The Variable-Pitch Screw Launch (VPSL) system, is a revolutionary ground-based electromagnetic launch technology that leverages magnetic coupling and variable-pitch leadscrews to accelerate payloads to very high exit velocities (e.g., >11,000 m/s) at a fraction of the cost of traditional chemical rockets. In a paper authored by Phil Swan and Alastair Swan of the Atlantis Project, details are presented on how VPSL overcomes limitations of existing mass drivers, such as the switching constraints of linear motors and rail wear in railguns. Phil Swan appeared on The Space Show last January to discuss the concept with Dr. David Livingston.
The capital cost of a VPSL system scales with the square of exit velocity (ΔV2), a significant improvement over the exponential cost growth of chemical propulsion (exp(ΔV/ΔVe )) and the cubic scaling (ΔV3 ) of some linear motor components in mass drivers. The authors present results from a parametric model that estimates a $33 billion USD capital cost (2024 dollars) for a human-rated system capable of accelerating vehicles to escape velocity for Mars missions, positioning VPSL as a game-changer for cost-effective space exploration.
As humans begin to explore and develop space beyond low Earth orbit (LEO), missions to the Moon, Mars, asteroids and beyond will demand significantly higher delta-v than those needed for LEO operations, especially for human round-trips, which nearly double the velocity requirements. High delta-v missions also reduce crew exposure to cosmic radiation and optimize provisions, but the rocket equation—where fuel mass grows exponentially with delta-v—makes traditional rockets increasingly expensive. VPSL is presented as a scalable, infrastructure-based solution that mitigates these costs, offering both economic and environmental benefits. By reducing reliance on chemical propellants, it aligns with global climate goals, marking a pivotal shift toward sustainable spaceflight.
As a starting point for economic considerations the Swans provided a historical context for exploration costs (in 2020 USD) of the Apollo Program ($257 billion), Space Shuttle ($197 billion) and the International Space Station annual costs ($500 million per-person-year; total of $150B to date); with an estimate that the Artemis Program will cost $93 billion through the end of FY2025 (likely over $100 billion by the time Artemis III returns to the Moon according to ChatGPT). Since the dawn of human spaceflight these programs demonstrate the immense financial burden associated with traditional (chemical rocket) spaceflight, yet their broader benefits—economic stimulus, technological innovation, and geopolitical prestige—justify the investment. The aim of VPSL is to reduce these costs dramatically.
The analysis then moves to a cost comparison of all rocket systems using empirical data that show an exponential relationship between launch cost and delta-v reflecting the “tyranny of the rocket equation” where higher velocities require exponentially more fuel, driving up costs for missions beyond LEO, which will become increasingly important as global space agencies push out into the solar system toward high delta-v destinations.
The paper contrasts the economics of rockets with mass drivers where the latter scale as the cube of the velocity (ΔV3) due to increased power demands at higher velocities. VPSL avoids this by converting electrical energy into rotational energy in screws, then transferring it magnetically to the payload, minimizing expensive pulsed-power electronics. For example, scaling a traditional mass driver from 100 m/s to 10,000 m/s increases costs by a million-fold as ΔV3 dominates, but a well designed VPSL mitigates this issue.
The specific implementation of a VPSL system is presented with an architecture targeting a 22-year Mars outpost program, with launches during Mars transfer windows. The payload is human-rated, assuming fit crews and acceleration couches, and is designed with sufficient capacity for life support, power generation systems, and rocket propulsion for in-space maneuvering as well as decent to the Martian surface.
This VPSL system includes a 774 km submerged floating underwater section, an 83 km underground ramp curving upward, and a 122 km aeronautically supported elevated tube with the exit aperture at an altitude of 15 km. The entire 979 km launching conduit would be evacuated to minimize drag with air locks at both ends, and face East to take advantage of the Earth’s rotation. For a Mars transfer orbit the exit velocity was calculated to be 11,129 m/s taking into account the Earth’s rotation.
VPSL outperforms rockets for high delta-v missions, leveraging fixed infrastructure costs and low marginal launch costs. It’s quadratic cost scaling and sustainable design make it a transformative option when compared to rockets for high delta-v missions.
I reached out to Phil Swan after his appearance on The Space Show to discuss VPSL and he graciously agreed to participate in an interview with me via email to dive deeper into some of the challenges for implementation of the architecture of the Mars mission. His outstanding responses below are backed up with rigorous engineering reasoning and I thank him for his time collaborating with me on this post.
Many of my interview questions arose from public feedback he received from over 125,000 YouTube views of his presentation on VPSL at the International Space Development Conference last May (Section F of the paper). This approach will hopefully help ascertain what actions are needed to realize the system as well as further engineering development needed to advance it’s technical readiness level. The first two questions involve funding mechanisms for implementation.
