here is the discounted present development cost of the system, including revenue generated. Thus it is the amount of money you would need today to finance all the future development at the maximum point, after accounting for any positive income that can cover later costs.

We do not have any absolute measure to convert net development cost to an evaluation score. In the absence of one, we will assume civilization is willing to spend the same fraction of total output on space programs. Since world real GDP grows about 3% annually, we will set an increase of net development cost based on the present value multiplier of 12 as setting an equivalent time horizon of 12 years. This gives 12x3% or 36%, as a nominal value, to which we assign a midpoint score of 50%. We will assign 0% net development cost a score of 75%, and 36% decrease in cost will score 100%. A 108% cost increase will score 0%, and the scale will extend beyond 0 and 100%. The score values are arbitrary. What matters is how they relate to the scores of the other selection criteria. This implicitly defines exchange ratios between criteria, like "a 1% increase in thrust is worth $5 million in development cost".

A better scoring of development cost could be obtained later by surveying people after explaining what the program benefits would be. We can also do a sensitivity analysis on the scoring. This is calculating how the results of the conceptual design change as a result of changing the selection criteria or their scoring. If you get the same design result for a wide range of scoring criteria, the design is said to be Robust or Insensitive to changes. This is desirable. If small changes cause very different design results it is likely not an optimal design, or more work needs to be done to distinguish alternatives or reduce uncertainty.

  • Low Recurring Costs

The previous measure reflects maximum net development cost at any point in time. A good conceptual design would also have a low ongoing operating cost. The major current recurring cost for space projects is launch to Earth orbit. Total launch cost is the product of how many kg you need to launch and cost/kg of the transport system. Improvements such as stronger materials or using space resources reduces the required kg for a given project. A number of component costs make up the recurring cost/kg, but the biggest improvements will come from different design and technology used for transport to orbit.

For this example we will measure cost reduction relative to current (2012) values as the baseline. Current space hardware and propellant masses for typical missions determine the total kg required. New designs and technology result in different masses for the missions. The mass improvement ratio is then current mass/new mass. For launch cost we will assume the quoted cost of US $1566/kg for the Falcon Heavy as the baseline. For new transport systems the marginal operating cost to Low Earth Orbit after paying for development is used. The launch cost ratio is then ($1566/kg)/(new launch cost). Total cost reduction is then the product of (mass improvement ratio) x (launch cost ratio). We will assume equal percentage reduction in cost has constant customer value, and therefore convert the logarithm of total cost reduction to a score. Half the maximum potential improvement, a ratio of 148:1, has a natural log of 5 and is scored at 100%. Other cost reductions are scored at 20% x ln(cost reduction).

Program Requirements Approach


The next part of developing the design is defining requirements. Program requirements will state how the general program goals will be met in terms of measurable features, parameters, and values. They will be documented at the end of the conceptual design stage. At this point (the start of conceptual design) we can start by identifying categories of requirements under each of the program goals. The requirements analysis step will then examine and select individual requirements from them.

Improving Life on Earth


This program goal was listed first because, for the time period we can reasonably plan a program, the majority of humans will live on Earth. The quality of our life here is therefore of high importance. Historically, and in the near future, almost all of the people working on space programs, the offices and factories, and sources of materials will also be on Earth. If new methods and technologies are developed to reach challenging space mission requirements, they can feed back to the toolbox civilization has in general to work with, and thus improve life here. An example of where this has already happened is in the Systems Engineering methods developed to manage complex aerospace projects. These methods can be applied to all forms of complex projects on Earth.

To relate this general goal to particular requirements, we can look at the impact of new space technologies on quality of life measures already in place. An example would be high efficiency closed loop food production. This could lead to improved food security or lower farm wastes when applied on Earth. Thus during conceptual design we should look for methods and technologies with Earth applications, determine what quality of life impact they could have, and give preference to those with the greater potential applications and impact. A fusion rocket would then be preferred over a regolith engine because functioning fusion reactors can also help solve energy security on Earth. A device that uses raw rock as reaction mass for propulsion does not have that kind of Earth application (that we know of).

Quality of life measures include physical measures such as GDP, life expectancy, and pollution levels, as well as social measures like education, leisure time, and civil rights. One step in establishing program requirements will then be to select the appropriate measures. The next step would be to set how much of our program should have potential impact on these measures, and what level of improvement is desired.

