Emissions standards concerning flare gas emissions are getting ever tighter, so refining and petrochemical processes always have to be made more efficient. However, that is not the only flare-related challenge facing the oil and gas sector today.
Despite the economic downturn, major investments are still being made in new plant and revamps. Such activities can put extra stress on both pressure relief systems and flare systems, as can the addition of new processes or re-routing atmospheric vents to the existing flare system. Typically, distillation towers are a significant contributor to the peak relief loads used for flare sizing and a particular problem here is that traditional ways of calculating the relief load from these towers often lead to the overdesign of flare systems and the consequential addition of potentially unnecessary new flare capacity.
In an effort to solve the problem of overdesign, dynamic simulation is being to help better calculate an accurate relief load figure. Nevertheless, there are cases where the plant upgrade stills require an increase in flare capacity.
However, two standards - API Standard 521 on Pressure-relieving and Depressuring Systems, and AMSE Section VII Code Case 2211-1 on boiler and pressure vessel engineering - offer an alternative solution to conventional pressure release devices. Instead, they allow for the use of safety instrumented systems (SIS). These focus on preventing the causes of the overpressure, rather than the controlled removal of the inventory. Because of the safety issues involved, such SISs must also be high integrity protection systems (HIPSs), too.
HIPS are getting more attention because they can still do away with costly investment in plant equipment, even when a dynamic simulation shows that extra flare capacity is needed.
In an effort to illustrate this, Invensys carried out a dynamic simulation study for a customer in order to work out the peak relief rate for an integrated de-isobutaniser tower and de-butaniser tower in the alkylation unit of a refinery. The simulation model considered plant behaviour without the HIPS and also considered a HIPS that would cut off the supply of steam to the column re-boilers when the pressures in the columns reached a certain level.
In the simulation, the two towers produced a distillate and bottoms product each. Overhead vapour from each tower passed through a cooling water condenser system. In addition, both towers had a single thermo-siphon steam re-boiler. The de-isobutaniser has two feed streams. The liquid feed primarily consists of i-butane, butane, and heaviers, while the vapour feed is a mixture of butane and i-butane. The feed to the de-butaniser is pressure driven flow from the bottoms of the de-isobutaniser.
The study looked at three approaches to evaluate the peak relief loads: the traditional unbalanced heat load approach; a rigorous dynamic simulation approach; and a rigorous dynamic simulation using a HIPS on the re-boiler steam. Two scenarios were tested on the model: the first involved a total power failure, but with steam still being supplied to the re-boilers; the second involved loss of cooling water to the condensers with the feed to the towers and steam to the re-boilers remaining unaffected.
API 521 requires that a simulation model is run for a maximum duration of half an hour after any process upset before credit for operator intervention can be taken. At the end of this time, the peak relief rate calculated by the rigorous dynamic simulation without the HIPS was 20 per cent lower than that estimated by traditional unbalanced heat load calculations. Importantly, the simulation also found that the HIPS system substantially reduced the peak relief loads on both columns and it completely removed the relief load from the de-isobutaniser for both the tested cases.
It was also important to be able to determine the optimum HIPS set point for the two towers. Setting it too high could have led to there not being a significant reduction in the relief load; setting it too low could have led to an increased rate of nuisance trips of the re-boiler when there are normal disturbances to the process. The dynamic simulation model helped achieve this objective as well based on sensitivity runs.
In the end, the customer was able to use the results to help make the decision on whether to invest in a complicated and costly re-design of the flare piping network.
Note that the dynamic simulation model can also be used to model even more integrated process units in order to find the optimum configuration of multiple HIPS for safety shutdown systems.
In conclusion, the dynamic simulation models effectively established the effectiveness of HIPS on reducing the relief rates from the unit and identified the optimum set points required for the HIPS. However, bigger and more complex integrated units with multiple interactions can make it difficult to determine the optimum individual HIPS set points and often require a combination of dynamic process modeling and rigorous sensitivity analysis to help determine them.
Finally, Invensys cautions that using any SIS demands an excellent understanding of all the regulations and standards involved.
Ian Willetts is director, Invensys Operations Management. Abhilash Nair is principal consultant, Invensys Operations Management. For more information visit www.invensys.com