How to size a vacuum pump: pump-down time, conductance and effective speed
Every vacuum pump enquiry we receive can be reduced to three numbers: the volume to evacuate, the pressure it has to reach, and the time allowed to get there. Everything else — the gas, the temperature, the pipework, the leaks — is a correction to those three. This guide sets out the arithmetic in the order a plant engineer needs it, then applies it to real Dr.Vakuum machines so the result is a model code, not a formula.
1. The pump-down equation
For a closed chamber with no gas load, the pressure falls exponentially:
t = (V / S) × ln(p₀ / p₁)
- V — chamber volume in m³ (include the pipework up to the pump)
- S — pumping speed in m³/h at the inlet of the chamber
- p₀ — starting pressure, normally 1,013 mbar
- p₁ — target pressure in mbar
A worked case: a 2 m³ vacuum vessel must reach 10 mbar in 4 minutes.
S = (V / t) × ln(p₀ / p₁) = (2 / 0.0667 h) × ln(1013 / 10) = 30 × 4.62 ≈ 139 m³/h
A 140 m³/h single-stage rotary vane pump — a V140 at 132 m³/h on 50 Hz — looks like the answer. It is close, but two corrections usually push the real requirement up, and skipping them is how pumps end up "too small" on site.
Why does the simple formula underestimate the pump?
Because it assumes the pump's speed is constant down to the target and that all of it arrives at the chamber. Neither is true. Pumping speed falls as the pressure approaches the machine's ultimate, and every metre of pipe between pump and vessel throws some of the speed away. The two corrections below handle those effects; applied together they typically add 30–100 % to the first estimate.
2. Correction one: the ultimate-pressure floor
A pump cannot reach its own ultimate pressure in finite time. Replace the pressures with their distance from the ultimate, p_ult:
t = (V / S) × ln[(p₀ − p_ult) / (p₁ − p_ult)]
For a target far above the ultimate this changes nothing — a 0.5 mbar pump reaching 10 mbar barely notices. For a target close to the ultimate it changes everything. Ask a 33 mbar liquid ring pump to reach 40 mbar and the term (p₁ − p_ult) is 7 mbar, not 40: the last few millibar take longer than the first nine hundred. This is why a liquid ring pump is the right machine for a 60–200 mbar process and the wrong machine for 35 mbar, even though 35 mbar is inside its catalogue range.
The practical rule: choose a pump whose ultimate is at least a decade below your working pressure if pump-down time matters. For 10 mbar, pick a 0.5–1 mbar machine; for 0.1 mbar, pick a two-stage pump at 5 × 10⁻³ mbar such as the DV60.
3. Correction two: conductance, or why the pipe is the pump
The speed printed on the datasheet is at the pump inlet. The speed at the vessel — the effective speed S_eff — is always lower:
1 / S_eff = 1 / S + 1 / C
where C is the conductance of the connecting pipework, in the same units as S. Conductance adds like electrical resistance: pipe sections in series add as reciprocals, and a valve or a filter is another series element.
In the viscous flow regime (above roughly 1 mbar for ordinary pipe sizes) the conductance of a round pipe is approximately
C ≈ 135 × d⁴ × p̄ / L (m³/h, with d and L in cm, p̄ in mbar)
where p̄ is the mean pressure in the pipe. The fourth power on the diameter is the whole lesson: doubling the pipe bore gives sixteen times the conductance. Halving it costs you fifteen-sixteenths.
Back to the worked case. Suppose the V140 is connected through 6 m of DN25 pipe (d = 2.5 cm, L = 600 cm) and we care about the speed near the end of pump-down, at p̄ ≈ 15 mbar:
C ≈ 135 × 39 × 15 / 600 ≈ 132 m³/h
1 / S_eff = 1/132 + 1/132 → S_eff ≈ 66 m³/h
Half the pump has gone into the pipe. The 4-minute target becomes 8 minutes. Moving to DN50 pipe gives C ≈ 2,100 m³/h and S_eff ≈ 124 m³/h — nearly the whole pump. On a short-cycle machine, changing the pipe is almost always cheaper than changing the pump.
Below about 1 mbar the flow becomes molecular, conductance stops depending on pressure, and the d³/L form applies; that is the regime where turbomolecular backing lines are sized, and it is covered in our note on high vacuum for coating and analytical instruments.
4. Gas load: the pump that never finishes
Pump-down time assumes a sealed, dry, empty vessel. Real processes add gas continuously, and once the inflow equals what the pump removes the pressure stops falling. The steady-state pressure is
p = Q / S_eff
with Q the total gas load in mbar·L/s (or mbar·m³/h, as long as the units match). Gas load comes from four places, and sizing means estimating each:
- Process gas — the vapour you are actually removing: water from a dryer, solvent from an evaporator, air drawn through a hold-down table. Usually the largest term, and often known from the process mass balance (kg/h of water → m³/h of vapour at the working pressure).
- Leaks — real holes. A well-built industrial system leaks 10⁻³–10⁻² mbar·L/s; a system with many flanges and sight glasses can leak a hundred times that.