Understanding the Earth


Understanding our home planet is mainly a scientific enterprise. It includes gathering data, developing theories and models, and then testing those ideas against reality. After you have gained a better understanding, then deciding what to do with that knowledge is a social and political issue outside this program's scope. It may lead back to new projects such as orbital sunshades to reduce temperatures on Earth, but the choice to start such projects should be based on sound knowledge. This group of requirements will address gaining that knowledge.

We want to gather data about the Earth and other planetary systems. The reason for the latter is other planets are natural experiments in what happens under other circumstances. We cannot gain knowledge by experimenting with the one planet we live on, but we can test our theories and models by looking at other plants. Gathering data has dimensions of time and space. We want to know both the current properties of planets and their environment in three dimensions, and the history of those properties across time. We can then set program requirements in terms of increased detail in both. An example would be "Map all Solar System objects to a resolution of 1 kilometer or better." A large pile of data is not very useful by itself, so sufficient resources to organize it and develop theories is also needed, after which a new cycle of data collection to test the theories would happen. Program requirements would then include the scientific staff to make use of the data and guide later cycles of data collection, and presenting the results of the accumulated knowledge in a form that educates people on choices they may need to make.

Reducing Space Hazards


This class of program requirements includes identifying hazards from space, followed by preparing for or preventing the damage from them. Known hazards include asteroids and comets, solar flares, and stellar explosions. Types of requirements can include the ability to deflect hazardous objects, and limiting damage from events like flares that cannot be prevented.

Increasing Biosphere Security


A single uncontrolled biosphere is inherently insecure from natural and human-made variations. Since nearly all of us currently depend on this one biosphere, we want to set program requirements that improve the security level. The requirements groups of understanding the Earth and reducing space hazards help in reaching this goal. This group goes further to counteracting undesirable variations and adopting the idea of backups from computer technology. Specific requirement might include providing safe storage of biological samples and testing of biohazards away from the Earth, active control of biosphere parameters, and establishing artificial biospheres in space or on other planets.

Expanding Resources


All civilizations require resources to function. This group of program requirements can include identifying scarce physical, material, and energy resources and setting quantity and cost goals for increasing their availability. Physical resources include quality living, growing, and working space in terms of dimension and environment. Material resources are all the raw matter and specialty compounds and equipment needed. Energy resources are needed in various forms for different purposes. All of these are interlinked in terms of flows of resources for different tasks.

Long Term Survival


Requirements in this area would include examining long term resource depletion. For example, continued use of nuclear fission for power would eventually deplete the Earth of Uranium and Thorium. If that were a critical part of keeping civilization operating, then eventual collapse would occur. There are alternative sources of power, so this particular example is not fatal, but it shows the idea of looking for items that would run out in the long term. The Sun also gets 1% brighter per 100 million years from stellar evolution, and the Earth permanently loses about 3 kg/s of Hydrogen from water dissociation, equivalent to 850,000 tons/year of lost water. These long term changes will eventually make the Earth uninhabitable. So program requirements can be set to either counteract depletion and changes to the Earth, or enable moving elsewhere within the timescale of the problem becoming critical.

Increased Choice


Existence on Earth under our current civilization imposes natural and human-made restrictions. Measures in this area look at lifting or eliminating these restrictions. Some examples are freedom of location - on Earth you are restricted by national governments from living anywhere you want. Another is freedom of gravity - you cannot choose to live under a different gravity level right now.

Increased Opportunity


Most of the Earth (the good parts at least) are already claimed by someone. By making new unclaimed or under-used areas accessible, that increases opportunities for people who want to start something new without first having to pay off previous owners. Measures in this area would include increased area or resources which are made available.

Design Approach


Having set up a way to measure how good the design is and establish requirement, we next need an approach to formulate the System Concept, a high level description of what the combined space system is and how it works. There is no single magic bullet (or magic rocket) that can meet all the program requirements by itself. If there were, someone would have used it by now, or at least be pursuing it seriously. Therefore we take the approach of leveraging multiple good ideas, which allows the savings to multiply together. This will result in a complex program of multiple systems, which need to be combined for best results.

To meet the cost criteria of affordable development we do not build everything at once. Instead the ideas get applied in incremental projects and systems which build on each other. This allows some return from the early parts to help pay for the later parts. The early parts are smaller in scale than what comes later, which further reduces initial development. This will result in a program which is extended in time.

