Since then, other variant designs involving building an artificial structure — or a series of structures — to encompass a star have been proposed in exploratory engineering or described in science fiction under the name "Dyson sphere". These later proposals have not been limited to solar power stations — many involve habitation or industrial elements. Most fictional depictions describe a solid shell of matter enclosing a star (see diagram at right), which is considered the least plausible variant of the idea (see below).
Dyson is credited with being the first to formalize the concept of the Dyson sphere in his 1959 paper "Search for Artificial Stellar Sources of Infra-Red Radiation", published in the journal Science. However, Dyson was inspired by the mention of the concept in the 1937 science fiction novel Star Maker, by Olaf Stapledon, and possibly by the works of J. D. Bernal and Raymond Z. Gallun who seem to have explored similar concepts in their work.
While it is believed that some of the design variants commonly described — specifically those based on the Dyson shell — are impractical, if not physically impossible, design variants of the sphere based on orbiting satellites or solar sails do not require any major theoretical breakthroughs in our basic scientific understanding for their construction. Deployment of spacecraft and satellites using photovoltaics might be seen as the first small steps towards building a Dyson swarm (see below for differences between these sub-types). However, creating and deploying energy gathering spacecraft and satellites in the numbers needed to create a solar system sized integrated energy gathering system are well beyond our present-day industrial needs or capabilities. It is also likely that there are unforeseen industrial scaling difficulties in such a construction project, and that our current understanding of industrial automation is insufficient to build the self-maintaining systems needed for the sphere's upkeep.
The variant closest to Dyson's original conception is the "Dyson swarm". It consists of a large number of independent constructs (usually solar power satellites and space habitats) orbiting in a dense formation around the star. This approach to the construction of a Dyson sphere has several advantages: the components making it up could range widely in individual size and design, and such a sphere could be constructed incrementally over a long period of time. Various forms of wireless energy transfer could be used in order to transfer energy between constructs.
Such a swarm is not without drawbacks. The nature of orbital mechanics would make the arrangement of the orbits of the swarm extremely complex. The simplest such arrangement is the Dyson ring in which all such structures share the same orbit. More complex patterns with more rings would intercept more of the star's output, but would result in some constructs eclipsing others periodically when their orbits overlap. Another potential problem is the increasing loss of orbital stability as adding more orbiting constructs increases the probability of orbital perturbations of other constructs.
As noted below, such a cloud of collectors would alter the light emitted by the star system, but as can be seen here, it is unlikely that such an alteration would be complete, and that some of the star's natural light would still be present in the system's emitted spectrum.
A spherical shell Dyson sphere in our solar system with a radius of one astronomical unit, so that the interior surface would receive the same amount of sunlight as Earth does per solid angle, would have a surface area of at least 2.72x1017 km2, or around 550 million times the surface area of the Earth. This would intercept the full 4x1026 watts of the Sun's output; other variant designs would intercept less, but the shell variant represents the maximum possible energy captured for our solar system at this point of the Sun's evolution. To put this figure in perspective, it is approximately 3.3x1013 times the total energy consumption of humanity in 1998 which was 1.2x1013 W.
There are several serious theoretical difficulties with the solid shell variant of the Dyson sphere:
Such a shell would have no net gravitational interaction with its englobed sun (see Shell theorem), and could drift in relation to the central star. If such movements went uncorrected, they could eventually result in a collision between the sphere and the star — most likely with disastrous results. Such structures would need either some form of propulsion to counteract any drift, or some way to repel the surface of the sphere away from the star.
For the same reason, such a shell would have no net gravitational interaction with anything else inside it. The contents of any biosphere placed on the inner surface of a Dyson shell would not be attracted to the sphere's surface and would simply fall into the star. It has been proposed that a biosphere could be contained between two concentric spheres, placed on the interior of a rotating sphere (in which case, the force of artificial "gravity" is perpendicular to the axis of rotation, causing all matter placed on the interior of the sphere to pool around the equator, effectively rendering the sphere a Niven ring for purposes of habitation, but still fully effective as a radiant energy collector) or placed on the outside of the sphere where it would be held in place by the star's gravity. In such cases, some form of illumination would have to be devised, or the sphere made at least partly transparent, as the star's light would otherwise be completely hidden.
