Case Study: Designing the Return — ISRU and the Mars Ascent Vehicle

"Travel light and live off the land." — Robert Zubrin, the guiding principle of Mars Direct

Executive Summary

Getting a crew to Mars is a solved problem in physics; getting them back is where mission designs live or die, because the return demands a rocket launched off another planet, and the rocket equation makes every kilogram of return propellant carried from Earth ruinously expensive. In this case study we design the return leg of a crewed Mars mission: we choose the ISRU-and-pre-deployment architecture, size the Mars ascent vehicle two different ways, size the propellant plant that fuels it and the power source that runs the plant, and lay out the risk-inverted timeline that never sends a crew without a fueled ride home already waiting. Where the first case study audited a real landing, this one builds a mission element from requirements — the return-problem answer that separates a Mars mission that closes from one that does not.

Skills applied

• Turning the return problem into a testable requirement, and choosing the ISRU architecture (§34.5, Ch. 29).
• Sizing a Mars ascent vehicle with the rocket equation, two architectures compared (§34.5, Ch. 3).
• Sizing an ISRU propellant plant's throughput from the propellant demand (§34.5).
• Sizing the plant's power source and justifying fission over solar (§34.5, Ch. 25).
• Building the risk-inverted, pre-deployment timeline set by the synodic window (§34.1, §34.6).

Background

The requirement. Return a crew of four from the Martian surface to Earth. The driving requirement of the whole mission (§34.6) is return the crew alive, and it flows straight into a design rule for this leg: no crew departs Earth until a fueled return vehicle is confirmed on the Martian surface. That one rule, borrowed from Zubrin's Mars Direct and adopted by NASA's DRA 5.0, forces two decisions at once — make the return propellant on Mars by ISRU (§34.5), and pre-deploy the ascent vehicle and its plant one synodic window (~26 months) before the crew flies, so it can make and store the propellant in advance.

The chemistry, recapped. The plant runs the Chapter-28 reactions on Martian feedstock: Sabatier, $\text{CO}_2 + 4\text{H}_2 \rightarrow \text{CH}_4 + 2\text{H}_2\text{O}$, on atmospheric carbon dioxide ($\sim 96\%$ of the air), plus electrolysis of the product water, $2\text{H}_2\text{O} \rightarrow 2\text{H}_2 + \text{O}_2$, to recover oxygen and recycle hydrogen. The output is liquid methane and liquid oxygen — methalox — the propellant of Chapter 17's Raptor and the natural product of Mars ISRU.

Phase 1: The architecture decision

The first design choice is what the ascent vehicle has to do, and it splits into two clean options:

• Option 1 — two vehicles (ascent + rendezvous). The MAV lifts the crew only to a low Mars orbit ($\Delta v_{\text{asc}} \approx 4.1\ \text{km/s}$), where it docks with an Earth-return vehicle pre-positioned in Mars orbit that carries the crew home. Less propellant to make on the surface, but a rendezvous in Mars orbit is a critical, life-or-death maneuver far from help.
• Option 2 — one vehicle (surface to trans-Earth). The MAV flies the crew directly from the surface onto a trans-Earth trajectory, combining ascent and trans-Earth injection ($\Delta v \approx 4.1 + 2.1 = 6.2\ \text{km/s}$). No rendezvous — the simplest operations — but far more propellant to manufacture, and a much bigger vehicle to land. This is essentially the Starship approach: the ship is the MAV.

We will size both and let the numbers inform the trade, exactly the discipline of Chapter 29.

Phase 2: Sizing the Mars ascent vehicle

Take a $5\ \text{t}$ crew capsule (crew of four, cabin, life support, samples) as the payload the MAV must deliver, an ISRU methalox engine at $I_{sp} = 360\ \text{s}$ ($v_e = 3.53\ \text{km/s}$), and a $5\%$ margin on each delta-v. The rocket equation (Chapter 3) does the sizing.

import math
G0 = 9.80665

def size_stage(dv_kms, isp_s, payload_t):
    """Propellant and wet mass (t) to give payload_t a burn of dv_kms at isp_s."""
    ve = isp_s * G0
    mr = math.exp(dv_kms * 1000 / ve)          # rocket-equation mass ratio
    m_wet = mr * payload_t
    return m_wet - payload_t, m_wet

