Case Study: Spiral Nozzle Retrofit in a Wet Scrubber Absorber
A pulp mill wet scrubber lost particulate removal to clogging hollow cone nozzles. Spiral nozzles, filtration and a lower pressure baseline fixed it.
A scrubber does not fail in a dramatic way. It fails quietly, one nozzle at a time, and the first hard evidence is usually a number on a stack monitor. That was the shape of the problem at an integrated pulp mill in the Nordics, where the wet scrubber on the recovery boiler and lime kiln flue gas train was supposed to be the last barrier before the stack. The scrubber was not short of liquid: the recirculation pumps ran at design duty, the slurry sat at the right pH, and every level was in service on paper. What was missing was the spray. Vane-type hollow cone nozzles were plugging on the solids the loop carried, the cone from each blocked nozzle collapsed into a dribble, and gas slipped through untreated patches of the tower cross-section.
Three problems arrived together, which is why the fix took a program rather than a purchase order. Maintenance was pulling and rodding nozzles on a cycle that had crept from monthly to weekly. Spray levels were coming out of service to clear blocked headers, and each outage cut particulate removal at the moment the boiler was at load. Every excursion on the stack particulate monitor carried a reporting obligation, whether or not the mill agreed with the number. The retrofit that ended the cycle was not exotic: large free passage spiral nozzles, filtration ahead of the spray header, and a pressure baseline set from the pump curve instead of from the old datasheet. This case walks through the reasoning, the arithmetic, and the published ranges the mill used as its acceptance band.
Case Snapshot
The reference configuration below is the generic shape of the installation the sizing was built from. Every value is a realistic working figure for a mill of this class. The mill stays anonymous because stack compliance data is commercially sensitive in its market.
| Item | Value |
|---|---|
| Industry | Pulp and paper, integrated mill with chemical recovery, Nordic (anonymized) |
| Duty | Wet scrubbing of flue gas from a recovery boiler and lime kiln, particulate and acid gas removal |
| Tower class | Spray tower absorber, single train, roughly 40 m² cross-section, three spray levels |
| Original setup | Vane-type hollow cone nozzles, 1/4 to 1/2 inch, header at 1.2 to 1.4 bar |
| System chosen | One-piece spiral nozzles, hollow cone on the absorption levels, full cone on the quench spray, 3/8 to 1 inch bodies |
| Operating point | Slurry at 0.7 to 1.0 bar at the header, 10 to 20 percent solids by weight, pH 5 to 6 |
| Free passage | 6 to 18 mm on the spiral bodies, against 25 to 100 percent of orifice on vane designs, at the low end in small bodies |
| Filtration | Duplex basket strainers on the spray header, 4 to 6 mm perforation, blowdown on differential pressure |
| Materials | 316L for the alkaline loop, duplex grade for the chloride-loaded section, ceramic-coated grooves on the abrasion side |
| Compliance | Stack particulate and SO2 limits, continuous opacity monitoring |
The tower class matters more than the exact tonnage. A single-train spray tower with three levels over roughly 40 m² is the most common shape in this duty, and the logic below transfers across the useful range of flue gas flows. Everything was read off the existing pump curve and tower drawing.
The Challenge
The liquid in an FGD or recovery boiler scrubber is not water. It is a recirculated slurry that carries everything the gas threw at it. Published limestone scrubbing references put the slurry at 10 to 20 percent solids by weight, at a pH of about 5 to 6, with a density near 1,100 kg per cubic metre. The same references put the duty at a liquid-to-gas ratio of roughly 5 to 15 litres of slurry per cubic metre of flue gas, at a tower spray density of about 15 to 30 cubic metres per square metre of cross-section per hour.
At a pulp mill the solids load has its own character. Carryover from the recovery boiler brings sodium salt dust. The alkaline loop precipitates calcium carbonate scale, which grows as sheets rather than as clean particles. That population of solids is why the original nozzle choice failed: the nozzles were sized for flow, and they were the wrong geometry for the water.
