Kessler@home
Kessler@home is a beta-stage BOINC volunteer computing project. It studies the long-term stability of low Earth orbit (LEO), the region where most satellites operate. Volunteers' computers simulate possible futures of the orbital environment, and the results are combined to estimate how likely a runaway collision cascade, known as the Kessler syndrome, is under a range of future scenarios, and when such a cascade would occur if it did.[1]
The project is in a beta test, which began on 4 November 2026. Work is currently limited to beta tasks, and the project is still checking results and performance across platforms before a wider release.[2]
Kessler@home is administered and developed by Ian&Steve C., a veteran BOINC user who serves as the project administrator and project developer. The project is independent and has no named sponsor.[1]
The project's name refers to the Kessler syndrome, a scenario first described in 1978 by NASA scientists Donald J. Kessler and Burton G. Cour-Palais. The project is not yet listed among the projects in the BOINC Manager, so it must be added by its web address.[2]
Background
In 1978, Kessler and Cour-Palais published "Collision Frequency of Artificial Satellites: The Creation of a Debris Belt".[3] They argued that collisions between orbiting objects could produce new fragments faster than atmospheric drag removes them, so that the debris population could grow on its own. The term "Kessler syndrome" was coined by John Gabbard, a NORAD analyst, and became widely used after a 1982 Popular Science article.[4]
In 2001, Kessler and Phillip Anz-Meador presented a "critical number" model. It estimated the number of intact spacecraft above which the environment in a given altitude band becomes unstable.[5] A later reassessment by Hugh G. Lewis and Donald J. Kessler found that the model had identified regions near 900 km and 1400 km where the number of intact objects in the February 1999 catalog exceeded the critical number.[6]
Several space agencies run long-term evolutionary models of the debris environment, including NASA's LEGEND, ESA's DELTA, the UK's DAMAGE and JAXA's LEODEEM. A comparison study by the Inter-Agency Space Debris Coordination Committee (IADC) found that their 200-year forecasts broadly agree. Under launch rates from 2001 to 2009, with 90% of satellites and rocket bodies disposed of within 25 years and no explosions, the population of objects 10 cm and larger still grows by about 30% over 200 years. A catastrophic collision occurs every 5 to 9 years, mostly between 700 and 1,000 km altitude. The growth is slow, over decades to centuries, and is not a single sudden event.[1][7]
Method
Scenarios and realizations
Kessler@home uses two central terms:
- A scenario is a complete set of assumptions that are held fixed, such as the starting population, future launch traffic, how reliably operators dispose of their satellites, how often a small debris hit disables a satellite, how much debris is removed, and how fast the upper atmosphere thins.
- A realization is one random draw of how events unfold within a scenario. It is a single simulated future of up to 200 years.
A BOINC workunit is one scenario plus one random seed, which is a whole number that starts the random-number generator. The same seed always produces the same future, bit for bit, on any computer.[8]
What one realization computes
- Orbit propagation. About 36,000 objects 10 cm and larger are tracked individually in 5-day steps. The starting population is built from the public catalog as of October 2026. Each object moves under Earth's gravity, atmospheric drag that follows the solar cycle, and, for objects whose orbits reach high above LEO, the pull of the Sun and Moon.
- Collisions. Objects are placed in 10 km cubes, and a collision probability is computed for each pair that shares a cube. The method follows the approach used in NASA's LEGEND, the UK's DAMAGE and MIT's MOCAT-MC.[9] Thin constellation layers are treated analytically using the Kessler 1981 collision rate.[10]
- Fragmentation. Breakups follow the NASA Standard Breakup Model, which sets the number, size, speed and area-to-mass of the fragments. Fragments 10 cm and larger are added to the tracked population. Smaller fragments, from 1 to 10 cm, are counted as a density per 50 km altitude shell.[11]
- Launches and disposal. Future launches follow one of three traffic files. Satellites operate for their design life, give or take a quarter, and then attempt to dispose of themselves. Operators never react to a worsening environment, so a runaway result means "if nothing is done".
- Runaway test. A run is stopped and scored as a runaway once its tracked objects pass 2 million. Launches stop in 2076 for this test.