- Outgassing and desorption — water released from walls and product. Dominant below 1 mbar on anything that has been open to humid air, which in Thailand is everything.
- Permeation and virtual leaks — trapped volumes behind bolts and in blind holes. Only matters in high vacuum.
A dryer example: 20 kg/h of water evaporating at 50 mbar. At 50 mbar and 40 °C, one kilogram of water vapour occupies about 29 m³, so the pump must remove 580 m³/h of vapour at 50 mbar — and the vapour must not condense inside it. That requirement, not pump-down time, sizes the pump: a liquid ring machine such as the LT250-200 at 525 m³/h, or a condenser ahead of a smaller pump. The condenser is usually the better answer: condense 90 % of the vapour at 30 °C and the pump only sees 58 m³/h.
5. Which curve to read, and where
Catalogue speed is a nominal figure at atmospheric inlet. What matters is the speed at your working pressure, read from the machine's performance curve. Three habits avoid the common mistakes:
- Read at the working pressure, not at the nominal point. A single-stage vane pump holds nearly full speed down to about 10 mbar and then falls away; a liquid ring pump loses speed steadily below 100 mbar; a side channel blower loses speed from the first millibar of differential and is finished by 300–400 mbar below atmosphere.
- Check 50 Hz. Thai supply is 50 Hz. Datasheets written for 60 Hz markets show 20 % more flow than you will get.
- Derate for temperature and altitude only where relevant: a 40 °C plant room changes gas density by 5 %, which is inside the margin you should carry anyway.
Every series page on this site shows its curve with the models overlaid, and each model page highlights its own curve against the siblings — the V Series chart is a good place to start. Reading those curves is a skill in itself; we cover it in Reading a performance curve.
6. Worked examples by regime
| Duty | Volume / load | Target | First estimate | After corrections | Dr.Vakuum answer |
|---|---|---|---|---|---|
| Packaging chamber, 120 L, 20 cycles/h | 0.12 m³ in 15 s | 5 mbar | 153 m³/h | ~160 m³/h, DN40 short hose | V140 |
| Vacuum vessel, 2 m³ | 2 m³ in 4 min | 10 mbar | 139 m³/h | 180 m³/h with DN50 pipe | V301 or 2 × V140 |
| Tray dryer, 20 kg/h water | continuous | 50 mbar | 580 m³/h vapour | 60 m³/h after condenser | LM200-065 + condenser |
| Load-lock, 40 L | 0.04 m³ in 20 s | 0.1 mbar | 66 m³/h | 90 m³/h, two-stage | DV90 |
| CNC hold-down, porous MDF | leak ≈ 200 m³/h | 300 mbar diff. | n/a (leak-sized) | 312 m³/h at 250 mbar | 1SC529 |
| Central plant, 12 moulding machines | 0.9 m³ total, 70 % simultaneity | 50 mbar | 400 m³/h | 630 m³/h, VSD | DSI630 |
Two patterns in the table are worth noticing. Short-cycle duties are sized by pump-down time and want a pump with headroom above the target pressure. Continuous duties are sized by gas load, and the sensible engineering is usually to reduce the load (condense, filter, seal) before buying flow.
7. The margin question
Oversizing costs energy for the life of the machine; undersizing costs the process. Our working practice:
- +25 % on a clean, well-understood batch duty.
- +50 % where the gas load is estimated rather than measured, or the system has many flanges.
- Variable-speed drive rather than oversizing on any central system — the margin is there when needed and not paid for when it is not. The DSI Series dry screw pumps are designed for exactly this.
Recommended Dr.Vakuum machines by regime
| Working pressure | Technology | Series | Typical sizes |
|---|---|---|---|
| 1,013 → 400 mbar (high flow) | Side channel blower | 1SC / 2SC | 48–1,134 m³/h |
| 1,013 → 500 mbar (very high flow) | Roots blower, vacuum duty | RBV | 59–3,060 m³/h |
| 500 → 150 mbar, dry | Dry claw | CL | 75–300 m³/h |
| 500 → 90 mbar, dry, oil-free | Dry rotary vane | DRV | 12–400 m³/h |
| 200 → 33 mbar, wet or dirty gas | Liquid ring | LM / LT / LS | 30–3,000 m³/h |
| 100 → 0.5 mbar | Single-stage oil-sealed vane | V | 10–640 m³/h |
| 10 → 0.05 mbar, dry | Dry screw | DSI | 150–1,200 m³/h |
| 10 → 5 × 10⁻⁴ mbar | Two-stage oil-sealed vane | DV | 5–255 m³/h |
| 10 → 10⁻³ mbar, high flow | Roots booster + backing pump | MB | 280–4,140 m³/h |
| < 10⁻³ mbar | Turbomolecular / cryopump | T / CR | 600–4,200 L/s |
Sizing help. Send us the volume, the target pressure, the cycle time and what is in the gas. Our engineers size against the actual curve at duty, including your pipework, and tell you where a smaller pump with a bigger pipe would do. Talk to an engineer.