How Do I Detect a Vacuum Tank Leak by Pressure Rise?

After the fixture is characterized, a vacuum leak larger than the permitted equivalent diameter produces a pressure-rise curve that crosses the validated rejection boundary. Build that boundary from absolute pressure, total connected volume, gas temperature, ambient pressure, and either a characterized reference leak or a compressible-flow model.

Test Definition and Acceptance Variable

Before anything else, confirm what the stated vacuum means. A value described as under 5 inHg can mean absolute pressure or vacuum relative to atmosphere. Those interpretations produce different pressure ratios and potentially different flow regimes.

1. Define the external absolute pressure P_u, normally the pressure surrounding the tank.
2. Define the initial tank absolute pressure .
3. Define the total gas volume V, including the tank, manifolds, sensor cavities, and tubing that remain connected during measurement.
4. Define the gas temperature T and the permitted equivalent leak diameter d_crit. If 0.0005 in is the rejection threshold, document whether that value describes a sharp-edged circular hole, a characterized reference leak, or an equivalent diameter assigned to a crack.
5. Choose the reported variable: absolute pressure rise, loss of indicated vacuum, calculated mass flow, or comparison with a reference curve.

For a leaking evacuated vessel, tank absolute pressure rises toward ambient pressure. A vacuum-gauge reading may fall at the same time. Calling both responses “pressure decay” invites reversed limits, so store and analyze absolute pressure whenever possible.

Do not move on until the test specification states the initial absolute pressure, surrounding absolute pressure, connected volume, temperature condition, measurement interval, and exact pass/fail variable.

Pressure and Temperature Measurement Setup

Select a pressure sensor that resolves the expected change over the allotted test interval without operating near a range limit. An absolute sensor directly supplies the pressures required by the gas-law and compressible-flow calculations. If a gauge or differential sensor is used, record ambient absolute pressure so both sides of the leak can be converted to absolute pressure.

1. Install the sensor on the volume that remains connected during the test. Avoid an isolation arrangement that measures a trapped branch instead of the tank.
2. Place the temperature measurement where it represents the test gas or establish enough stabilization for gas and wall temperatures to stop drifting materially.
3. Evacuate the assembly to , isolate the vacuum source, and mark isolation as time zero only after valve motion and pressure transients have settled.
4. Log tank absolute pressure, ambient absolute pressure, and temperature on the same time base.
5. Confirm that the recorded pressure rises for a known ingress path and remains nearly stationary for a sealed reference condition.

At fixed volume, temperature changes also change pressure. The differential form of the ideal-gas relation is:

dP/P = dm/m + dT/T

The constant-temperature assumption removes the second term. If temperature drifts during evacuation recovery, the resulting pressure change can look like leakage. Do not move on until repeated sealed-reference runs show that thermal recovery and sensor drift are smaller than the intended pass/fail separation.

Fixture Isolation and Baseline Leakage

The measurement includes every ingress path connected to the test volume, not just the tank defect. Valve seats, fittings, seals, sensor diaphragms, tubing, and the isolation connection can dominate the response from a 0.0005 in equivalent opening.

Observed response	Likely cause	Deciding check
Pressure rises similarly with the tank and a sealed reference	Fixture leakage, valve leakage, sensor offset drift, or thermal recovery	Repeat with the test port blanked or with a verified sealed reference volume
Early pressure rise is steep, then rapidly flattens	Isolation transient, gas-temperature recovery, or changing compressible flow	Compare delayed measurement windows and the temperature record
Nominally identical tests shift with weather or location	Gauge pressure used without ambient-pressure correction	Recalculate with measured absolute pressures on both sides
Calculated leak changes after tubing is added	Connected volume was omitted	Recalculate using total isolated gas volume
Reference runs scatter near the rejection line	Insufficient resolution, unstable starting pressure, temperature variation, or fixture repeatability	Trend each input separately across repeated runs

1. Run the complete fixture with its test interface sealed.
2. Record the baseline pressure-rise curve using the same evacuation, isolation, stabilization, and logging sequence planned for production tests.
3. Repeat the run to quantify repeatability rather than subtracting a single baseline result.
4. Correct for baseline only when fixture leakage is stable. Repair an unstable fixture instead of treating its changing response as a constant offset.

Do not move on until the blanked-fixture distribution is separated from the intended rejection boundary by a margin established from repeated measurements.

