MRI Quench Room Oxygen Hazard Calculator

This calculator is a planning and training aid. It is not a substitute for a quench pipe that vents outdoors, for a calibrated oxygen monitor, or for the emergency procedures written by your MR medical director and equipment vendor. If a quench vents into an occupied scan room, evacuate first and calculate later.

Introduction to oxygen displacement during a superconducting magnet quench

A clinical superconducting magnet holds its field because the niobium-titanium windings sit in a bath of liquid helium at roughly 4.2 K. A quench is the sudden loss of superconductivity: the winding becomes resistive, the stored magnetic energy dumps into the coil as heat, and the helium bath boils. The gas has to go somewhere. On a correctly installed system it goes up the quench pipe and out of the building. When that path fails — an iced-over termination, a joint pulled apart during roof work, a pipe that was never fitted on an older installation — the helium vents into the scan room instead, and the hazard changes from a noisy inconvenience into an asphyxiation risk.

Helium is chemically inert and is not a poison. It kills by displacement: every cubic metre of helium that enters a fixed room pushes a cubic metre of breathable air out through the door, the waveguide penetrations, or the pressure relief panel. The gas that remains behind still contains 20.95 percent oxygen locally, but the room average falls, and it falls extremely quickly because of the expansion ratio involved. One litre of liquid helium becomes roughly 751 litres of gas once it warms to room temperature. A 1,500 litre cryostat therefore contains a little over 1,100 cubic metres of gas — more than ten times the free volume of a typical scan room. There is no ventilation system in a hospital that can keep up with that.

This calculator exists to turn that qualitative statement into a number you can put in a risk assessment. It takes the cryogen inventory, the release rate, the room volume, the fraction of the release the quench pipe actually captures, and any ventilation that is still running, and it reports the resulting oxygen fraction, how long you have before the room drops below the OSHA action level of 19.5 percent, and how long the room then has to be purged before anyone can go back in. It also reports the number that usually settles the argument in a design review: the quench-pipe capture fraction you would need in order for the room to stay above the threshold at all. For most rooms that number is well above 99 percent, which is precisely why the pipe, and not the ventilation system, is the engineering control.

How to use this quench model with your own site survey figures

1. Enter the liquid helium released from the magnet. Use the cryostat capacity from the magnet data sheet unless you are modelling a partial release; a clinical 1.5 T or 3 T system typically holds 1,000 to 2,000 litres. The unit selector accepts US gallons if that is how your inventory is recorded.
2. Enter the peak release rate in litres of liquid per second. Vendor quench data usually gives the bath emptying time; divide the inventory by that time. Forty to sixty litres per second is a representative figure for a fast clinical quench.
3. Enter the temperature of the warmed helium gas. This sets the expansion ratio. Room temperature (20 °C) is the right choice for a displacement assessment, because the gas warms to ambient long before anyone can leave the room.
4. Enter the free air volume of the scan room in cubic metres or cubic feet. Use the internal RF-cage dimensions and subtract the magnet, patient table and cabinets; the free volume of a typical single-bay scan room is 80 to 130 m³.
5. Enter the fraction of the helium captured by the quench pipe. Zero models a complete venting failure, which is the design-basis emergency. One hundred models an intact pipe. Intermediate values model a partial blockage or a leaking joint inside the room.
6. Enter any room exhaust or make-up airflow that keeps running during the event. Many building automation systems shut the air handler down when they see a quench, so zero is the conservative default; enter the emergency exhaust rate only if you have verified it stays on.
7. Enter the alarm and recognition delay — the time between the oxygen monitor reaching its set point and a person actually starting to move — and the oxygen action threshold your policy uses.
8. Run the calculation, then read the capture-fraction sensitivity table underneath. Compare at least the intact-pipe and total-failure rows before you draw any conclusion about staff safety.

Every result is recomputed from your inputs; nothing on this page is a stored lookup. The permalink button encodes the whole scenario in the address bar so that a scenario can be pasted into an email or a drill report, and the CSV button exports the same figures for a quench safety file.