SSP: There didn’t appear to be a funding mechanism proposed for the VPSL system although there were a few references to features that would provide incentives for investors. Do you envision the project to be funded by private venture capital, governmental sources or a combination through public/private partnerships?
PS: Our funding strategy is designed to attract private investment through a phased development approach, where some liquidity and financial flexibility is offered by allowing employees and early-stage investors to sell shares to later-stage investors as key technical and engineering milestones are met, similar to staged investment rounds in deep-tech ventures. It would be like many other tech startups where for many years the company’s primary focus is growth as opposed to profits. While we anticipate private venture capital to play a significant role, we are also exploring potential government grants or public-private partnerships to support critical advancements. Revenue generation from early-stage prototypes and other technologies we develop along the way may provide additional funding streams, but the most significant returns will come when we enable affordable interplanetary spaceflight.
SSP: The $33.3B price tag included capital and operations costs but I did not see research and development included. While your calculations show that VPSL costs are very competitive and environmentally beneficial when compared to rockets, this system will require significant development costs to reach TRL 9. Do you have an estimate of the R&D budget?
PS: We anticipate the R&D budget to be 10% of the total estimated capital and operational costs. Our research and development efforts thus far have led to substantial reductions in the estimated costs, so strategic investment in R&D can drive down capital expenditures and improve overall system profitability. For example, a while ago our R&D work led to an improvement where we placed grapplers on both sides of the screws instead of just on one side. This innovation dramatically reduces the forces transmitted to the brackets that support the screws. In this sense, R&D serves as a cost-reduction mechanism. If we do the right amount of R&D and focus it on the most important problems, it could end up paying for itself.
SSP: The remainder of interview questions probe deeper into issues identified through public feedback in Section F of your paper. With respect to constructing a 979 km long vacuum tube and designing fast-acting doors to maintain vacuum while allowing high-speed exit of the vehicle, what are the specific engineering requirements and cost estimates for designing and maintaining fast-acting airlock doors capable of sealing a vacuum tube after a vehicle exits at 11,129 m/s, and how do these compare to existing vacuum systems like LIGO (Laser Interferometer Gravitational-Wave Observatory)?
PS: To exit the tube, the vehicle will pass through an already open fast-acting door first, and that door will start closing immediately. The other end of the airlock is covered with a burst disk. The ambient air pressure at the airlock’s altitude (15km) is around 12000 Pa and the pressure inside the tube is 5 Pa. When the vehicle breaks through the burst disk, the rarified outside air will start travelling into the tube at the speed of sound. The fast-acting door needs to finish closing before the ambient air rushing into the tube reaches it. The math in the model estimates that to meet these requirements the airlock needs to be at least 288 m long if the fast-acting door is engineered to close in 1 second. I should add that the fast-acting door can be backed by a second slower door that is designed to achieve a better vacuum seal.
After the vehicle exits, a new membrane needs to be stretched over the end of the tube to from a new burst disk, and then the airlock needs to be pumped down again from 12000 Pa to 5 Pa. Our current model estimates that it will take 10 minutes and cost 312 dollars to pump the air out of the airlock each time we cycle it.
For LIGO, the exterior pressure is roughly 100,000 Pa and its interior pressure is 1.33 × 10⁻⁷ Pa to 2.67 × 10⁻⁷ Pa – which is a vacuum that it has maintained for 25 years. That’s a ratio of ~7e11 to 1. For VPSL, the exterior pressure is 12000 Pa and it has an interior pressure of 5 Pa for a ratio of only ~2.4e3. So, in one sense, LIGO’s vacuum engineering problem is eight orders of magnitude harder than the problem for VPSL. So, what we’re proposing here falls comfortably within established engineering capabilities. But, VPSL introduces operational dynamics that LIGO does not face – such as repeated venting and sealing at the airlocks and high-speed vehicle interaction. So, in another sense, we will be facing some new challenges that LIGO doesn’t have to deal with.
SSP: To address skepticism about sourcing materials robust enough to endure the high speeds, heat, and magnetic forces cost-effectively, you asserted that the choice of steel and aero-grade aluminum would have sufficient engineering margins when compared to rockets. What are the maximum stresses, thermal loads, and electromagnetic forces experienced by steel screws and aluminum tubes at peak speeds, and can existing manufacturing processes scale these materials to a 979 km system without cost escalation?
PS: This question assumes that extreme forces or heat are unavoidable, but that’s not how we approached the problem. From both an engineering and architectural perspective, we began with the constraints of existing materials and designed a system that stays within those limits.