Conventional hollow cone nozzles form their pattern inside the body. A swirl chamber and a set of vanes spin the liquid into a ring before it leaves the orifice. Published FGD nozzle design data makes the consequence explicit. On a full cone design with an internal vane, the free passage, meaning the diameter of the largest sphere that can pass the nozzle, runs from 25 to 100 percent of the orifice diameter depending on the design. A vane hollow cone in the sizes this mill was running sits at the low end of that band. The narrowest point in the flow path is inside the body, upstream of the exit, and invisible from outside.
| Nozzle design | Free passage against orifice diameter | Narrowest point | Where debris collects |
|---|---|---|---|
| Vane full cone | 25 to 100 percent, typically at the low end in small bodies | Vane gaps and swirl chamber | Inside the body, upstream of the exit |
| Whirl tangential hollow cone | 80 to 100 percent, no internal parts | The orifice | The orifice itself |
| Spiral | 30 to 100 percent, no internal parts | The exit | The ramp surface, and only as deposit |
| Large free passage spiral | Up to 100 percent of the orifice, up to 38 mm in published designs | The exit | The ramp surface |
Reference data from published wet scrubber nozzle design literature. Free passage, not flow rating, is the number that decides whether a nozzle survives slurry service.
The failure mode was progressive and silent, which is the worst combination. A lodged particle reduced the open area, local velocity rose across the remaining passage, and the next particle was more likely to stick. Flow through the nozzle fell slowly, the cone narrowed, and the wetted footprint shrank until a dry wedge opened between two adjacent nozzles. At that point the gas had a path through the tower that never met a droplet. Published design guidance for wet limestone absorbers notes that above 90 percent removal the process is limited by droplet surface area rather than reaction speed, and the limit tightens toward 95 percent and beyond. A shrunken spray footprint attacks exactly that margin.
The maintenance response made the problem worse in a way that is easy to miss. Nozzles were pulled on a fixed cycle, so a partly blocked unit was sometimes reinstalled, because a nozzle that still passed water looks acceptable on a bench. The strainer was doing its job and was the wrong size for the problem: a 20-mesh screen opens to about 0.9 mm, so fines larger than the internal swirl passage but smaller than the mesh passed the screen and lodged in the nozzle. The strainer protects the pump. The nozzle needs a different number.
Turning up the pump was tried, and the physics refused to cooperate. Flow through a hydraulic nozzle scales with the square root of pressure. Raising header pressure from 1.4 bar to 2.8 bar buys about 41 percent more flow, not 100 percent, and a bridged solid stays bridged regardless.
The cost landed in three places at once. Spray level outages cut the liquid-to-gas ratio in the tower, so particulate removal fell until the level was back in service. Operators worked in a plume of slurry mist while clearing headers. And on two occasions the stack particulate monitor crossed the permit line during a level outage at high boiler load, which triggered the reporting process the mill was trying to avoid. The availability number the mill tracked was not mechanical availability. It was the fraction of operating hours in which all three spray levels delivered design coverage, and it was drifting down.
The Solution
The fix followed the same decision order the mill uses for any dirty-water nozzle duty, and the order is the point. Pattern, then free passage, then flow and pressure, then filtration, then layout and materials. Changing the order is how projects end up with a finely atomising nozzle that blocks in a week.
Pattern first, because the duty is gas contact
The three absorption levels needed a hollow cone. The ring of liquid produced by a hollow cone puts the spray mass at the perimeter of each nozzle’s footprint, and staggered levels of overlapping rings build the full cross-section coverage a tower needs, while the open centre lets gas pass without a large pressure penalty. Published spray tower references describe this as the standard arrangement for absorber spray banks, and it matches the duty: bring as much wetted surface into the gas path as the cross-section allows.
The quench section ahead of the tower is a different question, and it wanted a full cone. That spray has to fill an area rather than ring it, because the job is to cool the gas and knock down coarse carryover before it reaches the absorption levels. Both patterns come off the same spiral body, which is why the same spare part number now covers two duties at the mill. The comparison between the two is worked through in the spiral full cone versus hollow cone guide.
Free passage next, because the solids decide the number
The sizing rule the mill adopted is short: sample the loop, find the largest particle the slurry routinely carries, and set free passage at least 50 percent above it. The slurry analysis put routine solids at the coarse end, with scale sheets and salt agglomerates in the 3 to 5 mm band, so the mill specified bodies with a 9 to 12 mm passage and left the 18 mm bodies on the shelf for the highest-solids points. Nothing narrower than the exit sits in the flow path of a spiral, so anything that enters the body can leave it.