Multiple Ideas


To develop our system concept we will use the following ideas:

  • Use less of or eliminate conventional rockets - They have been in use for 50 years and had a lot of engineering development and optimization. Therefore using another conventional rocket is unlikely to bring much improvement, and other projects are already attempting to do so. Instead we will try to use some of the other hundred or so transport methods and variations identified in Part 2. This gives us the possibility of greatly improving on the performance and cost limits of chemical rockets which are imposed by their chemistry.
  • Design for re-use, repair, and recycling - It should be evident that these features will reduce hardware and supply cost, and yet many launch vehicles and satellites are used once and disposed of. For human crews on, for example, the Space Station, oxygen and food supplies are similarly used once and disposed of. Some space hardware is designed for maintenance and repair, but much of it is not. Therefore we will try to incorporate these multiple use and long life features to get more use out of hardware and supplies.
  • Use the material and energy resources of space - Again, it should be evident that bringing everything from Earth is a limiting factor, and the farther you go in space, the higher the cost of doing so. Solar power is so overwhelmingly useful that it has been used by almost every space project, but other material and energy resources have not yet been exploited. Therefore we will try to design for using them to leverage what we bring from Earth.
  • Build multipurpose facilities - One-time missions, as have often been designed to date, tend to not leave anything useful behind. Thus the next mission is exactly as hard and expensive as the last one. Therefore we will try to design facilities that can be used multiple times or on a permanent basis. An example would be landing a solar array on the Moon that is used to recharge a rover vehicle quickly, and then later can also be used to power an extraction plant. This makes more sense than having a solar array attached to the rover, and then another solar array for the plant.
  • Use diverse modular designs - Monolithic, or single piece, designs require replacing the entire item if your needs change or you have an upgraded technology you can use. For long term and complex projects you are not able to predict all the changes and upgrades that might be required. Therefore we will try to use modular designs where possible to make it easier to change things. Modular designs can also start smaller and be added to in steps, reducing initial costs and the size of transport systems. It is not required to use one method to do everything. On Earth we use industrial delivery systems like pipelines to deliver large quantities of goods cheaply, and reserve more expensive and safe methods to deliver people. Specialization of this sort is acceptable when it makes cost and technical sense.

Using the above ideas does not mean using them blindly where they are not appropriate. It means incorporating them where it makes sense, in an optimized amount. Past programs have tended to not use them enough, or at all. This has led to high cost and limited performance, which our new program tries to correct.

Incremental Projects


Rather than attempt to do everything at once, we will take the approach of designing and building our space program elements in progressive increments. The increments will establish new locations and improve measurable parameters in several dimensions as they get added:

  • Working environment range - Starting with temperate Earth locations provide production, habitation, transport, and other system elements that work in the given environment. Then extend their working range to hotter, colder, wetter, drier, and higher and lower pressures. Later extend the range of working environments to space locations with the additional variables of gravity and radiation levels and increased range of temperature and pressure.
  • Time and energy range - Time has components of communication time, travel time, and stay time. Energy has components of potential and kinetic energy to reach a given location. New locations will increase the range of communications and control distance, and require longer travel and stay times. They will also require greater energy changes to reach.
  • Performance levels - These are measures like cargo capacity, industrial output, efficiency and closure, and how many humans are supported. They are designated as performance requirements for particular program elements and systems. Each location starts with a given performance set, and it is improved in stages to higher levels.

Increments which are far away in time or require parameters far beyond current experience become uncertain to design for. New technology may get developed in the interim times, and meeting untried parameter levels may be difficult. Therefore at some point it is no longer useful to plan future items in detail. Instead, the options for these items can be laid out, and a plan to develop needed technology and reduce uncertainties put in place. Therefore program engineering work is not done once and finished, but is a continuing effort. At any point in time the program will have a baseline documenting current status and future plans, but that baseline will get updated on a regular basis.

Initial Program Concept


Based on the above discussion we will describe a starting point for the program concept. We emphasize that this is only a starting point, the final concept is the endpoint of the conceptual design stage.

Program Description

Our basic concept for the program is to expand the range of human civilization to new locations while meeting the above program goals and requirements. Any kind of civilization seems to require the ability to produce food and other physical items, to provide shelter, and to move people and items from place to place. There may be other basic requirements, but we will start with these. Therefore for each new location we set up and then expand in stages the production, habitation, and transport elements. We start from existing civilization as it is, start a new location, and once it is sufficiently developed, travel to and repeat the process at the next one.