Scenario sweep
Eleven settings are varied across all combinations of their options. They cover launch traffic, the traffic volume, the layout of planned 1,150 km constellation shells, disposal reliability (90% to 99.9%), the disposal rule, the rate of fatal small-debris hits, debris removal (none to 50 objects a year), the amount of small debris today, solar activity, the explosion of old rocket bodies, and the CO2-driven thinning of the upper atmosphere.[1] The sweep contains 302,400 scenarios and about 170 million realizations.[1]
Statistical precision
If the runaway probability is and it is estimated from realizations, its standard error is
This error is largest when . With realizations in the first pass,
This is 2.5 percentage points. At 95% confidence the interval is roughly percentage points. Scenarios near the boundary between stable and runaway, or where a rare outcome matters, receive a second pass of 2,000 or more realizations.[8]
The project notes that a 2017 study found about 60 runs are enough to pin down the average outcome of a single scenario, but a probability needs many more. A runaway probability near one half needs hundreds of runs to be known to a few points, and a rare outcome needs far more.[12][8]
Data sources
Orbits come mainly from the public satellite catalog maintained by CelesTrak, with the rest from Jonathan McDowell's GCAT. Masses, sizes and breakup histories come from GCAT, and average cross-sections come from ESA's DISCOS database as rounded derived values. Solar activity is taken from the historical record of the Sun's 10.7 cm radio flux (F10.7).[8]
Reproducibility
The simulations are chaotic, so a tiny numerical difference can lead two runs to diverge completely. BOINC checks results by running each workunit on two computers and comparing them. The application is therefore built to give identical results on every computer. It uses a counter-based random-number generator (Philox), in which each draw depends only on the seed, time step, object and purpose of the draw. It also uses correctly rounded mathematical functions from the CORE-MATH library. The results are designed to match across Linux, Windows and macOS, on x86-64 and ARM processors, and on GPUs.[8]
Validation
Before work was sent out, the application was checked against published results. Its collision rate on the 2009 population reproduces the IADC comparison study within the published range of agency models. Its 1 to 10 cm debris field is compared with ESA's MASTER-8 model. Its fragment clouds are checked against six real breakups: Fengyun-1C, Cosmos 2251, Iridium 33, Cosmos 1408 and two exploded Briz-M upper stages. The same scenario has also been run through MIT's MOCAT-MC model.[8] The project states that the application also follows the decay of these six debris clouds.[1]
Participation
Joining
During the beta test, joining requires an account on the project website. Account creation requires an invitation code, which is shown on the project's homepage. The BOINC Manager cannot create an account for this project. To join:[13]
- Create an account on the project website.
- Install BOINC if it is not already installed.
- In the BOINC Manager, choose Tools > Add project and enter
https://kesslerathome.org/in the Project URL box, since the project is not yet in the Manager's list. - Choose existing user and log in with the account's email address and password.
On a computer without the Manager, the boinccmd --project_attach command can be used with the account key from the account page.
Platforms
The project has applications for:[2]
- CPUs on Linux, Windows and macOS
- NVIDIA and AMD GPUs on Linux and Windows
Each computer can run up to 4 tasks per CPU thread and 10 tasks per GPU at a time.[2]
Replication
Every job is sent to two volunteers' computers, and the two results must agree exactly before they are accepted. For the beta test, 100 scenarios are each run with 10 random seeds, which gives jobs. Because each job is sent twice, the beta requires tasks.[2]
Project status
The beta test began on 4 November 2026 with the first work sent to volunteers. During the beta, the project is checking that results agree across platforms and watching run times, credit and errors. Work is limited to these beta tasks for now. Volunteers who encounter failed tasks or unexpected results are asked to post on the project's message boards.[2]
The project's homepage and About page still describe the science application as being in final checks and say that no work is being distributed. The news archive and the server status page, both dated 4 and 5 November 2026, report that work is being distributed.[1][14]
Timeline
| Date | Event |
|---|---|
| 1978 | Kessler and Cour-Palais publish the debris belt paper.[3] |
| 2001 | Kessler and Anz-Meador publish the critical number model.[5] |
| 4 November 2026 | The beta test begins with 1,000 jobs, or 2,000 tasks.[2] |
| 5 November 2026 | The server reports 2,734 tasks in progress and 71 users with credit.[14] |
Server statistics
The following figures are from the project's server status page on 5 November 2026 at 07:27 UTC. They reflect the beta test and will change as it continues.[14]
| Measure | Value |
|---|---|
| Users with credit | 71 |
| Users active in the last 24 hours | 71 |
| Computers with credit | 237 |
| Computers registered in the last 24 hours | 292 |
| Tasks in progress | 2,734 |
| Average runtime of recent tasks | 0.53 hours (minimum 0.01, maximum 13.41) |
| Estimated total performance | 19.48 TFLOPS (19,484.39 GFLOPS) |
Scientific publications
As of 5 November 2026, the Kessler@home team has not published any peer-reviewed papers, and none were found on arXiv or elsewhere on the web. The project's science page instead lists the literature on which its methods rest.[8]
Publications by the project
None yet.