# Option 1: MAV to low Mars orbit only (4.1 km/s + 5% margin).
p1, w1 = size_stage(4.1 * 1.05, 360, 5.0)
# Option 2: MAV surface -> trans-Earth directly (4.1 + 2.1 = 6.2 km/s + 5% margin).
p2, w2 = size_stage(6.2 * 1.05, 360, 5.0)
print(f"Option 1 (ascent only):   {p1:.1f} t propellant, {w1:.1f} t wet")
print(f"Option 2 (direct return): {p2:.1f} t propellant, {w2:.1f} t wet")
# Expected output:
# Option 1 (ascent only):   11.9 t propellant, 16.9 t wet
# Option 2 (direct return): 26.6 t propellant, 31.6 t wet

The two architectures ask the ISRU plant for very different amounts:

Sanity check. Both propellant fractions are ordinary for a launch stage — $71\%$ for Option 1, $84\%$ for Option 2 — and the jump from $12$ to $27\ \text{t}$ for the extra $2.1\ \text{km/s}$ is the rocket equation's exponential bite: adding a third more delta-v more than doubled the propellant. We will carry Option 1 as the baseline — its smaller plant and smaller landed MAV are easier to pre-deploy, and the rendezvous, while risky, is a maneuver spaceflight has done since Gemini. The trade would flip toward Option 2 for an operator (like SpaceX) willing to land a very large vehicle and eliminate the rendezvous entirely.

Phase 3: Sizing the propellant plant

The plant must make Option 1's $\sim 11.9\ \text{t}$ of methalox during the pre-deployment stay — say $420\ \text{days}$ of production before the crew leaves Earth (leaving margin within the ~26-month pre-deployment window). Methalox burns at a mixture ratio of about $3.5\ \text{kg}$ of oxygen per $1\ \text{kg}$ of methane, so split the total and find the average production rates:

prop = 11.9                       # t of methalox to make (Option 1, from Phase 2)
o2  = prop * 3.5 / 4.5            # oxygen fraction of methalox
ch4 = prop * 1.0 / 4.5            # methane fraction
hours = 420 * 24                 # 420-day production campaign
print(f"oxygen:  {o2:.1f} t  ({o2*1e6/hours:.0f} g/hr average)")
print(f"methane: {ch4:.1f} t  ({ch4*1e6/hours:.0f} g/hr average)")
# Expected output:
# oxygen:  9.3 t  (918 g/hr average)
# methane: 2.6 t  (262 g/hr average)

The plant must produce oxygen at about $920\ \text{g/hr}$ — a little under a kilogram an hour, continuously, for more than a year. Sanity check against reality: MOXIE on Perseverance made oxygen at roughly $8\ \text{g/hr}$ (§34.5). Our plant needs about $115$ times that rate, sustained. That is a serious scale-up — the central engineering challenge of Mars ISRU — but it is a scale-up of a demonstrated process, not a new invention. The physics is proven; the plumbing is the work.

Phase 4: Powering the plant

ISRU is power-hungry, and this is where the design reaches back to Chapter 25. The dominant cost is electrolysis. Splitting water to make the oxygen takes very roughly $10\ \text{kWh}$ per kilogram of O$_2$ (a Tier-3 planning figure that folds in real-world inefficiency), so the electrolysis alone draws about

$$ P_{\text{elec}} \approx \frac{9{,}300\ \text{kg} \times 10\ \text{kWh/kg}}{420 \times 24\ \text{hr}} = \frac{93{,}000\ \text{kWh}}{10{,}080\ \text{hr}} \approx 9.2\ \text{kW}. $$

Add the Sabatier reactor's heating, gas compression, and — the quiet power hog — cryogenic liquefaction to store the methane and oxygen as liquids, and the continuous demand roughly doubles to $\sim 20\ \text{kWe}$ (Tier 3). Now the power-source decision writes itself. Solar arrays could supply $20\ \text{kW}$ in Martian daylight, but the plant must run for $420$ days straight through a planet-wide dust storm that can dim the sky for weeks and coat the panels — a risk no mission wants on its critical "make the return fuel" path. So the baseline is a fission reactor (the Kilopower/KRUSTY class of Chapter 25), clustered or scaled to $\sim 20$–$40\ \text{kWe}$, which runs day, night, and dust storm alike. The return trip is too important to depend on the weather.