The first new locations will be easy ones on Earth, in temperate climates. New technology such as remote operations, automation, and resource extraction are first demonstrated there. Once built up, we transport new elements to more difficult locations on Earth, such as deserts, oceans, high altitude, cold, or underground. This expands the range of environments and the distances for remote control. Widening the range of conditions where people can live and work meets many of the program requirements, and the new locations should be self-supporting physically and economically.

After sufficient locations on Earth are established, the production capacity is used to build transport to orbit, and the remote operations and other technologies are further developed for space locations and the more difficult environments of vacuum, temperature, radiation, and lack of gravity.

To enable growth in new locations we design and build the following kinds of hardware:

Seed Factories

Developing a new location has always involved bringing a starter set of knowledge and tools. Historically that meant bringing animals, seeds, axes and hammers, and whatever else the technology of the time required to start building. For our future program, we want to use the best methods that modern technology allows. Our starter set should use automation and remote operation, and be able to not just make a fixed set of products, but make more equipment for itself to widen the range of outputs. So our concept is to use Seed Factories to establish production capacity at each new location. Unlike conventional factories which only produce a given set of products, a seed factory uses part of its production to make more equipment for itself. So over time it is able to make a wider range of products and use a wider range of local resources. The initial seed equipment, plus supplies and necessary components it cannot make yet, are delivered from the previous location. Over time, the production capacity will need fewer supplies and parts, and be able to make more items locally.

Besides self-expansion, the factory output is partly items needed to live and work in the particular location. In space that would include items such as parts and supplies for habitats, vehicles, mining equipment, water, and oxygen. The remainder of the output would be commercial items to trade for needed outside supplies. On Earth that could be any kind of product with a market. In orbit, an example would be building large communications satellites. Vehicles would carry copies of the seed equipment on to the next location to start growth there. Since habitats for humans are part of the production output, at first you may not have the ability to support many of them at a new location. Therefore the seed factories are designed for a mix of automated/robotic, remote control, and direct human operation to minimize the latter. The first seed factories are built in various locations on Earth. When they have sufficiently expanded they use their industrial capacity to build launch vehicles and space hardware. This is used to establish assembly and processing equipment in Earth Orbit. This evolves into a full seed factory, and from that grows to other locations in space.

Cyclic Systems

A linear system might dig up a resource to be used as fertilizer, use it once, and then it gets removed in the harvested crops and runoff. In a crowded Earth, or in space locations, where getting new supplies is hard, resources for once-through linear designs become expensive and unsustainable. Therefore our concept includes using cyclic systems which take old items and send them back to earlier production stages for repair, re-use, recycling, or reprocessing to new items. Because transport is an overhead cost, we prefer to do the cycling locally, but if it turns out to be more optimal to do the tasks at other locations we will do so.

High Efficiency Transport

New locations start with deliveries from previous locations, and will continue to need transport to deliver new equipment, supplies, and people. Once sufficient production is built up, delivery of finished products back to previous locations is needed. In addition, return of people and used items back to previous locations is needed before setting up permanent habitats and fully cycling systems. Thus transport is a necessary function, and we prefer to do that as efficiently as possible.

On Earth, machines like internal combustion engines are particularly inefficient, so we will look for more efficient replacements. In space transport, conventional rockets are even more inefficient reaching orbit than gasoline engines. Although chemical rockets are very efficient (~80%) as heat engines in themselves, when considered as part of a total transport vehicle, most of the work goes into accelerating fuel which is later burned and discarded. Therefore our concept is to replace conventional rockets as much as possible with a variety of higher efficiency transport methods. This includes several options for initial launch to orbit, and primarily electric thrusters once in orbit. Chemical rockets will not be entirely eliminated, especially at first, but are gradually replaced as the program evolves. Electric thrusters are about five to ten times more efficient in fuel use, and combined with the ability to extract fuel locally require dramatically less fuel from Earth, so we will try to use them heavily in the design. Like all rockets, chemical ones use fuel exponentially as a function of velocity. So even replacing part of their use will significantly reduce total launch mass. So our incremental approach still yields large gains even with smaller early steps.

continue to page 3