Literature underpinning the project
- Kessler, D. J. and Cour-Palais, B. G. (1978). "Collision frequency of artificial satellites: The creation of a debris belt". Journal of Geophysical Research 83(A6).[3]
- Kessler, D. J. (1981). "Derivation of the collision probability between orbiting objects: the lifetimes of Jupiter's outer moons". Icarus 48.[10]
- Johnson, N. L., Krisko, P. H., Liou, J.-C. and Anz-Meador, P. D. (2001). "NASA's new breakup model of EVOLVE 4.0". Advances in Space Research 28(9).[11]
- Krisko, P. H. (2011). "Proper implementation of the 1998 NASA breakup model". Orbital Debris Quarterly News 15(4).[15]
- Kessler, D. J. and Anz-Meador, P. D. (2001). "Critical number of spacecraft in low Earth orbit". Third European Conference on Space Debris.[5]
- Lidtke, A. A., Lewis, H. G. and Armellin, R. (2017). "Statistical analysis of the inherent variability in the results of evolutionary debris models". Advances in Space Research 59(7), 1698-1714.[12]
- Parker, W. E., Brown, M. K. and Linares, R. (2025). "Greenhouse gases reduce the satellite carrying capacity of low Earth orbit". Nature Sustainability 8(4), 363-372.[16]
- "A New Monte-Carlo Model for the Space Environment" (2024), describing MIT's MOCAT-MC model.[9]
See also
External links
- https://kesslerathome.org/ (official website)
- https://kesslerathome.org/science.php (the science behind the project)
- https://kesslerathome.org/server_status.php (server status)
- https://kesslerathome.org/forum_index.php (message boards)
- https://boinc.berkeley.edu/ (BOINC)
References
- ↑ 1.0 1.1 1.2 1.3 1.4 1.5 1.6 About Kessler@home. Kessler@home. Retrieved 2026-11-05.
- ↑ 2.0 2.1 2.2 2.3 2.4 2.5 2.6 News archive. Kessler@home. Retrieved 2026-11-05.
- ↑ 3.0 3.1 3.2 Kessler, Donald J..(1978).Collision frequency of artificial satellites: The creation of a debris belt. Journal of Geophysical Research.
- ↑ Kessler syndrome. Wikipedia. Retrieved 2026-11-05.
- ↑ 5.0 5.1 5.2 (March 2001).Critical number of spacecraft in low Earth orbit: Using satellite fragmentation data to evaluate the stability of the orbital debris environment. Third European Conference on Space Debris. Retrieved 2026-11-05.
- ↑ Hugh G. Lewis and Donald J. Kessler.Critical number of spacecraft in low Earth orbit: A new assessment of the stability of the orbital debris environment. Ninth European Conference on Space Debris. Retrieved 2026-11-05.
- ↑ (2013).IADC stability study summary. United Nations Office for Outer Space Affairs.
- ↑ 8.0 8.1 8.2 8.3 8.4 8.5 8.6 The science. Kessler@home. Retrieved 2026-11-05.
- ↑ 9.0 9.1 (2024). "A New Monte-Carlo Model for the Space Environment". arXiv: 2405.10430.
- ↑ 10.0 10.1 Kessler, Donald J..(1981).Derivation of the collision probability between orbiting objects: the lifetimes of Jupiter's outer moons. Icarus.
- ↑ 11.0 11.1 Johnson, N. L..(2001).NASA's new breakup model of EVOLVE 4.0. Advances in Space Research.
- ↑ 12.0 12.1 Lidtke, Aleksander A..(2017).Statistical analysis of the inherent variability in the results of evolutionary debris models. Advances in Space Research. pp. 1698-1714. DOI: 10.1016/j.asr.2017.01.004.
- ↑ Kessler@home. Kessler@home. Retrieved 2026-11-05.
- ↑ 14.0 14.1 14.2 Project status. Kessler@home. Retrieved 2026-11-05.
- ↑ Krisko, P. H..(2011).Proper implementation of the 1998 NASA breakup model. Orbital Debris Quarterly News.
- ↑ Parker, William E..(2025).Greenhouse gases reduce the satellite carrying capacity of low Earth orbit. Nature Sustainability. pp. 363-372. DOI: 10.1038/s41893-025-01512-0.