Runaway Reaction Relief: The PSV Sized for Fire, Not for Chemistry
A 25 m³ jacketed batch reactor holds 18,000 kg of a solvent-borne polymerization reaction mass. The vessel design pressure is 10 barg, the PSV is set at 7.0 barg, and its nameplate capacity is 9,500 kg/h of vapor at the set pressure. The relief header and scrubber are rated for 12,000 kg/h. The reaction mass has a heat capacity of 3.1 kJ/kg·K and a latent heat of vaporization of 340 kJ/kg at the relief condition.
Adiabatic calorimetry shows an onset at 118°C, an adiabatic temperature rise of 150 K, and a peak heat release rate of 5.2 MW at 135°C. The reactor is heated with 8 bar steam. Loss of cooling water is credible. A high-temperature trip closes the steam valve at 125°C, but it has a two-minute response and is not SIL-rated. The existing PSV was sized years ago for an external fire yielding 350 kW. The reactor is liquid-full at the relief condition.
1. Calculate the vapor relief rate corresponding to the peak heat release and compare it with the PSV capacity. State the assumptions.
Read the worked answer
q = 5.2 MW = 5,200 kJ/s. If the heat release is removed by vaporization at 340 kJ/kg, the vapor rate is 5,200/340 = 15.3 kg/s = 55,000 kg/h. The PSV is rated 9,500 kg/h at 7.0 barg. At 10% overpressure, absolute pressure rises from 8.01 to 8.81 bara, so vapor capacity rises by about 8.81/8.01 = 1.10, to roughly 10,450 kg/h. The shortfall is a factor of about 5.3.
The key assumption is that the reaction is tempered by boiling solvent, so most of the heat goes to vapor. Sensible heating of 18,000 kg at 3.1 kJ/kg·K absorbs 55.8 kJ/K; at 5.2 MW that is 0.093 K/s, so the two-minute trip delay adds about 11 K before the steam closes, not enough to prevent the runaway. The existing PSV, sized for a 350 kW fire, is inadequate for the runaway case.
2. Evaluate the credit for the high-temperature trip and explain why a larger PSV alone may not be sufficient.
Read the worked answer
The high-temperature trip is a basic control layer with a two-minute response and no SIL rating; it can reduce frequency but cannot be credited as a safety layer without proof testing and SIL verification. A single cooling-water failure plus trip failure is credible. In a LOPA, a cooling failure frequency of 0.1/year and a trip PFD of 0.1 gives 0.01/year, which is unlikely to meet a target of 1e-4/year for a catastrophic event without another independent layer.
A larger PSV alone is not sufficient because the reactor is liquid-full: the relief may be two-phase, so the vapor rating does not apply and the 12,000 kg/h header and scrubber are undersized. The correct path is layered. First reduce the hazard: lower monomer concentration, use semibatch feed, or add an emergency quench. Second, add an independent SIL-rated trip and a rupture disk sized for two-phase flow. Third, upgrade the PSV, header, knock-out drum and scrubber. A rupture disk alone may be acceptable only if it is sized with DIERS methods and vents to a safe location.
3. What data would you require before resizing the relief system?
Read the worked answer
Obtain heat release rate and adiabatic temperature rise at the relief pressure from calorimetry (ARC or APH), vapor-liquid equilibrium and latent heat at the relief condition, two-phase flow regime and void fraction, maximum mass in the reactor, certified PSV capacity at the relieving pressure and temperature, a relief-header hydraulic model, downstream scrubber and knock-out capacity, and the fire and tube-rupture cases. Also define the worst credible scenario and the credit for the high-temperature trip. Without these data, a bigger valve may still be wrong because the controlling case may be two-phase flow rather than vapor.