How to Spec a Spray Lance or Injection Quill for Vessel and Duct Duty

A dosing skid running at 6 bar pushed its contents back into the day tank overnight. The chemical injection quill on that line had a stuck-open nozzle check valve. The tank sat about 3 m below the injection point. Cleanout took a full shift and two tanker loads of neutralised waste.

Nothing was wrong with the tip. The failure lived in cracking pressure, insertion depth, and material choice. This guide is a decision path for how to spec spray lances or a chemical injection quill. Each section carries the number that settles one choice.

Two duty classes dominate, and they pull the design in opposite directions. Inside a vessel, the hardware sits below the liquid surface and injects into liquid. Depth, mixing and backflow protection rule everything there. Inside a duct, the same hardware fires into a hot gas crossflow.

Penetration, droplet size and evaporation time rule that case. The geometry looks similar on a drawing, but the sizing logic is not. Both cases share one discipline: fix the acceptance number before the part number.

The Snapshot

• Submerged vessel duty: set the tip 150 to 300 mm below the liquid surface, clear of the vapour space and the tank wall.
• Orifice flow follows Q(L/min) ≈ 0.65 · d(mm)² · √P(bar), so a 3 mm bore at 4 bar delivers 11.7 L/min.
• Check valve cracking pressure typically runs 0.3 to 1 bar (published range) and must exceed the static back-pressure at the injection point.
• Duct duty at 400 C: a 3 mm water jet into a 15 m/s crossflow reaches 0.34 to 0.56 m of penetration (published range).
• Replace the lance when flow exceeds the rated figure by more than 5 percent, since a 5 percent bore growth returns about 10 percent over-delivery.

Failures That Trace Back to Insertion Depth

A lance that stops short of the surface fires into the vapour space. The sheet breaks into coarse ligaments. It wets the tank roof and drains back down the wall as a film. In a heated vessel that film bakes onto the shell.

The wall then runs hotter than the bulk liquid. Coking nucleates along the wet-dry line where the film stops renewing itself. Once coke forms, the deposit insulates the wall and the metal under it overheats. The failure escalates from a fouling problem into a metallurgical one.

The opposite error is just as common. A lance pushed too deep throws its whole jet onto one patch of wall. Impact concentrates in a spot the size of a dinner plate.

Local mixing collapses and the injected chemical never disperses. The analyser downstream then sees a spike. The cure is depth control plus an offset angle, not a bigger pump.

Backflow is the third failure mode, and the most expensive. When a nozzle check valve fails to seat, line pressure drives process fluid up the quill. From there it reaches the dosing header, the metering pump, and sometimes the day tank.

Contamination of an entire chemical system is the result. One published range for the cleanup window on a continuous line is 4 to 8 hours of lost production. Add waste disposal and a pump rebuild to that figure.

All three failures share a root cause: the spec was written around flow rate alone. Flow rate tells you how much arrives. It says nothing about where the jet lands or how deep it penetrates. It also says nothing about what travels backwards when the pump stops.

The Four Hardware Families and Where Each Fits

Lance hardware splits into four families. Nearly all spray lances and injection quills land in one of them. Picking outside the family band is where most rework starts.

Family	Mounting	Typical duty	Insertion	Pressure band	Key limit
Straight spray lance	Flanged or threaded boss	Open-tank washing, quench, duct cooling	Fixed, 50 to 400 mm	1 to 10 bar	No retraction, no purge
Chemical injection quill	Threaded into a welded half-coupling	Submerged dosing into liquid or gas	Fixed, 100 to 600 mm	2 to 40 bar	Needs a check valve to hold header pressure
Retractable lance	Packing gland or ball-valve insertable seat	Duty where the tip must be pulled while the vessel stays live	Travel 300 to 1,000 mm	1 to 20 bar	Gland leakage sets the maintenance interval
Jacketed or purged lance	Flanged with a jacket or gas purge port	Hot ducts, sticky products, coking service	Fixed or retractable	1 to 20 bar	Purge flow must be continuous

A straight spray lance is the simplest option. It is a tube with one or more orifices, bolted into a vessel or duct wall. It suits open-tank washing, gas quenching, and light fouling duct cooling. The tip never needs to come out during a run.

A chemical injection quill adds a sealing thread, a defined tip position, and a valve. That nozzle check valve usually sits at the tip itself. Valve position matters more than valve brand.

Mounting the valve back at the header leaves a column of process fluid standing in the quill. A crystallising or polymerising product sets solid in that dead column and blocks the tube. Keeping the valve at the tip keeps the dead volume near zero.

