Start Here: The Quick Fixes That Leave the Shell Exposed
Three shortcuts show up on nearly every chlorine dioxide heater package, and all three leave a hole:
- "Size it for pump dead-head and move on." A liquid-full titanium shell blocked in with 160 F effluent still on the tube side will reach set pressure on thermal expansion alone, with the pumps off.
- "Use the API 521 fire equation and divide by the latent heat of water." That treats the relieving fluid as boiling water. An 11 g/L ClO2 solution decomposing exothermically adds its own heat and its own non-condensable gas, so the vapor rate is not Q/λ for steam.
- "Put a rupture disc on it, a disc handles anything." Not a gas-phase ClO2 deflagration. That pressure rise is a millisecond event and no nozzle on a 225 psig shell will keep up. You design that case out, you do not size for it.
Work the cases in the order below. Each check names the reading that decides the branch.
| Case | Trigger | Sizing basis | Branch |
|---|---|---|---|
| Blocked-in liquid thermal expansion | Block valves shut, hot effluent still flowing | API 521 thermal expansion, liquid | Almost always credible — size it |
| Pump dead-head | Downstream valve shut, ~200 psig | Liquid PRV, API 520 Part I | Already covered |
| Tube rupture | Effluent side breaks into shell | API 521 tube rupture / 10-13 rule | Check the design pressures |
| Thermal decomposition ("dis-association") | Bulk or film temperature above the stability limit | Adiabatic calorimetry + DIERS two-phase | Test, do not calculate |
| Gas-phase ClO2 deflagration | Vapor pocket above the decomposition limit | Not relievable | Prevent by design |
| External fire | Pool fire | API 521 wetted area + decomposition | Credible only with fuel present |
Check 1: Is the Shell Liquid-Full and Blocked In?
Close the shell-side isolation valves with effluent still running through the tubes and the shell becomes a closed liquid volume with a heat source. Liquid is nearly incompressible, so a few degrees of rise walks the shell straight to hydrostatic test pressure. This is the case that governs the small orifice on most bleach-plant exchangers.
Required rate, per the API 521 thermal expansion relation:
Take PHI as the exchanger duty with the shell side stagnant — the fouled-condition UA against the full 160 F to shell-temperature driving force, not the rated process duty. For a dilute aqueous ClO2 stream, water properties are the right first pass for alpha, G, and c.
Outcome: a fraction of a gpm, which lands on a D or E orifice. If the answer is more than a few gpm, the exchanger is oversized for the service and the trip strategy needs another look before you buy a valve.
Check 2: Can the Tube Side Overpressure the Shell?
Read the tube-side design pressure off the exchanger nameplate, including static head and the effluent pump shut-off. Compare it to the 225 psig shell MAWP using the API 521 10/13 screening rule: tube rupture can be dismissed when the low-pressure side design pressure is at least 10/13 of the high-pressure side.
Max exempt tube-side design = 225 x 13/10 = 292.5 psigBelow 292.5 psig: the case screens out. Note it in the relief basis and move to Check 3.
Above 292.5 psig: you now have a full-bore tube rupture into a liquid-full shell. Incompressible flow into an incompressible volume needs a very large orifice, and the relieving fluid is hot effluent mixed with ClO2. Cheaper answer: de-rate the effluent side with a restriction orifice or a relief on the effluent header so the shell never sees more than 292.5 psig.
Check 3: Can a ClO2 Vapor Pocket Form?
This is the branch where you stop sizing and start preventing. Heating drives the mechanism two ways at once. Solubility of ClO2 in water falls as temperature rises, so an 11 g/L solution at 115 F holds a higher equilibrium partial pressure over the liquid than the same solution at generator temperature. Separately, ClO2 decomposes to chlorine and oxygen, and the oxygen is a non-condensable that will not redissolve on cooling. Both effects feed the same headspace.
Gaseous ClO2 above its decomposition limit will self-propagate from a spark, a hot spot, or a pressure shock. Get the limiting concentration and the partial-pressure data from your ClO2 generator licensor for your actual solution strength — not from a generic table — then compute the equilibrium gas composition at 160 F. If the number is anywhere near the limit, the design answer is to eliminate the pocket:
- Orient the shell so the ClO2 outlet is the true high point, with no horizontal dead legs or blind nozzles above the liquid line.
- Run a continuous small vent from that high point to the mill ClO2 scrubber header, upstream of any isolation.
- Interlock the effluent supply on low shell flow so heat input stops when circulation stops.
- Trip on shell outlet temperature well below the 125 F instability point, with the sensor in the flowing stream, not a stagnant thermowell.