That choice is a trade, and the mill took it knowingly. A spiral forms its cone by deflecting a laminar jet along a spiral ramp rather than spinning the liquid in a pressurised chamber, so its droplet distribution is wider and coarser than a swirl-chamber nozzle of the same flow. The open path buys the one thing this tower needed: a spray level still at design coverage at the end of a campaign rather than at the start. The reasoning behind the geometry is set out in the spiral nozzle dirty water guide and the wider family comparison in the spiral nozzles overview.
Pressure and body size together, from the pump curve
Pressure and nozzle size cannot be chosen separately, and the mill had been treating them as two decisions. The header was measured rather than assumed, with a gauge at the nozzle manifold instead of at the pump discharge. That measurement alone explained part of the mystery. A slurry column of 1,100 kg per cubic metre rising 10 m to the top spray level costs about 1.1 bar of static head, so a pump discharge reading of 2.2 bar was consistent with a header pressure near 1.1 bar. The nozzles had been selected against the pump figure for years.
With the true header pressure known, the retrofit moved the design point down instead of up. Spiral nozzles form a pattern without a pressurised swirl chamber, and published FGD design data puts their typical operating band at 0.3 to 1.0 bar, against 0.5 to 1.5 bar for the traditional whirl tangential designs. The mill set a 0.7 to 1.0 bar baseline, which is where the recirculation pumps already sit on their curves. Body size went up to compensate for the flow lost at the lower pressure, and the number of spray points per level fell because each body now passes more. The two changes offset, and coverage held, because coverage follows spray angle and mounting distance rather than pressure.
Filtration sized for the nozzle, not only for the pump
The old 20-mesh screen was retained on the pump suction, because that is what the pump needs. A duplex basket strainer went onto the spray header with a 4 to 6 mm perforated basket, which sits below the smallest free passage in the new nozzle set. The logic is deliberate: any particle small enough to reach a spray point now has to be small enough to leave it. Blowdown runs on differential pressure, and the baskets change without taking the level out, which removed the last reason for a planned outage.
Materials and wear tracking
The alkaline section of the loop runs 316L. The chloride-loaded stretch of the flue gas path went to a duplex grade, because published FGD material guidance puts 316L on the edge in chloride service and pushes the specification toward duplex when the fuel brings chlorides with it. The abrasion side uses ceramic-coated grooves, with the coating thickness checked against the groove tolerance so that the coating cannot close the passage it was added to protect.
Wear monitoring is part of the design, not an afterthought. As a spray groove erodes, the discharge coefficient rises and the nozzle passes more slurry at the same pressure. A flow drift of 15 to 20 percent above rated is the replacement trigger the mill adopted, tracked per header from the pump data it was already logging. That converts nozzle wear from a visual inspection problem into a number on a chart.
The Engineering Behind the Choice
The engineering that carried this retrofit sits in three places: the surface area the spray presents to the gas, the power the pumps draw to produce it, and the geometry that decides whether a spray point is still working in month three.
Surface area and the coarser droplet trade. Published design targets for absorber spray banks put the volume median droplet diameter in the range of 1,500 to 2,500 µm. Droplets in that band are large enough to fall against the upward gas velocity in the tower. Published spray tower work puts the critical dropout diameter at roughly 800 to 1,200 µm at the gas velocities common in absorbers, which is the boundary between a droplet that settles through the spray zone and one that leaves with the gas and loads the mist eliminator.
A coarser spray costs surface area per litre, and the loss is predictable. For a given volume of liquid split into droplets, specific surface area scales with 6 divided by the droplet diameter. Moving the volume median diameter from 1,500 µm to 2,000 µm takes specific surface area from about 4,000 m² per cubic metre of liquid to about 3,000, a loss near a quarter. The mill did not try to win that back from the nozzle. It won it back from the tower: the levels, the overlap between them and the liquid-to-gas ratio all raise the wetted surface presented per cubic metre of gas, which is the number the absorption reaction sees. Choosing the coarse, reliable spray and paying for surface area with flow is the trade this duty rewards.
Pump power follows pressure at constant flow. The hydraulic power a recirculation pump delivers is proportional to flow multiplied by pressure differential. Holding flow steady, a drop in nozzle pressure from 1.4 bar to 0.9 bar takes about 36 percent off the hydraulic work the pump does. Real installations recover less, because pump efficiency and system losses do not fall with nozzle pressure, and the larger bodies and the strainer add their own drop. Published results from a 300 MW coal-fired absorber retrofit give the realistic shape of the outcome: spiral nozzles at low delivery pressure cut the number of nozzles to about one third of the conventional count and brought recirculation pump power to about 80 percent of the previous figure, a saving of roughly a fifth, while holding SO2 removal.
