Level swell

Level swell is the mechanism by which runaway chemical reactions vent a two-phase mixture. When a runaway chemical reaction generates gas or vapour, bubbles are formed throughout the bulk of the liquid. Because the bubbles are buoyant, they wijl tend to rise through the liquid in order to disengage at the surface. However, whilst they remain in the liquid, they occupy volume and so cause the 

---

The method uses the dnft-fhix level swell calculation models to take into account there being more vapor in the inlet stream to relief device than average for the vessel. 

Grolmes, M. A., A Simple Approach to Transient Two-Phase Level Swell, Multi-Phase Flow and Heat Transfer III. Part A Fundamentals. Proceedings of the Third Multi-Phase Flow and Heat Transfer Symposium—Workshop, Miami Beach, FL, April 18-20, 1983. 

Figure 4.3 LEVEL SWELL FOR A FLUID WHICH IS NOT INHERENTLY FOAMY (BUBBLY OR CHURN-TURBULENT FLOW)...

For a non-foamy system, the extent of the level swell and the fraction of gas/ vapour entering the pressure relief system, at a given gas/ vapour evolution rate, depends on the two-phase flow regime within the vessel. For non-viscous systems the two main flow regimes are bubbly-flow and churn-turbulent flow (see Figure 4.2 (b) and... 

Further information on level swell calculations and the determination of vessel flow regimes is given in Annex 3. 

The viscous systems that are of most concern are those which may give rise to laminar flow in the relief system. The level swell in the vessel may also be affected. These two topics are covered below. It should be noted that only moderately high viscosities, of 100 cP or more, may be sufficient to cause laminar flow or to change the level swell behaviour. Failure to take account of the effects of laminar flow could lead to the serious undersizing of the relief system. 

Considerable work has been. done at JRC Ispra to visualise level swell for viscous fluids17,8,91. This indicates that t moderate viscosity (around. 100 cP) foaming behaviour of the. fluid (if it occurs) dominates viscous effects. Many high viscosity fluids are foamy because they are not pure fluids. At higher viscosities, approaching 1000 cP, there appears to be much less foaming and the flow characteristics are dominated by. viscous effects. 

The main effect of the presence of solids on level swell will be in changing the liquid viscosity. The mixture viscosity Should be used in place of the liquid viscosity in level swell correlations. ... 

This is covered by Huckins14, DIERS /Grossel171, Keiter191 and Singh151, who also discusses the need to account for level swell in a quench tank. 

As described in Annex 3, the amount of level swell increases with gas or vapour superficial velocity. At bench-scale, the superficial velocity due to runaway will be very low compared with that at plant-scale, so that very little level swell occurs in bench scale tests, even for Inherently foamy fluids. However, some level swell at bench-scale is required to determine whether the reacting system is inherently foamy or not. For tempered systems, this can be achieved by rapid depressurisation of the test cell to give flashing and a consequent high superficial velocity, and this is one of the ways in which the DIERS bench-scale apparatus can be used. (Note that... 

A description of level swell is gjjven in 4.3.1. Level swell calculations do not apply to inherently foamy systems as these always vent a homogeneous two-phase mixture. 

Calculation methods are given here for cases (a) to (c). In section A3.4 below, references are given to a calculation method for case (d). The level swell calculation methods presented here use the drift flux correlations developed by DIERS[11. The DIERS correlations apply to a vertical cylindrical vessel, which is most often the case for chemical reactors. Modifications for horizontal cylindrical vessels are given by Sheppard[2,3]. 

Where it is uncertain whether thle system is inherently foamy, it is recommended that the worst case assumption is used (see 4.3.2(1)). For tempered systems, the worst case will be inherent foaminessl Where tempered systems are not inherently foamy, the level swell calculations described in this Annex may lead to a reduction in calculated relief system size. For untempered systems, the worst case is vapour/ liquid disengagement causing reduced mass loss from the reactor during relief. In this case, dynamic simulation (see A3.4) may be needed to take account of level swell in relief sizing. 

The different flow regimes during level swell (churn-turbulent, bubbly and droplet) were described in 4.3.1. In order to perform a level swell calculation, it is necessary to decide the flow regime. 

Figure A3.2 illustrates terminology used in level swell calculations. Annex 10 gives...

The amount of level swell is correlated with the superficial velocity, jg, of gas or vapour at the surface of the liquid. Superficial velocity is the volumetric flow of gas or vapour, divided by the vessel cross-sectional area (i.e., with no attempt to account for the fraction of the cross-sectional area occupied by liquid). Within a particular flow regime, level swell increases with increasing superficial velocity. 

Accurate prediction of the degree of level swell requires computer simulation141. However, conservative estimates can be made to predict superficial velocity for the following cases ... 

WORKBOOK FOR CHEMICAL REACTOR RELIEF SYSTEM SIZING Figure A3.2 DEFINITION OF TERMS IN LEVEL SWELL CALCULATIONS... 

The terminal bubble rise velocity, ( , is another correlating parameter for level swell. It can be calculated from the following equations, according to the flow regime. ... 

Level swell is characterised by the. void fraction within the swelled liquid, a. This is correlated for each flow regime as a function of 4V the dimensionless ratio of the superficial gas/ vapour velocity to the bubble rise velocity, i.e ... 

The other correlating parameter is C0. This is intended to take account of channelling of bubbles up the walls, rather than uniform distribution. DIERS[5] recommend the following level swell correlations. 

Figure A3.3 is a plot of average void fraction, a, versus the dimensionless superficial velocity, for the different flow regimes and values of C0. The correlations presented here may overestimate level swell for pure vapour pressure systems if there is a non-boiling region (in which static head suppresses boiling) at the bottom of the reactor. This is conservative for relief system sizing and is discussed further by DIERS151.

Equation (A3.6) can now be used to find the level swell. A value is required for the correlating parameter, C0. As two-phase relief is the worst case for relief system sizing, a value of C0 of 1.0 will be used, since this gives the highest predicted level swell ... 

SuperChems Expert 161 is a code developed by Arthur D Little Inc. for risk assessment consequence analysis, which also has a relief system sizing option. The code has a physical properties package that can handle highly non-ideal properties. It can also consider the effect of chemical reaction in the relief system piping. The code uses the DIERS drift flux methods for level swell and has the option of a rigorous two-phase slip model for the. relief system capacity. 

In order to use the sizing method, the reactor, void fraction, aD, at which total vapour/ liquid disengagement is expected at the maximum accumulated pressure, must first be evaluated. This may be done by level swell calculation (see Annex 3) or by small-scale experiment with the same vapour superficial velocity as will occur at plant-scale (see Annex 2). Equation (A5.7)can then be used to find the relief area ... 

The void fraction at disengagement has been estimated by level swell calculation (not shown here) as 0.9. 

Before using the method, the void fraction at disengagement must be evaluated at conditions corresponding to the maximum accumulated pressure during relief. This can be done by level swell calculation (see A3.3) or possibly by a small-scale experiment that uses depressurisation to achieve the same vapour superficial velocity as in the full-scale reactor during relief (see Annex 2). The required relief rate can then be calculated fromt31 ... 

This example is also used as a worked example for level swell calculations in A3.3.6. A reactor of volume 3.6 m3 contains 2610 kg of reactants under worst case runaway conditions. The relief pressure is 5.5 bara and the maximum accumulated pressure is 7.0 bara. 

---