For example, let’s start with the mechanical stresses. If we want a launcher for sending missions to Mars, this creates a requirement – we will need to launch vehicles at a speed of ~11,129 m/s relative to the surface of the spinning Earth. This is the speed at which the maximum mechanical stresses will occur.
The idea is that the spinning screws drive the adaptive nut. It’s basically a leadscrew and nut with a certain gear ratio. To figure out what that ratio needs to be, we first need to figure out how fast we can turn the screws without exceeding the stress limits of existing affordable materials. To ballpark that, we know that the yield strength for M2 High-Speed Steel can reach 1,300 to 2,200 MPa. But let’s assume we use a cheaper steel with a yield strength of 700 MPa and a density, ρ, of 7850 kg/m3. If we apply an engineering factor of 1.5, then we can set the maximum stress, σ, that we want to see in the steel to a value of 467 MPa. The rate that you can spin a spinning pipe without exceeding this level of stress is
[ref] where ω is in radians-per-sec, and ri and ro are the inner and outer radii in meters. Multiplying ω by ro gives the max rim speed of 404 m/s. This is a value similar to what the tips of airliner fans blades reach during takeoff.
From this value we can calculate the maximum slope of the screw flights, which is 11129/404=27. This means that the total force of the screw flights needs to be ~27 times higher than the force you need to accelerate the spacecraft, sled, and adaptive nut.
Since the coupling is magnetic, you can work out the coupling force across the “airgap” per square meter (see math in above linked paper). This works out to be 795775 N/m2, or less than 1 MPa (about 1/500th the internal tensile stress due to the centrifugal forces).
While you didn’t ask about this in the question, I feel that it’s important to mention that for this to work the screws and rails need to be very straight. To achieve that we will need automatic alignment actuators and something akin to an ultra-high-precision GPS system to achieve the required straightness.
You also asked about heating. This is a good question to use to validate the practicality of a launch architecture. For example, if a launcher was 1000 km long and it was made up of 1 million 1-meter segments, and each of those segments heated up by, say, 5 degrees each launch, then you could estimate how much energy was being dissipated as heat rather than being converted into kinetic energy—and it could be a lot. If each segment weighed one ton, heated up by 5°C, and had the heat capacity of water (about 4,200 J/kg·°C), then the total energy lost to heat would be:
1,000,000 segments × 1,000 kg × 5°C × 4,200 J/kg·°C = 21,000,000,000,000 J. That’s 2.1 × 10¹³ joules, or about 5.8 gigawatt-hours of energy lost to heating per launch.
By comparison, the kinetic energy of a 10-ton spacecraft (10,000 kg) in low Earth orbit at 7.8 km/s is:
(1/2) × 10,000 kg × (7,800 m/s)² ≈ 3.0 × 10¹¹ joules
So, the energy lost to heating in this example would be about 70 times greater than the kinetic energy delivered to the payload. In other words, such a launcher would not be very energy efficient.
In other architectures, this heat is generated because the segments rapidly convert energy from one form to another in the process of accelerating the vehicle, and such high-power conversions invariably generate heat. But the VPSL doesn’t rapidly convert energy from one form to another. The kinetic energy in spinning screws is directly channeled into the kinetic energy of the vehicle through what is essentially a magnetic worm gear. So, the screws and guideway will not heat up significantly during a launch because they are not heated up by the process of rapid high-power energy conversion.
Now there is still some friction that generates heat. Even a train on rails will generate some heat due to friction between its wheels and the rails, but the friction and heat generation associated with magnetic levitation systems is low enough that most people think of them as being “frictionless” – even though that’s not entirely true – maglev tracks and magnetic bearings are really just “very low” friction technologies.
SSP: Concerns were raised about potential eddy currents from the spinning screws and electromagnetic interactions causing energy losses and heat buildup which could reduce efficiency. In view of your acknowledgement that more engineering work is needed to quantify these interactions, have you calculated the magnitude of eddy current losses in a VPSL system at peak velocity, and have you designed experiments or computer code to run simulations or small-scale tests to determine how effective uniform magnetic fields and laminated components would be in reducing these losses?
PS: There are devices that are designed to use Eddy currents for braking, and there are technologies, such as magnetic bearings and maglev trains, that are designed to generate far less friction and wear than their mechanical counterparts. We’ve certainly designed devices to explore the limits of the low-friction high-speed magnetic levitation, but given the high speeds involved, we’ve chosen to implement these designs later on our prototyping roadmap. For one of them, we worked with a well-credentialed Ph.D. and an ASME Fellow in the field of rotordynamics and magnetic bearings. We shared our concerns with him about venturing into uncertain or poorly understood engineering territory. He reassured us that he was not aware of any engineering or physics reasons why our proposed technology would not work, and wrote us a letter of support where he stated, “I am confident in the merits of the proposed research.” That said, pushing beyond the speeds already achieved with maglev trains, the world-record-holding magnetic levitation rocket sled track at Holloman Airforce Base, energy storage flywheels, etc. certainly will involve doing more research and experimentation.