A retractable lance solves a different problem: maintenance without a shutdown. The tip travels through a gland or a ball-valve seat. An operator can pull it, clean it, and push it back while the vessel stays live. The trade is a dynamic seal that wears. Gland leakage, not nozzle wear, is the usual reason these units get rebuilt.

A jacketed or purged lance handles hot and sticky service. A steam jacket keeps the product above its crystallisation point. A gas purge keeps the orifice clear and shields the tip from radiant heat. Both add a second utility that must never be lost. Losing purge in a coking duct is how a tip becomes a plug.

Penetration Depth and Offset Angle in a Duct

Duct injection is a transverse jet problem. One dimensionless group decides most of it. The momentum flux ratio J = (ρ_jet · v_jet²) / (ρ_gas · v_gas²) compares jet momentum against crossflow momentum.

Penetration depth then scales with its square root: y_max / d = C · √J. Published correlations for round transverse jets put C in the 1.5 to 2.5 range. Treat the output as an order-of-magnitude check, not a guarantee.

Work a concrete case. A 3 mm orifice at 11.7 L/min gives a jet velocity of 27.6 m/s. That is 1.95 × 10⁻⁴ m³/s through a 7.07 mm² area. Flue gas at 400 C runs at 15 m/s with a density near 0.6 kg/m³.

Water is 1,000 kg/m³, so J = (1000 × 27.6²) / (0.6 × 15²) = 761,760 / 135 ≈ 5,643. Then √J ≈ 75. At C = 1.5 the jet reaches y/d ≈ 113, or 0.34 m.

At C = 2.5 it reaches y/d ≈ 188, or 0.56 m. That is the published range quoted in the snapshot. A 3 mm jet is therefore a near-wall injector in a wide duct.

Offset angle is the cheap correction. A jet aimed square into the crossflow loses its column early. The gas pushes the jet back onto itself and creates a recirculation pocket. Droplets then land on the upstream duct wall and run down as a wet streak.

That streak pulls dust out of the gas and builds a scale ridge. Tilting the injection axis 15 to 30 degrees downstream stabilises the trajectory. The jet bends once, in a predictable direction, and penetrates farther before break-up. Past about 35 degrees the jet starts skimming the wall instead, so the useful window is narrow.

Liquid-in-liquid injection flips the geometry. A vessel mixed by a side-entry agitator wants the quill aimed with the flow. Offset it 10 to 20 degrees from the agitator discharge. Counter-flow injection creates a stagnant zone on the far side of the quill.

The chemical then reads high at the probe and low in the bulk. A rotating spray nozzle used for rinse duty is a separate device. Position the quill so the rotating jet never sprays straight into its tip.

Orifice Flow, Droplet Size and Evaporation

Every quill starts with one equation. For water-like liquids, with a discharge coefficient near 0.95:

Q(L/min) ≈ 0.65 · d(mm)² · √P(bar)

Substitute a 3 mm bore at 4 bar. Then Q ≈ 0.65 × 3² × √4 = 0.65 × 9 × 2 = 11.7 L/min. Substitute a 2 mm bore at 9 bar: Q ≈ 0.65 × 2² × √9 = 0.65 × 4 × 3 = 7.8 L/min. The same relation runs backwards for troubleshooting. A lance rated 11.7 L/min that now measures 12.9 L/min has a worn bore.

Pressure changes rescale flow by the square root of the pressure ratio: GPM₂ = GPM₁ · √(P₂/P₁). A 2.0 gpm tip at 40 psi gives 2.0 × √(25/40) = 2.0 × 0.791 = 1.58 gpm at 25 psi. That is 5.98 L/min. Spray angle does not change with pressure. A low-pressure correction therefore never fixes a coverage problem.

Droplet size decides whether a duct injection works. A droplet must evaporate before it reaches the wall or downstream equipment. Large drops survive the trip, wet the wall, and build a deposit.

Published pressure-swirl data puts Dv90 at 120 to 250 µm for a 1 mm bore across the 3 to 10 bar band. A pneumatic air-atomizing nozzle at 3 to 5 bar air pressure reaches Dv90 of 30 to 60 µm. For evaporative duty above 300 C, aim for Dv90 under about 150 µm.

Residence time is straightforward geometry. A duct at 15 m/s through a 12 m straight run gives 12 / 15 = 0.8 s of travel. Published evaporation figures put a 100 µm drop at 0.15 to 0.3 s in 400 C gas.

A 200 µm drop needs 0.6 to 1.2 s. A 0.8 s window therefore tolerates Dv90 near 150 to 180 µm, with little margin. Anything coarser needs air atomisation, which is why that hardware appears in most evaporative duct duty.

Free-fall data explains the wet floor failure. A 100 µm water droplet settles at about 0.25 m/s. A 50 µm droplet settles at about 0.07 m/s (published range). In a slow duct a 250 µm drop can fall out of the stream within a few metres. Matching Dv90 to the residence time is the whole game.