Formula: expansion ratio, displaced volume and the room oxygen balance

Everything starts with how much gas the liquid makes. The liquid-to-gas expansion ratio is the ratio of the saturated liquid density at the normal boiling point to the density of the same helium once it has warmed to the temperature of the room:

Formula: E = ρ_liq / ρ_gas(T)

The NIST Chemistry WebBook gives the saturated liquid density of helium at 101.325 kPa as kg/m³ at a boiling point of 4.2143 K. The warm gas is very close to ideal at one atmosphere, so its density follows

Formula: ρ_gas(T) = (P M) / (Z R T)

with g/mol and a compressibility factor within 0.05 percent of unity. Fitting that expression to the NIST isobaric table between 280 K and 300 K reproduces every tabulated density to five significant figures and collapses to a single straight line:

Formula: E ≈ 2.561 × T

At 293.15 K that gives , and at about 295.7 K it gives the 1:757 figure that cryogenic gas suppliers publish for normal temperature and pressure. Older references — including the previous version of this page — used 1:700, which is roughly seven percent low and therefore understates the hazard. The calculator recomputes from the temperature you enter rather than assuming one value.

The gas volume produced by a release of litres of liquid, expressed in cubic metres, is

Formula: V_gas = (V_liq E) / 1000

and the portion that actually reaches the scan room, given a quench-pipe capture fraction , is . The unit chain matters here and is a common source of error: is litres of liquid, converts to litres of gas, and only the division by 1,000 converts litres of gas to cubic metres of gas so that it can be compared with a room volume.

Two bounding estimates of the mixed oxygen fraction

Once is known, the oxygen fraction left in a room of free volume depends on how the incoming gas mixes with the air that is already there, and the two limiting cases bracket the answer. If the helium behaves as a piston that pushes room air out ahead of it with no back-mixing, the surviving air fraction is simply the volume that has not yet been swept:

Formula: x_disp = x_0 (V_r − G) / V_r

This is the standard screening formula used in oxygen-deficiency hazard analyses, and it reaches zero when the released gas volume equals the room volume. If instead the room is perfectly stirred while the excess vents out through the relief path, the dilution is exponential rather than linear:

Formula: x_mix = x_0 e^−G/V_r

The screening result is always the lower of the two for , so it is the conservative one, and the calculator reports both. The pre-quench oxygen fraction defaults to 0.2095, the dry-air value of the U.S. Standard Atmosphere, not the rounded 0.21 that many quench spreadsheets use.

The time history while the helium is still flowing

To get a timeline rather than an end state you need the transient balance. Treat the room as a single well-stirred volume that receives helium gas at a volumetric rate cubic metres per second from a liquid release rate litres per second, and clean air at a rate , with an equal total volume leaving through the relief path so the room volume stays constant. The oxygen mole fraction then obeys

Formula: (d x) / (d t) = Q / V_r(x_in − x) − R / V_r x

which is a first-order linear equation with time constant and asymptote

Formula: τ = V_r / (R + Q) and x_∞ = (Q x_in) / (R + Q)

and

so that, starting from the pre-quench fraction ,

Formula: x(t) = x_∞ + (x_0 − x_∞) e^−t/τ

Inverting that expression gives the time at which the room crosses the action threshold :

Formula: t_thr = τ ln (x_0 − x_∞) / (x_thr − x_∞)

The expression is only meaningful while helium is actually flowing. The release lasts seconds, and the calculator refuses to report a threshold crossing later than that: if the inventory runs out first, the lowest oxygen fraction the room ever sees is and the threshold is never crossed. Ignoring that cap is the single easiest way to produce a reassuring but fictitious egress time, and it is the bug that was present in the earlier version of this page.

Recovery, mass, and the capture fraction you actually need

After the release stops, and the room becomes an ordinary purge problem, so oxygen climbs back toward the supply value with a time constant equal to one air change:

Formula: t_purge = V_r / Q ln (x_in − x_min) / (x_in − x_thr)

Recovering a room from pure helium back to 19.5 percent oxygen needs air changes regardless of room size, which is why the purge time is set almost entirely by the ventilation rate. With no ventilation the room never recovers on its own.

The mass of helium involved is best taken from the liquid side, because it is independent of temperature: , so 1,500 litres of liquid is 187 kg of helium. Finally, rearranging the screening formula for the capture fraction that just keeps the room at the threshold gives the design number:

Formula: η_req = 1 − (V_r(1 − x_thr /x_0)) / V_gas

Because is an order of magnitude larger than a scan room, this quantity is almost always above 99 percent. That is the quantitative reason the quench pipe, and not the air handler, is the control that keeps the room survivable.

Worked example: a total quench-pipe failure on a 1,500 litre 1.5 T magnet

Take a single-bay scan room with a free air volume of 100 m³ housing a 1.5 T magnet whose cryostat holds 1,500 litres of liquid helium. Vendor data gives a peak release of 45 litres of liquid per second, so the bath empties in 33.3 seconds. The oxygen monitor is set at 19.5 percent and, allowing for the alarm sounding and a technologist registering it, the recognition delay is 10 seconds. The building automation system drops the air handler on a quench signal, so ventilation during the event is zero. We first run the design-basis case: the quench pipe captures nothing at all.