🔧 Engineering Reality: the plant is on the mission's critical path, so it is redundant and it runs early. Because the crew does not launch until the tanks are full, the ISRU plant is the single most schedule-critical piece of hardware in the architecture — a slow or failed plant does not delay the mission, it cancels the crewed launch for that window (§34.1). So real designs pre-deploy the plant a full window early, run it with generous margin and redundancy, and confirm the propellant is made and stored before committing a single astronaut. This is the risk-inversion of §34.5 turned into a schedule: the return vehicle is fueled and waiting before the crew is even off the pad on Earth.

Phase 5: The timeline that ties it together

Assemble the pieces onto the synodic calendar of §34.1. The mission spans two launch windows:

The whole architecture closes because of one inversion: the expensive, leverage-heavy return propellant (§34.5) is never launched from Earth at all. Of the round-trip delta-v, the crew's vehicle carries from Earth only the outbound legs (TMI, TCMs, landing — about $4.6\ \text{km/s}$ margined, from the Checkpoint); the return's $6.5\ \text{km/s}$ is made on Mars. Delete ISRU and the mission must land that $\sim 12$–$27\ \text{t}$ of propellant from Earth — through the mass wall of §34.4, after paying to inject and launch it — and the design does not close on any near-term vehicle. ISRU is not an optimization; it is the load-bearing wall of a crewed Mars mission.

Discussion Questions

1. Option 2 needs $\sim 27\ \text{t}$ of propellant to Option 1's $\sim 12\ \text{t}$, yet its delta-v is only about a third larger. Explain, through the rocket equation, why the propellant more than doubled.
2. The design rule is "no crew launches until the return tanks are full." Trace how that single rule forces (a) ISRU, (b) pre-deployment a window early, and (c) a redundant, nuclear-powered plant.
3. Why is a Mars-orbit rendezvous (Option 1) considered a serious risk, even though rendezvous is routine in low Earth orbit? What is different about doing it at Mars?
4. The power estimate landed on $\sim 20\ \text{kWe}$ and drove the choice to fission over solar. What specific Martian environmental fact makes solar risky for this load in particular?

Your Turn: Extensions

• Option A (analysis). Re-size the Option 1 MAV if the crew capsule grows from $5$ to $7\ \text{t}$. Recompute the propellant, and then recompute the plant's oxygen production rate. By what factor did the plant's job grow, and is it more or less than the payload growth? (Note the amplification.)
• Option B (computation). Extend the plant code to sweep the production campaign length from $300$ to $600$ days and report the required oxygen rate (g/hr) at each. Hand-trace two points and show why a longer pre-deployment relaxes the plant's throughput — and what it costs in schedule. (Do not run it — add # Expected output:.)
• Option C (design). Design the hydrogen supply. The Sabatier reactor needs hydrogen the Martian air does not provide. Compare (a) bringing liquid hydrogen from Earth versus (b) mining and electrolyzing Martian water ice. List two pros and two cons of each, and pick one for a first mission, justifying it.

Key Takeaways

1. The return is the hard part, and ISRU is the answer. Return propellant carried from Earth is multiplied by launch, injection, and landing; making it on Mars deletes that stacked cost. ISRU is the load-bearing wall of a crewed Mars mission.
2. Architecture sets the plant size. An ascent-to-orbit MAV needs $\sim 12\ \text{t}$ of methalox; a direct surface-to-Earth MAV needs $\sim 27\ \text{t}$ — the rocket equation's exponential turning a one-third delta-v increase into a more-than-doubled propellant load.
3. Scale, not principle, is the challenge. The plant must make oxygen at $\sim 920\ \text{g/hr}$ — about $115\times$ MOXIE's demonstrated rate — sustained for over a year, on $\sim 20\ \text{kWe}$ of fission power that ignores the dust storms solar cannot.
4. The timeline inverts the risk. Pre-deploy the plant a synodic window early and launch the crew only once the return tanks are confirmed full — so no astronaut is ever sent to Mars without a fueled ride home already on the surface.