In addition to building physical prototypes, we plan to license advanced engineering software and bring on specialized talent to develop a multi-physics simulation using finite element analysis (FEA) techniques. These simulations will be validated through data collected from instrumented small-scale prototypes. They will give us more visibility on a wide variety of performance metrics.
SSP: Regarding fast-acting components, to ensure operational reliability and test real-world applicability of existing technology to VPSL’s extreme speeds, how reliably can electromagnetic grappler pads and actuators maintain synchronization and stability at speeds up to 11,129 m/s, and what are the failure rates of similar systems (e.g., magnetic bearings) under comparable conditions?
PS: It becomes easier to maintain synchronization as the vehicle approaches the muzzle of the launcher because the screw geometry changes more slowly at the muzzle end. Near the beginning, the geometry changes quickly and the grapplers need to reposition more rapidly, but the forces that they need to manage are also much smaller. If you haven’t yet seen Isaac Arthur’s video, “Mass Drivers Versus Rockets”, you should check it out. It has some good clips that show how the screw geometry changes and how the grapplers reposition during a launch.
Compared to ball and roller bearings, magnetic bearings exhibit extremely low failure rates in industrial use due to the lack of mechanical contact. Although, I suppose there must be some failures due to, for example, defective solid-state electronics in the controllers, power surges, corrosion of wires, fouling of sensors, etc.
Getting the failure rate to the level we need it to be at is a well-understood engineering exercise – like perfecting jet engines or building fault tolerance into hard drives. You need to test, iterate, and apply good engineering practices—refinement, redundancy, early fault detection, and so on. We will be building upon a substantial amount of experience that already exists within other industries – we’re not starting from scratch here.
SSP: You mentioned that to maintain investor confidence, you had a roadmap for developing the technology using a combination of physical prototypes and simulated “digital twin” prototypes. To address scalability physics and ensure the system can handle larger payloads effectively, how does magnetic field strength and consistency vary across a 979 km screw system compared to a small prototype, and what payload mass thresholds trigger performance degradation in digital twin simulations?
PS: Magnetic fields are not generated by the launcher’s guideway or screws’ flights (there are fields inside the magnetic bearings and electric motors that support and spin the screws though). Magnetic fields are generated by the adaptive nut and the sled. The strength of the fields between the grappler pads and the screw flights does peak as the vehicle approaches the muzzle end of the launcher. The strength of the fields between the sled and the guideway’s rail is constant during forward acceleration, and then it jumps up to its peak when the vehicle is on the ramp. Some of the small-scale prototypes will explore the same peak field strengths so that we can avoid surprises later as we scale up.
The system’s cost is expected to scale linearly with payload mass and payload mass will not trigger performance degradation. But if we were to go in the other direction, and scale down too far, that may introduce challenges – particularly with respect to vehicle stability and thermal protection during reverse reentry.
In the paper we said that cost scales with the square of delta-v – which is a lot better than the way that chemical rocket cost scales with the exponent of delta-v. However, we haven’t really explored how cost will scale at speeds much beyond 11,129 m/s. If we try to go much faster than that we’ll probably start running into material limits. Switching to more exotic materials will likely alter the cost-versus-delta-v relationship. We certainly do not want to suggest that the technology can scale up to the speeds needed for interstellar travel or anything like that.
SSP: To validate the economic and efficiency claims of VPSL when compared to existing rockets using energy data, have you done a detailed breakdown of the system’s $33.3 billion capital cost compared to the lifecycle cost of a chemical rocket program delivering 9.6 million kg to Mars, showing how much energy is saved by regenerative braking in real-world conditions?
PS: Yes, the capital and operating costs are computed by code within the digital twin and this includes the power savings from regenerative braking. While there has been some analysis of chemical rocket costs, much of our discussion in the public sphere revolves around addressing and correcting overly optimistic claims—particularly those made by Elon Musk—which are often repeated uncritically by some space enthusiasts. For example, this paper attempts to demonstrate that based on empirical evidence there is clearly an exponential relationship between cost and delta-v. As a widely circulated quote attributed to Peter Diamandis says, “Our brains are wired for linear thinking in an exponential world, and its causing us a great deal of strife.”