Tank Mass-Balance Model

For a rigid tank containing air at constant temperature, the ideal-gas relation connects pressure rise to accumulated mass:

m = P V / (R T)

dm/dt = V/(R T) · dP/dt

dP/dt = R T/V · m_dot(P_u, P, T, A, C_d)

Here, P is tank absolute pressure, R is the specific gas constant for air, A is leak area, and C_d is the discharge coefficient representing contraction and losses. For a circular equivalent opening:

A = pi · d^2 / 4

The mass inflow is not constant throughout the test. As tank pressure rises, the pressure ratio across the opening changes and the driving differential decreases. Tank density rises because mass enters the fixed volume; it does not rise because pressure “drops.” The vacuum indication decays while absolute pressure rises.

A physical crack cannot generally be mapped to diameter from area alone. Crack length, gap, wall thickness, entrance shape, surface condition, and flow path affect C_d. Equivalent diameter therefore means the diameter of a defined reference restriction producing the same response under the specified test conditions.

Check the measured initial slope against dm/dt = V(dP/dt)/(RT). Do not move on until the sign, pressure basis, temperature units, and total volume produce a positive ingress rate of credible magnitude.

Compressible Leak-Flow Selection

Use the external side as upstream because it has the higher absolute pressure. Let r = P/P_u and let gamma be the ratio of specific heats. Compare the instantaneous pressure ratio with the critical ratio:

r_crit = (2/(gamma + 1))^(gamma/(gamma - 1))

For r <= r_crit, use the choked-flow relation:

m_dot = C_d A P_u sqrt(gamma/(R T))
        · (2/(gamma + 1))^((gamma + 1)/(2(gamma - 1)))

For r > r_crit, use the subcritical relation:

m_dot = C_d A P_u sqrt(
  2 gamma/(R T (gamma - 1))
  · (r^(2/gamma) - r^((gamma + 1)/gamma))
)

Choked flow makes mass inflow insensitive to downstream tank pressure while the upstream state remains fixed. Once the ratio crosses the critical value, inflow decreases as tank pressure approaches ambient. The test may therefore begin with a nearly linear pressure rise and later bend toward ambient pressure.

Choose gas properties at a defined temperature and use one coherent unit system. Determine C_d from the restriction geometry or calibration; do not silently assign an ideal-orifice value to a crack. Do not move on until every sampled pressure ratio selects the intended branch and both equations meet continuously at the transition.

Pressure-Time Curve Generation

Generate the rejection curve by integrating the mass-balance equation because flow changes with tank pressure. A direct numerical update is:

P[i+1] = P[i] + delta_t · R T/V
                     · m_dot(P_u[i], P[i], T[i], A, C_d)

1. Set from the measured tank absolute pressure at the start of the evaluation window.
2. Set A = pi d_crit^2/4 for the critical equivalent diameter.
3. At each sample, calculate P[i]/P_u[i] and select choked or subcritical flow.
4. Calculate mass inflow and update tank pressure.
5. Repeat to the end of the specified window.
6. Reduce delta_t and repeat. Accept the step size only when further reduction does not materially move the predicted rejection boundary.

Create additional curves only for parameters that represent real uncertainty, such as connected volume, temperature, ambient pressure, or calibrated flow coefficient. A family of arbitrary curves can obscure the actual decision limit.

Compare the modeled curve with a test using a characterized reference leak at the critical level. Do not move on until the model reproduces both the initial slope and curve shape closely enough for the required classification.

Pass/Fail Boundary and End-to-End Verification

Use a characterized reference leak when equivalent hole size is the contractual result. It captures non-ideal restriction behavior and the complete instrument chain more directly than geometry alone.

1. Run a sealed reference to establish fixture baseline and drift.
2. Run the critical reference restriction from the same initial absolute pressure, ambient pressure, volume configuration, and temperature condition.
3. Run a restriction below the rejection level and one above it when such references are available.
4. Choose the evaluation feature: pressure rise over the defined interval, time to reach a defined absolute pressure, fitted mass-flow value, or departure from the complete critical curve.
5. Set acceptance limits from repeated distributions, including fixture baseline and measurement variation. Do not classify directly from one ideal calculation when the reference and model disagree.
6. Challenge the test with a known pass condition, the critical condition, and a known fail condition. Confirm correct classification, then repeat after reconnecting the test item to expose setup sensitivity.

The completed test is verified only when recorded absolute pressure, ambient pressure, temperature, start condition, connected volume, and reference identity reconstruct the same pass/fail result from the saved data.

Frequently Asked Questions