Apr-2017

The design temperature of flare systems

A key objective when setting flare system design conditions is to maintain the integrity of the system during fire relief

PAUL DAVID
Paul David Process Ltd

Article Summary

A great deal has been written about the minimum design temperature of flare and blow-down systems because cold temperatures can cause brittle failure or over-stressing of flare piping. Less has been written about setting the maximum design temperature of flare system piping.

American Petroleum Institute (API) Standard 521,1 which is used throughout the industry by process engineers, gives only general guidance on how to approach the task of setting the (mechanical) design temperature. The standard advises that the extremes of temperature of the fluids entering the header should be considered, heat transfer analysis may be performed, it is common to exclude the fire case, and careful analysis is required. There is considerable scope for interpretation by individuals and companies of how this is to be implemented. The author’s experience is that company standards differ in their approach and in the level of detail they offer, sometimes giving rise to more questions than answers. The refinery process engineer, faced with tables of relief case data, has to place a number in the box on the line list that says ‘Design Temperature’. It is a number for which he or she will be held accountable.

The design temperature of process equipment is usually the higher of: the maximum normal operating temperature plus a margin (typically of the order of 25°C); or the highest temperature expected during start-up, shutdown or upset. Upset conditions include the operation of the pressure relief valve and for equipment containing a saturated liquid the temperature is higher at relief pressure than at normal pressure. The selection of relief valve body and flange rating are normally based on the equipment design conditions.

The fire contingency is not normally considered when setting the design temperature of the equipment or its associated pressure relief valve (PRV). The fire relief temperature for a heavy hydrocarbon mixture, with a wide boiling range, can be very much higher than the equipment design temperature. This can lead to the somewhat uncomfortable result that the PRV fire case operating temperature is higher than its design temperature.

If during the fire the vessel PRV opens to relieve vapour, the temperature of the fluid entering the flare system may be much higher than the vessel design temperature. Such a possibility gives rise to the question: should the design temperature of the flare system be higher than the design temperature of the vessel it serves?

A fire on a hydrocarbon processing unit usually means that a loss of containment has already occurred. When a vessel containing hydrocarbon is subject to heat input from a major fire, further failure (for instance at the vessel flanges) should not be unexpected. Although this results in an escalation of the fire, the incident is still contained in the same plant area.

If hot vapour relief to the flare system causes sufficient stress at any point in the flare header (by thermal expansion) then significant damage can occur. This may cause loss of containment at a location remote from the original fire. The result would be a major escalation of the incident. It is therefore not illogical that parts of the flare system should have a higher design temperature than that of the equipment from which the fire relief stream originates. However, the header piping in on-plot flare systems is highly constrained (by many piping branches) and likely to be large in diameter. Setting an unreasonably high design temperature will result in major mechanical design problems.

There is no industry wide practice for setting design temperature based on fire case data. This scenario requires careful consideration to avoid either an inadequate specification or an infeasible specification. API Standard 521 states that it is common practice to exclude the fire relief scenario when specifying the design temperature of the flare headers. It may be a common practice but, in the author’s experience, it is certainly not ubiquitous. To consider fire case for the majority of refinery PRVs but to ignore it when specifying the disposal system to which they discharge might be considered questionable. After all, a fire on an oil refinery is a reasonably foreseeable contingency.

The following sections assume that the fire case will not be ignored and consider some of the issues that will therefore arise.

What is the fire case relieving temperature?
Wetted wall vessels
Many refinery vessels containing wide boiling range hydrocarbons will have very high calculated fire relief temperatures. The heavier the hydrocarbon and the higher the PRV set pressure, the higher will be the relief temperature. If we take the average boiling point as indicative of the temperature a vessel might be expected to reach, then a distillate stream with an average boiling point of 300°C might be over 400°C at a relieving pressure of 5 barg. The design of the flare sub-header piping would be very challenging at this sort of temperature.

Actually, the relieving temperature is unlikely to reach 400°C and it is therefore unlikely that the flare piping would ever reach this temperature. Hydrocarbons tend to start cracking if the temperature exceeds a value of around 350°C; if the liquid in a vessel is boiling at 350°C it is likely that some cracking is occurring at the vessel walls. The lower molecular weight materials produced by cracking will tend to reduce the effective vapour pressure of the liquid and make it unlikely that the temperature will continue to rise in the way predicted from the feed stream boiling range.

Depending on hydrocarbon type and molecular weight, it may well be that the hydrocarbon in the vessel will become supercritical at the relieving pressure and will no longer boil. Prediction of what happens inside the vessel is now even more difficult since cracking will still occur (or increase since the wall temperature is likely to increase). The formation of light hydrocarbons will tend to increase the critical pressure of the mixture and also cool the vessel contents since thermal cracking is endothermic.

If the calculated fire relief temperature is higher than 350°C it should be treated with extreme caution.

Vessels containing vapour
When vessels containing only gas or vapour are subject to fire heat input, very high initial relieving temperatures can be calculated depending on the ratio of normal operating to relieving pressure. In some cases the calculated relief temperature will be infeasibly high and failure of the vessel would have occurred before relief. In any case the mass relief rate is likely to be low and the relief relatively short lived. Failure of the vessel is likely unless the vessel is effectively cooled by fire-water. In either case the relief will cease. For these reasons, relief from a vessel containing vapour only is unlikely to heat up a significant portion of the flare piping and is unlikely to determine the design temperature of the flare system.

What is the temperature downstream of the pressure relief valve?
Overall, the flow through a relief valve is considered to be approximately isenthalpic – the nozzle flow is isentropic, but this does not continue throughout the valve. Therefore it is usual to expect a temperature drop across the relief valve due to the pressure falling at constant enthalpy. This is typically of the order of 1.5°C per bar drop across the PRV for a hydrocarbon stream.

The flow velocity at the relief valve outlet is almost always higher than at the relief valve inlet. Rigorous flow simulators will indicate a further temperature drop below the stagnation temperature which would be calculated by a process simulator at the valve outlet. (Stagnation temperature is the temperature that would occur if the flow was brought to rest isentropically.) This additional temperature drop will generally not amount to more than a few degrees and should not be accounted for in simple heat transfer calculations to estimate the wall temperature. Due to the velocity profile, the velocity of the fluid next to the wall is zero and the fluid temperature at the wall would be nearer the stagnation temperature than the bulk flowing temperature for adiabatic flow.2 The flow is not actually adiabatic, since heat transfer through the pipe wall occurs, but the principle that the velocity related temperature drop is not fully experienced at the wall remains.