Personally, I haven’t felt it was in our best interests to publish a study that emphasizes how prohibitively expensive a permanently manned outpost—or a city—on Mars would be using chemical rockets. While some people argue that attempting to settle Mars is fundamentally misguided, I personally don’t share that view since I believe in the potential of launch infrastructure.
But if you think that rockets are the only option available to us, then right now the cost-per-kg to Mars is on the order of 1.2 million USD. While many are excited about Starship and the potential of full reusability, we’re far more cautious about its ability to fundamentally change the cost of spaceflight for the delta-v’s and mission durations required for one-way and round trips to Mars. We’ve shared some of our reasoning and the available data on this – see: https://youtu.be/Apu6nDahjB4 and https://youtu.be/GvqAM9p4hss. In the absence of a game-changing development, sending a million tons to Mars with chemical rockets will cost on the order of ($1.2 x 106 /kg )(1 x 109 kg) = $1.2 x 1015, or 1200 trillion dollars. This isn’t the kind of problem that will “fix itself” anytime soon through experience curve effects.
SSP: Related to your preferred site of Hawaii’s Big Island, the ongoing legal, cultural, and logistical hurdles encountered by Caltech and the University of California in getting approval to build the Thirty Meter Telescope (TMT) seem insurmountable. Native Hawaiian groups and environmentalists who consider Mauna Kea a sacred site have caused a decade of delays. The telescope’s future is still uncertain so a project of the scale of a VPSL system seems very challenging. While your plan to engage with the community in a respectful and productive manner by clearly communicating benefits to the indigenous people like economic opportunity and cultural legacy make sense, it has likely already been tried by the TMT team. Have you identified specific alternate coastal sites with high elevation, low latitude, and access to large bodies of water that may not present such difficult environmental and cultural challenges?
PS: The summit of Mauna Kea is a culturally sensitive area. For many Native Hawaiians, the numerous telescopes located there are seen as an incursion on sacred land. Additionally, the U.S. military has used portions of the mountain’s slopes for training exercises, causing ecological damage. As a result, the local population is particularly sensitive to further disruption and, in many cases, would prefer the mountain be restored to its original, undisturbed state.
The VPSL system’s acceleration segment would be located offshore and underwater, while the ramp portion would be on the island but almost entirely contained within a tunnel. The tunnel would exit well below the summit—away from the existing observatories—through a small opening situated to avoid culturally significant sites. The elevated, evacuated launch tube would be a temporary structure, deployed every two years for a few weeks during Mars transfer windows.
A potential path forward could involve a three-way agreement: the launcher could be used to deploy multiple space telescopes. These offer a path to eventually phase out the existing summit observatories without impacting the scientific community that relies on them. In return, the Hawaiian community would agree to permit the construction and limited use of the launcher, for example during Mars transfer windows and on a few other occasions.
Over time, the Hawaiian people may come to see the launcher not only as a less intrusive alternative but as a source of enduring pride—an opportunity to contribute to humanity’s next great era of exploration. Rather than diminishing their culture, it could elevate it, building upon the proud legacy of the Polynesian navigators who first discovered and settled the islands. This vision, however, must be informed by dialogue with Native Hawaiian leaders and cultural practitioners – not just outreach – to ensure the project is shaped in a way that reflects and respects their values. In this way, Hawai‘i’s role in space exploration could be seen as a modern extension of their deep tradition of voyaging and discovery.
But, if Hawaii choses to pass on the opportunity, there are many alternative sites around the world that would suffice. Developing and characterizing alternative sites simply hasn’t received priority yet.
SSP: What are the projected environmental impacts (e.g., land use, wildlife disruption) and cultural consultation costs for siting a VPSL system on Hawaii, and how do these compare to alternative sites like desert-mountain regions in terms of construction feasibility and community acceptance?
PS: We don’t expect there to be a significant amount of environmental disruption but with an ecology that’s very sensitive, we will need to be careful. The launcher is underwater and should not impede marine life. The ramp is within a shallow tunnel, so it shouldn’t affect ecologies on the surface, but we’d need to come up with a good plan for dealing with the excavated material generated during tunneling. I expect that birds would tend to avoid the elevated evacuated tube. Vehicles will exit the system far offshore and at an altitude of 15 km, so they shouldn’t generate a lot of noise. Rockets, on the other hand, generate a lot of noise and a lot of pollution from their exhaust. By eliminating the need for rocket launches, VPSL’s net benefit to the environment would be enormously positive.
To close out, we view VPSL not just as an engineering challenge, but as a test case for a new kind of sustainable, infrastructure-led approach to spaceflight – one grounded in realism, openness to critique, and collaborative development.