There is a metallurgical trap here too. Titanium is the correct choice for wet ClO2, and it is the reason the exchanger is all-titanium. Titanium is not safe in dry chlorine — it ignites. A decomposition pocket that off-gasses Cl2 and O2 and then gets swept or dried is exactly the condition to avoid, in the shell and in the relief discharge piping.
Check 4: Get a Real Decomposition Rate
There is no hand calculation that returns a defensible gas generation rate for this system. Decomposition kinetics for aqueous ClO2 depend on concentration, pH, temperature, light, and catalytic contamination — iron, copper, manganese, and organics carried in by washer effluent are all accelerants, and a tube leak puts them directly into the shell. The route to a number is adiabatic calorimetry on the actual solution:
- Run a low phi-factor adiabatic test (VSP2, APTAC, or ARC class) on 11 g/L solution, plus a spiked sample representing effluent in-leakage.
- Extract onset temperature,
dT/dt, anddP/dtat the maximum credible starting temperature of 160 F. - Feed those into DIERS/AIChE two-phase vent sizing to get the required area at 21% accumulation.
- Have the testing lab and a licensed process safety engineer own that result. This is where a mechanical designer hands off.
The same data settles the fire case. If a pool fire is credible — and in a washer building with no stored flammables and non-combustible construction it may not be — the API 521 wetted-area heat input drives the solution past its onset temperature long before it boils dry. The relieving fluid is then a decomposing two-phase mixture, and the boiling-water vapor rate under-predicts it. Document the fire-case exclusion with the fuel survey rather than sizing a valve on the wrong fluid.
Set, Size, and Verify
- Set pressure: at or below 225 psig MAWP. Check the operating margin first — 200 psig dead-head against a 225 psig set is 89% of set, right at the conventional-valve limit. Expect simmer, and simmer in ClO2 service means seat leakage into the discharge line. Either knock the dead-head down with a restriction orifice or specify a pilot-operated or bellows valve.
- Accumulation: 225 x 1.10 = 247.5 psig for non-fire cases; 225 x 1.21 = 272.25 psig for fire/runaway; 225 x 1.16 = 261 psig if a second valve is fitted.
- Orifice: size each case separately in API 520 Part I — liquid for Checks 1 and 2, two-phase (omega or DIERS) for the decomposition case. Install for the largest area, not the sum.
- Materials: titanium body and trim, PTFE soft seat. Stainless will not survive wet ClO2. If a rupture disc goes under the valve for isolation, add a pressure-monitored interspace so a pinholed disc is caught before it fills with product.
- Discharge: to the caustic scrubber header, sloped, no low-point pockets, no dead legs, and no path that can dry chlorine out against titanium. Confirm the header can take the added rate on top of whatever else relieves into it.
- Verify: hydro the shell with the relief removed and a blind, then bench-test the valve at set with the tag on it. Commission the temperature trip by ramping the effluent supply and confirming it drops the hot side before the shell outlet reaches the trip setpoint. Prove the continuous vent path is open with a flow indication or a scrubber differential, not just a valve position.
Stop at Check 4 if the calorimetry has not been run. Sizing a two-phase relief for a self-accelerating decomposition without measured dT/dt and dP/dt is guesswork with a titanium shell and a chlorine release behind it. Send the case to your ClO2 generator licensor and to a testing laboratory, and have the final relief basis reviewed and stamped by a process safety engineer licensed in the plant's jurisdiction.
FAQ
Why does ClO2 off-gas when the exchanger shell is heated above 115 F?
ClO2 solubility in water drops as temperature rises, so the equilibrium partial pressure over an 11 g/L solution climbs with every degree. Thermal decomposition adds oxygen, which is non-condensable and will not go back into solution when the shell cools.
Why does a relief valve not protect against gas-phase ClO2 decomposition?
A gas-phase ClO2 deflagration develops pressure in milliseconds, far faster than any nozzle on a 225 psig shell can pass. That case is eliminated by keeping the shell liquid-full and continuously vented to the scrubber, not by adding relief area.
Why does the API 521 fire case under-size relief for a decomposing solution?
The standard wetted-area method assumes the relieving fluid boils and absorbs heat as latent heat. A decomposing ClO2 solution generates its own exothermic heat plus non-condensable gas, so the required area comes from DIERS two-phase methods driven by measured calorimetry, not from Q divided by the latent heat of water.
How do I get a decomposition rate for relief sizing?
Run low phi-factor adiabatic calorimetry (VSP2, APTAC, or ARC class) on the actual 11 g/L solution and on a sample spiked with washer effluent, then take onset temperature, dT/dt, and dP/dt at 160 F into a DIERS vent sizing calculation. Contamination with iron, copper, or organics from a tube leak shifts the onset down, so test that case too.