Visualise the lifeline of a proton-exchange membrane fuel cell — its water balance. Move the sliders for current density, temperature, pressure, membrane grade and humidification, and the tool computes the reaction water, electroosmotic drag and air-stream removal in real time, derives the membrane water content λ and the proton conductivity σ via the Springer model, and shows the flooding / dry-out boundaries.
Parameters
Active cell area A
cm²
Current density i
A/cm²
Cell temperature T
°C
Cell pressure P
bar
Membrane (Nafion grade)
Select thickness and dry λ
Air stoichiometry λ_air
Excess air supplied vs. stoichiometric O₂
Anode humidity RH_an
%
Cathode humidity RH_ca
%
Results
—
Reaction H₂O (g/s)
—
Electroosmotic drag λ_drag
—
Cathode λ
—
Proton conductivity σ (S/m)
—
Membrane R (Ω·cm²)
—
Water balance status
—
PEMFC cross-section — water flow animation
Anode (left) → membrane → cathode (right). Green: H⁺, blue: H₂O. Electroosmotic drag carries water across to the cathode, where it combines with the reaction water and is removed by the air stream. Membrane colour reflects water content λ.
Electroosmotic water flux from anode to cathode. F is the Faraday constant and I is the total current.
PEMFC Water Management — Humidification, Dry-out and Flooding
🙋
A PEMFC takes H₂ and air and makes water and electricity, right? If it makes water anyway, why do we need to humidify the inlet gases?
🎓
Great question. The key is the Nafion polymer membrane in the middle. It can only transport protons (H⁺) when water is dissolved in it — a dry membrane is practically an insulator and its resistance jumps by an order of magnitude. So during cold start, before reaction water builds up, or at low load when there is little water around, you have to humidify the inlet streams to keep the membrane "soaked". In production cars like the Toyota Mirai, miniaturising that humidifier has been a long-running development theme.
🙋
OK, so the membrane must stay wet. Then if I push the current density up and crank out reaction water, can I skip the humidifier?
🎓
In theory yes, but now the opposite problem appears. Above ~1 A/cm² the reaction water plus the electroosmotic-drag water starts to accumulate as liquid in the cathode flow channels and catalyst layer. That is "flooding". Air can no longer reach the catalyst and the voltage falls off a cliff. Push the current density on the left and you will see the water balance swing positive and trip the "flooding" verdict.
🙋
Yes — at 1.0 A/cm² it already says "flooding". So I just raise the air stoichiometry λ_air to dry it out?
🎓
That's the standard first move. Going from λ_air = 2 to 3 increases the air molar flow by 1.5×, so the water vapour carried out at the exit rises proportionally. But blower power scales roughly as λ_air cubed, so running at 3 all the time eats system efficiency. The real-world control is: cruise around λ_air ≈ 1.8–2.0, and pulse it up to 3 briefly as a purge when flooding is detected. Plug Power stationary units, Ballard buses and the Daimler GLC F-CELL all follow that recipe.
🙋
And on the dry side, what actually breaks?
🎓
More than just resistance. Higher membrane R means more IR heating, which dries it further — a vicious circle. And a dry Nafion membrane under mechanical clamping load can develop pinholes that let H₂ cross over. That is permanent damage, unlike flooding which clears once load drops. So designers keep a margin on the "slightly wet" side. Thinner membranes such as N212 (50 μm) respond faster but are more fragile when dry, which is why Mirai 2nd gen sticks with relatively thick N115-class membranes as insurance.
🙋
The cathode water content λ shows about 3.5 right now. What does that absolute number mean?
🎓
λ counts how many H₂O molecules sit on each SO₃⁻ group of Nafion — the range is 0–22. λ = 14 corresponds to vapour saturation (RH 100%), λ = 22 means liquid-water contact. Below λ ≈ 3 protons can barely move and conductivity is almost zero; λ ≈ 10–14 is the practical sweet spot. Your current cathode λ = 3.5 is fairly dry — humidify the anode to 100% and cathode to ~70% and λ_avg jumps to 8–10, more than doubling the conductivity. Try moving the sliders.
Frequently Asked Questions
The water budget inside the cell is generation + drag from anode − removal by the cathode air stream. Reaction water n_H2O = I/(2F), electroosmotic drag n_drag = I·λ_drag/F, and air-stream removal follows from the saturation pressure and air molar flow. This tool sums the three terms — positive means flooding-leaning, negative means dry-out-leaning.
The Springer (1991) empirical model is the standard correlation for Nafion membranes. It relates membrane water content λ (mol H2O per mol SO3⁻, range 0–22) to water-vapour activity and proton conductivity: σ(λ,T) = (0.005139λ − 0.00326)·exp[1268·(1/303 − 1/T)]. σ rises both with λ and with T. This tool uses the same equation to compute σ and the membrane resistance R.