At 20 °C the expansion ratio is 2.561 × 293.15 = 751, so the bath produces 1,500 × 751 / 1000 = 1,126 m³ of gas. That is 11.3 times the free volume of the room. Both mixed-fraction estimates therefore collapse: the screening formula gives zero oxygen as soon as the released volume reaches 100 m³, and even the optimistic well-stirred estimate gives 20.95 × e^−11.26, which is three ten-thousandths of one percent. The room is helium.

The timeline is worse than the end state suggests. The helium enters at 45 × 751 / 1000 = 33.8 m³ per second, so the mixing time constant is 100 / 33.8 = 2.96 seconds and the room passes 19.5 percent oxygen after 2.96 × ln(20.95 / 19.5) = 0.21 seconds. Against a 10 second recognition delay there is no egress window at all: by the time the alarm has been noticed the room has been an oxygen-deficient atmosphere for about ten seconds and is heading for zero. The mass involved is 1,500 × 0.12486 = 187 kg of helium. With the air handler off there is no purge at all; if an emergency exhaust delivering 12 m³/min can be started, recovery back above 19.5 percent still takes 2.67 air changes, which is 22 minutes.

Now ask the design question. The capture fraction that would just hold the room at 19.5 percent is 1 − 100 × (1 − 19.5 / 20.95) / 1126 = 99.39 percent. A pipe that captured 99 percent of the release would still let 11.3 m³ of helium into the room, taking it to 18.6 percent by the screening formula and 18.7 percent well-stirred, crossing the threshold 21 seconds into the release. A pipe that was 90 percent effective — a partial ice blockage, say — would admit 113 m³, more than the room holds, and leave a well-stirred fraction of 6.8 percent, which the published effects table below places in the range that causes loss of consciousness. There is no ventilation upgrade that competes with the pipe; the entire safety case rests on the vent path staying open.

It is worth saying plainly what this replaces. The earlier version of this page reported 26 seconds to threshold and a 16 percent equilibrium oxygen fraction for a similar scenario, and printed a mitigation table showing egress windows of 26 to 88 seconds. None of those figures were produced by its own equations, which give roughly one second for the same inputs, and it used an expansion ratio of 700 rather than the temperature-corrected value. A quench-response plan built on a minute of egress time when the true figure is a fraction of a second is not a conservative plan; it is a dangerous one.

Reading the oxygen number against published exposure effects

The 19.5 percent figure the calculator uses as its default threshold comes from OSHA, which defines an oxygen-deficient atmosphere in 29 CFR 1910.134(b) as one with an oxygen content below 19.5 percent by volume, and which treats an oxygen concentration below 19.5 percent or above 23.5 percent as a hazardous atmosphere in the permit-required confined space standard 29 CFR 1910.146(b). That is an action level, not the onset of injury. The table below is reproduced from the U.S. Chemical Safety and Hazard Investigation Board bulletin on nitrogen asphyxiation, which credits it to the Compressed Gas Association; the physiology is the same whichever inert gas does the displacing.

Two features of that table matter for quench planning. The first is that the onset of impairment is silent: there is no smell, no irritation and no dyspnoea to warn a technologist that the room has gone bad, which is why an oxygen monitor rather than human judgment has to be the trigger. The second is how narrow the usable band is. Between the 20.95 percent of ordinary air and the 10 percent at which a person can no longer walk out there is only about half of one room volume of helium. In the worked example above the room crosses that entire band in under two seconds.

Interpret the calculator output accordingly. A result above 19.5 percent means the modelled release stayed inside the regulatory band, not that the room is safe — the model is a room average, and it says nothing about the helium layer at head height. A result between 16 and 19.5 percent means an oxygen-deficient atmosphere in the OSHA sense, with impairment that the affected person will not reliably notice. A result below about 12 percent means a room that a person may be unable to leave unaided, and any egress window the calculator prints for that case should be read as the time available to someone who is already moving toward the door, not to someone who has to decide to.

Limitations of a perfectly mixed single-zone quench model

The most important limitation is also the least intuitive, so it goes first: perfect mixing is not the conservative assumption here. Helium gas at 20 °C has a density of about 0.166 kg/m³ against roughly 1.20 kg/m³ for air, so it is about seven times lighter. A quench plume rises immediately and spreads as a layer that grows downward from the ceiling rather than blending uniformly through the room. Everything this calculator reports is a room average, and while that layer is descending the oxygen fraction at standing head height is far lower than the average — locally it can be essentially zero while the reported average is still comfortable. The ACR Manual on MR Safety reflects this directly, instructing anyone leaving the room during a venting failure to walk as low to the floor as possible to keep their head below the accumulating gas. Treat the numbers here as a lower bound on severity and an upper bound on available time.