That published result is why the mill did not shop for the lowest pressure nozzle available. The pressure has to be high enough to make droplets in the settling band and spread them across the tower, and low enough to keep the pumps off their power limit. The window is narrow and it is set by the tower.
Coverage is geometry, and geometry is checkable. The wetted band under a nozzle follows the spray angle and the mounting distance. The width at distance d from a nozzle with included angle θ is 2 multiplied by d multiplied by the tangent of θ/2. A 120 degree spiral mounted 1.2 m above its target level wets a band about 4.2 m wide. The mill laid out each level so neighbouring bands overlapped by a margin, with the explicit purpose that a single partly fouled spray point could not open a dry gap straight through the level. Coverage, not flow, was the acceptance criterion for the layout.
Spray density sets the nozzle count. Published references put tower spray density at 15 to 30 cubic metres of slurry per square metre of cross-section per hour. On a 40 m² tower that is 600 to 1,200 m³ per hour across the absorption levels, or roughly 3,300 to 6,700 litres per minute per level. At about 70 litres per minute per spray point in the larger spiral bodies, the arithmetic lands between 48 and 96 spray points on a level before any overlap allowance. That explains the sensitivity of the original installation: with that many small orifices in slurry service, the tower is only as good as its most blocked level.
Flow changes with pressure, and the change is small. A body quoted at 100 litres per minute at 1.5 bar has a discharge coefficient of about 81.6 litres per minute per square root of bar. Dropped to 0.8 bar, the same body delivers about 73 litres per minute, 27 percent less flow for 47 percent less pressure. Going up one body size recovers the flow, which is why the retrofit ended with fewer, larger spray points rather than more small ones.
A wider passage also means a slower jet. Exit velocity is flow divided by exit area, and published FGD nozzle guidance lists slurry velocity through the nozzle as one of the factors driving erosion, along with fly ash content and limestone particle size. A body with a 9 to 12 mm passage at the same flow drops that velocity, so the nozzle that resists plugging also wears more slowly where erosion starts.
Where the reaction tank itself needs agitation rather than spraying, the mill’s process side uses liquid-driven mixing, and the same loop serves both duties. The guide to eductors covers how that arrangement is sized from tank turnover.
The Results
No site audit figures appear in this table. The values come from published reference ranges for wet scrubber and flue gas desulphurisation nozzle design, plus published retrofit results from a large absorber installation. They are the band this class of change delivers, not a promise for a specific tower.
| Metric | Published reference result | Why it moves |
|---|---|---|
| Free passage at the spray point | 30 to 100 percent of orifice diameter on spiral designs, up to 38 mm on large free passage versions, against 25 to 100 percent on vane designs and at the low end in small bodies | No internal swirl chamber or vanes, so the exit is the narrowest point |
| Plugging and cleaning interval | Routine rodding cycles extend to scheduled inspection, with published maintenance guidance recommending a pull and inspect interval from monthly down to quarterly | Smaller population of lodged particles because the passage passes what the loop carries |
| Absorber spray level availability | Published utility FGD availability is commonly reported in the high nineties of percent, with nozzle plugging a recognised cause of derate and unplanned spray level outage | Coverage stays at design, so the level is not taken out for clearing |
| Recirculation pump power | About 80 percent of a conventional nozzle set, a saving near a fifth, in a published 300 MW absorber retrofit | Spiral nozzles hold pattern at 0.3 to 1.0 bar against 0.5 to 1.5 bar for traditional whirl tangential designs |
| Nozzle count per spray level | About one third of the conventional count in the same published retrofit | Larger bodies at lower pressure pass more liquid per spray point |
| Droplet volume median diameter | 1,500 to 2,500 µm design band for absorber spray banks | Coarse enough to settle against the upward gas velocity |
| SO2 removal | 95 to 98 percent for wet limestone FGD in published references, set by liquid-to-gas ratio, pH and droplet surface area | Wetting the full cross-section restores the gas-liquid contact area |
| Stack particulate | Below about 10 to 20 mg per normal cubic metre in published wet FGD performance data with a working mist eliminator | Even coverage plus a demister that is not overloaded by coarse carryover |
These are published ranges, not site audit results. The outcome at any given mill depends on the solids analysis, the chloride load in the fuel, the tower geometry and the condition of the mist eliminator. A tower with a failed demister still shows stack particulate above that range no matter how good the spray pattern is, which is why the eliminator was inspected before a nozzle was ordered.