If assuming a radius of one AU, then the compressive strength of the material forming the sphere would have to be immense. Any arbitrarily selected point on the surface of the sphere can be viewed as being under the pressure of the base of a dome 1 AU in height under the Sun's gravity at that distance. Indeed it can be viewed as being at the base of an infinite number of arbitrarily selected domes, but as much of the force from any one arbitrary dome is counteracted by those of another, the net force on that point is immense, but finite. No known or theorized material is strong enough to withstand this pressure, and form a rigid, static sphere around a star. It has been proposed by Paul Birch (in relation to smaller "Supra-Jupiter" constructions around a large planet rather than a star) that it may be possible to support a Dyson shell by dynamic means similar to those used in a space fountain. Masses traveling in circular tracks on the inside of the sphere, at velocities significantly greater than orbital velocity, would press outwards due to centrifugal force. For a Dyson shell of 1 AU radius around a star with the same mass as the Sun, mass traveling ten times orbital velocity (300 km/s) would support 99 (a=v2/r) times its own mass in additional shell structure. The arrangement of such tracks suffers from the same difficulties as arranging the orbits of a Dyson swarm, and it is unclear how much energy would be consumed ensuring the velocity of the masses was maintained.
Also if assuming a radius of one AU, then there may not be sufficient building material in the Solar system to construct a Dyson shell. Dyson's original estimate was that there was enough material in the Solar system for a 1 AU shell 3 meters thick, but this included hydrogen and helium which are unlikely to be much use as building material, although additional building material might be manufactured if elements such as hydrogen and helium could be transmuted into heavier elements through nuclear fusion. Anders Sandberg estimates that there is 1.82 kg of easily usable building material in the Solar system, enough for a 1 AU shell with a mass of 600 kg/m²—about 8–20 cm thick depending on the density of the material. This includes the cores of the gas giants, which may be hard to access; the inner planets alone provide only 11.79 kg, enough for a 1 AU shell with a mass of just 42 kg/m².
The shell would be vulnerable to impacts from interstellar bodies, such as comets, meteors, and so forth.
Lastly, the shell would be vulnerable to the material in interstellar space that is currently being deflected by the Sun's Bow shock. The Heliosphere, and any protection it theoretically provides, would cease to exist.
A third type of Dyson sphere is the "Dyson bubble". It would be similar to a Dyson swarm, composed of many independent constructs (usually solar power satellites and space habitats) and likewise could be constructed incrementally.
Unlike the Dyson swarm, the constructs making it up are not in orbit around the star, but would be statites—satellites suspended by use of enormous light sails using radiation pressure to counteract the star's pull of gravity. Such constructs would not be in danger of collision or of eclipsing one another; they would be totally stationary with regard to the star, and independent of one another. As the ratio of radiation pressure and the force of gravity from a star are constant regardless of the distance (provided the statite has an unobstructed line-of-sight to the surface of its star), such statites could also vary their distance from their central star.
The practicality of this approach is questionable with modern material science, but cannot yet be ruled out. A statite deployed around our own sun would have to have an overall density of 0.78 grams per square meter of sail. To illustrate the low mass of the required materials, consider that the total mass of a bubble of such material 1 AU in radius would be about 2.17 kg, which is about the same mass as the asteroid Pallas.
Such a material is currently beyond our ability to produce; the lightest carbon-fiber light sail material currently produced has a density — without payload — of 3 g/m², or about five times heavier than would be needed to construct a solar statite.