Nozzle Check Valve Cracking Pressure and Drip Control

Backflow protection is a specification, not an accessory. A nozzle check valve needs three numbers written down. They are cracking pressure, reseal pressure, and leak rate.

Cracking pressure is the forward pressure that opens the valve. Typical published values sit at 0.3 to 1 bar. Light springs open near 0.3 bar and heavier ones near 1 bar. The selection rule is arithmetic.

Take the maximum static head at the injection point, then add process pressure and a margin. A quill injecting into a vessel holding 0.6 bar of nitrogen needs more than 0.3 bar. A 0.3 bar spring will chatter open and let gas push back.

A 1 bar spring holds. The same logic covers a quill injecting into a line below the header. A 3 m liquid column alone is about 0.3 bar, so a 0.3 bar spring sits right at that boundary and should be raised.

Reseal behaviour decides drip. A drip-free valve closes fully before the pump’s next stroke. Published drip-free criteria are usually written as no visible drip for 60 seconds after the stroke ends.

Rated pressure is the reference point for that test. Weeping tips are not cosmetic. In a hot duct, a weeping tip grows a deposit that eventually blocks the orifice and shifts the pattern.

Two failure modes dominate. The obvious one is a particle lodged on the seat. That holds the valve open permanently and lets the process side win.

The subtler one is crystallisation behind the seat. Product creeps into the spring cavity, dries, and gums the poppet. The valve then sticks open at low flow and shut at high flow.

Operators see erratic dosing and blame the pump. Spring fatigue is the long-term mode. A spring cycled past its design deflection every stroke loses preload over months.

The practical answer is a three-part routine. Check the seat for a full 360 degree witness mark. Measure cracking pressure against the original figure. Record the reseal drip test.

Any valve that cannot hold 0.2 bar above its rated value should be replaced. Adjustment is not the answer at that point.

Materials, Temperature Limits and Replacement Triggers

Material choice follows temperature and chemistry. Both need a number on the datasheet.

Material	Continuous temperature, published range	Handles well	Watch out for
316L stainless	Up to about 425 C in dry oxidising gas	Most neutral and mildly acidic liquids, steam, water	Chloride pitting above about 60 C, hot caustic stress cracking
Hastelloy C-276	Up to about 600 to 675 C	Strong chlorides, wet chlorine, mixed acids	Cost, and lower yield strength than carbon steel
PTFE	Up to about 260 C	Nearly all chemicals, low friction	Cold flow under load, poor abrasion resistance
PVDF	Up to about 150 C	Oxidisers, halogens, high-purity duty	Lower ceiling than PTFE, notch sensitivity

For a duct above 400 C with hot, oxidising gas, 316L is a reasonable default. Alloy C-276 answers the same duty when chlorides are present. That pairing covers most spray lances in duct duty. In a tank at ambient temperature, PVDF or PTFE liners often outlast bare stainless in aggressive chemistry. Support the quill so the liner carries no bending load.

Wear is a flow problem before it is a metallurgical one. Orifice wear widens the bore, and flow scales with the square of diameter:

flow increase ≈ (1 + x/100)² − 1

For 5 percent bore growth: (1.05)² − 1 = 1.1025 − 1 = 0.1025, or about 10 percent. For 10 percent growth: (1.10)² − 1 = 1.21 − 1 = 0.21, or 21 percent. That is why the replacement trigger is written on flow, not on appearance.

Measure the delivered volume and compare it with the rated figure. Schedule spray nozzle replacement when the deviation passes 5 percent. A lance that looks fine can be dosing 20 percent heavy.

Installation, Flushing and Pressure Test Sequence

Sequence matters, because each step validates the one before it. Skipping forward hides defects.

1. Verify the bore and orifice count against the datasheet before installation. Confirm d within 0.05 mm of the nominal figure.
2. Set the insertion depth from the vessel or duct wall to the tip. Do not measure from the boss face. Submerged duty: 150 to 300 mm below the low liquid level. Duct duty: tip flush to 20 mm inside the wall.
3. Check the offset angle with a protractor or laser pointer before final tightening. Duct duty: 15 to 30 degrees downstream. Vessel duty: 10 to 20 degrees with the flow.
4. Flush the quill and the connecting line with clean water or compatible solvent. Run at the working flow rate for at least 2 minutes, tip to drain. This removes weld scale and swarf that would seat on the valve.
5. Purge the gas path on a jacketed or purged lance. Confirm continuous purge flow at the design figure before applying heat. Loss of purge in a hot duct is the leading cause of plugged tips.
6. Hydro-test the assembly at 1.5 times the maximum working pressure. Hold for 10 minutes. No leak at the threaded joint, the weld, or the gland is acceptable.
7. Set the check valve and confirm cracking pressure within 10 percent of rated. Then run the drip-free test: no visible drip for 60 seconds after the stroke.
8. Commission at design flow and verify the delivered volume with a graduated cylinder or flow meter. Record the measured figure as the wear baseline.
9. Mark the baseline on a tag at the skid. Note the date, the measured flow, the orifice diameter, and the cracking pressure.