The mill defined its acceptance band from those ranges before the trial, so the result could not be argued about afterwards. The plan is short enough to reproduce:
| What was measured | Where and how | Acceptance basis |
|---|---|---|
| Header pressure at each level | Gauge at the nozzle manifold, not the pump discharge | Within 0.7 to 1.0 bar design band at design flow |
| Flow per header | Pump curve and header flow element, logged | No drift above 15 to 20 percent of rated, which signals groove wear |
| Spray pattern at each point | Visual and footprint check on a level outage, one level at a time | Full footprint with overlap, no dry gap between adjacent points |
| Solids in the loop | Sump sample, screened and measured for the largest routine fraction | Free passage at least 50 percent above the largest routine solid |
| Stack particulate and SO2 | Continuous monitors and reporting data | Inside permit, with level outages eliminated from the operating plan |
| Pump power at design duty | Existing motor data | Compared against the pre-retrofit baseline at matched load |
The pattern check is the step most plants skip, and the one that catches a spray point passing water in the wrong shape. A half fouled nozzle delivers roughly the right volume into the wrong footprint, so flow data alone will report a healthy level. The spiral nozzle overview covers it as part of a normal inspection round, which takes one planned level outage instead of an unplanned one.
Why This Case Matters
The mill changed three things, and only one of them was a nozzle. The nozzle change set the free passage. The filtration change made that passage worth something, because a coarse passage behind a fine screen is a coarse passage full of debris. The pressure change moved the design point to where the loop actually operates. Any one of the three alone would have shown an improvement and then drifted back, which is the pattern most clogging complaints follow.
The transferable discipline is the order of the decisions. Read the water first, then pick the geometry that passes it, then set pressure and body size together from the measured header pressure, then size filtration to the nozzle rather than to the pump, then lay out coverage with an overlap margin, and finally put a wear number on a chart and a pull and inspect interval in the maintenance plan. That sequence is the same for a scrubber, a quench tower, a cooling loop on an open circuit, or a dust suppression header fed from settled water. The wider selection path is in the spiral nozzle product range, which lists free passage next to flow and angle, because in dirty service those three columns are read together.
The wider family question comes down to one sentence: every other nozzle type trades uptime for fineness, and the spiral trades fineness for uptime. In a tower that must keep spraying for a campaign at a time on slurry carrying scale, grit and salt, that is the trade the duty asks for. Where droplet size is the control variable rather than coverage, another family is the right answer, and the spiral dirty water guide is honest about where the spiral stops being the correct tool.
What to send for a sizing, in one list:
- Slurry or liquor properties: solids content by weight, the size of the largest routine particle, pH, density and chloride level, because free passage and materials come from these.
- Measured header pressure at the nozzle manifold and the static head from the pump to the top level, because the pump discharge figure almost always overstates what the nozzles see.
- Flow required per spray level, or the liquid-to-gas ratio and tower cross-section, because spray density sets the total flow.
- Tower geometry: cross-section, number of spray levels and the mounting distance from each level to the target below it, because coverage comes from angle and distance.
- Gas conditions: flow, temperature at the spray, and the inlet particulate or SO2 load, because these set the removal target the spray has to support.
- Strainer or filter arrangement on the loop, mesh or perforation size, because filtration and free passage have to be chosen as a pair.
- Mist eliminator type and last inspection result, because carryover from a loaded demister can mask a spray problem.
- Pump curve and the power you will spend, because the design point is a pressure, not a nozzle model.
Each item changes the answer, so a complete list returns a sized configuration in one round instead of a trial.
Start with a slurry sample and the measured header pressure, and send both through the contact page. BoreJet comes back with the body size, the pattern, the free passage and the pressure band, sized to the loop you run rather than to the pump nameplate. That is the fastest way out of a rodding cycle that has become part of the shift routine.
Your Duty May Differ
Send the Vessel Drawing and the Pump Curve.
We size the head class, the count and the operating point for your tank, not a catalogue guess. A coverage test on the actual vessel beats any estimate.