Both kill performance, but in different ways. Flooding accumulates liquid water in the flow channels and catalyst layer, blocking oxygen and causing a sudden voltage drop at high current. Dry-out spikes the membrane resistance, raises IR losses across the whole current range and, when severe, leads to pinholes and irreversible degradation. Flooding usually recovers if you reduce load; dry-out damage often does not, so designers tend to bias slightly to the wet side.
Raising λ_air increases the cathode air molar flow, and the water vapour carried out at the exit rises proportionally — it dries the cell. λ_air ≈ 2 is a common balance point. Going to 3–4 helps clear flooding but the blower power scales steeply and system efficiency drops. The standard practice is to keep λ_air near 2 and pulse it up briefly as a purge when flooding is detected.
Real-World Applications
Fuel-cell vehicles (FCVs): The Toyota Mirai (2nd gen, 134 kW), Honda Clarity, Hyundai Nexo and Daimler GLC F-CELL all place the humidifier and water-management controller at the heart of the on-board package. Mirai 2nd gen combines a 3D channel design and a self-humidifying-friendly membrane to shrink — or partially remove — the external humidifier. Sub-zero start (down to −30 °C) is also an extension of water-management strategy.
Stationary co-generation and back-up power: Toshiba H2One, Bloom Energy (SOFC but with similar challenges), Plug Power (forklift power packs), Ballard (buses and logistics vehicles) and Cummins-Hydrogenics commercialise PEMFC stacks. Stationary systems can recover heat and water more easily than vehicles, so self-humidifying designs without an external humidifier are becoming standard.
Self-humidifying and advanced membranes: To eliminate the cost and bulk of an external humidifier, materials such as Nafion 3M PFIA, Aquivion (Solvay) and Pt-particle-dispersed self-humidifying membranes are now in production. They generate water on the cathode by exploiting H₂ crossover, simplifying the system and improving low-temperature performance simultaneously.
CAE water-distribution analysis: Commercial CAE codes (COMSOL Multiphysics, ANSYS Fluent, STAR-CCM+) provide PEMFC modules that solve 3D liquid-water distribution inside the gas-diffusion layer (GDL) and droplet formation in the channel. A 0-D balance like this tool is the starting point for setting boundary conditions and for sanity-checking those 3D models.
Common Misconceptions and Pitfalls
The biggest misconception is "humidify to 100% RH and water management is solved". In practice, a 100% RH inlet condenses on any cool spot and seeds flooding. Production systems use asymmetric "humid-dry" humidification — typically anode 100%, cathode 50–70% — to deliberately set up a water-activity gradient and pull water back through the membrane (back-diffusion). The default settings here follow that recipe.
Next, treating Springer's λ_drag as a constant 2.5 as the textbook does. The real electroosmotic-drag coefficient depends on the membrane water content: ~1 in a dry membrane, possibly >3 in liquid-saturated conditions. This tool uses the original Springer simplification λ_drag = 2.5·λ_dry/22, but at low RH or with thin membranes (N212) the field value can differ. For research use, adopt a temperature-dependent correlation such as Zawodzinski's.
Finally, "smaller membrane R always means better performance" is an oversimplification. R drops with thinner membranes, but thin membranes show more H₂ crossover, higher pinhole risk and steeper water-content gradients that fatigue mechanically. The real optimum sits inside the triangle of membrane resistance, crossover and mechanical strength — and "thinner is better" is a dangerous shortcut.
How to Use
Set cell active area (10–250 cm²) using the slider; typical PEM fuel cells operate at 50–100 cm².
Adjust current density (0–2.0 A/cm²); higher values increase water production from the ORR but amplify membrane drying risk.
Configure cell temperature (30–80 °C) and operating pressure (1–3 bar abs); higher temperature enhances proton conductivity but accelerates water evaporation.
Monitor Reaction H₂O generation rate, electroosmotic drag coefficient λ_drag (typically 0.8–1.4 H₂O/H⁺), and membrane resistance; green status indicates balanced water content.
Worked Example
A 100 cm² PEM stack operating at 1.2 A/cm² (120 A total), 65 °C, and 1.5 bar produces approximately 0.045 g/s H₂O at the cathode. With electroosmotic drag λ_drag = 1.1 and cathode hydration state λ = 14, proton conductivity reaches ~0.12 S/cm. Membrane resistance stabilizes near 0.18 Ω·cm² when back-diffusion balances water transport, maintaining the cell voltage at ~0.72 V under these humid conditions.
Practical Notes
Flooding risk increases above 1.5 A/cm² if cathode gas diffusion layer (GDL) lacks adequate drainage; reduce pressure or raise temperature to boost evaporation.
Membrane proton conductivity drops sharply below λ = 8; maintain cathode humidity by increasing operating pressure to 2.0–2.5 bar or reducing current density.
Electroosmotic drag consumes ~1 H₂O per proton in standard Nafion; lower drag (0.8–0.9) suggests membrane thinning or degradation.
At very high current density (>1.8 A/cm²) and low pressure (<1.2 bar), membrane drying dominates, causing voltage loss of 50–100 mV per 0.05 unit drop in λ.