A verification step without a recorded number is not a verification step. The baseline tag is what turns a future flow reading into a maintenance decision.

Common Specification Mistakes

• Specifying on flow rate alone. Many spray lances are bought this way. The duty also needs depth, angle, and droplet size.
• Leaving the nozzle check valve at the header. That creates a dead column of product in the quill. Put the valve at the tip.
• Choosing cracking pressure below the static back-pressure. A 0.3 bar spring in a 0.6 bar vessel will chatter and pass gas. Add head plus margin.
• Ignoring the boss face when setting depth. Measure to the tip, never to the flange face. Field depth errors of 50 to 100 mm are routine.
• Aiming square at a duct crossflow. A 15 to 30 degree downstream tilt buys penetration for free. A square jet bends into the wall.
• Using a coarse pressure-swirl tip for evaporative duty. Dv90 of 200 µm and a 0.8 s residence time do not coexist.
• Sizing shaft power on feel. Use kW = Q(L/min) × P(bar) / 612. A 20 L/min lance at 100 bar needs 3.27 kW.
• Treating wear as a visual item. Orifice wear is invisible at 5 percent bore growth but already costs 10 percent over-delivery.
• Running without purge on a jacketed lance. In sticky or coking service the purge is not optional.

Acceptance Criteria at Handover

Handover needs four measured numbers. Delivered flow within 5 percent of rated at design pressure. Check valve cracking pressure within 10 percent of the datasheet value. No visible drip at the tip for 60 seconds after the stroke ends. Coverage confirmed on the target with a strip, test card, tracer or probe.

Add a mechanical check that no one enjoys but everyone needs. Confirm the quill cannot vibrate loose. Cantilevered spray lances in a duct see crossflow excitation. A thread with no locking feature can back out over months. A tack weld on the boss or a locknut with a witness mark is cheap insurance.

Document the baseline, the date, and the acceptance numbers in the maintenance file. The next inspection then compares against a real figure instead of memory.

FAQ

How deep should a spray lance sit below the liquid surface?

Land the tip 150 to 300 mm below the lowest expected liquid level. Closer to the surface, the jet breaks into the vapour space and causes wall film. Deeper than 300 mm adds no mixing benefit and makes retraction harder.

What cracking pressure should a nozzle check valve have?

Set cracking pressure above the maximum static back-pressure at the injection point. Published spring ranges run 0.3 to 1 bar. A 3 m liquid column alone is about 0.3 bar, so a 1 bar spring is the safer pick.

Can a spray lance replace a rotating spray nozzle for tank cleaning?

No, they solve different problems. A lance delivers a focused jet or a submerged dose. A rotating spray nozzle distributes rinse water over a full vessel wall. Use the rotating unit for cleaning and the lance for dosing, and keep their patterns from crossing.

How do I know when to replace a worn lance?

Compare measured delivered flow with the rated figure. Replace when the deviation passes 5 percent. Five percent bore growth already produces about 10 percent over-delivery. Recheck at each planned maintenance window.

Why does a lance aimed straight into a duct wet the wall?

A square jet bends back on itself in the crossflow. Droplets then land on the upstream wall and run down as a wet streak. Tilting the axis 15 to 30 degrees downstream stabilises the trajectory and raises penetration depth.

Is a lower spray angle better for duct coverage?

Not below about 15 degrees. A very wide angle spreads the jet thin and loses penetration. A very narrow angle concentrates wear on one spot. Match the angle to the duct width with width ≈ 2 · distance · tan(θ/2). At 30 degrees and 1.5 m: 2 × 1.5 × 0.268 = 0.80 m.

The Bottom Line

Depth, angle, and cracking pressure decide whether a lance works. Flow rate alone will not.

Most spray lances in vessel and duct duty fail because gas or product travels backwards past a weak check valve. That is an upstream problem, settled before the pump stroke.

Set cracking pressure above the measured back-pressure and keep the nozzle check valve at the tip. Log the baseline flow at handover. Send the fluid, the pressure, and the target depth to contact us and the numbers will be checked against your duty. Related reading: full cone nozzles, solid stream nozzles, spray nozzle flow rate calculation, and tank rinse nozzles.