Seawater cooling circuit: 316L failure and a proposed duplex substitution
A coastal data centre in Tseung Kwan O uses once-through seawater cooling. Titanium plate heat exchangers transfer heat from a closed chilled-water loop to seawater drawn from a nearby intake. The first design used 316L stainless steel for the seawater-side pipework, flanges and strainer bodies, with seawater at 1.2 m/s and 30 °C, chloride about 19,000 mg/L, and intermittent sodium hypochlorite dosing for biofouling control. After 18 months, several 316L spools show pitting and crevice attack under gaskets and flange faces, and one strainer body has perforated. The contractor proposes replacing the 316L with 2205 duplex stainless steel throughout, retaining the existing carbon steel structural supports and galvanised bolting, and reducing the design velocity to 0.6 m/s to reduce erosion. The operations team wants a 15-year life with no unplanned shutdown.
1. Explain the corrosion mechanism, calculate PREN for 316L and 2205, and state whether 2205 is adequate at 30 °C with chlorination.
Read the worked answer
Mechanism. 316L depends on a passive chromium-oxide film. In chlorinated seawater at 30 °C the film breaks down locally at inclusions and in crevices - under gaskets, at flange faces, and beneath silt or biofilm where oxygen is depleted and chloride concentrates. The result is pitting and crevice corrosion exactly where it was found. Residual chlorine raises the corrosion potential, which makes pitting more likely, not less; hypochlorite dosing controls biofouling but does not protect the alloy.
PREN = %Cr + 3.3 × %Mo + 16 × %N:
- 316L (17 Cr, 2.1 Mo, no deliberate N): about 24.
- 2205 (22 Cr, 3.1 Mo, 0.17 N): about 35.
A higher PREN means a higher critical pitting temperature. At 30 °C, 2205 is a genuine improvement and is widely used in seawater, but it is not immune: crevice corrosion under gaskets and deposits remains the controlling risk, and chlorination erodes the margin. For a 15-year, no-unplanned-shutdown duty the defensible options are super duplex 2507 (PREN about 42) for wetted parts, or a non-metallic solution (GRP, PE or PTFE-lined carbon steel, or titanium for the exchanger). 2205 is acceptable only if the detailing, velocity and chlorination control are right and the operating temperature stays well below its critical pitting temperature; that case must be argued, not assumed.
2. Critically review the contractor's proposal and state the corrections you would require before accepting the alloy change.
Read the worked answer
Reject the proposal as written; accept the alloy change only with the following corrections.
- Velocity. Reducing to 0.6 m/s is the wrong direction. Seawater needs a scouring velocity, typically above 1.0–1.5 m/s, to stop settlement, biofilm and crevice formation; slow and stagnant zones are where stainless fails. Erosion-corrosion of stainless is not the controlling limit until several m/s. Restore 1.2–1.5 m/s, check the low-flow case, remove dead legs and provide drain and flush points.
- Galvanic isolation. 2205 must be electrically isolated from carbon steel supports, galvanised brackets and any carbon steel pipe: isolation kits, non-metallic wear pads, lined clamps. Galvanised bolts in seawater will waste quickly; use PTFE-coated or duplex fasteners, or isolate them.
- Gaskets. PTFE or EPDM, full-face where possible, with crevice-free joint geometry.
- Fabrication. Qualified WPS and PQR, purged root, controlled heat input, ferrite check on duplex welds (typically about 35–65%), pickling and passivation of welds, NDT of butt welds.
- Testing. Hydrotest with low-chloride water, not seawater; drain and dry immediately afterwards.
- Strainers. Duplex or rubber-lined baskets with differential-pressure alarms rather than a slow-velocity workaround.
3. How would you frame the whole-life decision, and what QA and monitoring records would you require?
Read the worked answer
Whole-life decision. Compare over 15 years including failure risk and downtime, not first cost. 2205 is roughly 1.5–2× carbon steel, super duplex more, and titanium has the highest first cost but often the lowest whole-life cost in chlorinated seawater. For a data centre, an unplanned shutdown costs far more than the pipe, so the value case for the higher alloy or for a lined or non-metallic system is usually strong. Present the comparison as a matrix: first cost, expected life, failure mode, consequence, spares and lead time, and maintenance regime.
QA records. Mill certificates with heat numbers and verified PREN; positive material identification on installed spools and welds; WPS/PQR and welder qualifications; ferrite readings; pickling and passivation records; NDT reports; hydrotest records; and a corrosion-control dossier that records velocity and chlorination operating limits.
Monitoring. Corrosion coupons or electrical-resistance probes, periodic ultrasonic thickness checks at known crevice locations, and a residual chlorine and velocity log. The operating limits must go into the O&M manual so that a future maintenance team does not save money by slowing the pumps or over-chlorinating, which is exactly how the first failure was invited.