However, there has been some speculation about the creation of ultra light carbon nanotube meshes through molecular manufacturing techniques whose density would be below 0.1 g/m². If production of such materials on an industrial scale is feasible, and such materials could be used in light sails, the average sail density with rigging might be kept to 0.3 g/m² (a "spin stabilized" light sail requires minimal additional mass in rigging). If such a sail could be constructed at this areal density, a space habitat the size of the L5 Society's proposed O'Neill cylinder—500 km², with room for over 1 million inhabitants, massing 3 tons—could be supported by a circular light sail 3,000 km in diameter, with a combined sail/habitat mass of 5.4 kg. For comparison, this is just slightly smaller than the diameter of Jupiter's moon Europa (although the sail is a flat disc, not a sphere), or the distance between San Francisco and Kansas City. Such a structure would, however, have a mass quite a lot less than many asteroids. While the construction of such a massive inhabitable statite would be a gigantic undertaking, and the required material science behind it is as yet uncertain, its technical challenges are slight compared to other engineering feats and required materials proposed in other Dyson sphere variants.
In theory, if enough statites were created and deployed around their star, they would compose a non-rigid version of the Dyson shell. Such a shell would not suffer from the drawbacks of massive compressive pressure, nor are the mass requirements of such a shell as high as the rigid form. Such a shell would, however, have the same optical and thermal properties as the rigid form, and would be detected by searchers in a similar fashion (see below).
The Ringworld, or Niven ring, could be considered a particular kind of Dyson sphere. Larry Niven, who first developed the concept, described it as "an intermediate step between Dyson Spheres and planets". The ringworld could perhaps be described as a slice of a Dyson Sphere (taken through its equator), spun for artificial gravity, and used mainly for habitation as opposed to energy collection. Like the Dyson Shell, the Niven ring is inherently unstable without active measures keeping it in position with regards to its central star – a fact recognized by Larry Niven and addressed in the sequels to his novel on the concept, Ringworld.
A bubbleworld is an artificial construct that consists of a shell of living space around a sphere of hydrogen gas. The shell contains air, people, houses, furniture, etc. It was invented to answer the question "what is the largest space colony that can be built". However, most of the volume is not inhabited and there is no power source.
Theoretically, any gas giant could be enclosed in a solid shell; at a certain radius the surface gravity would be terrestrial, and energy could be provided by tapping the thermal energy of the planet. This concept is explored peripherally in the novel Accelerando (and the short story Curator which is incorporated into the novel as a chapter) by Charles Stross when Saturn is converted into a human habitable world.
Stellar engines are a class of hypothetical megastructures, whose purpose is to extract useful energy from a star, sometimes for specific purposes. For example, Matrioshka brains extract energy for purposes of computation; Shkadov thrusters extract energy for purposes of propulsion. Some of the proposed stellar engine designs are based on the Dyson sphere.
A black hole could be the power source instead of a star in order to increase energy to matter conversion efficiency. A black hole would also be smaller than a star. This would decrease communication distances which would be important for computer based societies as those described above.
The existence of such a system of collectors would alter the light emitted from the star system. Collectors would absorb, and re-radiate, energy from the star. The wavelength(s) of radiation emitted by the collectors would be determined by the emission spectra of the substances making them up, and the temperature of the collectors. Since it seems most likely that these collectors would be made up of heavy elements not normally found in the emission spectra of their central star — or at least not radiating light at such relatively "low" energies as compared to that which they would be emitting as energetic free nuclei in the stellar atmosphere — there would be atypical wavelengths of light for the star's spectral type in the light spectrum emitted by the star system. If the percentage of the star's output thus filtered or transformed by this absorption and re-radiation was significant, it could be detected at interstellar distances.
Given the amount of energy available per square meter at a distance of 1 AU from the Sun, it is possible to calculate that most known substances would be re-radiating energy in the infrared part of the electromagnetic spectrum. Thus, a Dyson Sphere, constructed by life forms not dissimilar to humans, who dwelled in proximity to a Sun like star, made with materials similar to those available to humans, would most likely cause an increase in the amount of infrared radiation in the star system's emitted spectrum. Hence, Dyson selected the title "Search for Artificial Stellar Sources of Infrared Radiation" for his published paper.
SETI has adopted these assumptions in their search, looking for such "infrared heavy" spectra from solar analogs. As of 2005 Fermilab has an ongoing survey for such spectra by analyzing data from the Infrared Astronomical Satellite (IRAS).