The model also ignores the thermal transient. Helium leaves the cryostat near its boiling point, and cold helium is much denser than warm helium, so the first seconds of a release can behave quite differently from the buoyant layer that follows; the white cloud people describe is condensed atmospheric moisture, not the helium itself. Cold-related injury is a real and separate hazard during a venting failure — the ACR manual lists frostbite and hypothermia alongside asphyxiation — and none of it is modelled here.

Pressure is not modelled either, and in a real venting failure it may be the first thing to matter. Delivering 34 m³ of gas per second into a sealed radiofrequency enclosure raises the room pressure fast enough to blow out the pressure relief panel, distort the door frame, or make an inward-opening door impossible to push open. The oxygen balance used here explicitly assumes that the excess volume leaves the room freely through a relief path; if it cannot, the room is a pressure vessel and this calculation no longer describes it.

Several further assumptions are worth stating explicitly. The release rate is treated as constant until the inventory is exhausted, whereas a real quench peaks early and tails off, so the early part of the timeline is conservative and the late part is not. The capture fraction is a single constant, so a vent that ices up progressively during the event cannot be represented; run the calculation twice, once at each end of the range, instead. The ventilation rate is likewise constant and the make-up air is assumed to be ordinary outdoor air. The expansion ratio uses an ideal-gas fit to the NIST isobaric table that is exact between 280 K and 300 K at one atmosphere and degrades outside that range, and the liquid density is taken at the normal boiling point, so a pressurised or subcooled bath will differ slightly. The whole inventory is assumed to be released; magnets that retain part of the bath, and systems with a separate nitrogen jacket, need their own figures.

Finally, the calculation says nothing about the other hazards that make a quench dangerous: the magnetic field persists for a minute or more after the field starts collapsing, ferromagnetic objects brought in by responders remain lethal projectiles, and the decision to quench a magnet is itself a clinical judgement. The governing documents for an actual installation are NFPA 99, the Health Care Facilities Code, which covers siting and construction of the MR suite; IEC 60601-2-33, the particular standard for the basic safety and essential performance of magnetic resonance equipment, which covers quench-related emergency provisions; and the vendor's own siting manual. A calculator is a way to understand why those documents require an external quench vent. It is not a way to argue that one is unnecessary.

Questions engineers ask about quench oxygen displacement

Sources for the constants and thresholds used here

Helium properties: National Institute of Standards and Technology, NIST Chemistry WebBook, SRD 69 — Thermophysical Properties of Fluid Systems (helium). Saturated liquid at 101.325 kPa: T = 4.2143 K, density 124.86 kg/m³. Gas at 101.325 kPa: 0.16812 kg/m³ at 290 K, 0.16527 at 295 K, 0.16252 at 300 K, from which the expansion ratio used here is derived (webbook.nist.gov/chemistry/fluid). Oxygen-deficiency thresholds: U.S. Occupational Safety and Health Administration, 29 CFR 1910.134(b), Respiratory Protection — Definitions (osha.gov 1910.134) and 29 CFR 1910.146(b), Permit-Required Confined Spaces — Definitions (osha.gov 1910.146). Physiological effects table: U.S. Chemical Safety and Hazard Investigation Board, Safety Bulletin: Hazards of Nitrogen Asphyxiation (No. 2003-10-B, June 2003), table "Effects of Oxygen Deficiency on the Human Body", attributed by the Board to the Compressed Gas Association, 2001 (csb.gov). The Compressed Gas Association publishes the same material in its oxygen-deficient atmospheres bulletin, formerly SB-2 and now CGA P-76. MR-specific practice: American College of Radiology, ACR Manual on MR Safety (2024), sections on the Cryogen Venting Zone and on Emergency Magnet Off (Quench), including the instruction to leave the room staying as low to the floor as possible during a venting failure (acr.org MR safety). Installation requirements: National Fire Protection Association, NFPA 99, Health Care Facilities Code, and International Electrotechnical Commission, IEC 60601-2-33, Medical electrical equipment — Part 2-33: Particular requirements for the basic safety and essential performance of magnetic resonance equipment for medical diagnosis. Both are consulted for the installed quench-vent arrangement; neither is quoted numerically on this page because both are paywalled standards. Pre-quench air composition: NOAA, NASA and U.S. Air Force, U.S. Standard Atmosphere, 1976, dry-air oxygen fractional volume 0.209476.